Saturday, November 14, 2020

A high-current DC (and AC) noise filter for UPS or RV use

A friend of mine (Glen, WA7X) acquired a 16 kVA UPS (for free!) a year or so ago - a commercial system consisting of four hot-swappable 4 kVA modules:  With his current load, he only uses one of the four modules, the rest being available as spares or providing room to grow.  Using this as a battery back-up system for important devices in his house (computers, etc.) it's active all of the time as it is an "online" UPS - that is, the inverter pulls power from the battery bank, but the battery bank is always being charged.

Figure 1:
Whiteboard diagram of the dual AC mains filter for the
UPS - See text for details
Click on the image for a larger version.

AC-side filtering:

When he first installed the UPS, he discovered that being a commercial device, it was only a "Class A - commercial" device under FCC part 15 - and it trashed the 20 meter amateur band and caused interference on a few others.  This, however, was easy to remedy as he'd asked me for advice and built a larger version of UPS noise filters that we'd implemented together in the past:  See the article "Containing RF Noise from a Sine Wave UPS" - link.  

Being capable of many kVA, the filtering for this UPS had to be built from scratch rather than using (expensive!) commercially-available filter modules, but this was easily done using readily-available ferrite toroids and bypass capacitors.

Figure 1 shows the general diagram, crudely sketched on a white board in his shop after our consultation.  The inductors are 12-14 turns of 6 AWG on FT240-31 cores, each half (phase) being an equal number of turns for best common-mode suppression as depicted in Figure 2.  Because the UPS outputs 240 volts, the 50+ amp capability of unbundled 6 AWG wire is sufficient for the envisioned load on this UPS.

Figure 2:
The inside of the dual mains filter, built
into a standard NEMA box.  The capacitors
- mostly obscured - are connected to the
blocks with the ground side bonded to the
Click on the image for a larger version.

The filter uses suitably-rated parallel 0.01uF and 4700pF capacitors:  Those across the AC leads (which could have been as large as 0.1uF or so) help force the RF energy to be common-mode across the bifilar choke while the capacitors to ground on the "outside" (non-UPS) side of the filter shunt the remaining RF - which is already at higher impedance due to the choke - to the common-point ground.  Shown in red on the drawing in Figure 1 are large 43 Mix slip-on beads on the "UPS" side of the filtering to better-suppress the high frequency (VHF) components:  Ideally, one would run both conductors through each bead for net zero flux on the core, but larger diameter devices were not available at the time of construction.

The filter pictured in figures 1 and 2 completely solved the RFI problem:  One has to get within a few inches of the UPS cabinet to hear magnetically-coupled RF energy with a portable shortwave radio.

DC-side filtering:

It wasn't a huge surprise, then, when he added more battery capacity external to the UPS - 120 volts DC - and the racket on 20 meters and other bands reappeared.  Because RF is RF, the filtering method for the DC leads is exactly the same as required for the AC leads:  Common-mode choking, bypass capacitance and single-point grounding techniques. 

Considering that the UPS is capable of up to 16 kVA, the DC filter needed to be able to handle more than 100 amps at the 120 volt (nominal - about 138 volts, actual) input.  Looking about, he found a pre-made set of 6 foot long, 2 AWG, very flexible "inverter cables" at Harbor Freight (cost:  $35) that were conveniently available - easily capable of handling about 100 amps - more than enough because he was not ever expecting to load the UPS to its capacity.

Because of the size of the wire, standard FT-240 (2.4 inch/61mm O.D.) cores aren't appropriate, so Glen obtained some "Monster" size toroids (Mix 31) from  These cores are about 4" (102mm) in diameter and it was possible to wind 7 bifilar turns of the 2 AWG wire onto them, yielding about 170 uH - more than enough inductance to provide adequate choking on the HF bands.

Because they were on-hand, the same capacitors were used:  0.01uF and 4700pF capacitors in parallel:   With a DC system, much larger-value capacitors (e.g. 0.1-10uF) of appropriate voltage could have also been used if lower-frequency attenuation were required.  Like the AC choke, large slip-on ferrite beads (31 mix in this case) were slipped over each of the 2 AWG wires on the "UPS" side to help suppress the higher-frequency energy.  Because of the current involved, 200 amp screw terminal strips were procured - both to terminate the connections to the wire comprising the inductances, but also provide connections to the "outside" world.

Figure 3:
The completed DC noise filter.  The bifilar-wound choke on the "monster" 31-mix core is wound with 2 AWG welding/inverter cable:  The slip-on ferrites on the "UPS" side of the DC are clearly visible.
Mostly obscured are the bypass capacitors, connected to the screw-type terminals.
Click on the image for a larger version.

There are a few caveats to making a filter like this work:

  • The "ground" lead must be as near zero length as possible.  This box was bolted directly to the box containing the AC input/output filter described above  (which, in turn, is bolted to the UPS cabinet) to establish a single point ground where the RFI on the AC in/out leads and the DC leads come together:  Connecting the two boxes with just a few inches of wire caused noticeable degradation in its performance!
  • The cables connected to the UPS must be considered to be "dirty", carrying a lot of RFI, and must be kept as short as possible.  Additionally, one must keep other wires away from these "noisy" leads to prevent interference from being re-coupled into them!
  • The external battery bank itself has its own fuse, at the battery bank:  Do not even think of connecting a high-current power source like this without some sort of short-circuit protection!

As with the AC filter, this one appears to be completely effective with no conducted noise being detected on the leads of the external battery connection.

Where might these techniques be applied?

The filters shown above are simply "scaled up" versions of those described previously on this blog (links below) to handle higher voltage and current.  A few instances where these techniques might be useful include:

  • Adding higher battery capacity to an existing UPS.  You may own a UPS that will power your gear, but simply has too little battery capacity for the desired run time - and adding external battery capacity safely (e.g. fused, insulated) is one way to do this.  As in this case, adding more capacity caused radiation of RFI which had to be suppressed.
  • Suppress noise from an existing UPS.  Many modern UPSs are likely to create RFI - and these pages show how that might be mitigated.
  • Suppress noise from an RV power system.  Many RV (recreational vehicles) have power converters (AC to DC for charging batteries) and inverters (DC to AC for running mains-voltage devices) that are likely to generate RFI.  The techniques described on these pages show how it is practical to prevent the conduction/radiation of RFI on both AC and DC leads.

Thanks to Glen, WA7X, for supplying the pictures:  I just scribbled down diagrams and notes and gave him a few capacitors - he's the one that actually built the thing!

Related links:

Links to other articles about power supply noise reduction found at

The large, ferrite toroids and beads used on this project were obtained from - link.

* * *

This page stolen from


No comments:

Post a Comment


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!