Sunday, June 30, 2024

Reducing QRM (interference) from a Renogy 200 watt (or any other!) portable solar panel system

Figure 1:
Renogy 200 watt folding panel, in the sun
Click on the image for a larger version.
A year or so ago I got a 200 watt foldable solar panel system.  This unit - made by Renogy - consists of two glass panels in metal frames equipped with a sort of "kickstand" assembly to allow it to be angled more favorably with the sun to improve its output.  I use this panel when "car camping" to charge the batteries to run the sorts of things that one might bring:  Lights, refrigerator, amateur radio transceivers and who knows what else.  

On that last point, I've done some "in the field" operating on the HF amateur bands while the battery is being charged and noticed that the charge controller (and not the panel itself!) produced a bit of "hash" on the radio - mostly in the form of frequent "birdies" that swished around in frequency as the solar insolation and temperature varied - as well as a general low-level noise at some frequencies.   

This problem is not specific to the Renogy panel's charge controller, but common to almost any panel+controller combination that you will find.

Nearly all "portable" and RV solar power systems cause QRM:

You will find similar systems built into RVs and campers and these are also well known (notorious, even!) for generating RFI.  The techniques described here to quiet interference from these devices applies equally to those as well - but note that one may have to "scale up" the inductors/capacitors to accommodate higher voltages and currents that may be found in those systems.

By placing the solar panel with charge controller and the battery being charged some distance away from the antenna, this interference could be reduced, but that fact that it was even there in the first place annoyed me, so I did what I have done many times before (see the links to other blog entries at the end of this article) and mitigated it by making fairly easy, reversible modifications to the panel's controller.

Portable solar panels and RFI

In my travels, I've been around other users of portable solar panels of various brands and I have yet to find any commercially-available portable panel+controller combination that does NOT produce noticeable RFI on HF/VHF among the half-dozen or so brands that I have checked. 

In comparison with most of the others that I've been around with radios, the Renogy is comparatively quiet - producing less overall QRM with fairly long wires between the panel/controller and battery - than the others - but I decided that I could make it even quieter!

Where does the QRM come from?

It is NOT the solar panel itself that produces the radio frequency noise, but rather the charge controller attached to it.

Modern charge controllers electronically convert the (usually higher) voltage from the solar panels down to something closer to the battery voltage and this is typically done using PWM (Pulse Width Modulation) which means that these devices contain high-power oscillators:  This is true for simple "PWM" types of charge controllers as well as those using "MPPT" techniques.  It's this oscillation / switching action that produces a myriad of harmonics that can extend through the HF spectrum - and even into VHF/UHF!

The "Antenna" in this case consists of two parts of the PV system as depicted in the drawing below:

Figure 2:
A typical solar charging system showing the separation of the two major components that can radiate interference:  The panel itself, connected to the input of the charge controller and the wires and load connected on the output side.
If there is even a slight amount of differential between the two at radio frequencies, the system will radiate.
Click on the image for a larger version.
 
In other words:
  • The wires connecting to the load.  Typically a battery being charged - which can be connected to other things (e.g. vehicle, inverter, etc.)  The wires connecting the panel to these other things - and those devices themselves - act as part of the "antenna" that potentially radiates noise.
  • The solar panel itself.  The solar panel consists of large plates of metal - not only the silicon of the panel, but any metal frame and wiring:  This large area of conductive material offers a suitably large aperture to permit radiation of HF RF.

The "load" and solar panel constitute two different parts of the charge controller's system when it comes to radiation of RF:  The panel is connected to the INPUT of the PWM circuitry while the wiring is connected to the OUTPUT of the PWM circuitry, effectively forming a dipole antenna.  To a degree, the electrical lengths of these two conductors - which can include power cords or even a vehicle - overall can broadly resonate, affecting certain frequency ranges more than others.

The reason for the generation of the interference is due to the fact that the PWM circuitry (which is operating at a frequency of 10s or 100s of kHz) uses square waves, rich in harmonics.  As the voltage input (from the panel) and the output (to the battery/load) are different parts of the PWM circuit, they necessarily have different waveforms on them.

Figure 3:
Charge controller with additional filtering showing added
bifilar-wound chokes on both the input and output leads.
Click on the image for a larger version.

While this device does have some filtering to provide a degree of input impedance reduction (fairly high capacitance) and smoothing of the PWM waveform of the output (more capacitors and likely some inductance) the extent to which this filtering is implemented is suitable for the purpose of providing clean DC power to the load and maximize power conversion efficiency.  This filtering - and likely the controller's circuit board itself - was likely not intended to provide the high degree of RF suppression needed to make it quiet enough to avoid the conduction of RF energy onto its conductors which is then picked up by a nearby receiver.

Containing the RF energy
 
As the controller itself is potted with a silicone material, it's not practical to modify it directly to make it RF-quiet - and there is no need to do so:  Instead, we must take steps to eliminate any differential RF currents that may exist between DC Input and DC output terminals.

Ferrite alone is NOT the answer!

One may presume that the answer to this problem is the implementation of RF device such as snap-on or toroidal ferrite devices - and you would be partially correct.  Any practical inductor - such as that formed by the introduction of a ferrite device onto an existing wire - will have rather limited efficacy in quashing RF currents.

Snap-on devices (e.g. those through which a wire passes) have very limited usefulness at HF frequencies (<30 MHz) - especially on the lower bands - as they simply cannot impart a significant amount of reactance in the conductor onto which they are installed.  At higher frequencies (VHF, UHF) they can have a greater effect - but their efficacy will usually be disappointing at HF.

Using a device that can accommodate multiple turns through its center such as a toroid (or even a larger snap-on device) it may be possible to get up to a few hundred ohms of reactance on a conductor across a fairly wide frequency range - but even this will be capable of reducing the amount of RF by 10-20 dB (2-3 "S" units) at most:  Depending on the intensity of the RFI from the solar controller, this may not be enough to quash the interfering energy to inaudibility - particularly in a remote and otherwise "RF Quiet" location.

To be sure, it's worth trying just the ferrite devices by themselves to see if - in your situation - it reduces the RF interference from the controller to your satisfaction, but remember that the location where you are likely to be using this panel is probably far quieter (RF-wise) than your home QTH:  A "quiet" Solar charging system may seem quiet enough at your noisy home QTH, but could still be noisy in the middle of nowhere.

The addition of capacitors to the circuit can improve the efficacy over ferrite alone by orders of magnitude.  Consider the diagram below:

Figure 4:
Diagram, including additional filtering.  L1 and L2 are the bifilar chokes seen in Figure 3, above while the capacitors (C1a, C1b, C1c and C2a, C2b and C2c) and their implementation are described below.
Click on the image for a larger version.
 
Ferrite devices L1 and L2 are comprised of bifilar-wound inductors on the DC input/output lines, respectively.  These inductors will suppress common-mode RF energy that may appear - but these alone are not likely to be quite enough.
 
In order to force the RF energy to common mode to maximize L1/L2's effectiveness, capacitors C1a, C1b do so for the "external" connections (e.g. those connected to large devices, long wires) while C2a and C2b do so for any RFI emanating from the controller itself.
 
C2c - which is placed between the DC input and output of the charge controller - effectively shunt RF energy differences between the in/out terminals to minimize the differential currents.  Figure 3 shows C2a placed between the two positive terminals, but it could have been placed in any combination (+ to -, - to -, etc.) and been just as effective since the capacitors C2a and C2b effectively short the + and - terminals together at RF frequencies.  If your OCD bothers you, could could add additional capacitor combinations, but the three shown above for C1 and C2 proved to be adequate.
 
The real work for our filtering magic is actually done by C1c.  As seen from the diagram it's shunting RF currents that might appear on the "external" sides of L1 and L2 - which will have been significantly reduced in amplitude by L1 and L2 anyway:  The low impedance of the C1c at RF (a few ohms) coupled with the high RF impedance of the conductors through L1 and L2 work together to make sure that differential RF currents that might exist between the input and output of the charge controller are minuscule, and thus there is effectively no RF energy that can be radiated.
 
Figure 5:
Three 0.1uF monolithic capacitors placed across the
controller's terminals (C2a, C2b, C2c).
Click on the image for a larger version.
Implementation

A glimpse of what was done may be seen in Figure 3.  Some 14 AWG paired copper wire (red/black) was wound on two FT140-43 ferrite toroids - about 6 bifilar turns in this case:  Individual wires could have been used other than "zip" cord - just be sure that the two parallel conductors are laid in parallel to maximize the effectiveness of the bifilar configuration.  Two of these wire/bifilar devices were constructed - one for the DC from the panel and the other for the output to the battery/load.  "Spade" lugs were installed on one end of the red/black wires - two lugs per wire/bifilar assembly.  (FT240-43 or FT240-31 toroids could also have been used, but the FT140-43 is a fraction of the cost, half the diameter, and perfectly suitable for this application.  The FT240 size may be more appropriate if such a filter network is constructed for a higher-current system with larger-gauge wire.)

On the solar controller itself, small 0.1uF, 50 volt monolithic capacitors were installed (C2a, C2b, C2c) to form part of the filter circuitry:  Minimal lead length is important for maximum effectiveness.  While monolithic ceramic capacitors are preferred because they are small (and will fit more easily in tight spaces) and have very low ESR (Effective Series Resistance) one could use disk ceramic capacitors instead.  Film/plastic capacitors are less effective at higher frequencies.

Figure 6:
Terminal strip with capacitors C1a, C1b and C1c.
As described, these capacitors do much of the "bypassing"
of RF differential currents between the input and output.
Click on the image for a larger version.

As can be seen from this picture and Figure 5, the terminals are the "clamp" type and are connected in the same manner as the lugs on the cable on the bifilar toroid assembly. - and also note that this "modification" is completely reversible as nothing at all was changed on the controller itself.

The other end of the red/black wires were soldered to a four-position screw terminal strip as seen in Figure 6 - similar to the one on the back of the charge controller.  As with the terminal strip on the controller, three 0.1uF 50 volt capacitors were soldered (C1a, C1b, C1c) on the back for RF bypassing.  It is possible to have connected the capacitors under the clamps as was done on the controller, but soldering them to the back means that they would not be prone to falling out or being lost if the cables were changed.
 
With these connections made, the wire on the toroids and the connections to the added terminal strip were covered with "Shoe Goo" - a robust rubber adhesive (which may be used to fix shoes, as the name suggests) both as mean of strain relief and to provide electrical insulation.
 
The reader may have noted that we have physically brought together the input/output cables again at this terminal strip - and this was intentional.  By keeping the leads with the bifilar inductors as short as possible and then bringing them back together, we can use the shortest-possible leads on our capacitors to effectively "short" the input and output cables together at radio frequencies, making it impossible for the wires to radiate effectively at HF.  With this, the RF energy is contained within the area of the charge controller itself and the terminal strip/cables and since this is a very small aperture at HF, it can't radiate effectively and additional metallic shielding is unneeded.
 
At VHF/UHF frequencies - where the physical size of the controller+bifilar chokes is a larger proportion of the size of the wavelength (plus the fact that the components used won't work as well at these frequencies) means that some RF energy could radiate, but testing shows that the amount of VHF/UHF RF energy conveyed by the panel and cables was reduced below the point of detection more than a few feet (a meter) or so away from the system.

Spectrum analysis plots

Using a Tiny SA Ultra spectrum analyzer, I coupled the supplied telescoping antenna to the output (battery/load) cable by holding it in parallel with it.  While inductive coupling would have been preferable - and more repeatable and sensitive - this quick test gives a general indication of the nature of the energy being emitted by the charge controller and the reduction afforded by the added filtering.

Take a look at the "before" trace with no filtering:

Figure 7:
"Before" (no filtering) analysis plot with the telescoping antenna of the analyzer held against the DC output cord.
Click on the image for a larger version.
 
Figure 7 spans from DC to 30 MHz with each vertical division representing 5 MHz.  As can be seen, there is a peak of about -90dBm at around 7 MHz (40 meters) and several other peaks across the HF spectrum.
 
Figure 8 is the "after" trace with filtering:
 
Figure 8:
The same plot/conditions as in Figure 7, except after the described filtering was applied.
While it would have been preferable to have better-coupled to the wires from the panel/controller to measure the RFI, this wasn't done at the time in the interest of time resulting in the upper trace showing ambient (off-the air) signals and some local RFI rather than what the panels are producing.
In tests with a portable radio, however, neither the panel nor the output cable (to the battery/load) carried any audible RF interference after the installation of the filtering.
Click on the image for a larger version.
 
Noting the 7 MHz area, we now see that the signal level is around -105dBm - about 15dB lower than in the "before" trace, without filtering - and as we are limited by the background noise energy in this plot, it's likely that the reduction was far more than this.  
 
At the time that these plots were taken, I covered the panel with a moving blanket to "turn off" the solar generation while coupling to the output wire in the same manner as the traces above and there was no difference when I did so compared to the "after" trace of Figure 8.  In other words, the filtering reduced the conducted emissions to levels well below the ambient signal level.
 
Again, these weren't rigorous tests and not as sensitive as they could have been (particularly at the higher end of the HF spectrum).  As the noise floor represents what was in the general area (a slightly RF-noisy location) the plots were unable to resolve the noise floor from the charge controller at higher frequencies that were audible in the field from the power converter, in a truly RF-quiet location.  As it would be easy to reverse this modification I may re-do these plots, this time coupling more effectively into the cable to more accurately show the amount of signal reduction.

Conclusion:
 
Prior to the modification, getting within several feet/meters of the solar panel with a portable shortwave receiver equipped with SSB revealed drifting "birdies" from the controller's normal operation and holding the antenna against either the panel or the output cable made this orders of magnitude worse.

After the modification these "birdies" were inaudible on the cables:  It took holding the portable receiver's antenna within a few inches/cm of the charge controller to hear its operation.  By the addition of these nine components (two bifilar inductors, six capacitors and the terminal strip) the RF energy is confined to the (small!) physical space of the controller itself and is no longer being introduced differentially to the panel and output cable, causing it to be unable to radiate effectively at HF, making it very quiet and "Radio Friendly".

While the supplied charge controller for the Renogy panel was a simple PWM type rather than an MPPT (Maximum Power Point Tracking) and is thus somewhat less effective at extracting all-possible energy from it, there is no reason why this sort of filtering could not be applied to either types.
 
This shows how a typical portable solar panel+charge controller can be made to be RF-quiet and "POTA" or "SOTA" compatible.  This (reversible!) modification has rendered this panel completely quiet across the HF spectrum and inaudible on VHF/UHF frequencies as well at distances of more than a few feet (a meter or so) as well.

* * * * *
Related articles:

This page stolen from ka7oei.blogspot.com
 
[End]