Resolving Hum Issues In Vacuum Tube Amplifiers
ฝัง
- เผยแพร่เมื่อ 28 ส.ค. 2024
- Transcribed from: "Transistor Audio and Radio Circuits" 2nd edition, Mullard
"It should be remembered that the input sensitivity of an amplifier is typically 3mV at 1KHz, on the magnetic pick-up position, and the output voltage is of the order of 20V, implying a voltage gain approaching 10^4 (40 dB) It is therefore essential to keep the output separated and screened from the input. The magnetic field from the mains transformer may cause hum, so the transformer should be remote as possible from the input.
Earthing
Currents of several amps magnitude circulate in the power supply and output stages. It is important that no wiring carrying these currents is included in the input circuit, otherwise hum or instability, due to the small but significant resistance of the wires, will result.
The earthing arrangement for stereo amplifiers is considerably more complex than for mono, since the single power supply and common earth for the two signal inputs make it more difficult to avoid earth loops. "
Thanks for posting this informative video, David. There's so much superstition regarding good grounding techniques that it's refreshing to actually see a definitive, empirical study.
Doug - good to hear from you. Thanks. I agree that the discussions never end about hum and grounding techniques. This method may not be perfect but 75 dB of S/N and I hear only the slightest hiss from my midrange horn. I see you have been posting again lately. You do very nice videos.
@@ElPasoTubeAmps Likewise, David. So much of what we hear is supposition or opinion, it's nice to actually see some data and hard, factual evidence. Good job, as usual:) Some day you should apply your investigative methods to another oft-heard statement that mismatched output tubes yield more even-order harmonics, and thus produce more pleasant tone.
@@UncleDoug That's an interesting one and should be easy to make. I have had to come to the conclusion that we like the artifacts of a tube amplifier. Why else would we listen to them - especially the SET. Its harmonic profile is that of a triangular wave and people love it and the second harmonic is the dominate harmonic. I don't buy this idea that all even harmonics are good and all odd harmonics are bad. I don't know where that came from. On the other hand, how can I argue with a musician - the creators of music. I just try to peacefully accept some of this and keep going. Thanks for the idea.
That was an interesting reminder about how important attention to detail is. It can make the difference between mediocre, good and great. Thanks!
Yep, I agree. And the issue may be due to several things so picking them out one at a time can be necessary to make progress. Much like my radio story of ignition noise and alternator whine, I reduced the noise on my 20 meter ham radio receiver by working with and basically repositioning each ethernet switch in the house and especially in the basement. I use a SA and tweaked each switch, watching the noise level go down and after I did all of them, the noise was reduced dramatically so the point being, looking for one single point of noise may not turn out the way we want it to. You said it right, mediocre, good or great, takes a methodical approach.
Appreciate your explanation of the Hum Pot Control.
Thanks, I have dealt with hum pots and filament grounding all my building life. In many cases, hum is minimized by the pot being to one side or the other (one leg or the other of the filament winding being grounded) but it also may may be anywhere in between and only with a pot can we find this perfect spot. If we can measure S/N and get it to around -75 dB, the rest is academic. We can't hear any annoying hum at that level so it really isn't necessary to peruse the -90 dB level, but we often do...
Just think about how much -90 dB is - that is a ratio of one billion to one. 10 watts of audio, 10 nano watts of noise and hum. At -75 db that is a ratio of 31.6 million to one. That would be 316 micro watts of noise and hum for a 10 watt signal. Sorry... just had to go there. It would be interesting to do an experiment where we determine the threshold of noise and hum that we can actually hear, at an absolute power level and would become annoying taking speaker efficiency into account so it can be applied to all speakers.
Brilliant experiment. Even though you _added_ capacitance, a lot of it, presumably to the output side of the PI filter, you also managed to add a lot of hum at various points. Can you please leave your brain to science? Best experimenter on TH-cam!
And that was it in the Gibson amp here too we have been discussing privately. It was humming badly, 100Hz so downstream of the rectifier. I had changed the power transformer (PTX) and misrembered where to reconnect everything. What I ended up with went like this:
1. Star earth with mains ground, PTX centre tap (CT), and another ground runner going to pin 1 of one of the 6L6s.
2. From here ran various other grounds, e.g. to the output stage and phase splitter cathode grounds, and some shielding.
3. From here also ran another runner to pin 1 on the other 6L6.
4. From there went the PSU capacitor grounds and input cathode grounds. This was original!
5. The input sockets are separately grounded at the chassis by their nature, and this ground is carried to the volume control grounds.
6. The speaker jack is also grounded at the chassis from an otherwise floating OTX.
So you see the loop. All the stuff coming from (2) above is inside the PSU loop. So it hummed. All I had to do was move the CT to (4) and the 100Hz hum vanished. Can't even see it on the spectrum analyzer now,let alone hear it. The PSU currents must go around their own loop. You just better not have any other grounds connected *within* that loop, or they are going to acquire the 100Hz sawtooth and send it up through their cathodes or whatever they are connected to.
I know.this grounding is still not ideal. I should keep the CT connected to the PSU cap ground, but disconnect that from pin 1, and run this CT/cap ground junction back to the star separately from everything else.
EJP
Hi David,
Excellent video and thought for any power supply or audio designs. 120 Hz hum can sometimes be tough to tame completely.
73,
Glenn WA4AOS
Hi Glenn - just scrolling thru my videos and saw you comment. I don't know if I have your direct email anymore but thought I would just say,
Hi, and hope all is well with you. I am still around and doing OK. 73 WA4QGA My direct email is good on QRZ.
I'm DEFINITELY learning from your videos 👍.
Food for thought. Thanks for the look.
Excellent advice on a difficult issue! I have a HP8903 as well and love it. Cheers!
Interesting observation. I felt like random grounding for audio amplifiers would not be a good idea, but didn't have a measurable way of proving it.
I would like to see the effect of the EM field on transformers and inductors and what kind of separation should the components have. How close is too close on a chassis. And while you're at it, methods of mitigating its effects like using a transformer with bell caps or installing a metallic shield plate between the transformer, and say, a tube.😉
Thanks
I made some measurements a few years ago concerning the orientation and distance the power transformer is from the output transformer. It is measurable but not extreme. I don't know if I made a video of it or mentioned it in a video but in the case of an open frame power transformer placed near the output transformer with the fields parallel, a small amount of 60 Hz hum will be induced in the output transformer. Makes sense... Potted transformers and the such, I don't think I tested them or how metal covers and end-bells would affect the coupling but it seems an inch or so between transformers is sufficient to make the coupling nominal - and keep the magnetic orientation of the cores perpendicular if possible. We do the best we can in our homebrew equipment. 🙂
Well thanks for that analytic test procedure! I have to overworke a preamp b+ supply I think!
1x or 2x is not a problem. Usually problem is 3x, what you have there as 180 Hz.
Thanks 😁
I've heard that the first filter capacitor, rectifier, and transformer should be as close to each other as possible, and then the other nodes should have the filter capacitor close to the stages that are supplied from it.
also have seen 2 ground buses in amps, one for the power supply, and other one for the signal ground (which grounds at the input)
have you tried designing an amp like this? if so what were the results?
Thank you!
Thanks for the insights. I'm curious to know what you think of the modulation distortion on the 1kHz peak in the FFT, where the mains hum gets into the output tubes and modulates as 120Hz sidebands around the fundamental and harmonics of the 1kHz signal. I believe it's largely down to mains hum on the screen grids. What effect do you think it has on the sound? Thanks!
How does it make it sound? That is the trillion dollar question, for sure. I don't know how 60 Hz mains hum and 120 Hz full wave power supply hum will affect sound and I think it is so complex it can not be solved analytically. As an example, studying power supply design with a choke input filter or a Pi filter, the bottom line is, it has to be built and examined empirically as just the loading of the choke and the inductance swing is so complex, it defies a definitive analytical solution. Certainly the sum and differences of mixing frequencies in voice and music is way beyond a mathematical solution. With that said, how can we possibly justify the appealing sound of a single ended triode amplifier where the harmonic profile resembles a sawtooth waveform instead of a sine wave? And amplifiers with THD at the 0.01% level sound sterile? I think it like the taste of beer. We like what we like for reasons again, as complex as the sum and difference of mixing signals. Hope I am making sense. I never thought about hum on the screen grids but the screen grid is very sensitive so you might have something there.
On the heater lifting the heater about 45 volt above ground per the RCA red book (1941) has help me a great deal on signal tubes and guitar amps . Refference the signal ground 10 ohm a way from the safety ground also helps on the 10 ohm resistor 5 watt min. Nice testing . 83db to 86db is the real world best snr for tube the 90 is for me a pipe dream . Mid 70's if very good period. Try a diode on the center tap of the high voltage secondary and look at the imd effect. Look at 6cj3 diode s cheep and 250 ma all day long .
I remember that Dynaco "floats" their input RCA jacks across a 10 ohm 1/2 watt resistor. I am sure it helps and they can prove it but I have also seen parasitics occur in these amplifiers and it burn that 1/2 watt resistor open and then the input ground is floating for sure... your 5 watt minimum is a good idea from my experience. I have seen similar things on other amplifiers not just Dynaco. Yes, I agree that 80 dB S/N is good and actual getting consistent 90 dB readings depends on everything from fluorescent lights to the quality of the BNC patch cables. I can only read 90 dB S/N (usually...) when I use the high quality HP 3 foot (1 meter ?) patch cables and not the cheap ones. I have heard of that diode in the CT before but I have never checked it out. I will have to try that some time and see if I can measure differences in IMD. Thanks for your comments.
@@ElPasoTubeAmps The 5 watt rating is so the fuse blows as it should not the 10 ohm resistor . As Ronny said trust but verify. Too often the real cause of a problem is dismissed before it is tested to see if it is it because it take time to test all of it .
can you build a microphone tube amplifier ?
Yes.
Awesome Video! I have grounded my amplifier as Star with everything running back seperately. Its a busy terminal! and I do still have a noticeable amount of 100hz. Where are you adding that extra capacitance into the B+ ? Directly on the output tube Plate connection? or somewhere else?
And as for your input ground, is it connected directly to the chassis and at the same point as all your amplifying stage cathodes? meaning only two ground points on the amp- connected by the chassis only?
Thanks for another interesting video!
Hoping I can improve my S/N after watching this!
I was connecting the positive lead of the outboard capacitor at the power supply Pi network. My goal was to just get some 120 Hz power supply current, as small as it may be, flowing back thru the chassis to make the point of the video. We also have to consider that this Pi network already has a large capacitor grounded as I think is the correct way so the point being, if this was the main electrolytic capacitor of the power supply being grounded at or around the signal input, I would bet the S/N would get bad really quick.
The RCA input connectors are grounded right to the chassis. I generally just connect components off the tube sockets to either the tabs on the tube socket or mount a terminal strip or ground terminal under the screws that hold the tube socket to the chassis.
I don't know this to be true but even in a star or buss ground, I would make a separate star or buss ground point for the power supply components and not mix them with the input signal ground points. Put the power supply ground point around the power transformer and/or AC input away from the signal ground point.
@@ElPasoTubeAmps Great! thanks for the info. I will look at doing some seperate ground points for each channel perhaps instead and see how that goes!
Hello. What should I need to do this same FFt analysis? Software sound card, variable resistor.. Anything else? Thanks.
An internal sound card will probably work but the USB sound card is likely to be better. You can download the FFT program for free. It has a built in oscillator you can use as the input to the amplifier. You need a load resistor for each channel, assuming stereo, usually of 8 ohms with a power rating equal to or greater than the output power level of the amplifier. Some hookup cables and you should have all you need - that I can think of right now. Good luck and have fun.
@@ElPasoTubeAmps thank you very much!
Hi David: what are you using as a spectrum analyzer on your computer?
The one called TrueRTA trueaudio.com/rta_abt1.htm is sufficient for a FFT display. The oscilloscope on it is no good.
There is another one called SpectraPlus but it is very expensive.
I have PT either hum/buzz when the B+ kicks in 30 seconds later with HV delay. It's quiet when turn on until B+ goes to output tubes then the hum/buzz sounds at PT. Do you have any suggestion where to look at? It's not that bad but can be fixed then great. Thanks
I am just guessing but it might be a physical thing with the transformer. Maybe tighten up the bolts on the transformer? Squeeze it or press it with your hand to see of the tone/volume changes? Hopefully it will be something simple.
@@ElPasoTubeAmps Thank you so much, it's strange all the screws loose on both OPT and PT, tighten them up, buzz smaller than before, won't hear much unless put my ears next to the amp.
Very interesting, I'm dealing with 100 hz hum problems on a mosfet amplifier , I've managed to get it down to 1 mv into 8 ohms (measured on an 8903), equates to 0.12 microwatt ! But it is still audible if i stick my head right next to the speaker, just shows how sensitive my speakers are
Doing the math, 0.12uW compared to 10 watts (I don't know what your amplifier is rated at) is a S/N of 79 dB. That isn't bad. This -90 dB S/N goal can be hard to obtain and sometimes just as hard to measure as the jumper cables can make a difference along with some unknown bit of noise creeping in on the measurement. I am as much of a -90 dB S/N and 20 to 20KHz frequency response geek as anyone and I try to obtain that level but it doesn't always work out perfectly for me. If anything the SET amplifier projects taught me is, it seems we love some bit of THD in our music created by our amplifiers. Why else would the vacuum tube industry be doing as well as it is in this age of solid state devices that perform "perfectly"?
@@ElPasoTubeAmps odd and even order harmonics.
@@ElPasoTubeAmps thanks for the reply, the mosfet amp is home made, it's rated at 100 w per channel, but I'm only getting 94 at clipping (not important) I've been working on it again today in 80 degree heat (I know that's cold where you live) anyway I've improved it a fair bit mainly by re routing cables as far away from the toroidal transformer as possible, so the new figures are, 0.6 mv hum into 8 ohms gives 0.045 uw, it's still audible but only if you stick your ear right up to the speaker. Did a s/n test with the 8903 which comes out at 89db not quite the magic 90! This was done at 94 w output. It still annoys me I can't get it completely silent! Best wishes from Nick
I hope to get a new scope soon with built in spectrum analyzer so will be interesting to see the various tones if it will read that low
@@ptronix I think you've won the war. If you do the math of 94W and 0.045uW, that is 93.2 dB. Can't get much better than that. I have Klipschorns which, I am sure you know, are very sensitive and no matter the amplifier, with my ear to the midrange horn, I can hear a tiny bit of hiss but nothing out in the room. Sounds like you have a made a very nice amplifier.
I’m tearing into a Deluxe Reverb today that sounds like it’s got a short somewhere in the power supply. The power transformer hums like a SOB!
I would measure the current coming into the amplifier right at the AC input. In idle mode, I would expect the current to be between one and two amps - just a reasonable amount - and see if the current rises when the transformer hums more. Maybe lose screws holding the transformer together?
@@ElPasoTubeAmps it draws .6A at idle. Which is ok. The power transformer is really tight. I used larger 10/32 stainless hardware to mount it up as a first attempt to fix the hum. I’m literally tearing down the amp right now. Hopefully I find something. It is Mercury iron throughout, so if the pt is just bad I can get a free replacement.
When you connect that capacitor at differen points in the amplifier, doesn't the long groundlead then just act as an antenna that picks up AC hum and injects it into that point,
or rather into the point where the other end of the capacitor is connected?
So your basically poking around and injecting a 120Hz (among others..) signal into the amplifier?
That's why the effect gets stronger when you get near the input of the amplifier.
If the circuit was high impedance it would pick up surrounding AC hum but this is big capacitor in a very low impedance circuit. I agree they get stronger near the input and even as definitive as the moment was measuring it, every homebrew amplifier is going to have its on character. I can only guess it is 120 Hz currents induced into the chassis along with the ground loops of the audio portion. Seems a good idea to keep it away from the input but I seem to remember a topology (buss ?) Where everything is connected to a copper wire running the circumference and then ground it to the chassis at the input. I did that once. It was a disaster and hummed like hell. It didn't last long.
@@ElPasoTubeAmps This effect would have been much more exaggerated, if you would have been able to lift the installed filter cap ground points and had those all connected to the spots you connected the introduced test cap.
@@srtamplification Absolutely agree with you. Most of the "work" of filtering the power supply ripple was already being taken care of by the originally installed capacitors. Lifting both of the power supply capacitors from ground and moving that ground point around would have been much more dramatic. I don't think any one topology of grounding scheme is bad, i.e. star, buss, etc but... in any and all cases, it is my belief and experience that keeping these chassis ground points separate for power supply components and low level signal input is a good idea.
@@ElPasoTubeAmps Yes. Much of what type of ground system to use depends on the frequency of the circuit. A multi-point or what I believe you refer to as a bus ground, is very common in high Hz applications. My experience with audio amplifiers have been much the same as yours. Ground all power supply components to one point close to the source and all audio components close to the signal input.
Yes yes yes 100% correct. The center tap and all power supply electrolytics must be grounded at a single point. Then a ground buss may run from that single point across rhe chassis
Totally agree - the comment at the top, transcribed from the Mullard document agrees with us also. Thanks for your comments and confirmation.
Can this method be used on linear amplifiers, regarding hum on the audio?
I assume you are speaking of an RF amplifier for ham radio? I would think that hum on a transmitter output is likely to come from either power supply noise in the transmitter audio section, not the linear amplifier, or RF is getting back into the microphone. RF into the microphone is much more common that we thing. What amplifier are you using?
@@ElPasoTubeAmps a homebrew GS-35B GG triode HF linear amp.
You should increase your r a age rate on your display it’s hard to watch
To much noise in the room...
you have hum in your amplifier because it likes a tune lol
No, the amplifier hums because it doesn't know the words to the song! :-)