Wednesday, October 15, 2014

The "Pointless" 10 meter DSB QRP transmitter

Figure 1:
The slightly-modified "Shake Light" and the transmitter.  The flashlight portion
still works!
Click on the image for a larger version.
The annual "Homebrew" meeting of the Utah Amateur Radio club meeting was coming up and while I was going to show off what I'd done thusfar on my mcHF transceiver - link I wanted something else that was a bit more "fun" and a very quick build, preferably one that could also give a signal to hear on the transceiver at the same time.

After a bit of brainstorming, something simple and pointless occurred to me - the so-named 10 meter DSB (Double SideBand) QRP transmitter in the title of this article.

I'm not sure where the idea came from but the germ was, no doubt, related to a previous article in this blog, "A Mechanically-powered Capacitor Flashlight" (May 13, 2012) - link  where I describe the inner-workings of the "shake lights" that were the objects of the annoying commercials that had been playing at the time.

These commercials have (thankfully!) disappeared from TV and the flashlights largely gone from the stores, but I bought several of them on clearance a few years ago.  As noted in the article linked above, while these particular flashlights actually do have electronics in them that will generate electricity in response to mechanical motion, not all of them do as some are apparently electrically inert, being marked as a "Toy" as the article "What Light Through Yonder Flashlight Shakes - Link describes.

The flashlights that I got on surplus aren't as good as the somewhat expensive ones mentioned in the article above, but I was curious how well they might power a simple, low-power transmitter.

The transmitter itself is very simple and is depicted in Figure 2, below.

Figure 2:
The diagram of the "Pointless" 10 Meter QRP DSB Transmitter.
C4 is connected directly in parallel with the 0.22 Farad capacitor already inside the "Shake Light" which means that the flashlight's "run" time is also increased!
If a 50 ohm "antenna" load is used instead of the telescoping antenna, the 3.9uH inductor would be omitted.
Click on the image for a larger version.

This transmitter is almost as simple as it could be!
  • The RF energy comes from a 28.570 MHz crystal oscillator module that I found in my box of "Crystals and oscillator modules."  At 5 volts it consumes 5-6 milliamps and it seemed to function down to about 3 volts.  I picked this frequency because it was in an amateur band on a frequency where SSB (or "DSB", in this case) was legal.  The fact that it was 10 meters was a plus since that made the antenna a bit more efficient.  In testing other oscillators, I noted that some drew over 20 mA, so I stuck with the lowest-power device that I found.
  • U1 is a doubly-balanced mixer that I pulled from a piece of scrapped satellite equipment.
  • T1 is a 1k-to-8 ohm transformer from Radio Shack, part number 273-1380, but about any similar impedance transformer would probably work.
  • The Microphone is a standard electret microphone available about anywhere.  I think that it came from a junked telephone handset, but I'm not sure.
  • The 1 Farad, 5.5 volt capacitor came from a junked piece of computer equipment that had used it for a memory backup power source.
  • Some telephone cord - the sort with embedded cloth woven in - used to connect the flashlight to the transmitter.  This cord is very lightweight and able to flex repeatedly without fatiguing and breaking - plus it was free, having been cut from a junked telephone handset.
  • The "Q1" microphone amplifier circuit consisting of all of three components (not counting the coupling capacitors) to boost the audio.
  • No RF bypass capacitor on the V+ lead is shown as the oscillator module seemed to have one internally, so I didn't bother...
How it works:

Assuming that capacitor C4 is already charged up to something around 5 volts, pressing switch SW applies power to the oscillator, the microphone, and microphone amplifier, Q1.  Q1 boosts the audio from the microphone by a factor of about 100 or so and transformer and T1 converts the rather high impedance from the amplifier down to something closer to the 10's of ohms that U1, the mixer, wants to see on its diodes.

Figure 3:
View of the back side of the transmitter.  The 1 Farad capacitor,
C4, dominates the view!
Click on the image for a larger verion.
When current from the audio transformer upsets the diodes in U1, the mixer, it gets out of balance and RF gets through:  The more out-of-balance from the higher audio level, the more RF gets through.

Interestingly, if the audio goes positive, the phase of the RF goes one way, but if the audio goes negative, the phase of the RF goes the other way, providing symmetry in both amplitude and phase.  In the process it generates "Double Sideband" (DSB) which is the same as single sideband except that both USB and LSB are generated at the same time.

The output of the mixer goes through L1 which, in this case, is a 3.9 uH choke.  For this device I used a telescoping whip about 26.5 inches (67.3 cm) long and this amount of inductance "approximately" resonates it at this frequency - very approximately, but better than if I'd just fed the output of the mixer directly to the whip!  The RF output power has not been measured, but it is likely on the order of 1-2 milliwatts PEP, at most!

It is difficult to tell from the picture in Figure 3, but the entirety of the plastic-bodied BNC antenna connector is actually insulated from the grounded aluminum bracket to which it is mounted and both its "shield" and center are connected together.  This allowed me to "bypass" any matching network on an antenna that I might try and use the telescoping whip antenna that I'd happened to grab from my box of VHF/UHF whip antennas.  It was observed that the addition of the 3.9uH inductance added to the range when using the whip:  If you were to connect the BNC connector to a 50 ohm load (e.g. an antenna that was already resonant) then the shell of the BNC connector would be grounded and 3.9uH inductor would be omitted.

Note that there is no low-pass filter anywhere on this design - other than the very slight effect of L1 - so caveat emptor!

A few comments about the design:

Messing with the mixer:

When I used the RMS-11X mixer - which is surface-mount - you'll see from careful scrutiny of Figure 4  that I used tinned (#26 AWG) hookup wire to mount it upside-down ("dead bug") on the prototype board, bridging the three common "ground" connections and taking them to the bottom of the board along with the other three connections, namely the LO (Local Oscillator), RF (typically the signal input or output) and the IF.

As is typical I connected the "LO" port of the mixer to the crystal oscillator via C3, the capacitor being used to block the DC component of the oscillator and reduce the current consumption since it is a DC short and the output of that oscillator is a square wave with a DC component.  Meanwhile, the mixer's "RF" port was connected to the antenna with the "IF" port going to transformer T1 as the audio source.

I was, however, surprised when I got no modulation at all and decided to investigate.

Figure 4:
The Top of the "Pointless 10 meter DSB QRP transmitter.  U2, the oscillator,
is the square can while U1, the mixer, is the white, square block to the left
of it, just above the transformer.
Click on the image for a larger version.
Typically, these doubly-balanced mixers are connected such that the connections to the diode "ring" inside of them are brought out via the "IF" port - which makes since since, in many cases, the "IF" can include audio - or even DC - but when I measured the "IF" port I found a DC "short".  Re-checking the data sheet, I verified that I'd not messed up the pin-out - something easy to do when you are using a device upside-down - but then I remembered something in the back of my mind from a previous time that I'd used one of these RMS-11X mixers:  Checking another connection, I found the (approximately) tell-tale 0.225 volt bi-directional "Shottky" diode drop on the "RF" pin instead - something that I'd apparently noticed in the past, but only dimly just remembered!

Why the data sheet shows the connection in the what that it does is beyond me.  Perhaps it is an error, but it could also be because the "RF" port is sometimes so-designated because it may that which that has the greatest amount of isolation to the "LO" port over most of the frequency range of the device - a fact implicitly bolstered by the fact that "DC" is not included in the "official" frequency range of any of this mixer's ports.

Swapping the positions of the "antenna" and "audio" leads was easy enough and upon doing so I suddenly had modulation - but not very much - so I added the Q1 amplifier, a simple common-emitter, self-biasing design that draws around a milliamp, but provides a lot of gain:  Now, instead of having to yell at the transmitter to drive it into clipping and not get much power, I can talk fairly quietly, up-close, drive it into clipping, and not get much power!

Fighting with Farads:

As noted in the May 2012 blog post, the only reason that these "shake lights" were usable to the consumer at all was that they actually contained a pair of lithium coin cells!  If you removed those coin cells, the magnet and coil assembly did actually work to provide power, but it took a few minutes of shaking to bring the voltage from zero up to the 2.4 volt or so just to get the LED in the original flashlight to shine very dimly - and another couple of minutes to get enough charge before it would shine brightly for, perhaps, a few 10's of seconds.

Even though this entire circuit drew less than 10 milliamps I noted that it dragged the original 0.22 Farad capacitor from a 5 volt charge (provided by an external power supply) instantly down to about 2.25 volts due to its internal resistance, requiring the addition of C4, a physically-larger capacitor with much lower internal resistance.  Dropping only about 0.15 volts under load, this meant that not only would I get better run time, but charging would likely be a bit more efficient since less power would be lost in resistance within the capacitor itself.

How long does it take to charge the entire thing from zero volts?

I'll admit that I've never run this completely down and then tried to charge it up again using the shake light (I'm not a teenager anymore!) but I did do a bit of experimentation and have determined that for every 5 seconds of talking, it takes about 60 seconds of shaking to restore that amount of charge on the capacitor, once it it has already been charged to 4-5 volts.  Based on those numbers it would probably take at least 15 minutes of vigorous shaking with a completely discharged capacitor before one would start to get usable results!

Is this transmitter, powered by a "shake light" practical?

No, not really!

If you wanted a means of mechanically powering low-power transmitter (or even a transceiver!) - one that could produce much power power than this - you would be far better-off using one of those crank-powered flashlights and scavenging power from its mechanically superior means of energy production, instead!

Concluding thoughts:

This transmitter project was built in just a couple of hours in a single evening using parts entirely from my junk box:  The only part that I had to buy (but I had on-hand, anyway) was the 1k-to-8 ohm Radio Shack audio transformer.  The entire reason for building it was to demonstrate that the mcHF transceiver was capable if "hearing" a signal - and to get some laughs - at the UARC Homebrew meeting.

One of these days I'll see how far it will transmit.  I suspect that it will go, perhaps, a couple of thousand meters (or yards) using the whip antenna and a trailing wire for a ground/counterpoise - particularly if the antenna matching is optimized, but being that it is capable of only a few thousandths of a watt output, it will never break a DX pileup on its own!

Practically speaking, an NE602 could have been used instead of the diode-ring, doubly-balanced mixer with many other changes since the impedances for the '602 are radically different from the 50 ohms of the diode mixer.  One difficulty would have been placing a '602 on 10 meters:  Fundamental-mode 10 meter crystals are pretty rare, so an overtone crystal and oscillator would have to have been configured - plus the additional matching components and the likely need of at least one RF amplifier stage to make its output power comparable to that of the circuit above:

Although "different" and, perhaps a bit more current-efficient, the NE602 version would not have been "simpler"!


This page stolen from

Wednesday, October 1, 2014

What to do if your IFR 1200 Service Monitor display just shows "@" signs on the display - and if your deviation meter goes non-linear...

Figure 1.  A properly-working IFR-1200 in receive mode.
A few months ago the old IFR FM/AM-1200S service monitor on the bench at work started getting flaky.  If you turned it on, cold, it would just display a string of "@@@@@@@@@@@@@" ("at" signs) on the display and would not do anything else.

If it was allowed to sit in this state for a while and warm up, another power cycle could often bring it to life, but eventually this trick stopped working and it was completely dead.

Because we had another IFR 1200 kicking around, we used that for a while - but it had a nagging problem with its deviation meter in that it didn't read quite correctly so, eventually, I decided to tear into them both and fix the problems:  More on the deviation meter problem, later...

Fixing the "At sign" problem:

Figure 2:  The display showing "at" signs, indicating
that the CPU cannot properly start.
Click on the image for a larger version.
Pulling the Processor board I noticed immediately that there were exactly FOUR electrolytic capacitors - at least two of them were the typical power supply bypass capacitors, but one of them was related to a line that had something to do with the CPU's startup/reset - in this case, C6, the one near the upper-right corner.  This particular capacitor, if it becomes leaky, tends to hold the CPU's RST line high and the internal pull-down resistor cannot take it out of reset and you get it out of the "@@@@@@" mode.

Since, on old equipment, you can almost always blame the capacitor - and be right much of the time - I decided just to replace them all.  The picture below shows the board and the four capacitors:

The four capacitors are easy to spot:  Three of them are clustered together near the bottom-left corner and the other (C6) is just to the left of the CPU, the large chip in the upper-right corner of Figure 3.

All four of these capacitors should be replaced while you are at it.  Note that your circuit board may also be of the "blue" variety and if so, it may be a bit fragile:  When removing the capacitors use a temperature-controlled soldering iron and proper techniques as these types of boards (e.g. the "blue" ones) are reportedly easy to damage.  I didn't have any trouble when working on this board, but this is good advice, nonetheless!

Figure 3
The Processor board from the IFR FM/AM-1200S (the older "A-3" version).  The culprit is C6, the capacitor just to the left of the CPU near the upper-right corner.  The newer version of the board looks completely different and may/may not have a similar problem.  If it does, it would be due to C3, a 4.7uF capacitor along the top edge of its board.
Click on the picture for a larger version.
I replaced these capacitors with 47uF, high-temperature 105C devices.  It is important to note that the capacitor that you use to replace it must be either:
  • Shorter than the height of the tallest chip+socket on the board, or
  • If the capacitor is taller, you must use extra lead length and lay it on its side, using spaghetti tubing to insulate the extra leads.
The reason for this is that there is very limited clearance between this board and its neighbor:  If the capacitor is too tall it will hit the next board, either making the installation of this (or the adjacent) board impossible, or the aluminum end of the capacitor will short out components on the other board!

By the way, the capacitor that causes the problem with the version of the board shown in Figure 3 is C6, the one in the upper-right corner!

Another reason why your IFR-1200 may be throwing "At" signs:

There is another reason why your IFR-1200 may be throwing "At" signs:  Its SRAM backup cell may be dead, or nearly so!

If you look at Figure 3 you will see a lithium coin cell in the lower-left corner:  Measure the voltage between the top of the cell and the ground plane of the board:  If it is lower than about 2.8 volts you should really replace it while you are at it and if it is lower than 2.6 volts, that is too low for reliable operation.

What happens is this:  When you power off the IFR the backup battery is supposed to hold the contents of the RAM, but when the battery gets too low (below approx. 2.6 volts) there is a reasonable chance of corruption at some point.  The way the software is written it seems possible that the SRAM could be corrupted in a way that the CPU  cannot start up and detect this condition and the unit "hangs".

The only way to "fix" this condition once it occurs is to clear the SRAM's contents, which may be done one of two ways:
  • Under anti-static conditions, remove the SRAM chip.  This is the chip to the left of the windowed chip in Figure 3 that, in the picture, is labeled "TC5517APL-2".  The SRAM chip in your unit may have different nomenclature.  When this chip is removed, stick it in a "bug rug" (conductive foam) or set its pins on a wet paper towel for a few seconds to completely discharge it as its intrinsic capacitance can actually hold its (corrupted!) contents for a while - even out of the socket!
  • Momentarily (for no more than 1 second) short pins 14 and 28 of the SRAM chip to clear the memory.
Unless you replace the cell, this problem is likely to happen again, so you have two options:
  • Remove the cell completely.  You will get a "Checksum Failed" type of error whenever you power it up - and occasionally you may not be able to power it up without turning it off again and waiting a minute or two - but you shouldn't have to take it apart.
  • Replace the cell.
Clearly, the latter option is preferred.  In this unit the cell was a "2325" type lithium coin cell with tabs on it and it had lasted well over 20 years.  This sort of tabbed cell may be ordered directly from Digi-Key (you will have to specially-order it with solder tabs!) but you could substitute it with the more readily-available 2025 or 2032 type of cell, particularly if you were to use a plastic coin cell holder - keeping in mind the available clearance between the two boards.

While the 2325 cell is a 160-210maH cell (typically), the 2032, even though it is smaller, has approximately the same specifications (190maH) and the 2025 has around 160maH if you choose a cell with a known, good brand.  If you used one of these smaller cells, instead of 20 or so years, it may last only 15 or so - still a reasonable lifetime!

Note that you cannot solder wires to one of these coin cells without damaging it, reducing its lifetime significantly:  The only real way to attach one of these cells to the circuit would be to spot weld tabs to it or to use a holder.

* * *

About that problem on the "other" IFR-1200S:  Non-linearity on the deviation meter.

It was noticed that the deviation reading on the other IFR-1200S wasn't "linear".  In other words, if I did a first Bessel Null with a 905.8 Hz tone and set the deviation meter for a reading of 2.18 kHz, it would read higher than 5.00 kHz at the second Bessel null of the same 905.8 Hz tone.

Interestingly, the deviation as displayed on the oscilloscope was correct, but the deviation meter was "off" by an ever-increasing, non-linear amount as the deviation went up.  I then noticed that this was NOT true if I switched to the "Medium" bandwidth mode - which is actually the "Wide" filter with a low-pass filter in the audio path.

What it turned out to be was one or more of the 10.7 MHz crystal filters on the IF on the "10.7 MHz Gen/Rec" board - and the reason that I determined this was that I noticed that the audio coming from the FM demodulator was variously distorted in the "FM Narrow" mode, but not distorted at all in the "FM Mid" or "FM Wide" modes.

Because, as we know, the "Mid" and "Wide" share the same filter (a fairly wide ceramic filter) we could rule out the demodulator itself as the culprit for the source of the distortion.  I also noted that the actual amount of distortion varied with the amount of deviation:  Because the location and amplitude of the sidebands of an FM signal vary with the amount of deviation (and modulation frequency) this also told me that there was something in the signal path that was asymmetrically disturbing this signal as it passed through - and it could only be one thing:  The "FM Narrow" IF filter!

Fortunately, I was able to use an off-the-shelf 2-pole ECS 10.7 MHz 15 kHz wide monolithic crystal filter to replace it with, retune the filter for lowest distortion (as indicated on the IFR's own distortion meter) and this fixed the problem!  Fortunately, these components are fairly cheap and as of the time of writing, still readily available.

For reference, these are Mouser part:  520-107-15B  - This comes as a matched pair of two 2-pole filters, which is exactly what you need - in other words, you buy "one" of these items and get two devices that have been matched at the factory. The Digi-Key part number for the exact, same item is:  X704-ND.

Because these new filters operate at 1.8k ohms instead of the higher impedance of the original crystal filters I had to parallel the input/output (R4 and R10) resistors with resistors to achieve the 1.8k source/termination impedance to properly match these filters:  For this I used 3.0k resistors in parallel with R4, R80, R6 and R81.

The specified ratings for the above pair of filters are 15 kHz at the -3dB points and at least -40dB at +/- 25 kHz.

The input/output termination transformers for these filters should be readjusted (using a non-metallic tool to prevent breakage of the core) according to the manual, although adjusting for minimum distortion with a 1 kHz tone set for 5-6 kHz deviation (using the unit's own distortion or SINAD meter, which also measures distortion/noise, but presents it differently) and then do similar for the "discrimination" coils on the demodulator, as well as slightly adjusting the various bandpass transformers along the signal path as well.  Please note that you will need to check/readjust the deviation for all three ranges of the deviation meter when you are done.

After I did this the deviation meter, once recalibrated according to the manual, behaved normally!

The two filters related to the FM signal path are YFL1 and YFL2.  These are the two metal-can crystal filters located closest to the top of the board (nearest to the connectors) in the center, un-sheilded area.

Note about the AM signal path and filter: 

The "AM" signal path uses a still-narrower filter that is not used in FM demodulation.  If this goes bad a suitable substitute may possibly be one of the narrowband FM filters of the same product line as the above.

It won't have the exact same bandwidth as the original filter, but the input/output impedance of these filters appears to be the same as the original.

The Mouser part number for the single device is 520-107-7.5A while the matched pair is 520-107.7.5B.  The Digi-Key part number for the X701-ND for the single device and X702-ND for the matched pair.

The devices noted above, when used as a pair, are rated at 7.5 kHz bandwidth at the -3dB points and -40dB at +/- 14 kHz with an input/output impedance of 1.8k.

The AM filters are YFL3 and YFL4 and these are the "crystal cans" located just below YFL1 and YFL2 noted above.  There may be small chip ceramic capacitors connected directly to the leads of these crystal filters on the bottom side of the board:  When you remove them, note exactly where they went.  Unless you have the test equipment to "sweep" the AM IF filter precisely, to not worry about putting these capacitors back into place as they were matched with the original crystal filters and the properties of those specific units when they were new.

Since the aforementioned replacements are slightly wider in bandwidth, and since AM is generally more "forgiving" than FM in terms of and phase imbalance across the passband when measuring things like distortion (but not necessarily frequency response) - and the fact that many amateur radio operators would not really be using it for AM very much anyway - just plopping in new filters to replace obviously-defective units will probably result in reasonable performance, anyway!


This page stolen from