Saturday, February 6, 2016

Using "Ultracapacitors" as a power conditioner and ballast for transient high-power loads (or "How to run your HF rig from D-cells" - sort of...)

The problem of "high impedance" power sources:

The title serves to illustrate a problem frequently encountered when trying to power a device that operates with a high peak current:  Your energy storage - or your power source - may have plenty of capacity, but not enough current capability!

One such example of a power source that has plenty of capacity, but rather limited power capability is that of the "cigarette lighter" in a typical vehicle:  As long as the engine is running, you can pull power - but not too much:  More than 10-15 amps is likely to blow a fuse, and even if you were to replace the original fuse with, say, a 20-30 amp fuse (not smart!) the rather light gauge wiring would likely result in a voltage drop that would render a typical 100 watt HF rig unusable on voice peak.


For another, more extreme example let us consider a set of alkaline "D" cells, referring first to some online data:
  • The "Energizer" D-cell data sheet - link
  • "Duracell" D-cell data sheet - link.  Does not include direct amp-hour rating, but such may be inferred from the graphs presented.
  • "'D' Battery" Wikipedia page - link.
Please note:  Manufacturer's links sometimes change - you may have to resort to an internet search if one or more of the above links do not work.

One thing that jumps out is that a "D" cell has somewhere between 5 and 30 amp-hours per cell, but if you study the graphs, you'll also note that this apparent capacity drops like a rock with increasing current.  Why is this?

At least part of this is due to internal resistance.  If we examine the data for a typical alkaline "D" cell we see that on a per-cell basis that the internal resistance is 0.15-0.3 ohms per cell when it is "fresh", but this increases by 2 or 3-fold near the end of life of the cell (e.g. <=0.9 volts/cell) and increases dramatically - and very quickly - at still-lower voltages.  Interestingly, the manufacturer's data used to include graphs of internal cell resistance, but these seem to have disappeared in recent years.

If we take a general number of 0.2 ohms/cell and expand that to a 10-cell series battery pack we get a resistance of 2 ohms which means that if we attempt to pull even one amp we will lose 2 volts - and this doesn't take into account the contact and wiring losses related to these batteries!

If you look at the graphs that relate battery capacity to discharge current you will notice something else:  If you draw twice the current, your apparent capacity - in "run time" decreases by more than half and if you convert this to amp-hours, the more current drawn, the fewer available amp-hours.

These two facts together tell us two things:
  • We cannot draw too much current or else resistive losses will cause excess voltage drop.
  • Higher current consumption will cause a marked drop in available amp-hour capacity.
This second point is often referred to as "Peukert's Law" (Wikipedia article here - link).  While Peukert's law was derived to describe this effect with lead-acid cells, a similar phenomenon happens to a greater extent with cells that have a higher internal resistance such as alkalines.

As you may have inferred from the title of the article, our particular application implies a usage where our typical, resting (or average) current consumption is quite low but our peak current consumption could be very high.  Clearly, with a string of "D" cells, alone, we cannot tolerate the latter!

A power source with good longevity:

What we need is a low-impedance power source, and as it turns out almost any type of rechargeable cell - whether it is lead-acid, NiCd, NiMH or lithium - has a lower impedance than alkaline cells so the obvious question that one might ask is why not use one of those other types?

The obvious problem with lead-acid is that of longevity:  If you own a lead-acid battery that is more than three years old (and certainly more than five!) it is likely lost a significant percentage (at least 30%) of its rated capacity - probably more unless it has been treated really well (e.g. controlled temperature at or below 70F, 21C) and always kept at a proper floating voltage when not in use (e.g. 13.5-13.7 volts for a "12 volt" battery at nominal room temperature.)

On the other hand, modern alkaline cells will retain the vast majority of the capacity for at least 5 years, just sitting on a shelf - a period of time that is beyond the likely useful lifetime of either lead-acid or most rechargeable lithium-ion cells!  What's more is that alkaline cells are readily available practically anywhere in the world and they come "fully charged".  To be sure, there are other types of "primary" (non-rechargeable) cells that have excellent shelf life such as certain types of lithium, but these are much less-available and would likely "break the bank" if you were to buy a set with comparable capacity!

A low-impedance voltage source:

An advantage of Lead-Acid, NiCd, Lithium Ion and some NiMH cell types is that they have quite low internal resistance compared Alkaline cells:  Even an aging lead acid battery that is near the end of its useful life may seem to be "OK" based on a load test as its internal resistance can remain comparatively low even though it may have lost most of its storage capacity!

One could ask, then, why now simply parallel Alkaline cells, with their ready availability, long life and high storage capacity with one of these other cell types and get the best of both worlds?  In theory you could - if you had some sort of charge control circuitry that was capable of efficiently meting out the energy from the alkaline pack and using it to supplement the "other" storage medium (e.g. lead-acid, lithium-ion, etc.) but you cannot simply connect the two types in parallel and expect to efficiently utilize the available power capacity of both types of storage cell - this, due to the wildly different voltage and charge requirements.

Even if you do use a fairly small-ish (e.g. 7-10 amp-hour) lead-acid or lithium-ion battery pack, even though its internal resistance may be low compared to that of alkaline packs, it likely cannot source the 15-20 amp current peaks of, say, a 100 watt SSB transceiver without excess voltage drop, particularly if it isn't brand new.

This is where the use of "Ultracapacitors" come in.

In recent years these devices have become available on the market and for reasonable prices.  These capacitors, typically with maximum voltages in the range of 2.5-2.7 volts per unit, may have capacitance values as high as several thousand Farads in a reasonably small package while at the same time offering very low internal resistance - often in the units of milliohms.  What this means is that from this capacitor one may pull many 10's of amps of current while losing only a small percentage of the energy in heat:  Indeed, many of these capacitors have current ratings in the hundreds of amps!

What this means is that we can use these capacitors to deliver the high, peak currents while our power source delivers a much lower average current.

The difference between peak and average current - and power:

With a simple "thought experiment" we can easily realize that with our transmitter, one second at 100 watts (about the) same total power as ten seconds at 10 watts.  If, in each case, we averaged the power over ten seconds, we would realize something else:  In both cases, the average power is 10 watts per second.

Clearly, the power requirements for a radio operating at 10 watts are different than at 100 watts:  For the former you'll likely need 4-5 amps of current, but for the latter you'll need 18-20 amps (at 12 volts or so, in each case.)

Here, we have a problem:  If we have a given resistance somewhere in our DC supply, we lose much more power at, say, 5 amps than at 18 amps according to the equation:

P = I2R

In other words, the ratio of the power loss is equal to the square of the ratio of the current differences - in other words:

182 / 52 = 12.96 (or 13)

That means that power losses at 18 amps are 13-fold worse than those at 5 amps.

Clearly, the best way to mitigate this is with heavy cables to minimize resistance, but what if your power source is, by its very nature, fairly high in resistance in its own right?

This is where the "ultracapacitors" can be useful.  By acting as a reservoir to handle peak currents, but rely on the battery - whatever form it may take - to make up the average.

With SSB (Single SideBand) transmission we have an almost ideal situation:
  • There is no RF power when we aren't saying anything
  • The amount of RF power is proportional with voice peaks, and
  • Speech has comparatively rare peaks with a lot of empty spaces and lower-energy voice components interspersed.
  • SSB is 6-12 times more power efficient in conveying voice than FM and occupying 1/3-1/4th of the space (bandwidth) as one FM signal.
In other words, when we are transmitting with SSB, the average power of a hypothetical 100 watt transmitter will be much less than 100 watts.

Compare this with 100 watt FM transmitter:
  • When you are "keying down" it is always putting out 100 watts, no matter what your voice is doing.
  • 100 watts of FM is less effective in conveying voice than even 10 watts of SSB and it takes at least 3 to 4 times as much bandwidth as an SSB signal.
One obvious take-away is that if you are in an emergency, battery power communications situation where you need to communicate on simplex and find that it takes 20-50 watts of FM power, you are probably making a big mistake sticking with FM in the first place as you could do better with 5 watts of SSB or less, but I digress...

For the purposes of this discussion the point that I was really trying to make was the fact that the use of these "ballast" capacitors is appropriate only for relatively low duty-cycle modes such as SSB or, possibly CW:  If you tried this with FM or digital modes such as PSK31 the long duty cycle (e.g. key-down) would quickly drain the energy stored in the capacitors and they would "disappear", putting the load back on the battery bank.

This technique is not new:  For many years now one has been able to buy banks of capacitors intended for high-power car audio amplifier installations that provide that instantaneous burst of current required for the "thud" of bass without causing a similar, instantaneous voltage drop on the power supply.  Originally using physically large banks of "computer grade" electrolytic capacitors, these systems are now much smaller and lighter, using the aforementioned "Ultracapacitors".

There are also devices on the amateur radio market that do this:  Take as an example the MFJ-4403 power conditioner (link).  This device uses ultracapacitors in series to achieve approximately 5 Farads of parallel capacitance across the output leads of the device.

Now, a few weasel words:
  • The device(s) described on this page can/do involve high currents and voltages that could cause burns, injury, fire and even death if improperly handled if reasonable/proper safety precautions are not taken and good building techniques are not followed.
  • The device(s) described are prototypes.  While they do work, they may (likely!) have some design peculiarities (bugs!) that are unknown, in addition to those documented.
  • There are no warranties expressed or implied and the author cannot be held responsible for any injury/damage that might result.  You mileage may vary.
  • You have been warned!

How this may be done:
Figure 1:
The capacitor bank/power conditioner with 53-1/3 Farads,
in a package slightly larger than a 7 amp-hour, 12 volt
lead-acid battery.  How much energy is actually
contained in 53.33F at 13 volts?  Theoretically, about
the same as just one alkaline AA cell.
Click on the image for a larger version.

For a variety of reasons (physics, practicality) cannot buy a "16 volt" ultracapacitor:  Any device that you will find that has a voltage rating above 2.7 (or, in some cases 3.something volts) is really a module consisting of several lower-voltage capacitors in series.  What is more, you cannot simply throw capacitors in series as there is no guarantee that the voltage will always divide equally amongst them unless there circuitry is included to make this so.

Another consideration is that if you have such a device - a large bank of capacitance - that is discharged, you cannot simply connect it across an existing power source because a (theoretically) infinite amount of current will flow from the power source into the capacitors, if their voltage is lower, to force equilibrium.  Practically speaking, if you were connect the battery bank "suddenly", if the resistance of the wires themselves didn't serve to limit the current you'd likely blow the fuse, trip the breaker and/or cause the power supply to go into some sort of overcurrent (or shut down) mode - none of which are at all helpful.

Finally, this capacitor bank will be (theoretically) capable of sinking or sourcing hundred of amps if applied to a high current source/shorted out, so some sort of protection is obviously needed.


The diagram in Figure 2, below, provides these functions.
Figure 2:
Schematic diagram of the capacitor bank/power ballast/conditioner and charging circuit - see the text for
a circuit description.
Click on the image for a larger version.


Circuit description:

Capacitors C1-C6 are "Ultracapacitors":  Their exact values are not important, but they should all be identical values and models/part numbers.  In the example of the MFJ-4403, six 25 Farad capacitors are used yielding a total of 4-1/6 Farads while the drawing in Figure 2 depicts six 350 Farad capacitor being used to yield a total of 58-1/3 Farads.  The greater the capacitance, the more energy storage, but also the longer the "charge" time for a given amount of current and, of course, the larger the size and the higher the initial cost of the capacitors themselves.

Zener diodes D1-D6, each being 2.7 volt, 1.3 watt units, are placed across each capacitor to help equalize the voltage.  As is the nature of Zener diodes they will be conducting at lower voltage than nominal, increasing dramatically at their rated voltage - which, itself, can vary significantly.

Originally, I experimented with the use of a series-connected resistor, diode and green LED across each capacitor to equalize the voltage as depicted by components Ra, Da and LEDa.   In this circuit the LED, a normal, old-fashioned indicator-type "non-high brightness" LED was used, taking advantage of its 2.1 volt threshold voltage along with the 0.6 volt drop of an ordinary diode with a 5.1 ohm resistor in series to provide a "knee".  While this circuit did work, providing a handy, visual indication of a "full charge" state of each of the six capacitors, it did not/could not safely conduct enough current to strongly force equalization of the capacitors' voltages.  The Zener diodes, with their maximum current of more than 400mA, as compared to 15-25mA for the LED-based circuit, seemed to be more effective.  I left these in place after constructing the prototype since there was no reason to remove them and the illumination of the LEDs serves to indicate that the capacitors are charging up with the equal illumination being generally indicative of equal charge distribution.

It will be noted that the "top" of the capacitor bank is connected to the positive side of the power source, "straight through" at all times via TH1 and TH2, current-inrush limiters.  Because of this, when this unit is "off" it is, for all practical purposes, transparent, consuming no current.  These devices act as self-resetting fuses, limiting the current to a sane amount, somewhere in the 30-50 amp region, if the output (or input!) is shorted:  Ordinary "slow blow" fuses could be used here, but if so, the advice is to keep spares on hand!

It is only the "bottom" of the capacitor string that is connected/disconnected to enable or disable the "ballast" (and filtering) capability of this circuit:  When "off" the bottom of the capacitor bank is allow to float upwards.

If the capacitors are discharged when switch SW1 is turned ON while power is applied the first thing that happens is that the gate of Q1 is turned on via resistor R3.  When this happens current flows through R1, but when it exceeds approximately 0.6 volts - a voltage commensurate with approximately 2.5 amps of "charge" current - transistor Q3 begins to be turned on, drawing down Q1's gate voltage.  When this circuit reaches equilibrium 2-3 amps will flow through Q1, R1 and TH3 and thus charging the capacitor bank comparatively slowly and preventing a "dead short".
Figure 3:
The "control board" with the charge control/regulator circuit.  In this prototype R1 was implemented using a pair of 0.5 ohm 15 watt resistors since I didn't have a single 0.22-0.25 ohm, 1-5 watt resistor on-hand.
As can be seen, Q1 and Q2 are bolted to the lid of the die-cast box for heat-sinking.  In the lower-left corner can be seen the heavy wires that connect to the input/output and to the capacitor bank, along with the 30-amp current limiting devices, all connected on an "island" of copper cut out of the piece of circuit board material.
The rest of the circuit is constructed on a combination of the "Me Squares" and islands cut into the piece of circuit board.
Click on the image for a larger version.

While in this equilibrium mode the gate voltage on Q1 is necessarily reduced to keep it partially "off", to maintain the current at approximately 2.5 amps, and there will be a voltage drop across R3:  This voltage drop is detected by Q5, a PNP transistor via R4 which, if Q2 is turned on, will also be turned on which, in turn, turns on LED2, the "Charging" LED which also turns on Q4 which, in turn pinches off the drive to the main capacitor bank switch, Q2, forcing it off.

(In the event of a circuit malfunction self-resetting fuse TH3 limits the maximum current through Q1 to 5 amps before "blowing", at which point the current will be reduced to a few 10's to 100's of milliamps.)

Once the capacitor bank has become (mostly) charged and the voltage across it is nearly the same as the applied voltage, the charging current will begin to drop and as it does, transistor Q3 will start to turn off, causing less voltage drop across R3 in an attempt to make Q1 conduct more.  At some point, when Q1 is already "fully on" this will no longer work and the voltage drop across R1 will start to drop, turning off Q1 entirely.  When this happens the voltage at the "bottom" of R1 (the side connected to the gate of Q1) will be equal to that of "V+_SW" and this will cause transistor Q5 to turn off.

Once Q5 turns off, the "Charging" LED will also turn off, as will Q4 and the voltage on the gate of Q2, being pulled up by R5 and "slowed" by capacitor C8, will start to rise.  As the gate voltage on Q2 crosses the "on" threshold it will conduct strongly, connecting the bottom of the capacitor bank to ground with only a few milliohms of resistance.

If switch SW1 is turned off, the voltage at "V+_SW" drops and via R5, C8 and the voltage at the gate of Q2 drops, turning it off and disconnecting the capacitor bank.

Construction details:

The ultracapacitors were wired together outside the enclosure (a Hammond 1590 series die-cast box) using multiple pieces of #12 AWG bare wire, both for low ohmic resistance and mechanical support.  Additional pieces of wire were used on the capacitors' support pins for spacing and support when the two banks of three were arranged to be parallel to each other, terminals facing - but separated by a safe distance as seen in the pictures.  The balancing Zeners, etc. were also installed across the capacitors at this time.
Figure 4:
The bank of six, 350 Farad, 2.7 volt capacitors mounted in the bottom of the Hammond die-cast box.  The capacitors are connected together using multiple, folded pieces of #12 AWG copper wire that also provide mechanical support.
Not obvious from the picture, the heavy "bridge" between the left and right bank at the bottom of the picture in the center is insulated from the metal box, to prevent shorting, by a piece of clear plastic from a discarded "blister pack" product package that was heat-formed around the screw-boss and then secured in place with RTV.
On the left may be seen the"on/off" switch and the two indicator LEDs while the "power in/out" leads, using #12 flexible speaker wire, are visible on the lower-right.  Across each capacitor may be seen "Ra, LEDa and Da" depicted in the
schematic and discussed in the text, originally used for capacitor balancing.
Click on the image for a larger version.

Prior to mounting the capacitors in the bottom of the box the holes for the LEDs and switches were drilled/filed - this to eliminate the possibility of the tools damaging them.

The two parallel banks of three series capacitors were prepared and then placed in the bottom of the box, held in place with RTV (Silicone (tm) seal).

There is no circuit board:  The actual control circuitry is mounted on the lid, with Q1 and Q2 being bolted to the lid itself for heat-sinking using electrical insulating hardware (washer, grey insulating pads) that were scavanged from a dead PC power supply.  The circuit itself was constructed on a piece of glass-epoxy circuit board material.

Without making a circuit board, there are several ways that the circuit could have been constructed, but I chose to use a variation of the "Manhattan" style which involved islands of copper.  In some instances - such as for the large resistor(s) comprising the 0.25 ohm unit on Q1 and for the connections of the power in/out leads and TH1 and TH2, islands of copper were isolated on the board by first drawing them out with a pen and then slitting both sides of a narrow (1/16th-1/8th of an inch) trace with a sharp utility knife and straight edge and then using the heat from a soldering iron to aid in the lifting of the narrow strip to isolate the area of board.

For other portions of the circuit I used "Me Squares" available from QRPMe.com (link):   These are small pieces of glass-epoxy circuit board material with nice squares etched on them that one glues down using cyanoacrylate (e.g. "super") glue and then uses as points for soldering and connection.

The nice thing about these "Me Squares" is that they are very thin, look nice and are very flat - which makes them easy to solder and glue down, but one could also cut out squares of ordinary circuit board material and solder those down, instead, provided that they were also made flat on the back side and de-burred for maximal surface contact.  Finally, one could use the utility knife and isolate islands - or even use an "island cutter" tool - to produce isolated lands on the piece of circuit board material itself.

The main reasons for using this technique were that it was quick, and also that it was surface-mountable:  Had I wired it on perforated board or even made a conventional circuit board, I would have had to stand it off from the lid to insulate the circuit from it:  Using this technique the board itself was simply bolted flat to the lid and was quick to wire up.

For interconnects between the circuitry on the lid short lengths of #12 AWG stranded wire were used for the high-current leads connecting the capacitors and input/output leads and much smaller (#24 AWG or so) for the wires that connected to the switch and LEDs.  For the "outside world" connections 30 amp Anderson Power Pole (tm) connectors were used mounted in the standard manner.

A few caveats with this circuit:
  • This circuit consumes some current (at least a few 10's of milliamps - maybe more) whenever it is set to "on", even after the capacitors have equalized.  What this means is that it will slowly drain your battery if the switch is left in the "on" position and because of this it is recommended that one switch it to the "on" position ONLY when intermittent, high current is going to be needed.  In other words, if you are going to receive (only), leave it "off", turning it "on" only if you plan to transmit, knowing that it may take 10's of seconds for the capacitors to charge.
  • If the power source cannot deliver the amount of current required during the "charge" cycle - or if it briefly "blinks" - the source voltage will sag and it will likely switch from "charge" mode to "operate" mode and connect the ultracapacitors directly across the power source.  If the power source has limited current in the first place, this will simply mean a "dip" in voltage while the capacitor bank charges, but if this occurs with a high-current source that has had a momentary glitch, a premature switch to "operate" mode could, in theory, blow a fuse.  This premature switching would likely happen if you had this connected via a cigarette lighter and started a vehicle while it was in a charge phase.
  • This capacitor bank should be treated like a battery:  If it is shorted out you will get lots of current - more than enough to burn open small wires and blow fuses - or even burn open large wires or small tools if you do an "oops" on the unfused/unprotected side of the circuit, or neglect to include any fusing/current protection!
  • Neither this capacitor bank (or any battery) should ever be placed on the output with any power supply that has a "crowbar" voltage protect circuit (such as is present on many power supplies such as the Astron "RS" and "RM" lines) as a power line or transient on the output could cause the crowbar to trigger and short the output of the power supply - including the capacitor bank or whatever battery you might have on the output.
  • This sort of device must be contained within a rugged enclosure/container.  The picture shows the prototype being built into a Hammond 1590 series die-cast aluminum box that is both very rugged and also provides heat sinking for Q1 and Q2.  In the unlikely event of a catastrophic failure due to a wiring or a capacitor going bad this enclosure will not melt and is likely to contain the "mess" - even if it were to get very hot!

  • This circuit should be disconnected/disabled when the vehicle is started since these capacitors can supply "car starting" current on their own and it is possible to blow fuses, pop breakers, damage switches, circuits, relays and other mischief if the current from the capacitor bank were to "back-feed" through wiring that was not designed for that sort of current!
  • It is the nature of power FETs such as Q1 and Q2 to have an intrinsic "reverse diode" - even when turned off.  Be aware that if it so-happens that the bottom of capacitor bank is more negative than the "ground" (Battery -) side, these diodes - particularly the one in Q2 - will conduct.
I have no doubt that this circuit could be improved a bit:  It was designed and put together over two evenings in preparation for the 2015 "Homebrew Night" meeting of the Utah Amateur Radio Club - including the time it took for the RTV to set up and paint on the enclosure to dry!

Possible uses:

While it is possible to use this to allow an HF rig to be powered from a set of D cells, it is more practical to use smaller lead-acid or lithium-ion packs as the primary power source and use the capacitor bank as the "ballast" to supply the peak currents.

Figure 5:
One end of the enclosure showing the on/off switch and
the two indicator LEDs.  The red "charge" LED indicates that the capacitor
bank is charging (at 2-3 amps) and the "power conditioning" capability
is not available until it has completed.  With completely discharged
capacitors this process takes a "minute or two", depending on the
voltage and the capacitance of the bank.
Click on the image for a larger version.
It may also be used in a vehicle to allow an HF transceiver to be powered from a cigarette lighter connection by reducing the average current drawn by it and thus keeping the average voltage higher to prevent the radio from shutting down/misbehaving on voice peaks.  When used in the comparatively "dirty" electrical environment of a vehicle it will go a long way to remove spikes from solenoids and motors, not to mention alternator whine.

I'm certain that this device could be improved, but it seems to function as it is.

Conclusions:

How usable is it, in the real world?

It depends a lot on how - and with which radio you plan to use it.  For example many older-vintage Kenwood HF transceivers will fail to function properly (e.g. operate with distorted audio, "FMing" of the signal, etc.) much below 12-12.5 volts while more modern, compact HF radios like the Yaseu FT-100 and FT-857 will happily run at 10-11 volts - perhaps at slightly reduced output power, but fine otherwise.  The upshot is that if you are considering a radio to be operated from "marginal" power sources, be certain that you have done your research and consider how your candidate radios operate at low supply voltage and how/if they degrade "gracefully" or not!

How about running an HF rig from alkaline "D" cells?  As it turns out I can happily transmit 100 watt (peak) SSB using my old Yaesu FT-100 with the device described on this page using 10 "D" cells in series.

To do this effectively, one must minimize the contact resistance of the battery contacts which pretty much rules out using cheap, spring-loaded battery holders which, by themselves, can have almost as much resistance as the cells themselves. Aside from spot-welding tabs onto the alkaline cells (the heat of soldering would likely cause some damage and slight loss of capacity) the best holders are aluminum with heavy bus bars and the fewest number of springs and contacts (e.g. multiple cells directly in series) such as the Keystone four cell holders, models 158 (two of them) and one two cell holder, Keystone model 186.

One cannot "key down" with a carrier without significant voltage sag, but the FT-100 seems to work OK on typical SSB with voice peaks - but under such heavy loads don't expect to get much longevity before the cells' internal resistance increases:  As noted previously not all radios (such as older Kenwood mobiles) behave so well at lower voltages, so do your homework!

As detailed in the article, a more practical use of this sort of device is as a "power conditioner" to help compensate for voltage sags due to resistance in the interconnecting cables, somewhat underrated power sources, aging battery packs and/or "small"-ish batteries.

How about a solar power source?  Assuming that voltage from the panel is regulated to a safe value (15 volts or below) the capacitor bank could, in theory, maintain voltage if the solar array provided at least the average current, but considering that solar illumination can vary wildly it would be recommended that additional storage capacity (remember that 53-1/2 Farads has only the theoretical storage capacity of a single "AA" alkaline cell!) be used as well, such as a 7-20 amp-hour lead-acid or lithium-ion based pack - but this sort of system could well be the basis of another article!

Tuesday, January 19, 2016

Homebrew Drake TR-3 power supply

Years ago I was a teenager with a novice license and I got for a combination Christmas/Birthday present a "brand new" (to me) Drake TR-3 transceiver - but it had no power supply.

What to do?

Fortunately, I had the original manual and in it, there was the schematic of the original Drake AC-3 power supply, reproduced below for convenience.

Figure 1:
A schematic diagram of the original Drake power supply.
Click on the image for a larger version.
It looked simple enough.  At first I puzzled over the plate and low voltage (250 volt) supply portions of the diagram, noting the seemingly-odd wiring of the diodes and capacitors, but the circuit description noted that these were voltage doublers and a quick check with my ARRL Radio Amateur's Handbook explained how this circuit worked.

Where to get the parts, then?

The chassis was no problem:  Several years prior I'd picked up a Dynaco 40 audio amplifier (or similar) with a blown power transformer (and melted tubes) for just a few dollars at a thrift store.  I'd built into this case my "Novice" transmitter, a crystal-controlled (no VFO!) unit using a12BY7A driving a 12DQ6 design, slightly modified from that which had appeared in (then old) 1970 ARRL Radio Amateur's Handbook that I'd borrowed from my Junior High School library and with this I used a Heathkit SB-303 receiver that I'd bought from my Elmer with lawn-mowing/misc. money.  I'd used that combination, along with a few random crystals that I'd managed to scrounge/warp (another story!) on 40 and 15 meters for 6-7 months.

Now that I had the TR-3, my plan was to scrap the homebrew transmitter and use its case for the power supply, taking me off the air.  Both itching to use my "new" TR-3 and knowing that I'd suffer "withdrawl" when I disassembled my only working transmitter, I wanted to complete this power supply project quickly.

Lining up parts:

Having collected parts for several years by then I rummaged around and found several power transformers, some of which I'd bought for $1-$2 each at a local thrift store, apparently having been removed from tube-type TVs or audio amplifiers (Who would bother doing that?  The store just had them on a shelf with stickers on them!)  Some of these transformers had clearly come from large, console tube-type color TVs as they were quite large and had quite a few windings on them - but these windings were unmarked aside from generally inscrutable wire colors.

In looking at the voltage requirements for the Drake TR-3 I saw that I would need around 650 volts at up to (approx.) 600 milliamps, peak, for the plate supply on the final amplifier, 250 volts for the "other" circuitry within the radio and a nominal -60 volt supply, adjustable down to -45 volts for the grid bias on the finals and, of course, 12-13 volts for the filament string which, for the TR-3 - designed to be a mobile radio - consisted of a series-parallel arrangement of both 6 and 12 volt tubes.

In rummaging around my transformer collection I expected, at first, that the plate supply would be difficult, but that turned out not to be so - the difficulty would be the 250 volt supply.

Not knowing how the thrift-store transformers were wired and not having faith that their color codes would follow any of the "standards" mentioned in the ARRL handbook I first took an ohmmeter and mapped out all of the wires, noting which seemed to be connected to which and, as well as I could, the resistances of these windings.  This, alone, was useful as I now had a pretty good idea as to which ones where the likely high voltage windings (highest resistance), where the center-taps likely were, which ones might be the filament windings (lowest resistance) and finally, which one might be the primary windings.

You probably noticed a lot of "likely" and "might be" statements in the above paragraph and this meant that I from resistance measurements alone, I could not be sure that I had properly identified everything.  There was only one thing to do:  Connect it to mains power.

Even as a teenager I knew enough to not connect it directly to a wall outlet, so I decided on a two-step process.  Having a 12 volt AC transformer kicking around, I used that as the proxy for 120 volts, knowing that worst case, I'd throw only a few amps into the wrong place, the current limited by this power-limited transformer - and that is exactly what I did with a large, black transformer that had probably seen previous service in a console color TV.  Some of my guesses weren't exactly right, but after just a few minutes I found a medium-resistance winding that I'd suspected to be the primary and upon applying 12 volts to it found a winding that was suspiciously 1.2 volts and another that was in the 15 volt range - plus a few other miscellaneous taps and windings.

Now I was confident that I could "light it up" from the 120 volt mains, so I rummaged around in my dad's electrical box and found a porcelain light socket and wired it in series with the (believed) primary of the transformer, pulled a 60 watt bulb from a reading lamp and plugged it in.

Sure enough, the 60 watt bulb did not light, indicating that there was no short or gross misidentification of a wire and there was a 12.6 volt winding suitable for the filament and another winding with 155 volts and shorting either of these out caused the bulb to illuminate to what appeared to be full brilliance.  Shunting the bulb I briefly shorted the 12.6 volt winding and noted that transformer hummed noisily while the lights in the room dimmed slightly, indicating, in a rough fashion, that it was capable of supplying quite a bit of current and I surmised that it would be able to supply the filament string for the TR-3.

The 250 volt supply:

I was now wondering what I could do with the 155 volt winding? - clearly intended for some "low voltage" stuff within the TV such as IF, audio and tuners but with bridge rectification, capacitive input, from the center tap the most I could hope to get would be 200-220 volts - and that was without a load.  Using the voltage doubler method was out of the question since not only would the voltage be too high, I didn't have any suitable capacitors on-hand and being a poor teenager, I didn't have any money left over at that moment to get any!  All I had were the original "can" capacitors from the Dynaco 40 (and possibly something else) and since the cans had to be grounded, they weren't suitable for use as a doubler since one of the capacitors (both of which had to be identical) had to be "floating".

What I really needed was a 200-ish volt winding on which I could use a bridge rectifier that, when filtered with a capacitor, would get me within the target range of 250 volts, under load.

Rummaging through my pile of junk I found just the thing... sort of...  What I did find was a fairly large isolation transformer with 120 volt primary/secondary capable of at least 100 watts.  In series with the other transformer I could obtain about 275 volts AC - but this would be too much, at least for a capacitor-input power supply.  As it happened, the same piece of equipment (I have no idea what it was supplied to be) that contained the isolation transformer also had several large chokes that I managed to "ring out" to around 6.2 Henries (at no current - it likely "swings" to a lower inductance when a DC current passes through them) and by placing this in series with the full-wave rectified output I would not get the 1.4x (or so) peak voltage from the series combination, but rather something lower. 

Very carefully, I constructed the "250 volt" power supply, wiring the two transformers' "high voltage" windings in "boost".  Upon turning it on I (somewhat expectedly) got a bit over 300 volts, so I did more rummaging and found, on the same piece of equipment that had yielded the isolation transformer and choke, a 20 watt, 750 ohm "slider" type of adjustable resistor and I placed that in series with the choke.  In doing a bit of quick math and knowing approximately how much current the TR-3 pulled from the 250 volt line, I figured that I could afford to burn a few watts:  I just hoped that its voltage would be adequately stable.

Comment:  Had I the correct capacitors, I could have voltage-doubled the output of the isolation transformer, but I would have still needed a pretty hefty transformer to supply the 12.6 volt filament rail, so my "transformer count" would have remained the same based on the parts on hand at the time.

The Plate supply:

The plate supply was an easy one.  Somewhere - I don't know where - I'd managed to scrounge a Triad P-3A plate transformer with a 5 volt rectifier tube winding (no use to me) and a high voltage secondary of 300-0-300 volts.  I wasn't quite sure of the current rating of this transformer as it just said "300-0-300" and "0.3 amps" on the nameplate so I made the assumption (probably wrong!) that I could wire it as 600 volts, use a bridge rectifier, and safely get 300 mA.  I did know that a 600 VAC source would yield 800-900 volts DC, unloaded - a bit much for the three 12JB6 tubes in the final amplifier which "want" closer to 650 volts under load

The solution?  Another choke.  That same piece of equipment that had already yielded the isolation transformer, 750 ohm adjustable resistor also had another, identical 6.2 Henry choke on-board so I threw that in series with the output of the bridge rectifier's output.

In reading about choke input power supplies I knew that while I'd get lower operating voltage, the unloaded voltage would likely be fairly close the that of a capacitor-input power supply.  From what I could gather from my ARRL handbook it appeared that the 12JB6 tubes would be able to handle 800-ish volts on the plates at idle without difficulty, but I was wondering how much it would drop under load - and how quickly?

A bonus of the choke-input was also that it offered a bit of relief to the plate transformer that I knew that I would be overtaxing:  At maximum power output, the three tubes in the final amplifier would be pulling around 600 milliamps at 650 volts or so (according to the Drake manual).  By using a choke input the "peak current" at the top of the AC cycle where the filter capacitor charging would occur was significantly alleviated, effectively improving the power factor from "awful" to "not so bad."  but the degree by which the effective capacity of the transformer would be increased was a bit vague.

It also occurred to me at the time that the 1N540 diodes that I used for the supply, rated for only 250 mA at 400 volts, were somewhat "marginal" for the plate supply when 600mA was being pulled, despite the fact that this current was being split between two legs:  The choke input and its reduction of the very high, peak repetitive currents has likely helped preserve these diodes over time.

Another point was that it somewhat "decoupled" the output of the power supply from the transformer:  Brief, high-current voice peaks would be immediately sourced from the capacitors first while being "stretched out" (averaged) as current and magnetic field builds in the choke.  I figured that that this should, in theory, work for SSB where the peaks are typically brief, but less-so for CW.  Time would tell.

The grid bias supply:

The final supply voltage needed was a low-current negative supply that was adjustable from approximately -45 to -65 volts.  In staring at the schematic I didn't see any obvious place from which I could tap a stable, isolated source of 70-120 volts of AC:  In theory, I could have coupled from one of the other transformers - most likely on the 250 volt supply - but I figured that I wanted the bias supply to be reasonably stable.

Going back to the junk box I found a small transformer that was probably from an old, tube-type UHF TV converter that had a 6 volt filament winding and a 120 volt secondary.  Interested only in the high voltage secondary knew that I could probably deal with 120-130 volt DC that I'd get from the 120 volt winding, half-wave rectify and resistively drop it to the desired -65 to -45 volt range.

This was pretty straightforward:  Just a resistive divider with some capacitive filtering.  When wired up, it seemed to perform as expected.
Figure 2:
The (recently reverse-engineered) diagram of the as-built power supply, built in 1983.
The 1N540 are old (1966 date codes) 0.25 amp, 400 volt diodes, of which I had many at the time - likely
marginal when 600 mA of plate current is being pulled!  Each leg of diodes
on the 600 volt supply could likely be replaced with a single 1N4007 - although I'd probably
use two in series.  D17 is a tiny, 200 volt glass diode of (presently) unknown designation.
Click on the image for a larger version.

About the mechanical construction:


As mentioned previously, the enclosure for this power supply had previously been used for my homebrew Novice transmitter - and before that, a Dynaco 40 audio amplifier (I think - the small PCB was marked as "Dynaco".)

On the deck of the chassis there really wasn't quite enough room for all of the components:  On the top deck, the transformers were cheek-to-jowel, abutting each other with no room for anything else, but I managed to squeeze everything in.  Underneath  I wired the rest of the components - diodes, filter capacitors, resistors - using parts that I scrounged from my somewhat limited junk box.

With the circuits seemingly operating under no load, I obtained the correct "Jones" connector for the TR-3, wired a cable harness to plug into the octal socket that happened to already be present on the amplifier chassis, plugged it in and turned it on.

Testing:

Figure 3:
The underneath of the power supply.  The inside of the bottom cover of the
supply (not visible in any of the pictures) is covered with two layers of high-
quality cloth "gaffers" tape for insulation.  The two 100uF, 450 volt plate
filter capacitors are visible along the bottom edge.  When I was building
the supply I accidentally drilled a hole in one of them, apparently without
damaging the foil inside and patched it with RTV (silicone) seal since
I couldn't afford to get another capacitor at the time!
That was back in 1983 and both capacitors are still "good".
Click in the image for a larger version.
Amazingly, nothing blew up!  The filaments lit up and the voltage strings came up to their proper voltages - within reasonable tolerances, anyway, at least once the adjustable resistors and rheostats were tweaked.

Connecting the output of the TR-3 to a 150 watt light bulb (I didn't own a large dummy load at the time) I keyed up in CW, dipped-and-loaded and observed the beautiful brilliance of RF-driven tungsten while noting that the at full load, the plate voltage sagged down from 850 volts to 625-650 volt range - approximately what the Drake manual specified that it should be.

I was on the air!

A curious problem - and a fix:

I was still a novice at this point so I used it only on CW - which was slightly awkward since the TR-3 has no sidetone.  At first I would tune the transmit frequency with my Heathkit SB-303 and use that to monitor myself, but this was quite awkward since it required another, separate speaker - very inconvenient if you were typically using headphones to avoid bothering others in the house.

I then noticed something about the power supply:  It hummed loudly.  With all of those transformers crammed into a steel box, this was to be expected, but what was particularly strange was that it hummed loudest when I was receiving, becoming nearly silent when I was keyed down.

What I also noticed was that if I left the power supply on for more than an hour or hour and a half, I started to smell hot enamel - the tell-tale indication that something transformer-related was getting very warm.

For several months I would operate CW by resting my feet on the power supply, on the floor, and use the absence of hum in lieu of the sidetone - and I got pretty good at it!  After an hour or so, when the power supply got too hot, I would then shut it off for a while an allow it to cool down before turning it on again.

After a few months of this I finally I decided that I would find out what was going on.  Without the power supply connected to the radio I discovered that if I disconnected either one of two transformers that were next to each other the hum would disappear:  Could their magnetic fields be bucking each other?

On a whim I reversed the primary and secondary leads of one large of the transformers (the one with fewest leads, of course) and the hum that was present when receiving was gone - it only hummed when keyed down, as it should:  I may have also done some rearrangement of the transformers on the top deck to better-separate their magnetic fields, but memory fails me on this point.

When I was done I could now could leave the rig and power supply on all day and it would barely heat up:  It took only a session or two for me to get used to the the "non-inverted" power supply hum CW sidetone!

Figure 4:
The top deck of the power supply showing the four AC
transformers, two chokes and two capacitors.  This picture
was taken after vacuuming out an inch or so of dust.
Starting from the bottom-let and working clockwise:  The plate
transformer, two filter capacitors for the 250 volt supply, the input
choke for the 250 volt supply, the filament/175 volt transformer,
the isolation transformer, the input choke for the plate supply, and
the transformer for the bias supply (the small transformer, wedged in.)
Click in the image for a larger version.
The next year I upgraded from Novice to Advanced class and used the Drake on SSB where it and the power supply worked very well.  Later, I even managed save enough money to buy a brand new, replacement AC-4 power transformer - the same as the one in the original Drake AC-3 power supply - but I never did get around to rebuilding that old power supply with its plethora of magnetics:  I still have the new-in-a-box AC-4 power transformer somewhere!

Final comments:

This power supply still exist as evidenced by the fact that the pictures on this web page were taken very recently.  Both this power supply - and the Drake TR-3, now over a half-century old - still work fine despite my having owned it decades (and some minor repair, replacement of a dried-out capacitors in the power supply) despite the fact that I've obviously abused the plate transformer.

As can be seen, some (many?) aspects of the construction techniques used are a bit "iffy" and were I to build it today - after the benefit of decades of experience and the availability of other materials I would certainly do it differently.

Is it dangerous?

Figure 5:
The power supply, assembled, showing the side with the grid bias
adjust control (far left) and the pilot light.
Click on the image for a larger version.
Not very - at least with the covers in place.  Some of the "flakiness" of the construction methods involved (e.g. excessive use of  unsecured "flying leads", components that could be better-secured, using some bits of electrical tape to insulate wire instead of heat-shrink tubing in a few places - heat shrink tubing was a bit rarer when this was built) would contribute more to reliability than issues with safety and I would most certainly not build it (exactly) this way again!

Despite the fact that this power supply been bounced around the country several times, subject to very high humidity for several years (hence some of the rust!) and, of somewhat "interesting" design it continues to function reliably.  If it does ever blow up I do have a spare AC-4 power transformer kicking around.

But then again, there's EvilBay...

Wednesday, December 30, 2015

Adding FM transmit to the mcHF transceiver

In previous postings I wrote about how FM reception - including squelch and subaudible tone detection - was accomplished on the mcHF.  As is often the case, it is usually more difficult to receive a signal than to generate one and this is arguably the case with FM as well.

In writing the code to generate FM I found the above to be true:  It was comparatively trivial to produce an FM signal, particularly applying "tricks" that I'd already done in the demodulation.

Producing the FM carrier:

One of the features that I added to the mcHF many code releases back was that of "Frequency Translation" in which the local oscillator was "off-tuned" from the receive frequency, + or - 6 kHz for the mcHF with the "baseband" signals for receive and transmit being shifted by 6 kHz (in the opposite direction) in software.

One reason that this was done was to improve the performance of the transceiver by removing its receive and transmit passbands from the immediate vicinity of the "zero Hertz hole" which not only improves frequency response, but it reduces other issues related to thinks like "1/F" noise and, perhaps most importantly, greatly reduces the likelihood that other receiver audio (e.g. audio amplifier energy) will find its way back into the microvolt-level audio paths via power supply and ground loops and cause feedback!

Applying this to transmit, we soon realize that if wished to produce a signal with a constant carrier, such as AM or FM, we would have to remove ourselves from this "zero Hertz hole" as it would be, by definition, impossible to produce a carrier in that hole as a the carrier is, in fact, represented by DC.  (For SSB, which purposely has its carrier removed, this "hole" is irrelevant...)

This explains another reason why this feature was added:  The eventual addition of AM transmission (and reception) several revisions ago, but this same feature used once again for FM, but in a different way:  Via the use of DDS (Direct Digital Synthesis) techniques.

The use of DDS techniques has been discussed here before - see the article "Generating low-distortion audio sine waves using a PIC and DDS techniques." - link.

Using DDS techniques to generate an FM carrier:

In this technique one generates a sine wave (or any other arbitrary signal, for that matter) by jumping through a "lookup table".  In the case of the mcHF, with its 48 kHz sample rate, if we wanted to generate a 6 kHz sine wave this implies that we would need to step through this sine wave table once every 8 samples.  This sounds easy enough - but how would one do this?

Take a look at this bit of code:

loop:
   accumulator = accumulator + frequency_word
   table_index = accumulator > (size of accumulator in bits - size of table index in bits)
   amplitude = sine_table[table_index]

To explain the above:

- The variables "accumulator" and "frequency_word" are both integers.  Let us presume 16 bits, unsigned, each, which means that each value would range from 0-65535.  Incremented past 65535, it would return to zero.
- "sine table" is a lookup table containing values mapped to a sine wave.  Let us presume that our sine table contains 1024 entries - a number that may be represented by precisely 10 bits.
- "table_index" is used to index the table.  It must be able to index all of the sine table (10 bits) so we will use a 16 bit value for this.
- "amplitude" is the result from the sine table.  This could be an integer or floating point value - whatever format your system ultimately requires.

To calculate the value of "table index" we need to take the top 10 bits of the "accumulator", which means that we can obtain by taking the accumulator value and shifting it to the right by 6 bits (e.g. size of accumulator in bits, minus the size of table index in bits, which are 16 and 10, respectively.)  By doing this we can see that as the value of "accumulator" increases, it also points farther along the sine table.  When the value of accumulator "rolls over" back to zero, the pointer into the sine table also resets back to the beginning.

To understand how the frequency is generated, let us now assume that "frequency word" is set to 1.  We can see that every 65536 times through the loop we will "roll" through an entire sine wave, but since our sample rate is 48 kHz, we know that the produced frequency will be:

48000 / 65536 = 0.732 Hz (approx.)

If we want to generate an arbitrary frequency, we would take the above ratio and use it to calculate the "frequency word" as in:

desired frequency / (48000/65536)

or, rewriting a bit:

(desired frequency * 65536) / 48000

or, reducing the fraction even more:

(desired frequency * 512) / 375

If we wanted to generate a frequency of precisely 6 kHz, the above equation would yield a "frequency word" value of 8192 - which just happens to be exactly 1/8th of 65536, which makes sense since we have already figured out that since 6 kHz is 1/8th of our sample rate of 48 kHz, it would therefore take 8 samples to produce such a sine wave!

Modulating our carrier:

We now know how to generate a carrier, but how to modulate it?

We know that FM is simply "Frequency Modulation", and we also know that by varying the value of "frequency_word" above, we can change the frequency, does this mean that if we superimpose audio on our value of "frequency_word" that we can modulate our signal?

Yes, it does.

Let us rewrite the above code a bit:

loop:
   accumulator = accumulator + frequency_word + audio[audio_index++]
   table_index = accumulator > (size of accumulator in bits - size of table index in bits)
   amplitude_I = sine_table[table_index]
   table_index = table_index + 256
   if(table_index >= 1024)
      table_index = table_index - 1024
   amplitude_Q = sine_table[table_index]

(Let us assume that "audio_index" is updated each time through the loop and represents one sample of the audio, also sampled at 48 kHz, to be modulated onto the carrier.)

Let us first take a look at the first line after the start of the loop where we added the term "audio".  Because our audio is already represented digitally as a numerical value that goes above zero for a positive voltage and below zero for a negative voltage, it would make sense that we could simply add this to our "frequency_word" value.

In other words (pun intended!) when the audio voltage increased above zero, our frequency would increase, but as it went below zero, our frequency would decrease - just as FM would.  What's more, because this is an exact numerical representation, our frequency change would be proportional to the audio applied - which is just want we want to occur for low-distortion, faithful representations of our audio.

Figure 1:
A demonstration of a typical FM signal modulated with a tone as
displayed on the mcHF's waterfall display.
(This picture doesn't have much to do with transmitting, but
I wanted to include some color in this posting!) 
There is another modification to the above code as well.  If you look, you will see that we do two look-ups in the "sine_table".  The first one is our original value, now called "amplitude_I" (In-phase) but we now see that we have taken our table index and added 256 to it which is precisely 1/4th the size of our sine table:  One quarter of a sine wave is, of course, represented by 90 degrees.  After "fixing" that value so that it is always lower than 1024, we look up into the sine table again and call this value "amplitude_Q".

What we have done here is generated two sine waves exactly 90 degrees apart from the same frequency synthesis operation.  As you will recall from your understanding of the "phasing" required to generate an SSB signal, you need both an "I" and "Q" signal for transmit and unlike the generation of quadrature audio for SSB which requires some fairly "hairy" math, we have handily done this for FM with almost no math at all!

Comment:  In reality one would employ modulus operators rather than "greater-than and subtract" -or even logically "AND" the table index value with 1023 (decimal) after adding 256 to it, either being a much quicker operation for a computer than in the example shown above.

Additional audio processing:

As was mentioned in the discussion about the demodulator, for amateur radio purposes we don't actually want to transmit an "FM" signal, but really a "PM" (Phase Modulated) signal.  For our purposes, a PM signal is really an FM signal in which the audio is pre-emphasized at a rate of 6dB per octave - which is a fancy way of saying that a signal voltage that causes +/- 1 kHz of deviation with a modulation frequency of 1 kHz would cause +/- 2 kHz of deviation with a modulation frequency of 2 kHz.  As noted in previous postings, this is done to improve the overall signal-noise performance of the system as it boosts the "highs" in the audio at about the same rate as the noise increases on weak signals.

There are some practical issues with this pre-emphasis that must be considered.  If you were to start your pre-emphasis at 1 Hz, by the time you get to 2048 Hz would have pre-emphasized your audio by 66 dB or so (if I've done my math right) - a ridiculous amount, and any audio content that might be present at higher frequencies would be boosted even more!  Such high frequency content would also cause high amounts of deviation which, in turn, would greatly expand the occupied bandwidth of the transmitted signal - something that is neither necessary or neighborly!

Clearly, this implies that we must do two things:
  • Limit the frequency range over which we do our pre-emphasis
  • Filter the the audio to the desired range for speech communications
If we start our pre-emphasis at around 200 Hz, instead, we can see that by the time we get to 3200 Hz we need to boost only by 36 dB - a far more reasonable value than the 66 dB mentioned above!

For speech we need only reproduce audio from around 250 Hz to something in the area of 2500-2700 Hz.  Our low-frequency limit is imposed by our desire to include the encoding of "subaudible" tones on our transmitted signal and it is important that we remove a reasonable amount of energy in that frequency range so that spectral content of the human voice - particularly that of the adult male - does not encroach in that area and cause reliability problems with decoding on the receiving end.

Fortunately, such tools are already at hand!

Pre-emphasis:

We already met the differentiator algorithm in our receiver as it was used to reduce the low-frequency, subaudible tones from received audio.  This algorith reproduced below.

loop:
  filtered = α * (old_input + input - old_filtered)
  old_filtered = filtered
  old_input = input

Where:
  "α" is the the equivalent of the time constant in that a "small" α implies an R/C circuit with a fast time-constant strongly affecting "low" frequencies.
  "input" is the new audio sample.
  "filtered" is the high-pass filtered (differentiated) audio

In the case of the receiver we used it as a high-pass filter with a cut-off below the speech range, but for transmit we can adjust the "knee" of this differentiator such that it is just above the speech range, instead.  As it turns out, an "α" value of 0.05 is suitable for our purposes.

Filtering:

Having done pre-emphasis, we still need to do filtering, but I'd already implemented a "transmit" filter on the mcHF for both SSB and AM and all I needed to do was redesign the filter to suit the FM audio characteristics.  I used MatLab and the filter designing plug in for this, but the "Iowa Hills" filter designer suite (free and easily found via a web search) could be used to produce suitable sets of coefficients.  As are most of the filters used on the mcHF, these filters were IIR since one can get a lot of "bang for the buck" in terms of good, "sharp" filtering with relatively few computation cycles.

With a fairly compact filter with fairly low computational overhead I was able to achieve >20dB of voice rejection in the upper frequencies used for subaudible tones and well over 50 dB of attenuation above 3200 Hz - much better than that achieved in a typical, analog FM transmitter.  At the low end, audio below 250 Hz was attenuated by at least 20dB with over 40 dB reduction for audio content below 200 Hz - this, to prevent "pollution" of the frequencies occupied by subaudible tones.

Comment:  Because I was using floating-point math, the order in which pre-emphasis or filtering is done is unimportant.  If fixed-point/integer math was used, instead, you would need to carefully analyze the signal path and the resulting values to assure that nothing "blew up" (e.g. exceeded the integer range or, at the other extreme, was so "small" that resolution was compromised and distortion/noise introduced) at the expected audio levels at all frequencies!

Limiting:

One necessary function employed in typical amateur FM transmitters is that of the limiter to "clip" the audio to an absolute maximum level.  This device improves overall intelligibility by allowing the designer to set the microphone gain to a somewhat excessive level, but the clipper forcing a maximum loudness.  The result of this is that somewhat low audio from soft-spoken users is boosted to promote intelligibility while those who have "hot" microphones and/or speak loudly do not cause excess amounts of deviation of the transmitted signal.

The mcHF does not have a clipper, per se, but it does have an audio compressor that was implemented many versions ago to improve usability on SSB.  Like a limiter, this device prevents the audio from exceeding an absolute maximum level and it also adjusts the gain upwards during quiet portions to reduce the "peak-to-average" ratio of the audio, thereby improving intelligibility.

I did experiment with both a "hard" and a "soft" limiter (or clipper) in software.  A "hard" limiter is one that sets an absolute ceiling on the amplitude of the signal present while a "soft" limiter, as the name implies, is less abrupt, more like the "knee" of a diode with some sort of logarithmic-like action.  Because they alter the waveforms, both of these methods generate harmonics and intermodulation products - the "soft" limiter being a bit less aggressive - which require that filtering be done.  Since we are low-pass filtering the audio, the higher-frequency harmonics outside the speech range will not contribute to the occupied bandwidth of the signal but the increased energy in the upper speech frequencies from harmonics of the lower-frequency audio components coupled with the pre-emphasis can somewhat broaden the signal.  Finally, because the signal is distorted by the clipping action, high audio levels that result in a lot of clipping are likely to result in audio that "sounds" degraded.

In comparing the sounds of the limiter/clipper to that of the audio compressor, I decided to use the latter as it was more "pleasing" to the ear and more versatile, overall since there are a number of available adjustments (e.g. the amount of audio into the variable gain stage and the decay rate of the variable gain stage.) As noted, I eventually decided not to use a clipper and used the already-existing compressor, instead.

Without either this compressor or a limiter, an FM transmitter would have the problem of their signal being "too wide" for loud-speaking operators and "too quiet" for those that were soft spoken - neither condition being desirable for communications!

Subaudible tone:

A desirable feature of a modern FM transmitter is that of the Subaudible Tone, discussed previously.  This signalling method consists of the generation of a low-level sine wave in the range of approximately 67 to 250 Hz that is superimposed on the transmitted audio which is used by the receiver to validate the presence of signal.  While this was traditionally used in commercial radio to allow several groups of users to "share" the same frequency, amateurs have typically used it as interference mitigation techniques to prevent the receiver - that of the repeater or the user - from responding to noise or signals for other sources.

For additional information about subaudible tone signalling, read the Wikipedia article - link.

Since it is just a sine wave, it is very easy to generate - and we already know how!

Re-using the DDS algorithm, above, we need only generate a single tone, unmodulated this time, and sum it with our transmitted audio.  We would of course, do this after we have done our pre-emphasis, filtering and limiting/compressing, placing this tone generator just before the DDS that produced our FM signal as the code snippet below illustrates.


[Filtering and limiting/clipping of audio already done]

if(tone_generator=TRUE) {
      tone_accumulator = accumulator + tone_frequency_word
      table_index = tone_accumulator > (size of tone accumulator in bits - size of table index in bits)
      tone = sine_table[table_index]
      audio = audio + (tone * amplitude)
}

[DDS that generates the FM signal as shown above]

As we can see, only if the tone generator is turned on do we go through the loop - something that we'd do to save processing power.  Included in the above code snipped is an additional parameter, "amplitude" which would be used to scale the value from the sine lookup table such that it yielded the proper amount of deviation on the transmitted signal - typically in the area of 15-20% of peak deviation.

In the case of generating the audio tone we'd need to make certain that we had enough frequency resolution to accurately produce the tone, and as we already calculated we know that with a 16 bit counter at a 48 kHz sample rate our resolution is approximately 0.732 Hz. Assuming no sample rate errors, this would imply that we could generate the desired frequency to within half that resolution worst-case, or approximately 0.366 Hz.

This frequency resolution is adequate for generation of these tones, again assuming that there are no additional error sources related to sample rate, but if you were not satisfied with that amount of resolution it would be a fairly simple matter to increase the number of bits used by the accumulator and frequency word to improve the resolution, just as was suggested for the frequency modulation.

For the calculation of the frequency words, all that was required was that the code include a table containing the frequency, in Hertz, of each of the subaudible tones:  Since we already know the sample rate and the number of bits - and therefore the maximum counts for our accumulator - we can calculate, on the fly, the needed "frequency word".

Tone burst:

There is one more tone signalling scheme occasionally encountered on FM repeater systems, and that is the "tone burst", sometimes called "Whistle-up".  Although it has largely disappeared from use, it is reportedly used in some areas in Europe.

In this system a short burst of a specific tone, typically 1750 or 2135 Hz, is transmitted to "wake up" a repeater for use, and once this is done, it may be used normally.  Once the repeater has again become dormant, a timer expires and it will no longer respond to signals until it, again, receives a burst.

For a Wikipedia article that includes a section about single-tone signalling, look here:  link

This is generated in exactly the same way as a subaudible tone, namely with a bit of DDS code that looks just like the above!  From a purely practical standpoint, unless one absolutely needed to generate both a subaudible tone and a tone burst at the same time, one could actually use the same bit of code - provided that the amplitudes of the different tones (subaudible, burst) were taken into account.

Unlike a subaudible tone, a tone burst is typically transmitted only at the beginning of a transmission and for a fairly short period - perhaps one second.  While one could rely on the user to time the duration of the tone burst, on the mcHF the duration of the burst was timed by counting the number of interrupt cycles called to process the audio, making the process semi-automatic:  The user needed only activate push-to-talk and then press-and-hold the button that produced the tone and the rest would be completed automatically.

"DCS" codes:

Not mentioned previously there is one additional signalling scheme sometimes found on amateur frequencies, and that is "DCS" (Digital Coded Squelch) which consists of a binary signal with a base frequency of 134.4 Hz modulated with a specific bit pattern.  This signalling scheme is quite rare in the amateur radio community - and even rarer on HF (10 meter) repeaters where this radio is likely to be used - so there has been no serious consideration in its support


How well does it work?

Generating a sine wave with a low-distortion audio generator and feeding the modulated signal into a communications test set (a.k.a. "Service Monitor") - a device specially designed to analyze the quality of communications gear - the modulation was tested at several audio frequencies and found that the distortion was at approximately the level of detection of the instrument to at least +/- 5 kHz deviation.

Testing was also done using speech at various levels, including attempts to overdrive the audio input and on a spectrum analyzer the occupied bandwidth was observed to be contained within the expected bandwidth mask for both the "narrow" (+/- 2.5 kHz) and "wide" (+/- 5 kHz) deviation settings with no audible distortion present nor were there any unexpected spectral components outside the frequency range typical of such an FM signal - even though the "accumulator" of the frequency-modulating DDS is only 16 bits and the audio represented by it will have even lower resolution (e.g. on the order of 12 bits, maximum.)

At the present time I can't think of any additional features that would need to be added to the FM mode so it is, for now, "good to go."

Thursday, December 17, 2015

Minimizing VHF (and HF) RFI from electronic ballasts and fluorescent tubes

Several years ago we started replacing the old, "iron" ballasts with T-12 fluorescent tubes at my work with T-8 tubes and electronic ballasts.  The program to do this was along these lines:
  • When a T-12 tube or ballast failed, immediately retrofit the fixture.
  • As soon as practical, retrofit those lights that are on all of the time, or "most" of the time.
  • Gradually retrofit other fixtures as needed/convenient.
The lamp sockets (known as "tombstones") for T-8 and T-12 sockets are, for all practical purposes, interchangeable.  (Yes, there can be some voltage rating differences...)  Since, in most cases, we were working with what had originally been 4-tube fixtures - two of which (e.g. one ballast) had been disconnected years ago to save energy (we didn't miss the extra light)  we had extra tombstones on-hand to replace those that were broken, discolored or damaged and in some cases we even replaced the bracket and tombstone with those of the brand new "official" T-8 variety.

Figure 1:
Typical "iron" ballast used for T-12 tubes.
Click on the image for a larger version.
We were also able to get T-8 ballasts of known-good brands (Advance, Triad, Sylvania, etc.) for very good prices from an number of sources - far lower than "retail" - and we were sticking only with these brands because we felt that they would offer good reliability and efficiency.  In the years since we have started, the majority of fixtures have been retrofitted and we have yet to replace even a single electronic ballast and we've only replaced a half-dozen tubes - most of those within a week or so of being installed - all of this while providing at least as much light with lower power usage and heat load.

Before we go on, here are a few "weasel words" of warning:
  • The devices discussed present hazardous/fatal shock hazards:  Do not even think about doing any such modification unless you have experience with such things.
  • There are no guarantees, expressed or implied, that the modifications will work for you, are suitable or will meet electrical code requirements in your area.
  • If you do such modifications you take any and all risks related to it, including damage to person or property, injury, fire, or liability.
  • All RF interference situations are unique:  There is no guarantee at all that even when performed as described that the steps described will mitigate problems that you might be experiencing.
  • Please use care and common sense!
  • You have been warned!

The rise of interference:

There is one place where we have had problems with the T-8 tubes and accompanying electronic ballasts and that is the electronics shop.

Soon after retrofitting the fixtures in the shop we noticed that we had difficulty hearing local 2 meter and UHF repeater when listening in that room, the noise floor having risen up by 10-20dB - depending on frequency and exact location within the room - with 120 Hz modulated noise and the fact that it disappeared when the lights were turned off pointed directly to the cause.


Differences between magnetic and electronic ballasts:


For a typical 4-foot (approx. 102cm) tube, good, old magnetic ballasts are, at least when powered from 120 volts, transformers that boost the voltage to that high enough to sustain conduction on each cycle of the 120 Hz sinusoid.  When starting, there are special windings that will activate the filament and/or boost the voltage such that this conduction - which is higher when the tube is "cold" (both physically and has not been lit) but when the current is established as the tube lights, these diminish. To limit the current through the tube to a safe value it is typical that there is a large amount of series inductance which, by virtue of its reactance, sets the maximum power consumed by the lamp.

For more general information about how fluorescent tubes and ballasts work, go to "Sam's F-Lamp FAQ" - link.

For various reasons (core, skin-effect, magnetic and resistive losses, etc.) these old "iron" ballasts are somewhat inefficient, but even so, this efficiency of the linear, fluorescent tube made the combination far more efficient than practically any other tungsten (incandescent) light source.  Nowadays, the skinnier "T-8" tubes are used along with electronic ballasts and these skinnier tubes can, for various reasons, produce at least as much light as the larger T-12 tubes, but in order to maintain such efficiency and to provide reasonable longevity they must be driven with an electronic ballast.

Unlike the old-fashioned "iron" ballast, modern, electronic ballast are essentially high-power oscillators that produce a lot of voltage (perhaps much as 1kV peak, open-circuit and while starting the tube) at a fairly high frequency - say, 30-60 kHz.  Using this higher frequency allows much smaller, lower-loss magnetics to be used and driving the tube with a higher frequency than that of the mains improves its "power to light" conversion efficiency even more.

Electronic ballasts and interference:

Even the old, "iron" ballasts with the fluorescent lamps could cause RFI (Radio Frequency Interference), largely due to significant nonlinearities in the conduction of the tube with respect to the applied voltage resulting in the generation of harmonics.  Furthermore, a malfunctioning ballast could generate interference in other ways, such as internal arcing. Typically, such interference, if present, occurred on the mediumwave (AM broadcast) frequencies and, occasionally, in the HF (shortwave) range and, most typically, either very short range - within a few feet/meters of the lamps themselves - or conducted via the power connections.  Usually, replacement of the defective tubes and/or ballast remedied this situation.

With electronic ballasts the situation is quite different.

With a high-power (10's of watts) electronic oscillator on-board there are already problems at hand:
  • With frequencies in the 10's of kHz, harmonics appearing in the mediumwave and HF are of much lower order.  In other words, the 30th harmonic of the switching frequency in an electronic ballast puts you squarely in them mediumwave range while the 10th harmonic from an iron ballast is still within the frequency range of hearing!
  • The rise/fall times of the waveforms from an iron ballast are quite slow in comparison with those from an electronic ballast which, by necessity of design, can be in the order of 100's or 10's of nanoseconds.  Having these high switch rates on the transistors inside the electronic ballasts puts energy squarely in the MF, HF and even the VHF or UHF range!
  • The length of a common 4-foot (1.2 meter) tube/fixture is well within the VHF/UHF range.  This means that the fluorescent tube itself is perfectly capable of acting as a radiator at many frequencies - not to mention various other parts of the fixture and contained wiring.
Unless one were to scrap the electronic ballast altogether and go back to using the old, iron ballast, if you experience interference issues there are a number of things that you will have to do.

Conduction of high frequency energy onto conductors and radiators:

In the case of our electronics shop the biggest problem appeared to be direct radiation of energy at VHF frequencies, apparently from the tube and connecting wiring.  One way to minimize this effect would be to reduce the harmonic energy emerging from the ballasts and finding their way onto the tube and wiring.

The most practical way to do this with minimal risk of deleterious effects (e.g. loss of efficiency, electric shock, fire, reduction of operational lifetime of the ballast and/or tubes) is to apply a bit of series inductance to the leads that feed the fluorescent tube.  While we are at it, we might as well apply some common mode filtering to the AC (mains) leads as well to minimize conduction of "grunge" that might find its way out via that path, as well.

Figure 2 shows the typical connection of the tubes and electronic ballast.
Figure 2:
Electronic ballast and typical connection to tubes.  In typical "room temperature" installations the filament heater is not used, except as a simple emitter of electrons:  In the example, above, "load sharing" is afforded by the individual, series capacitor for each tube and starting ionization for each tube is provided solely by high-voltage excitation from the ballast.
Not depicted, ballasts are available that have four wires to each tube to provide heater current for starting the tube(s) in cold environments, but most "indoor" ballasts have only a single wire connected to each end of the tube.
Click on the image for a larger version.

As can be seen, the tubes are in parallel with the output of the high-power oscillator, typically using a series reactance (capacitor) to limit tube current and to provide a mechanism for load-sharing and offer operational stability in the presences of the tubes' negative resistance characteristics.  Many electronic ballasts allow different numbers of tubes to be connected:  3-tube ballasts will allow 2 or 3 tubes while 2-tube ballasts will usually allow just one tube to be powered, and using these series reactances is a simple way of keeping the tube current more-or-less constant, despite the number of tubes connected - either at the time of installation or due to later tube failure.

Figure 2 also indicates something else:  On the wires marked "A" and "B" are carried the high-frequency currents from the ballast's oscillator.  As we discussed before, not only are these same currents very non-sinusoidal and contain many harmonics, but the tubes themselves tend to badly distort the currents, further-increasing the harmonic content on these leads.  Finally, at VHF and UHF frequencies, the very lengths of wires "A" and "B" and the tubes themselves, "T", can make them capable radiators in their own rights!

The distortion added by the tubes themselves is likely not significant one gets above MF and HF frequencies as the mechanism is generally too slow in its own right, but we still have the harmonics from the oscillator itself to contend with which can be carried, as current, by the ionization of the gas within the tube itself which will form an antenna and as we know, as long as there is some current flowing, we will have electromagnetic radiation.

Inductors used for reducing the harmonic content on the tube current:

Our best bet is to reduce the harmonic content on the tube current itself and the easiest way to do this with minimal intrusion and risk to the user is to insert a small amount of inductance onto the connecting leads "A" and "B".
Figure 3:
A pile of toroidal inductors pulled from some scrapped power supplies.
In the upper-left is a bifilar choke used for mains filtering, also found
in higher-quality switching supplies.
Click on the image for a larger version.

We had a number of junked switching power supplies in the "boneyard" and we pulled a pile of toroidal inductors (see Figure 3) from some of them and did a quick analysis.

The most common type of toroidal inductor is the sort typically found in PC power supplies, usually used for output filtering, wound in either a yellow or yellow-and-white core.  The permeability of these cores isn't extremely high, typically achieving somewhere in the area of 20-30 microhenries with 10-15 turns, but they are not particularly suitable for use much above the range of 100-200 kHz:  Higher than this they get quite lossy - but we don't really care too much about that since we are trying to quash this energy, anyway.

The "Yellow-White" cores appear to be "26 Mix" Iron Powder types that are specified for use from DC to approximately 800 kHz.  The aforementioned "Yellow" cores would have the color code for "6-mix", but when they were measured, their permeability was closer to that of "26 mix" indicating that whoever made them wasn't precisely following color convention!

The sizes (outside diameters) of these inductors varied, from approximately 1 inch (25.4mm) in diameter to over 2 inches (50mm).  If you were to translate these to "store bought" toroids these could equate to sizes from "T94-26" to as large as "T200-26" or "T225-26".

Another inductor that appeared to be useful was one that was simply painted light gray and it, too had permeability roughly equal to that of the yellow/yellow-white ones - that is, 10-20 turns yielded something in the area of 20-100 microhenries.  I've found no obvious reference to "light gray" painted toroids in the typical references so I don't know what "mix" they might be, but they appear to be vaguely similar to the "Yellow-White" ones in general properties.

When either the gray or the yellow/yellow-white inductors were broken, it appeared that they contained some sort of metallic powder, pressed and molded into shape and covered with paint. 

The third type of inductor, often used for filtering the the AC mains input, appeared to be some sort of very high permeability ferrite and putting 10-20 turns on these yielded hundreds of microhenries or even millihenries of inductance:  These are not suitable points "A" and "B" and should be set aside for now - but we will be using them later.  When broken these particular cores were either a black compound (e.g. ferrite) or, in a few cases, wound with some sort of thin, ferromagnetic tape as in the case of the red-cored inductors depicted in Figure 3.

Which toroids to be used where?

Let's take a look at a redrawn version of Figure 2:
Figure 4: 
A fluorescent ballast/fixture with added ferrite to minimize RFI by virtue of "isolating" the fluorescent tubes at
radio frequencies.
For "cold weather" ballasts that use filament current for starting, each "pair" of wires to each tube would be treated as a single conductor through L1-L3.
See text for explanations!
Click on the image for a larger version.
First, let us take a look at inductor L1 which is in series with the "common" lead "B" that connects to one side of both tubes.  With the inductors that we had we are able to put approximately 15 turns of the original wire into this core which yielded an inductance of 10-15uH (value not critical!)  In running the numbers we can see that at 30 kHz, a typical operating frequency of an electronic ballast, this inserts approximately 3 ohms of reactance into the circuit.  Considering that for a 32 watt, T8 fluorescent tube that the average current will be around 300 milliamps (see reference, here - link) we can calculate that if the current were perfectly sinusoidal (it is not!) the effective loss would be a volt.  Because this is a reactance, we would not experience I^2R losses in the same way that we would if we were dropping the same voltage with an ohmic loss (a resistor) but even if we were, this would amount to only 1/3rd of a watt or so.

If we were to translate this same inductance (15 uH) to the middle of the AM broadcast band (1 MHz) we would see a reactance of around 94 ohms - significantly higher than at the operating frequency, reducing the potential current through the tube, at that frequency, by a factor of around 30 which corresponds to 10's of dB reduction.  Since only a fraction of the oscillator's energy is at harmonics of this magnitude a negligible amount of power is being blocked by this inductance.

Taking this to an extreme, let us consider a frequency of 150 MHz with this same 15uH of inductance where we find the reactance to be approximately 14000 ohms - extremely high and, for all practical purposes, completely "blocking".  In reality the effective series resistance would not be nearly this high as not only would this particular core material ("26" mix) not be effectively offering such inductance at this frequency, but there would also be capacitive (shunt) coupling across the inductor/windings itself and between various conductors (wires) located under the cover of the ballast shield.  Nevertheless, the impedance would be significantly increased due to the inductance present and the core losses at this frequency.

Taking another look at Figure 4 we can see that we'd need to install similar filtering in the other leads connecting to the tubes as well (e.g. the "A" leads) and the similar math applies.  By installing such inductors in both leads we effectively "isolate" the tube at high frequencies by virtue of the series inductance while leaving the frequencies at which the tube is powered - around 30 kHz - virtually unaffected.

How do we know that this does not affect normal operation?  A bit of emperical measurement can (and did) verify this assertion:
  • A "Lux Meter" placed below the fixture showed no discernible difference in light output before and after modification.
  •  The added inductors did not get perceptibly warm:  If they did, this would indicate power loss!

A core not to use at points "A" and "B":

In an experiment I placed one of the high permeability ferrite cores (not one of the yellow cores) in place of L3:  It immediately got too hot to touch, indicating both high losses and a significant amount of inductance.  When checked on an inductance meter I found that approximately the same number of turns as I'd placed on this core when tested in the fixture (15 or so) yielded around 200 microhenries - a reactance of around 38 ohms at 30 kHz - a significant amount at 300 milliamps of tube current!

The upshot:  When trying an core, verify that it does not warm up by an appreciable amount before deciding to use it!

If it gets (noticeably!) warm after several seconds/minutes, it is not suitable and could pose a hazard due to heat:  This heat could, in theory, melt insulation which could pose an electrical/fire hazard or the heat itself could melt or cause combustion in its own right!

Don't worry - we still plan to use this core!


Additional modification:

Figure 5: 
Added inductances on the leads marked "A" in Figure 4,
above.  The inductor on the far left is a multifilar choke -
see the text below for an explanation.  There are actually three
small yellow/white toroidal inductors, one being obscured.
Click on the image for a larger version.
If you look at Figure 5 you will see, on the right side, the three, smaller yellow/white inductors (one is obscured.)  In this particular case this is a 3-tube fixture, the third tube used to provide extra light as it was placed over the workbench area.

Because it was directly over the workbench, it was typically much closer to the gear being used and tested which made the noise emitted by it a bit more problematic by proximity.  To reduce this noise still-further an additional inductance was added in the form of the large, yellow inductor on the left.  This particular core was taken from a scrapped PC power supply (almost all of them have such a device) and the three wires were wound, in parallel - keeping the same number of turns - through the core to fill it up to make, in this case, a "tri-filar" winding.  Doing this forced an additional series inductance on the "A" leads depicted in Figure 4 as "F1" but it forces the currents through the tubes to be equal - not too difficult a job considering that L1 and L2 (depicted in Figure 4) have already presented a high impedance (at high frequencies) on those leads already.

While doing the initial work on the first fixture to be modified the FT-817 was laying on the workbench immediately below it and it was noted that by grabbing a wire of the unmodified fixture I could couple some of the noisy RF (at around 144 MHz) into my body which would then be re-radiated as evidenced by the the noise from the FT-817 increasing.

As a rough indicator of the efficacy of the added inductances I observed that after installing L1-L3 I could grasp the wire "after" these inductors (on the "tube" side) and hear no difference, but if I did so on the "ballast" side of the inductors I heard the noise increase, indicating to me that they were doing their job in removing a significant amount of "grunge".  While I was doing this I noted that I could also hear "grunge" if I grabbed around the AC mains wires as well, indicating that some of this energy was being conducted via that route.  Being a commercial building, the mains input is routed via metallic conduit and it was noted by poking around in the "drop" ceiling and placing the antenna of the FT-817 near-ish the power conduit that relatively little of the noise was emanating from the conduit, but since I had a plethora of chokes on-hand I decided to try an experiment.

Using the high-mu (lossy) cores:

Using one of the same "lossy" ferrite cores that got hot when tested, I parallel-wound, in bifilar fashion, as many turns of the AC mains wire of the ballast as I could fit on the core.  As with the milti-filar core, above, it is important that these wires be kept parallel and that the exact same number of turns be used for each of the two mains conductors so that the "common mode" attenuation be maximized.

When wound in a bifilar fashion - and on leads that have only a residual amount of RF energy - we are using these ferrite cores in the manner intended with no fear of their getting hot:  Because both wires of the AC mains go through in parallel, their magnetic fields cancel and the core does not "see" it - it is only that small amount of RF energy that is not supposed to be there that is being blocked by the bifilar winding on the core!  It is important to note that in order for this type of winding to work, you must wind equal numbers of turns of both wires, preferably in parallel or gently twisted together:  It is by virtue of the equal inductance and coupling between these two wires that this sort of filter works.

After doing this I did the "grab" test with the wire again:  If I grasped the wire on the "ballast side" of the newly-added bifilar inductor I could hear an increase in the noise on the FT-817, but if I did so on the "mains" side I could not.  In other words, this choke was keeping the "grunge" from the ballast out of the mains!

While it may not have been particularly important on VHF/UHF to add this extra bifilar inductor on the mains power leads, it was pretty easy to do since I had the fixture "open" already - and I decided that every little bit helps!  Also, it is worth noting that on MF and HF that the it will likely be via the mains power leads that the majority of noise will be conducted while at higher frequencies like VHF and UHF, the wavelengths are small enough that energy can be emitted directly from the lamps and fixtures themselves as their very size is a significant portion of a wavelength!

Efficacy of the modifications:  Did they work?

Of the 6 fixtures in the shop, only the two directly above the workbench where they are closest equipment under test now have both the small chokes in each of the leads in series with the tubes (e.g. L1 and L2).  All of them have an choke on the "common" side of the tube (L3), the common-mode choke on all of the tubes (F1) and the bifilar choke on the mains side (F2).

Did this modification have any effect?  The quick answer is yes and here are the ways that it was measured.
  • A portable HF/VHF/UHF all-mode transceiver with a whip antenna (an FT-817) was placed underneath/near the fixture, the distance of the plastic diffuser.  The "S-Meter" reading of the noise of the modified fixture was compared with an unmodified fixture with the same complement of tubes and ballast type/model and the modified (filtered) fixture was found to radiate far less than the unmodified one based on observations of the S-meter and audible noise.
  • Two spectrum analyzers were set up on the workbench in the electronics shop connected to small whip antennas and sweeping in the range of 100-200 MHz.  After all of the fixtures in the shop were modified, the amount of noise indicated on the analyzers was 10-20dB lower than before the modification, depending on frequency.  It was not completely gone, but significantly reduced.
  • Using a portable handie-talkie transceiver, several VHF (2 meter) repeaters were checked for signal quality within the shop before and after fixture modification.  All but the strongest were inaudible in the shop prior to modification, but all were easily audible afterwards.  A slight amount of extra noise is apparent when the lights are turned on, but there is no trouble in finding a location that provides a suitably noise-free signal for testing/monitoring now!

Final comments:

The addition of screening

Prior to the installation of the chokes in series with the lamps some experimentation was done with metallic screening of the fixture.  The particular fixtures that we are using are light-gauge Lithonia fixtures that have spot-welded seams and appear to be likely to provide a reasonable RF-tight seal on the back side:  It was via the open front, through the plastic diffuser that RFI was radiating - apparently from the tubes themselves.

In an experiment we decided to cover the front of the fixture with aluminum foil.  While blocking the light, it would serve to help us determine if this was a viable means of containing the RFI within the enclosure.  We placed, lengthwise, two parallel sheets of foil over the diffuser, but it happened that there was a very narrow (1/4" inch, approx. 6mm) gap between the two strips running along the long dimension of the fixture.

We noted that with the foil in place that the interference actually got worse until we bridged the narrow gap with small pieces of scrap foil in 3-4 places.  We surmised that the two pieces of foil were acting as independent radiators until we connected the two, at which point the noise level was reduced - but it was still significant.  Clearly, the foil itself was coupled to the the noise energy within the box and re-radiating a significant percentage.

We then bridged the foil to the steel box itself, causing a further reduction in noise - but the amount of reduction depended strongly on where, exactly, we made this connection.  If this was done in each corner of the box (e.g. four places) there was a significant reduction.

Practically speaking one would not use aluminum foil to provide such shielding as it would clearly block the light.  Aluminum window screen would also not be recommended as there is not guarantee that there is an electrical connection across its entire face as each wire is not "bonded" to its neighbor - plus it, too, blocks a significant amount of light.

This leaves the use of so-called "expanded metal" or "hardware cloth" - both of which are electrically-connected across their plane as options.  Of the two, the "hardware cloth" - typically galvanized steel mesh is, by far, the cheapest and least "light blocking" alternative - plus, it is possible to solder to the zinc coating with relatively little difficulty.  This material is available in different mesh sizes and for VHF/UHF frequencies the larger sizes (around an inch or several centimeters) would be adequate.

The problem would be providing a firm, electrical bonding of this screen to the case of the light fixture.  On the "hinge" side of the diffuser cover one would use very short, wide metal straps to make the connection to the case, but for best efficacy it would also be preferable to make a similar connection on the side opposite the hinge where the cover opens - but aside from the installation of "finger stock" or very short plug/receptacles, this is rather complicated to do!

About LED replacements:

I am aware of LED drop-in replacements for T-12 and T-8 fluorescent tubes, but has, for the time-being, dismissed these owing to issues of cost, reliability, efficiency, and "light quality" (e.g. color rendition or "CRI").  For an in-depth report on this subject see "Performance of T18 and T8 Fluorescent Lamps and Troffers and LED Linear Replacement Lamps" - link.  Since this report was published significant improvements have been made in LED efficiency, but similar issues - particularly those related to quality and longevity - remain unless one uses devices made by reputable manufacturers.

As far as EMI/RFI issues related to LED replacements for T-12 and T-8 tubes:  Based on reports in the amateur radio press and on forums such as the EEVBlog (examples here - link and another link) I would expect similar - or worse - problems to occur, depending on the manufacturer - as they would also have issues with high slew rates on switching regulators and/or semiconductor-related transients.  While it is very likely that common-mode mains filtering would also be recommended on these devices as well, it may be more difficult to remove noise along the entire length of a series or series-parallel fed array of LEDs that was nearly 4 feet (1.2 meters) long as this is a significant portion of  a wavelength at VHF/UHF frequencies and could easily radiate on its own right!

If you do wish to upgrade an existing "troffer" or similar fluorescent fixture to LEDs, I would strongly recommend that you obtain JUST ONE and carefully analyze it before you risk ruining your HF, VHF and/or UHF reception with little means of mitigation!