Noisy AC Mains power line - filtering solutions?

[tech_builder]

Hi,

I am having some difficulty in my lab with noise on the AC mains. It is showing up in my lab power supplies (both linear). A lot of it is coming from a PLC (power line communications) unit, but even with that removed the noise is quite bad. Does anybody have any filtering recommendations for filtering PLC noise?

I did some online searching and found a company called Onfilter (https://www.onfilter.com/power-line-communication) who claim to have a solution to the very problem of PLC; however their solution costs quite a bit ($500US plus shipping for 10A version). Does anybody have experience with their products?

I see that there are many input filter solutions available on Digikey etc, but they all seem to filter at 100kHz and up, and I'm not sure if this is going to work for my situation as they seem aimed at EMI. I see there are 1, 2, and 3 stage filters, and they range in cost from about $25CAD and up.
Single stage filter ~$25CAD
https://www.digikey.ca/en/products/detail/te-connectivity-corcom-filters/6-6609959-6/759090
Three stage filter ~$100CAD
https://www.digikey.ca/en/products/detail/10VN1/CCM1649-ND/361444?itemSeq=349223681

There is also line conditioners by Tripp Lite and other similar brands. I tried a simple line conditioner by Monster (Power Center HT800) that has a filter on the input. It showed some attenuation on the signal, but not too much - making me leery to think some of these other solutions will be much more successful. I'm guessing the Monster device has a single stage filter similar to the one for $25 above, but I'd like to know what is in the TrippLite devices. Reading their data sheets doesn't give much information about this kind of stuff (says 75dB of attenuation, but no chart or other info). I guess I'm just not sure if it will be attenuating the right frequencies.

Attached is a snapshot of the noise coming out of one of my linear power supplies (Rigol DP832). I put a 1k resistor across the output at 5V to make the measurement.

Any help would be appreciated.

Thanks!

[Kean]

Do you have any LED lighting?  If so, the 50kHz noise could be from that.  It may not even be conducted through the PSU.

If you are sure it is conducted noise coming through the power line, then a power line filter should help to some degree.  A lot of devices, especially computers and test equipment, will already have similar filters.  The DP832 doesn't seem to though, probably for cost savings as it passed EMC without one.

Attenuation on the 10VN1 isn't great down at 50kHz.  I'm pretty sure you can find other models with 40dB or more loss at 50kHz.  Take a look at the Corcom 6FC10, it is only rated for 6A but has much higher attenuation at lower frequencies for both common and differential mode noise.  That model seems to be currently out of stock though, so maybe you'll need to look at a few more models.  Maybe 3EQ1 (3A only).

Also note that these filters can have leakage currents of several mA.  You can have problems with too many on a single circuit.

[AndersJ]

If your noise is ”everywhere”, you might be probing or measuring it incorrectly.
Or your oscilloscope has a problem.
Is the noise still there if you disconnect your power supply from the mains?

[srb1954]

I see that there are many input filter solutions available on Digikey etc, but they all seem to filter at 100kHz and up, and I'm not sure if this is going to work for my situation as they seem aimed at EMI. I see there are 1, 2, and 3 stage filters, and they range in cost from about $25CAD and up.

I would try going with an conventional mains filter initially and not worry too much about attenuating all the way down down to 10kHz. Normal mains filters are are cheap enough to experiment with.

Looking at your scope trace the noise spikes appear to be very narrow which implies a lot of high frequency content, probably hundreds of kHz and upwards. This could possibly be cleaned up by a normal mains filter leaving just some residual lower frequency noise, which could be handled by the PSU control loop.

A typical linear power supply has quite good low frequency noise rejection but this falls away for frequencies 100kHz and above as the loop gain of the regulator drops off. The effective noise rejection of typical linear PSU can be as little as a few dB for MHz noise, which is why you need the external filter on the oncoming mains to kill the higher frequency stuff.

[bob91343]

Try contacting the power utility company.  Often they have good solutions.

[nctnico]

A typical linear power supply has quite good low frequency noise rejection but this falls away for frequencies 100kHz and above as the loop gain of the regulator drops off. The effective noise rejection of typical linear PSU can be as little as a few dB for MHz noise, which is why you need the external filter on the oncoming mains to kill the higher frequency stuff.

This makes no sense at all since the mains transformer's core does not pass such high frequencies at all. It is more likely the OP picks up noise through ground loop and / or improper probing techniques. A first step is to power all equipment on the bench from the same grounded power strip (preferably from a grounded socket).

[tech_builder]

Hi All,

Thank you for your replies...and nctnico is the winner. It was sloppy probing. I connected to the power supply output with a BNC to banana jack adapter and the noise disappeared (left hand side of attached pic). I also tried twisting the power supply cables and that reduced the noise by maybe 20%. My concern is that this noise is going into my DUT (which is a linear load project) and possibly causing issues there. Would a choke of some kind at the DUT be effective in keeping this noise out?