Friday, February 20, 2015

A "quiet" 5 volt USB car power supply

For a previous, related article see the May, 2014 posting - "How USB car power adapters can ruin 2 meter mobile reception" - link.

In that post I wrote about those ubiquitous USB car power adapters that fit in a cigarette lighter and while these devices work well for power phones, GPS receivers and the like, they are terrible if you have any intention of listening on 2 meters while in your car - even if you are using an external antenna.

Unmodified, I found that the unit that I had in my car effectively reduced the sensitivity of my 2 meter transceiver - with its external, permanently-mounted antenna - by nearly 40dB.  To put that into other terms, a signal that was weak with this adapter turned off would have to be increased in signal strength by a factor of 10 thousand to sound the same when the adapter was plugged in with cables attached!

Figure 1: 
The completed USB car power adapter in the box.
Click on the image for a larger version.
Modifying the adapter as noted in the above link I managed to knock down garbage emitted from the adapter by 15-20 dB (a factor of 30-100) - a significant improvement - but when driving up some of the local canyons I found that repeaters that had been perfectly copyable along the entire route previously were no longer audible unless I unplugged the adapter.

Something had to be done!
 
I'd already done about as much
to the small cigarette-lighter USB adapter as I could, short of completely rebuilding it, but it was still too noisy.  The problem with this device was that of differential currents:  Between the input and output terminals of the device there were, on the circuit board, several amps of switching current floating around.

Even though there was a common "ground" shared between the input and output, this same "ground", consisting of just a few short and somewhat thin traces rather than a large, heavy and solid ground plan was suprisingly reactive at higher frequencies - such as those above a few 10's of MHz.  Even a very short length of circuit board trace can have a few 10's of nanoHenries of inductance, but if you are pushing amps of current and considering harmonic energy at 100+ MHz, you can soon see that these seemingly few nanoHenries can make what would seem like a "solid" ground plane act like the feedpoint of an antenna.  With the switching supply itself as the transmitter and the car wiring and the connected USB cables acting like the wires of a dipole antenna one can soon see where the interference was coming from!

Aside from having the voltage converter designed on a "proper" multilayer circuit board with very high quality components - not something that you'll get on a $5-$20 power adapter - the only other way to take care of the problem is to put the entire thing into a metal enclosure and filter/bypass all of the power leads going in and out.

Two approaches:

There are two approaches that I could have taken to accomplish this task.

 1)  Containing noise from the original adapter.

The easiest would have been to take the original USB power adapter and put it in a shielded enclosure and bypass all of the leads going in and out.  I would, of course, have lost the convenience of the small, self-contained device that plugged into the cigarette lighter, but I would have solved the problem.

The techniques described below could apply to filtering the original adapter, just as they did the approach that I ended up taking.

2)  Build another supply.

Figure 2:
The "DC to DC Converter Step-down Voltage LED Power Module 3A"
obtained from EvilBay.  (Ignore the slightly messy soldering - that
was the fault of myself and the often  "grainy" nature
of  lead-free solder!)
Click on the image for a larger version.
If you have been following this blog or read any of the other entries, you will not be surprised that I took this route - but I did not build it entirely from "scratch", but I went to EvilBay and found some inexpensive buck-type switching regulators for under $2 each and used them as a starting point:  I couldn't even get the components themselves for $2!

The devices that I found were described as "DC To DC Converter Buck Step-down Voltage LED Power Module 3A 12V To 5V 3.3V" (see Figure 2) and are based on the LM2576 switching regulator.  These also had a fixed 3.3 volt regulator on board, but I had no need for that so I left it in place, doing nothing with it.

It is important to note that a cheap, $2 switching regulator from EvilBay is not going to be any better than the original USB car adapter in terms of RF cleanliness and it may even be suspect in terms of reliability so a few things need to be done to the $2 board before it may be deemed to be reliable and useful.

The first things to consider are the electrolytic capacitors on the cheap switching regulator boards:  They are not to be trusted!  This should be considered to be true of any cheap switching supply that you get on EvilBay!
 
Figure 3:
The inside of the converter showing the  components for filtering
and the two switching modules.
Click on the image for a larger version.
In any switching regulator, one of the most important aspects of component selection is that of the quality of the capacitors:  They must be of the low ESR type, designed specifically for switching service and even more importantly, they must be of a known manufacturer, obtained from a reputable seller.

Upon inspection of these switching regulators I saw that the capacitors onboard were rated at 105C - a good sign - but I'd never heard of the maker.  Rather than removing the original capacitors, I simply paralleled them with 100uF low-ESR Nichicon units that I'd obtained from a reputable supplier (either Digi-Key or Mouser) and used a dab of RTV ("silicone") adhesive to secure them into position.  I would not recommend the use of "hot melt" (thermoset) glue for this:  It will be in a car - which will get hot and bounce around and they will break loose!  Putting these "good" capacitors in parallel would take a lot of the stress off the "unknown" capacitors on the board, prolonging their life.

Figure 4: 
A close-up of the added capacitors and the piece of copper added
for heat sinking and mounting of the regulator board.  As can be seen, a
small piece was added to help stiffen the right-angle board and
increase the heat conductivity.
Click on the image for a larger version.

The second thing to consider about these cheap switching regulator boards are their current/power ratings.

These units are rated at 3 amps output, and a quick check on the data sheet for the LM2576 indicated that this was about right.  Going on faith that the LM2576's on board were the genuine article - and not counterfeit devices - I put one of them on the bench supply and loaded the output to 3 amps and found that they held up fine, but that the heat-sinking - such as it was - was not adequate, at least if I were to put them inside a metal box with no air ventilation.

Fortunately, the solution was simple:  Add a bit of extra heat-sinking.

Using a bit of scrap copper, I constructed and then soldered an "L" bracket to the circuit board on the back side of the board, opposite to where the LM2576 was mounted.   This bit of copper would not only conduct heat away from the LM2576, to the aluminum body of the  die-cast box, but it also provides a very good local electrical "ground" connection for the board as well:  Figures 4 and 6 show details on this bracket.  When you solder this piece of copper to the board, be careful in the application of heat as you could easily "un-solder" the LM2576 from the other side:  I did this by accident on this board that you see in Figure 2 which explains the rather lumpy solder connections!


Filtering circuitry:

The schematic diagram in Figure 5, below, shows how the input and output filtering is connected.
Figure 5: 
Schematic diagram showing the filtering and interconnections.
The "5V/3A Buck Conv." are the voltage converter modules as described/modified.
Click on the image for a larger version.

How it works:

Referring to the diagram in Figure 5, above, L1, a 22uH inductor (the toroid in the upper-left corner of Figure 3, wound with red wire) offers impedance to RF that may ingress from outside and feedthrough capacitor FT1 shunts any remaining RF to ground.  L2 (the yellow-core toroid on the far left wound with reddish wire) blocks RF that may be present from the switching converters DC inputs, also allowing capacitor FT1 to do its job.  Capacitors C1 and C2 perform "bulk" filtering of high switching currents that may be present on the DC input lines, coming from the two DC-DC converters.

There are two identical DC-DC "buck" type converters and only the upper one will be discussed:

Inductor L3 (visible on the right side of Figure 3 on edge covered in heat-shrink tubing) blocks residual switching energy and RF that are on the DC line coming out of the switching converter and these are shunted to ground by FT2, a feedthrough capacitor and additional filtering is provided by C3.

As noted above, all of the electrolytic capacitors are of the "Low ESR" types and of well-known manufacturers (I use only Panasonic or Nichicon).  Again, when dealing with switching supplies, these special low-impedance capacitors are absolutely necessary in order for proper, long-term reliability and efficient operation of such power supplies and using any other type of capacitor will inevitably result in reduced operational lifetime and/or efficiency.

Figure 6: 
The back side of the switching regulator board showing the
added capacitors and the mounting bracket/heat sink.  The 100uF, low
ESR capacitors added to supplement the original capacitors
on the (cheap!) regulator board can be clearly seen, held in place
with RTV (silicone) adhesive.
Click on the image for a larger version.

Not mentioned in the above description are components R1, R2 and R3 which are self-resetting thermal fuses.  These typically look much like yellow disk ceramic capacitors (they may be either round or square) and when the current through them exceeds their ratings, they get hot (approximately 100C) and their internal resistance increases, effectively opening the circuit.  Unlike a fuse, when the current is removed they immediately cool down and return to their previous state and reset themselves.

These devices are inexpensive and have the obvious advantage of protecting their circuits like a fuse, but self-resetting after the fault has been cleared!

In the above circuit I happened to use "feedthrough" capacitors which may be seen on the metal barrier near the right edge of Figure 3 (the blue devices soldered into it).  While these devices are especially designed for passing DC and blocking RF, they are a bit difficult to find - but are not absolutely necessary.  Instead of feedthrough capacitors, good-quality "monolithic" multilayer capacitors (the small square ones - not disk ceramic) could be used instead, soldered to the wires with very short leads as they pass or through a hole in the solderable ground plate.

In looking at Figure 3 you will also notice that there are two metal barriers constructed of brass:  One in the upper-left corner, just above L2, and the more obvious one near the right side into which feedthrough capacitors FT2 and FT3 are soldered.  Perhaps a bit of overkill, these provide a (literal!) RF barrier into which the feedthrough capacitors are soldered,  Practically speaking, they provide a convenient place to which the important RF bypassing capacitors may be mounted to the aluminum box to which one cannot solder

Aspects of filtering - why this works:

It was noted earlier that the reason why the original USB power converter was so noisy was that there were many amps of switching currents floating around along the circuit board and even though it was "grounded" at DC, the fact that there was so much current and that the circuit board's traces had some inductance that was significant at VHF was the reason why it radiated badly!

In this circuit, we have taken pains to avoid the pitfalls that would cause it to radiate and if you take your own approach using your own switching converter - perhaps putting that USB power adapter that you already own into its own, shielded box, there are a few things to consider.

In this case, it is the combination of the box itself and the inductors and capacitors that work together to contain the switching energy within the confines of the ground plane of the interior of the box.  It is important to note that it is not the shielding of the box, per se that is the magic here, but the combination of chokes in the various leads and the "solid" ground plane which assures that the circulating currents stay between the input and output leads and do not appear across them where they can radiate.

Take, for example, the output inductors, L3/L4.  The job of an inductor is to resist the change of current so it will pass DC just fine, but it will block AC which means that any RF that gets out of the switcher will hit L3/L4, be blocked by it and whatever small amount of residual energy is left will get shunted to ground by FT2/FT3 and further filtered by C3/C4.

The point here is that on the output lead, L3/L4 will block the RF currents as they leave the switcher, breaking up the path for these currents on the output leads.  This not only prevents that energy from appearing on the output leads, but it also prevents any circulating currents between the input and output as well.

What about switching currents on the input lead?

This is handled by using a good quality capacitors for C1 and C2 which will shunt the vast majority of switching energy to ground.  Residual RF switching energy is then blocked by L2 and whatever little gets through is shunted to ground by FT1 and then there's yet another inductor, L1!

Sources of components:

I happen to have a pretty good junk box of components - particularly the toroidal inductors used.  If you don't have such inductors laying around, junked PC power supplies will likely have what you need in the form of toroidal inductors and small, solenoid-wound chokes wound on ferrite.

The values given (22uH, 47uH) are not at all critical:  Anything from 4.7uH to 100 uH would likely work fine as this range would be more than enough to choke off RFI - but the higher values (22uH and higher) would be somewhat preferable if you have an HF rig in your vehicle that might be bothered.  The most important rating on the chokes to consider is that they be wound with reasonably large wire - say #18 AWG or heavier:  If a couple of amps is pulled through the choke, you don't want it to drop more than a tenths of a volt at most, particularly on the output lead!

As mentioned above, you really do need to get good quality electrolytic capacitors for this and I would recommend places like Digi-Key or Mouser in the U.S. and Panasonic or Nichicon brands. I would strongly suggest that when you go looking for capacitors that you get only those that have "low impedance" and/or "low ESR" in their specifications.  Another thing to look for is their temperature rating:  If they are only 85C, they are probably NOT low ESR or low impedance type.

The die-cast box is approximately 4-5/8" x 2-1/2" x 1-1/2" (12 x 6.5 x 3.75 cm) in size and I obtained it from Jameco Electronics, but similar boxes are readily available surplus and on EvilBay.  A suitable enclosure could be also be constructed using pieces of copper-clad circuit board material, and this would have the advantage of being able to solder directly to it, or one could use a much less-expensive folded aluminum "Bud" type utility box.  The important point is that internally, everything must be connected together on a very heavy, solid ground plane, preferably  without any mechanical joints in the box itself between those internal ground connections.

If you were to just put all of these same components into a plastic box and connect their common point grounds together with thin pieces of hookup wire, you would be risking having RF currents circulating along that thin piece of wire and there being differential voltage across it and having the problem, once again, of RFI escaping the switching regulators!  The only way to make a plastic box work for this sort of thing would be to construct a "box within a box" with the internal one being constructed entirely of metal.

Getting the power out and connecting it to the devices in the car:

Up to this point nothing has been said about getting the power out of this box.

Since the female USB connector is ubiquitous, I decided that this was a good approach so I found some inexpensive "USB Extension cables" on either EvilBay or Amazon (I forget which) and when they arrived, I cut them up, using only rather short portion of the cable with the female USB cable to minimize voltage drop, verifying the power connections in the USB connector using an ohmmeter and a USB pinout diagram that I found on Wikipedia.

It should be mentioned that some devices, particularly cell phones of various brands, particularly those named after fruit, may require that the "data" lines be connected to resistors that "program" the charging current before they will accept a charge:  Not having one of those types of phones I connected nothing to the "D+" or "D-" lines and found that my Android phone charged normally - although it may be that it would charge faster if I would have connected those lines to resistors.  Some phones will pull several amps while charging, hence the use of a pair of 3 amp converter boards!  (e.g. one to run the GPS receiver in the car, the other to charge phones...)

Other devices simply connect to the female USB sockets as normal.

One of the devices powered by this box is my Garmin GPS receiver which has a "special" power cord:  If there is not a resistor across one of the pins of its mini USB connector it will search for a computer when it powers up, delaying its start up.  Since this resistor is built into its power cord I simply removed the cord from the original Garmin power adapter and connected it to the new power converter box and it was happy, booting up immediately, bypassing the check for the computer!


How well does it work?

One of the tests that I ran on this USB supply was to connect a 100 MHz oscilloscope to the input and output leads of the power supply.  With the sensitivity of the 'scope set to maximum I can see just a few millivolts of ripple from the LM2576 regulators, but this energy is confined only to the switching frequency of these devices and the first few harmonics.  Since it is at such a low level it is very unlikely that even if I were to power a sensitive receiver from this regulator that was tuned to the switching frequency that I would even be bothered by it!

The 'scope showed absolutely no evidence of switching energy at higher frequencies so I connected my FT-817 directly to the DC input and output of the supply via a 0.01uF blocking capacitor and only at the lower frequencies (say, 160 meters and lower) could I detect the harmonics of the switching regulators.  If I placed a wire connected to the FT-817's antenna port near one of the power wires going into/coming out of the box, I could hear nothing of it at all!

After all of this, it needn't really be said that this device is completely "clean" at 2 meters and 70cm as well as the FM broadcast band!

What I did hear a little bit of "grunge" from is one of the devices that I run from it (a gps-based dashcam) but breaking that device open and adding a small choke and capacitor on its DC input lead fixed that problem - but a couple of turns of the power lead through a ferrite core would have probably quashed this as well.

Using a "Linear" regulator:

It is worth noting that a much simpler 5 volt USB power supply could have been built using a linear, 3-terminal regulator such as a 7805 or one of its variants.

While this would certainly satisfy the problem of there being switching energy, it would introduce the problem of power conversion efficiency.  In the car environment, wasting a few watts of electricity isn't too much of a problem, but the difficulty is getting rid of heat.  For example, if you were charging your telephone and it was pulling 1.25 amps, running the numbers tells us:

14 volts (typical vehicle voltage) - 5 volts (output) = 9 volts to drop across the regulator

9 volts * 1.25 amps = 11.25 watts of heat to dissipate

11 watts of heat does not sound like much, but it actually takes a fairly large heat sink to do this - and this heat sink must have free air circulation around it.  In other words, if you build a power converter based on this, if you put it into a box, the heat sink cannot be enclosed within the box unless there are a lot of holes and the box itself is located where there can be good convection air circulation!

Alternatively one could build the regulator into a metal box, using the enclosure itself as the heat sink.  Still, it would be a good idea not to bury it somewhere where it could not get some air across it or was exposed to engine heat or in the direct path of air from the heater vent.

In contrast, the aforementioned switching power converter is capable of approximately 6 amps and if one assumes that it is 85% efficient - a reasonable value - one can see that it would not produce much more heat than the above example, worst case!


Final comments:

I could have used the above techniques to clean up the original cigarette-lighter USB power adapter, if I had:
  • Removed the circuit board from its case.
  • Put it in a metal box, grounding it firmly.
  • Supplied DC input power via the L/C (coil/capacitor) filtering as done above.
  • Filtered the DC output power via the L/C filtering as shown above.  I would not have been able to use the USB connectors of the original power adapter, directly, but rather I'd had to have wired in female USB connectors as was done above.
 It would have taken less effort, but it would probably have worked just as well!


[End]

This page stolen from "ka7oei.blogspot.com".

7 comments:

  1. Excellent post, thanks. I've got a lot to learn about this subject. I've just started looking into it because I installed a 555 timer circuit in my car to give an intermittent windscreen wiper function, but noise on the car 12v supply when the engine is running is causing the timer circuit to give random results. I definitely need some kind of filtering but I don't think my problem is RF... maybe just voltage spikes.

    ReplyDelete
    Replies
    1. For your 555 circuit, I doubt that it is RF: I'd just put a 10-47 ohm resistor in series with the power supply of your 555 with a 47-220 uF capacitor across the 555's power bus, "after" the added resistor: This should suppress any glitches that make their way in from the car's electrical system.

      If you use a relay to trigger your wiper motor it is recommended that it be on the "non-555" (e.g. not filtered) side of that added resistor and that there is the normal "catch diode" across the relay's coil.

      Delete
  2. Thanks for your reply! If you can spare the time, could you help me out with these follow-up questions to your idea:

    Does the resistor have any effect on the voltage supplied to my 555 timer circuit, or is the 555 such high impedance that it doesn't make any difference? My relay coil measures about 300 ohms, so I guess I'll need to be sure the relay still operates with the RC filter resistor that I choose.

    About the relay not being on the filtered side - at the moment I have the relay coil connected to the 555 timer output with a diode across the coil. I'm not sure how to activate the relay without it being on the 555-side (filtered side).

    Finally, I read that somebody had success cleaning up their supply by using a large capacitor across their circuit (4700uF). What would be the difference in filtering between an RC circuit and the idea of using just a big capacitor? I much prefer your RC circuit suggestion because I think the 4700uF cap across my timer circuit would cause serious arcing on my switch contacts every time I switched on the circuit.

    Thanks again KA7OEI.

    ReplyDelete
    Replies
    1. You mention that your relay's coil is around 300 ohms, so if you put a 10-15 ohm resistor in series with the power supply for the circuit - including the power to the relay - will drop the voltage only few percent under that load.

      The reason for the resistance is to improve the relative efficacy of the capacitor to be added across it. Without it, spikes/surges would have only the resistance/inductance of the wiring connecting the circuit - along with the added bulk capacitance to suppress spikes.

      With the resistor an additional amount of loss is added, acting like a sponge for the spike, but with the much higher impedance a smaller capacitor may be used and to even greater effect. Because the timing of the 555 is based on ratiometric voltage measurements its timing is not affected (much) by changes in voltage.

      Having said that, changes in supply voltage during the timing cycle will affect the timing slightly - as will happen when the relay pulls in and consumes a bit of current - but this will, at worst case, require a few degrees of adjustment of the timing control.

      Delete
  3. Much appreciated KA7OEI. I'll let you know how it goes.

    ReplyDelete
  4. Hi, just to update you: the RC filter, along with moving the relay coil off of the filtered supply and on to the noisy supply, worked great! No more random firing of the relay by the 555 timer, and the delay time in the car is only 2% faster than it was on the bench power supply. A very good result, thanks for your help.

    ReplyDelete
    Replies
    1. Great to hear that, Daz! Resistors+capacitors are a great combination in cleaning up short-duration spikes like those that can appear in the automotive environment - especially when the current is low enough that a few 10's of ohms can be put in series without affecting the voltage.

      Delete





PLEASE NOTE:


While I DO appreciate comments, those comments that are just vehicles to other web sites without substantial content in their own right WILL NOT be posted!

If you include a link in your comment that simply points to advertisements or a commercial web page, it WILL be rejected as SPAM!