So I took a minute to unhook my feedback loop this morning and do some checking up on the amplifier just to make sure it's working right.
I've got nice symmetrical drive, right up to the grids of the power tubes where I'm getting about 70 volts of signal which seems correct given that my power tube cathode are sitting at about 40 volts. Unfortunately my power tubes are ild and weak and they only give me about 15 watts before clipping. I have the whole test bench set up with rms volt meters on the input and output and I'm going to try and dial in the feedback which is a totally new experience for me. As it stands I get a low frequency oscillator when I apply my feedback. I'm going to check that it is actually going in as negative feedback but I'm pretty sure it is. Perhaps I need to change something else to get it stable.
I'm open to suggestions because I've never done any stability testing or feedback tuning before. I understand the reason for it but haven't developed the working knowledge of it yet...
I've got nice symmetrical drive, right up to the grids of the power tubes where I'm getting about 70 volts of signal which seems correct given that my power tube cathode are sitting at about 40 volts. Unfortunately my power tubes are ild and weak and they only give me about 15 watts before clipping. I have the whole test bench set up with rms volt meters on the input and output and I'm going to try and dial in the feedback which is a totally new experience for me. As it stands I get a low frequency oscillator when I apply my feedback. I'm going to check that it is actually going in as negative feedback but I'm pretty sure it is. Perhaps I need to change something else to get it stable.
I'm open to suggestions because I've never done any stability testing or feedback tuning before. I understand the reason for it but haven't developed the working knowledge of it yet...
Are you sure of your output matching - compatibility between B+, Ra-a, and the 6L6? If looking for low power, go back to ohms law first and be sure 15 watts isn’t all you should expect.
When tuning feedback, I usually set the dominant pole in the input stage around 15 KHz. If high levels of NFB (>25 dB) it may need to be set lower to get stable. If much less is used you can usually go higher. Higher is better, up till the point where it can’t be made stable. It’s a cliff, so watch it. Then adjust the cap across the main feedback resistor to give the prettiest square wave response (Minimize ringing).
If you accidentally get it backwards it won’t blow anything out. Except your ear drums, if speakers are connected. Use a dummy load.
When tuning feedback, I usually set the dominant pole in the input stage around 15 KHz. If high levels of NFB (>25 dB) it may need to be set lower to get stable. If much less is used you can usually go higher. Higher is better, up till the point where it can’t be made stable. It’s a cliff, so watch it. Then adjust the cap across the main feedback resistor to give the prettiest square wave response (Minimize ringing).
If you accidentally get it backwards it won’t blow anything out. Except your ear drums, if speakers are connected. Use a dummy load.
Perhaps I could get some help with this part. I put a tone through the amp with feedback disconnected, ran it into my hp distortion analyzer and Set the level to show 0db. Shut the power amplifier off, and reconnected the feedback. With negative feedback applied I get a power output level about 12db lower than without any feedback. That's 12db of global feedback, correct? The fact that the gain goes down tells me I'm applying negative feedback, and the low frequency oscillation is probably due to a low frequency time constant in the coupling caps is that correct? What's the next thing to investigate to get this thing fixed, swapping coupling cap values? Where would you start?
I am not sure what tubes you are using and what the operating parameters are, but 12-15 watts could be about what the original Acrosound gave. You may have to change a few things to get more watts.
A quad II using the KT66 is specified for approx 15 watts for a a pair of KT66's which are similar valves. People have been known to substitute 5881's and 6L6gc's when they cant find or afford expensive KT66.
Oscillation can be due to a reversed feedback connection, and check you earths returns are nice and clean with low impedance paths as a starter.
A quad II using the KT66 is specified for approx 15 watts for a a pair of KT66's which are similar valves. People have been known to substitute 5881's and 6L6gc's when they cant find or afford expensive KT66.
Oscillation can be due to a reversed feedback connection, and check you earths returns are nice and clean with low impedance paths as a starter.
Low frequency oscillations are a either a coupling cap/output trafo issue or insufficient bypassing/decoupling on the power supplies. The former is solved by appropriately staggering the poles. The latter requires bigger caps in the supply and/or regulation of anything derived from the main B+.
Could a cathode bypass cap of insufficient value cause this type of oscillation?
Am I measuring 12DB of feedback, is that correct?
Am I measuring 12DB of feedback, is that correct?
Yes to both. One of the low frequency poles you need to deal with is set by the bypass cap. I normally use fixed bias (for other reasons) so that one doesn’t normally pester me. 12 dB of feedback isn’t crazy high, and you should be able to stabilize it without doing anything insane. Obviously, HF was already at least ok.
I think I'll start by changing the cathode bypass cap on the first stage.
I didn't have the value that the schematic asked for and fudged in one that's probably too small. Thanks for your help.
I didn't have the value that the schematic asked for and fudged in one that's probably too small. Thanks for your help.
Changed cathode bypass resistor and still have a low frequency oscillation.
How do I go about "staggering the poles"
I don't understand what is being referenced by "staggering the poles"
How do I go about "staggering the poles"
I don't understand what is being referenced by "staggering the poles"
Making sure the RC time constants don’t all line up on top of one another. You will never get one pole completely ”dominant” on the LF range, but you can get one of the coupling cap time constants higher than the rest. That is usually needed.
Unintentional positive feedback through the power supply is also quite possible. Dont discount this as a source of trouble till you’ve ruled it out.
Another method I’ve used with Williamson designs is to remove the feedback at very low frequency - allowing it to revert to open loop gain at “DC”. You shouldn’t have to do that here.
Unintentional positive feedback through the power supply is also quite possible. Dont discount this as a source of trouble till you’ve ruled it out.
Another method I’ve used with Williamson designs is to remove the feedback at very low frequency - allowing it to revert to open loop gain at “DC”. You shouldn’t have to do that here.
Last edited:
That's great general information but how do I test for it?
Where do I look? What numbers do I need to look at specifically?
Is it the coupling caps and grid resistors that form these "poles" could you point me to some reason the topic if you don't want to take the time to explain?
Where do I look? What numbers do I need to look at specifically?
Is it the coupling caps and grid resistors that form these "poles" could you point me to some reason the topic if you don't want to take the time to explain?
Do I plug the .22 coupling cap value into a calculator with the value of the resistor after that?
If so I get a pole at 1.5 hz and another at 5.9 hz
How far apart do those numbers need to be ideally?
If so I get a pole at 1.5 hz and another at 5.9 hz
How far apart do those numbers need to be ideally?
This doesnt answer your question about the amplifier poles.
On the attached schematic they had a tube rectifier that you have replaced by two diodes, they will not drop the voltage that the original rectifier would drop. Have you corrected for that?
What are your voltages around the circuit and is the HT voltage in spec ?
In terms of the PSU the diagram has 20uF before and after the choke do you have those in place I can see something near the diodes but its hard to tell.
The Transformer heater winding from the diagram has a 100ohm pot across the power transformer's 6.3V ac output that can be be used for hum reduction. Do you have that in your circuit. and made adjustments?
The feedback capacitor is shown as 120mmF that 120pF in new money I believe. you have the amp running on the bench so apologies for the basic questions.
On the attached schematic they had a tube rectifier that you have replaced by two diodes, they will not drop the voltage that the original rectifier would drop. Have you corrected for that?
What are your voltages around the circuit and is the HT voltage in spec ?
In terms of the PSU the diagram has 20uF before and after the choke do you have those in place I can see something near the diodes but its hard to tell.
The Transformer heater winding from the diagram has a 100ohm pot across the power transformer's 6.3V ac output that can be be used for hum reduction. Do you have that in your circuit. and made adjustments?
The feedback capacitor is shown as 120mmF that 120pF in new money I believe. you have the amp running on the bench so apologies for the basic questions.
Yes, your observations are correct.
No hum pot installed yet.
All power supply caps are 20 uf.
I used a 150 pf cap because I did not have a 120 pf.
Somehow, now the oscillation from earlier is gone. It sits still and plays nicely up to about 18 watts. I think the bypass cap on the first cathode was way too small. I put on a 220 (schematic calls for 250) and it now behaves itself.
No low frequency oscillation.
I'd still like to know how to calculate the "poles" and what those numbers mean.
I've got it to quit oscillating but I think I could still get it sounding better with some instruction, I'd likely learn something too.
B+ is a healthy 450v DC and the cathode sits about 40v above ground.
No hum pot installed yet.
All power supply caps are 20 uf.
I used a 150 pf cap because I did not have a 120 pf.
Somehow, now the oscillation from earlier is gone. It sits still and plays nicely up to about 18 watts. I think the bypass cap on the first cathode was way too small. I put on a 220 (schematic calls for 250) and it now behaves itself.
No low frequency oscillation.
I'd still like to know how to calculate the "poles" and what those numbers mean.
I've got it to quit oscillating but I think I could still get it sounding better with some instruction, I'd likely learn something too.
B+ is a healthy 450v DC and the cathode sits about 40v above ground.
“poles” is engineering-speak and would really take too long to properly explain. They correspond with frequencies determined by the entire amplifier’s loop gain. They can be approximated by calculating individual (1/2piRC)’s but in reality they interact with each other. With a complete amplifier model they may be explicitly calculated, and they are complex numbers (with the j). The loop gain must end up with less than 180 degrees of phase shift for stability. It’s hard even for dedicated college engineering students -don’t feel bad. I got a “B” in controls. I never got B’s. But I learned a lot about making amps stable. With relatively low levels of feedback (ie, 12 dB) you often don’t even need to go through calculations, just use common sense and follow working examples.
If it “went away on its own” I would suspect power supply feedback. There are additional loops through the supply that have their own poles - and the same thing applies for stability criteria. The power supply impedance is NOT zero although it is assumed to be. Caps can take a while to form, and their values increase during that period. When it was first switched on they were “too small” and oscillation resulted. Quite common, and happens all the time in old vintage equipment where the caps are dried out. They may never recover and need replacement.
220 vs. 250 uF is normally a don’t care, BTW. If power supply caps are only 20 uF instead of 200-400, it tends to invite supply related oscillations. If the value falls by half you could be in trouble. If a big one falls by half its not as noticeable.
If it “went away on its own” I would suspect power supply feedback. There are additional loops through the supply that have their own poles - and the same thing applies for stability criteria. The power supply impedance is NOT zero although it is assumed to be. Caps can take a while to form, and their values increase during that period. When it was first switched on they were “too small” and oscillation resulted. Quite common, and happens all the time in old vintage equipment where the caps are dried out. They may never recover and need replacement.
220 vs. 250 uF is normally a don’t care, BTW. If power supply caps are only 20 uF instead of 200-400, it tends to invite supply related oscillations. If the value falls by half you could be in trouble. If a big one falls by half its not as noticeable.
Have you got a 10 turn pot? You need a 20k linear pot to dial in your NFB, your signal analyser, dummy load & a scope. Use two terminals of the pot (so as a rheostat or variable R ) measure across the wiper and one end, set to 20k. Connect the pot between your two FB points IE cathode of V1 and + of OPT. One end of OPT connected to ground. Put in a 1khz sine in the IP at about 0.5 to 1v RMS. Turn the volume pot up until you have about 3.5v RMS OP, adjust scope so your sine fills all the divisions.
Start turning the pot, is the sinewave getting bigger or smaller? If the former you are feeding the amp positive FB. Change the polarity of the OPT or phase spliter. Do again, monitor your OP voltage & THD & your scope, you should see OP V drop as well as the THD, stop, increase the volume until you are back at 3.5v RMS or whatever V you started at, what's the THD now? You can decrease the FB resistor until the amp goes into parasytic oscillation, back off a bit, then replace the pot with a fixed R, or apply just enough FB so that THD, stabilty & frequency response are all optimal; your call.
I did this here this weekend in this video -
I was in a rush to get it filmed so it might not be as clear as I would have liked but hope it gives you an idea of what to do. There is further info on Patrick Turners old site which has been archived in this thread - https://www.diyaudio.com/community/threads/rip-patrick-turner.371896/ his post on how to test an amp is well worth a read. It will give the bare bones but to test and setup an amp for the best F response, THD & stability takes a few hours of mucking about with test gear, hours of knob twiddling which is the next best thing to sex : )
Andy.
Start turning the pot, is the sinewave getting bigger or smaller? If the former you are feeding the amp positive FB. Change the polarity of the OPT or phase spliter. Do again, monitor your OP voltage & THD & your scope, you should see OP V drop as well as the THD, stop, increase the volume until you are back at 3.5v RMS or whatever V you started at, what's the THD now? You can decrease the FB resistor until the amp goes into parasytic oscillation, back off a bit, then replace the pot with a fixed R, or apply just enough FB so that THD, stabilty & frequency response are all optimal; your call.
I did this here this weekend in this video -
Andy.
I ended up having something miswired on the gain stage and upon further work I determined that 2 of the acrosound transformers I used had open plate windings on one plate. I had to buy 2 more, but now I've got all 4 amplifiers working with 12db feedback and they sound really great. I swapped them into my home system in place of my fully restored mc30s and I didn't notice any drastic differences.
For new years eve, I took the whole rig down to the bar with 2 turntables, a mixer and a microphone and DJed into the new year using my home made 6l6 power amps.
It sounded great, and I barely had the volume cracked, we could have dumped a lot more power into that room...
For new years eve, I took the whole rig down to the bar with 2 turntables, a mixer and a microphone and DJed into the new year using my home made 6l6 power amps.
It sounded great, and I barely had the volume cracked, we could have dumped a lot more power into that room...
Last edited: