Reducing/Eliminating EMI from Power Inverter

Bassplayer1985

Subscriber
Hey guys,

Figure i'd post here for this topic. So in another thread I just purchased a 1500w sine wave power inverter with a small 210Ah battery bank. The inverter was cheap but works exceptionally well for the price point. Great surge capability, and a furman line conditioner makes my pro audio equipment run great with no interference. For the use with a HAM shack forget it. I dont have a HAM setup, but I'm looking for ways to clean up the AC lines so it doesnt act like a huge antenna radiating the EMI from the inverter itself to my equipment.

As with all sine inverters (or any DC to AC inverter in general) is the EMI/RFI that radiates from it. I know there isn't too much one can do to negate emissions from the case itself, but I know with the use of ferrite beads and collars one can reduce the EMI/RFI traveling on the AC lines.

For the record, this inverter does not have any FCC 15 certification, its a china made unit. I'm not expecting miracles, but how does one go about choosing the correct ferrite bead/collar on the AC and DC lines to reduce EMI?

I know not all ferrite beads are made the same. There are ohms, millihenries, center frequencies, db attenuation etc. The AM radio band is the worst, from 550khz all the way to 1700khz. The interference is coming at 50khz intervals. Oddly enough the FM band is not affected much at all. this may explain why my pro audio gear does not have any problems with a proper power line conditioner.

Thanks for the input.
 

I like oldstuff

Subscriber
Just an attempt at a band aid as your abilities are waaaay beyond mine. Sometimes it astounds me as to the difficulty I have deciphering a simple non electronic logic based schematic on the old Kohler gen I have. It's been years since I've been in the power equipment biz and having the need to mentor techs.

Since that era I've been a Sales Engineer of large induction heating equipment from 5kW to 7 mW from 200hZ to 27 mHz.* emphasis on the sales end of it. Dealing with harmonics and other gremlins in electronics and control circuitry, and depending on 'dirt' in the sine, putting MOV's across various components has often helped the spikey issues caused by scr's, Mosfets and IGBT's doing their thing. And yes, we've had those mov's go bang in a big way also!

Having said that, there is often no substitute for the mass of an iron core. Toroidals do work well if we get the turns ratio and core mass permeability right.
 

Bassplayer1985

Subscriber
I got a call/email into customer service at Corcom. Looking at their EMI filters. Looking at the specs it may help with the radiated emissions that are traveling through the AC lines. They even have DC EMI filters for the battery lines.

http://www.cor.com/Series/PowerLine/DK/

"Having said that, there is often no substitute for the mass of an iron core. Toroidals do work well if we get the turns ratio and core mass permeability right."

Thats what I seem to be running into. I found this site for HAM shack users, they have every conceivable ferrite bead and material one could need. I am new to figuring out henries, inductance, and resistance to find the right ferrite material. I know what frequency range i want to clean up so thats a good start.

http://www.amidoncorp.com/

http://www.karinya.net/g3txq/chokes/

Can anyone point me in the right direction on how to calculate what exact ferrite material I need?
 

Bassplayer1985

Subscriber
Okay as im figuring this out on my own here are my results so far.

After speaking at length on the phone with a technician from Corcom he said that my application and subsequent EMI hash problem is common. He recommended the FC series power entry filters as they are specifically made to work in high noise environments such as frequency converters (DC to AC, AC to AC, DC to DC, AC to DC) Much of the EMI happens between 0.5mhz to 30mhz and these filters compensate in that range. Common mode noise attenuation can be 60-75db loss! Differential mode noise is equally as good.

http://www.cor.com/Series/PowerLine/FC/

My plan is to get a 16FC10 for my inverter, meaning 16 amps rated but they do make them up to 50 amps I think. Its about $110 for the 16FC10 so not exactly cheap.

First cheap method im trying is some type 31 ferrite toroids from Fair-Rite. This mix targets 1mhz to 300mhz according to spec. The trick is to wrap each lead wire (ground, line and neutral) on its own torroid as many times possible for max impedance at targeted frequency range. My first problem is I dont know exact frequency intervals the inverter operates at, best guess is intervals of 50khz with subsequent harmonics above that.

FYI type 75 is for lower frequencies down to 200khz or 0.2mhz. Type 31 has better permeability for the AM band and some of the upper freq band up to 300mhz though there is a noticeable drop off in attentuation towards 300mhz. The better mix would be to use type 43 ferrite to have better attentuation performance from 300mhz and below.


Thats all I got for now, waiting on some fat type 31 ferrite toroids to come in the mail. If results look promising ill order more for DC lines and eventually the corxom filter. Again all my efforts focus on reducing EMI transmitting on the AC lines to equipment. Radiated emissions cannot be helped unless I enclosed the entire inverter unit in a type 31 ferrite case which makes it heavy and bulky.
 

Bassplayer1985

Subscriber
Success! At a fraction of the cost too. I got the corcom filter at a hilarious steal at about 1/3 of the retail price.

The Corcom 16FC10 has GREATLY reduced the conducted RFI/EMI on the AC lines. In addition a combination of grounding the case to an 8ft ground rod, and also bonding the negative DC terminal to the case/ground rod. This basically gives a solid ground reference of zero voltage potential of the case/battery.

***GROUNDING THE BATTERY ON A MODIFIED SINE INVERTER WILL PROBABLY DESTROY IT! GROUNDING THE BATTERY ONLY WORKS ON PURE SINE INVERTERS THAT DO NOT HAVE A LIVE NEUTRAL CONFIGURATION TO MY KNOWLEDGE!!

Here is my little EMI/RFI filter setup, much like the ones I've found on the internet. This procedure can also help quiet inverter generators too. I do intend adding type 31 ferrite collar beads to the DC lines for extra filtering.

In conclusion, a solid ground reference & properly deployed filtering is key to reduce/eliminating/controlling EMI/RFI, both radiated & conducted. Conducted emissions you will see a big difference. Radiated isn't going to be too much different but it certainly will help!

Heres a video too. When I eventually get my yamaha inverter generator I'll try the same setup and see how it performs. For now I'm pretty satisfied with the results!

http://youtu.be/0f03izSQXSw

 

Attachments

len k

Subscriber
If you have some hanging arround, just for kicks you might try a flat braided wire to the ground rod. Less AC resistance, skin depth issue. Also locate inverter/filter as close as possiable to ground rod.

Didn't see your post here, I usually not on the solar side of the Stak.
 

Bassplayer1985

Subscriber
I dont have any flat braid wire lying around but its worth a try. The ground rod is literally within 5 ft of my setup. Yeah for some reason very few people visit this solar/wind sub forum yet I thought this subject i was working on was better suited under this sub forum.

Oh well its linked to the general discussion.
 

Power

Active member
Hi Bassplayer.
Some suggestions:
Ungrounded, the inverter case acts as an antenna, radiating hash.
All wiring from the case to the panel ground bus or ground rod acts as an antenna. Using the ground terminal on a grounding outlet is poor practice, especially if Romex wire was used in the outlet circuit.
Due to the skin effect at those frequencies, it is best to use braid, and run it direct and as short as possible to either the ground rod or ground bus in the panel. We run direct to a nearby ground rod with Amateur radio transmitters.

The Kill A watt PCB acts as a small antenna. I would put it downstream of the 16FC10.
Large diameter copper braid can be opened up and slid over the DC wires, and AC up to the 16FC10. Ground the braid at one end only. That will considerably reduce hash radiated by power wiring, and may obviate the need for most of the ferrites.

I like the way you researched and investigated the issue, and found a reasonable solution.
 

Bassplayer1985

Subscriber
I will take those suggestions into consideration. Copper braid that is a 1/2 inch diameter is not exactly cheap for the DC lines. I'm pretty happy with the results I've gotton so far. I like the copper braid for the ground lead to the ground rod. Researching and confirming what you guys said the skin effect is favorable with braid vs round wire. I agree grounding through a grounded outlet is not the best method. In my defense it was a quick setup. The ground rod is literally on the other side of the wall where the main panel is. Also my water main is very close to my current setup with a connection to the ground rod too.

I got two type 31 ferrites coming in the mail I'll put on the DC lines and see how it works overall. I'm sure it will only help what I have accomplished already. At this point I don't want to sink anymore cash into this issue. Further testing has yielded no adverse function with any of the devices I plan to run with it. (19inch LCD, DVD, digital tv antenna, CFL & LED lights, and a 900 watt microwave) eventually I'll relay this setup to home backup looking into solar panels and charge controller after i get my inverter genny.

Thanks for the feedback guys!

bass
 

Bassplayer1985

Subscriber
Ferrite torroids came in today. (Fair-Rite #2631626202) and found 25ft of copper braid wire at 1/4inch wide for under 20 bucks at hardware store. I put toroids on the DC lines and retested. Results were slightly better but only marginal. Skipping DC ferrite beads as they didn't improve much.

Instead, I wrapped 12 AWG single conductor wire 14 times on three separate torroids. Results were FANTASTIC! I'll put the torroids in a plastic housing from home depot soon enough with some appropriate NEMA 15-P ends but setup now is: (battery/case grounded with copper braid wire to ground rod)

Inverter---> 3 torroids--->16FC10 Corcom--->Equipment


For those that are inquiring, further testing I have found the AM frequency intervals at which the MOSFETS driving the sine wave approximation is. Those are 680khz and 855khz, pretty common switching frequencies from what I've discovered. As a result the harmonics/overtones of those frequencies radiate/conduct away from the inverter. So my use of type 31 ferrite beads was exactly what was called for to fix this problem. The corcom device is merely extra added filtering.:)

Cheers.
 
Top