Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

General discussion area for tube amps.

Moderators: pompeiisneaks, Colossal

strelok
Posts: 288
Joined: Wed Mar 26, 2014 5:46 pm

Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by strelok »

So as promised in my other thread here is a technique seldom seen for the reduction of heater noise. While working on my little 6v6 Express this weekend I found that a fair bit of noise was coupling into the preamp from the filaments. In this particular case it was the result of a rather large amount of common mode noise coming in off the AC line. This of course gets coupled into the secondaries of the PT through the inter-winding capacitance and shows up in the filaments. I'm not sure how egregious the AC power is where I'm at, compared to other places, this is the first time I've ever really paid much attention to it. But you can see it plainly on a scope from Hot or Neutral to ground without the need for any additional filtering. Usually its between 30-50kHz amplitude modulated by the 60hz fundamental. Probably ringing from some bridge rectifier pulling large current pulses through the inductance of the power system somewhere.

Anyways, there's several ways to deal with it that I'm sure most everyone here is familiar with, elevating the heaters, DC supply, and even a regulated supply. I was feeling lazy that afternoon and didn't feel like trying to find a space for any of that. First thing I tried was simply putting some 1uF caps across the 100R heater balance resistors already in the circuit. That's worked fairly well for me in the past but in this case it was insufficient. It was one of those, put one cap in it makes one side better and the other worse, put both in and its back to how it was before. I just happened to have a spare 1mH common mode choke left over from a power supply project I completed recently, heater current is 1.8A, choke was rated for 2A so why not give it a try. It definitely helped but more capacitance was needed to really get the cutoff frequency low enough to fully suppress the noise. I ended up settling on 220uF per leg of the transformer to ground. This completely cleaned up the heater waveform on the scope and knocked out and absolutely massive amount of noise. The amp is now in that scary quiet category where it has a bit of hiss makes you think 'oh thats probably a bit loud' and then you turn up your guitar volume, hit a chord and the whole room vibrates.

There are some practical limitations to this approach however. 220uF has an impedance at 60Hz of ~12 ohms. That results in about half an amp of ripple current in the caps and will shorten their life if they aren't rated to handle it. Really a choke with 10 or even 100x more inductance is needed, so we can use correspondingly smaller caps to get the same level of attenuation. Compared to alternative solutions this can get expensive (relatively speaking) depending on the amount of current needed. 100mH is pretty much out of the question unless you want to roll your own (not hard but kinda a pain especially on a toroid). 10mH rated around a few amps is pretty common though. I'm probably going to order one along with some 22uF caps to give a try.

You might think why not just put it on the primary side where the currents are lower and its potentially an option. In my case the AC mains current at full tilt for this amp is about 800mA so less than half. The big problem is that any caps from AC mains to ground need to be Y2 rated safety capacitors and these are typically quite bulky and only come in fairly low values.

The best approach would probably be to just do the preamp. Maybe even skip the PI depending on the number of tubes involved, and since its the secondary side, we can use basically any appropriately rated cap we want. As to its effectiveness I can't say how it compares to something like heater elevation as I haven't tried that in this specific circuit. The upshot is we get a nice well defined source impedance and can consistently and easily guarantee a certain amount of attenuation of the incoming noise. FWIW this is now the quietest guitar amp I own by far.

Anyway, I don't know how useful that will be to anyone here, but maybe it was at least interesting. :lol:
User avatar
bepone
Posts: 1642
Joined: Mon Mar 09, 2009 4:22 pm
Location: Croatia
Contact:

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by bepone »

the problem in the power transforner should be overtaken by making electrostatic shield between secondary and heater winding, which need to be located at the top (last winding), and gnd this winding
User avatar
bepone
Posts: 1642
Joined: Mon Mar 09, 2009 4:22 pm
Location: Croatia
Contact:

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by bepone »

otherwise, making only V1 DC supply for heaters is simple.and cheap(7812) and fixing all of the problems.. V1 matters only, sometimes V2
User avatar
TUBEDUDE
Posts: 1691
Joined: Thu Jan 04, 2007 5:23 pm
Location: Mastersville

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by TUBEDUDE »

Are you using a toroidal power transformer? I'd be surprised to see 50kHz on the secondary of an EI transformer
Tube junkie that aspires to become a tri-state bidirectional buss driver.
R.G.
Posts: 1254
Joined: Tue Dec 02, 2014 9:01 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by R.G. »

First off, good investigative work! AC power line noise is commoner than people think; some buildings have "sick" AC supplies.

Your post started me on some random thoughts.
- You see the noise on the hot and neutral to ground. Do you get the same result on the AC socket itself? If yes, this indicates that the building may have either ground impedance problems, or high power switching power supplies injecting noise. This itself could be a clue in tracking down the source of the noise, as in does it vary by time of day, etc. Could be a PC using a lot of power. I have a NAS backup system with a total of sixteen hard drives running, so there's a fair amount of AC power chopped in the power supply. Used to be that appliances didn't contribute much noise, but today's washers, driers, refrigerators, cooktops, ovens, and so on have moved to switching power supplies as well.
if no, it would indicate that the amp is the source of its own noise. Good on you for figuring out that it could be the house wiring ringing!
- I'm guessing that you didn't get to this point yet, but have you tried a packaged EMI power line filter before the amp? I keep a salvaged "brick" style filter in the equipment bone pile for this kind of test. A packaged EMI filter might let you get further into figuring out whether the noise is generated outside the amp or by ringing and/or oscillation from inside the amp.
- You mentioned "Probably ringing from some bridge rectifier pulling large current pulses through the inductance of the power system somewhere"; if the amp has solid state rectifiers, these cut off so sharply that they can excite RF in the wires to/from the rectifiers. Using fast-soft recovery rectifiers like FREDs can cure this. However, rectifier pulse noise tends to be squarks of RF noise, not 60Hz modulated constant ringing.
- As mentioned in the thread already, a shield inside the PT can help with AC power line noise. Does your PT have a shield? The mechanism of this shield is not just electrostatic shielding. It's also an L-C-L filter, as the Ls are the inherent leakage inductances of the windings to the shield. This helps too. It's not as effective as the concentrated inductances and caps of a packaged filter, but every bit helps. Note that if the amp doesn't have three-wire grounding, neither the shield inside the transformer or an external EMI filter will be as effective. Also applies if the building safety ground is high impedance.
- Were the 220uF caps on the heater half-winding polarized or NP? I'm guessing from your posts NP, but this was just one of the things that caught my eye.
- I'm a bit surprised at the need for such large caps when what you're trying to filter out is 30kHz. I would suspect that the ESR and ESL of the caps is poisoning the rejection of the high frequencies; 220uF might be well into its inductive region at over 30kHz. Smaller film or ceramic caps might do as well.
- You mention maybe just doing the preamp; it might be interesting to just do the first preamp tube(s).
User avatar
nworbetan
Posts: 154
Joined: Sat Jul 15, 2017 11:34 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by nworbetan »

strelok wrote: Mon May 20, 2024 4:10 am Usually its between 30-50kHz amplitude modulated by the 60hz fundamental.
This probably isn't going to change your noise reduction strategy, but I think the noise you're dealing with is probably not amplitude modulated, or if there is some amplitude modulation happening then the description would be more complicated than the tiny quote I cherry picked. This quote from R.G. is the beginning of a more complicated description of what you probably meant.
R.G. wrote: Mon May 20, 2024 2:30 pm - You mentioned "Probably ringing from some bridge rectifier pulling large current pulses through the inductance of the power system somewhere"; if the amp has solid state rectifiers, these cut off so sharply that they can excite RF in the wires to/from the rectifiers. Using fast-soft recovery rectifiers like FREDs can cure this. However, rectifier pulse noise tends to be squarks of RF noise, not 60Hz modulated constant ringing.
But to be clear, 1) something similar to amplitude modulation caused by the high current pulses through the rectifiers, and 2) the high freq ringing excited by the pulses, could be thought of as two related issues happening at the same time in the same place. In some transformers with some loads 1) may lead to 2), and it sounds like your 2) might be especially egregious for some reason.

I grabbed a couple images from the web for examples of what you described, but probably isn't what's actually happening:
carrier_signal.jpg
am_modulated_wave.jpg
With a 30-50kHz signal being modulated by a 60Hz signal, there would be 500-833 cycles of the carrier signal for every cycle of the 60Hz signal.
You do not have the required permissions to view the files attached to this post.
R.G.
Posts: 1254
Joined: Tue Dec 02, 2014 9:01 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by R.G. »

The only thing I could come up with for 60Hz modulated 30-50kHz is the pulse modulation generated by a power factor correction circuit trying to modify the power taken from the AC mains on an instant-by-instant basis for a high power switching power supply. Maaaayyybe...
I considered but didn't mention power-line carrier communications stuff. I'm not current on how this last works these days. In X-10 days, it was high frequency bursts near the zero crossings, but the technology may have moved on.
User avatar
nworbetan
Posts: 154
Joined: Sat Jul 15, 2017 11:34 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by nworbetan »

It's really easy to either over or under-explain things when you're talking to complete strangers.

To be extra clear: I was and am still agreeing with your assessment that the rectifiers are probably triggering a HF ringing. I think if we could see a picture of the scope trace it wouldn't look like an "amplitude modulated" signal, but a much simpler signal that's mostly just the sum of the 60Hz and the HF being mixed together.

The squashed peaks of the 60Hz line voltage when the transformer voltage drops because it can't handle the high current pulses as easily as the no-load current... yes that would resemble amplitude modulation a little bit, but whether it's truly amplitude modulation or just something that looks like amplitude modulation is probably debatable and I'm not super interested in splitting that hair.
strelok
Posts: 288
Joined: Wed Mar 26, 2014 5:46 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by strelok »

Thanks for all the replies, I'll try to address everyone's points in order:

-Electrostatic shield would definitely be effective in combating this issue. Unfortunately I don't know of anyone making off the shelf PT's with them included. I'm at the point now where I could fairly easily design my own but that's a lot of turns to do on a cheap amazon winder by hand. Also EI lamination's are only sold in bulk AFAIK so unless I wanted to do several it wouldn't really make sense from a cost perspective either.

-DC supply is certainly a good option. I'm just lazy and this was the easiest to implement with the space I have to work with and the components on hand. Kind of a 'just make it shut up now' so I can get on with dealing with studying stability like I had planned without all the noise in the way. At this point I was already on like number 3 or 4 of side tracks of a side track down the rabbit hole. :lol:

-Its an EI transformer. I mean they have large inter-winding capacitance's so its not all that surprising to me given the amount of noise present. I've been measuring the AC at various places today, work, home, etc and work seems to be particularly bad where I'm doing most of this stuff.

-Thanks, I've started doing SMPS design lately so its something that's present on my mind at the moment. I actually took my first supply in today for conducted emissions testing. Went pretty well, its about 10-12dB over in a couple spots so its going to need an improved filter as well. Nice that I get to put the parts for both projects on one order.

-The noise is just present on the power line without the amp hooked up. Its pretty much always there and I'm not really sure whats causing it. There's not really anything in the building when I'm working on this stuff that's drawing any appreciable power. So unless its just a horrifically bad switching supply (Typically CM noise in SMPS is mostly in the MHz as the strays are much smaller) I suspect its either a problem with the wiring in the building or its coming from outside. The building is old and the wiring is shitty and given that this is a residential neighborhood the wiring is my #1 suspect. I once got a light tingle from a neutral line while changing a light fixture there that scared the crap outta me, had 10-20V on it. Good point if your ever to working on AC lines to check the neutral to ground and not just the hot to make sure there's no voltage present in the event that stupid was done in the past (as it often has). Tomorrow I'll go around unplugging things and probe some different circuits with the scope to see what I can find. Given that its an audio company, probably shouldn't have noisy AC lines....but hey it makes for good proof of noise immunity. :lol:

-That's sorta what got me to look at the incoming AC was the fact that it was long bits of ringing rather than just a fast spike. The frequency was also too low. It actually had me chasing my tail for a bit thinking it was an oscillation and only when I started disconnecting stuff to isolate it did it become apparent that wasn't the case. I felt rather foolish when I put my probe on the filament and just saw it sitting there with the amp half torn apart. :lol:

-I actually did try one of those power line filters, unfortunately the frequencies in question are just too low for it to have much of an effect. Most of them are intended for 100kHz and up where the switching noise is. Sort of what I was alluding to with the title was that probably the reason you don't see people do this much is that for audio-ish frequencies the values of the components just get too big to have much of an impact. But then again I'm dealing with a pretty bad case.

-The 220uF's where electrolytic. Not something I'd leave in there for any length of time but worked well enough for a test. I'm trying to find some NP in 22uF but there aren't too many options that I like in a small enough size. I think I might up the CMC to 20mH instead of 10mH and that gives a bit more options for 10uF film caps.

-I was too honestly. I started at 1uF and went to 33uF then 220uF. Going up just made things better. Guess that's an indication of just how bad the noise is in this case. Also you might be surprised at how low impedance modern electrolytic's are. The 220uF are used for reducing switching ripple at the output of a power supply and they're good out to a few hundred kHz before the ESL starts to bring the impedance up significantly. If I remember right the ESR is below 0.05 ohms. Funnily enough I removed the choke today and it seems the caps where really doing most of the work. Will be interesting to see how lower value caps and the larger value choke fair.

-Yeah I think you could probably get away with just doing the first preamp tube or two. In that case it would make selection of the choke a lot easier.

-Sorry I meant to say it *looks* like an amplitude modulated waveform. Hence the rectifier theory. Eg a full wave charging a capacitor input supply turn on near the crest of the AC waveform and draws a large pulse of current. This sets off some sort of ringing upstream and its poorly damped so it takes a while to die off. Then shortly after the next crest comes by. Its just a theory as it was about the only thing I could think of that would make it look like that. Removed from the 60hz (by looking at the noise getting into the preamp) it does look somewhat like AM modulation but the envelope is a lot more complicated. I'll see if I can remember to take a pic of it. It can be hard to describe this sorta thing in words, but I think we're essentially on the same page. Its definitely a bizarre case and I'm curious to dig into it further to find the culprit.
User avatar
nworbetan
Posts: 154
Joined: Sat Jul 15, 2017 11:34 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by nworbetan »

strelok wrote: Tue May 21, 2024 5:22 am The noise is just present on the power line without the amp hooked up. Its pretty much always there and I'm not really sure whats causing it.
Can you post a pic of how noisy the wall voltage is without the amp plugged in? You might've missed the bullseye when you described it, but it could still be the origin of the problem.
User avatar
didit
Posts: 1002
Joined: Thu Sep 25, 2008 3:37 pm
Location: Canada

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by didit »

strelok wrote: Tue May 21, 2024 5:22 am Unfortunately I don't know of anyone making off the shelf PT's with them included.
Hammond does for several of their higher end MI replacement power transformers. Also, does in other product lines.

..
User avatar
bepone
Posts: 1642
Joined: Mon Mar 09, 2009 4:22 pm
Location: Croatia
Contact:

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by bepone »

strelok wrote: Tue May 21, 2024 5:22 am Thanks for all the replies, I'll try to address everyone's points in order:

-Electrostatic shield would definitely be effective in combating this issue. Unfortunately I don't know of anyone making off the shelf PT's with them included. I'm at the point now where I could fairly easily design my own but that's a lot of turns to do on a cheap amazon winder by hand. Also EI lamination's are only sold in bulk AFAIK so unless I wanted to do several it wouldn't really make sense from a cost perspective either.

-DC supply is certainly a good option. I'm just lazy and this was the easiest to implement with the space I have to work with and the components on hand. Kind of a 'just make it shut up now' so I can get on with dealing with studying stability like I had planned without all the noise in the way. At this point I was already on like number 3 or 4 of side tracks of a side track down the rabbit hole. :lol:
every better winding company which is following audio designs.. lets say Pieme Elektra Italy, also custom wound from any local winder are normal priced and doable..

your LCLCLC filter are normal way to filter anything, but they are expensive way of filtering at the end..
strelok
Posts: 288
Joined: Wed Mar 26, 2014 5:46 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by strelok »

nworbetan wrote: Tue May 21, 2024 3:19 pm
strelok wrote: Tue May 21, 2024 5:22 am The noise is just present on the power line without the amp hooked up. Its pretty much always there and I'm not really sure whats causing it.
Can you post a pic of how noisy the wall voltage is without the amp plugged in? You might've missed the bullseye when you described it, but it could still be the origin of the problem.
Oh I'm 95% certain that's the origin of the problem. Unless there's something else buried in that noise that's the real cause which I highly doubt .

Also good to know there are people making PTs with shields in them, would be worth consideration for future builds.
strelok
Posts: 288
Joined: Wed Mar 26, 2014 5:46 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by strelok »

I did take some pics of the noise today:
IMG_5927.JPG
This is the filament waveform without any filtering. I forgot to take one of the line all by itself but they're practically identical. The line being obviously just scaled up from what you see here.
IMG_5929.JPG
This is the output of the second gain stage with all controls dimed. Pretty nasty.
IMG_5931.JPG
Zoomed in view of second stage output again with all controls dimed.
IMG_5934.JPG
Another view of second stage output, dimed controls.
IMG_5936.JPG
Filament with 220uF caps to ground. No CMC, 1mH doesn't really seem to make much difference one way or the other.
IMG_5938.JPG
Output of second stage with caps in place and controls dimed. Note the vertical scale.
IMG_5940.JPG
I tried elevating the heaters to no avail as you can see in this pic. Again 2nd stage output, no caps to ground (aside from the bypass for the elevation obviously), controls dimed.
IMG_5943.JPG
Same as above but the elevation voltage was increased from ~25 to ~35 volts.

Its a shame you can't get 220uF in non polarized form in a non gigantic form factor as that would be the easiest way to deal with this problem lol. I did pull V1 and it goes away entirely, indicating that just a CM filter on the V1 filaments would likely be sufficient. I may give that a try or I might try DC filaments. Not sure yet.

I also found that the outlet feeding my bench wasn't properly grounded and given the buildings wiring that's going to be a real PITA to fix. However it isn't the real culprit, it definitely makes the problem worse, but the noise is still present on properly grounded circuits. Moving the amp to one of those definitely yields quieter results but its still far from as quiet as I'd like.
You do not have the required permissions to view the files attached to this post.
R.G.
Posts: 1254
Joined: Tue Dec 02, 2014 9:01 pm

Re: Common Mode Filtering - A slightly impractical approach to reducing heater induced noise

Post by R.G. »

Interesting. It would be revealing to see a second trace with the filament voltage, to be able to spot the relative positions of the peaks and valleys of the noise and the AC mains waveform. It seems like the noise magnitude envelopes are similar to full wave rectified sine(ish), but inverted, with the round ends nearest the zero volts line. This is kind of, maybe, consistent with the noise from a switching-type power factor corrector - I think. Just guessing, could be confirmation bias ( :D ) but it might be. If you have a single biggish transformer supplying AC mains voltage to several units, somebody on the same feed line could do this.
Post Reply