SKYLINER build - inspired by ODS #124 and #060

[Stephen1966]

... I did not need to tweak it further, but i did set it twice. First when finishing the amp, and again after playing the amp for a few months. It is not as critical as some believe imo.

I totally get that. Of course, I want to learn as much as I can about this amp before stuffing it away in the head and if it's possible to dial the amp in to something like it's ideal operating condition, then I'm interested. Geez, I can't even get the same mains wall voltage from day to day though, so there comes a point where you have to say, it's as close as it will get. And be done with it

[Richard1001]

I'm sorry Richard, I really am an idiot - I didn't mean to make you cry but I'm glad you see the funny side. I corrected myself and hoped you noticed it before replying but I was a bit late out of the gate.

It's starting to make sense as I understand the long-tail pair better. Because of the common tail resistors (the long tail), the non-inverting (input) side of the PI tube is lower in gain than the inverting (feedback) side thus the need for dissimilar fixed resistors to equalise the gain from both anodes. Result, the voltage needs to be higher on the non-inverting, input side of the tube because if the internal resistance of the tube is a pseudo constant, the increased voltage, increases the current and thereby increases the gain. Bringing both sides of the tube into a balance. The trimmer is matching this to any imbalance in the power tubes as well so reading the signal at the grids of the power tubes wouldn't give the full picture, and for that you have to go to the plates (V8 and V9 just happen to be on the primary blue and brown ends of the OT in my case) or the cathodes of the power tubes which takes in the addition of the screen grid currents as well.

If I inject the signal at the power amp in, bypassing the preamp, I guess it is possible to look at the load voltage of the output (about 20Vrms as we said before) to find that 50% output. I guess it is also possible to look for the point where the sine wave starts to clip and then dial it back a tad, using that as the operating point. The first, expressed elsewhere, offers us the option of setting the amp at what we consider nominal operating volumes and lower voltages at the output if we normally use the amp quieter. The second option, looking for clipping, given the amps ability to stay clean might result in a very hot optimised signal which we might never approach in working conditions.

I do have the cathode resistors in my build and above, you will see that I can tie the cathodes together just by connecting the opposite pairs of bias outputs, it's electrically equivalent to tying the cathodes together directly at the resistors. I would be curious to see if we could connect the probes to the cathode pairs and then add the signals together in the scope to achieve a ~0V trace. Similar to the way you describe adding the signals from the plates. A bit also like the SVT method for Harmonic Distortion Meter. I'm not clear whether the signals for the scope would be in DC or AC in the cathode method though, but given the low voltages and currents there, it would be safe to flip between and see which yields the best results.

Little correction: the input side of the long tail PI is inverting the signal.

And don't connect opposite pairs of output tube cathodes together, but only the two tubes of which the anodes are allready tied together.

You want to measure AC with your scope to set the AC balance. There is a little flaw using the Ampeg methode on a regular guitar' amp. On the Ampeg there are two bias controls to set the bias current for each side exactly the same. So using the DC voltmeter methode there is 0 volt at zero signal.

Most guitaramps have only one bias control to set the bias for both sides of tubes. So unless the tubes match exactly (allmost never) you start out with a small difference in voltage on either side, so with zero signal you allready measure the bias difference on your DC voltmeter. To set the AC balance you therefor do not set it to zero, but to the same reading you got at zero signal.

[Stephen1966]

Little correction: the input side of the long tail PI is inverting the signal.
...

Thanks. You're right. I see it now.

[Stephen1966]

Here's a hypothetical... What would happen if someone made the schoolboy error of forgetting to install the cap across the Master pot?

I will give you a hint... I forgot to install the cap across the Master pot.

I am getting a sawtooth waveform at the plates with no signal and while the line into the Master is clean as a whistle - I followed the trace all the way from the input jack - the wiper of the Master, leading out to the PI appears jumpy and almost with several waveforms dancing around. Almost like the signal from the power tubes was feeding back.

[ijedouglas]

Here's a hypothetical... What would happen if someone made the schoolboy error of forgetting to install the cap across the Master pot?

I will give you a hint... I forgot to install the cap across the Master pot.

I am getting a sawtooth waveform at the plates with no signal and while the line into the Master is clean as a whistle - I followed the trace all the way from the input jack - the wiper of the Master, leading out to the PI appears jumpy and almost with several waveforms dancing around. Almost like the signal from the power tubes was feeding back.

If you are referring to the bright cap on the master volume pot, I don't think it will make a difference (other than to allow some highs to pass at low volume). I don't think it is there to block anything coming back from the power section. I typically leave that cap off.

[martin manning]

Here's a hypothetical... What would happen if someone made the schoolboy error of forgetting to install the cap across the Master pot?

I will give you a hint... I forgot to install the cap across the Master pot.

I am getting a sawtooth waveform at the plates with no signal and while the line into the Master is clean as a whistle - I followed the trace all the way from the input jack - the wiper of the Master, leading out to the PI appears jumpy and almost with several waveforms dancing around. Almost like the signal from the power tubes was feeding back.

The MV cap is a bright cap. You can experiment with the value to see what you like. Some people omit it.

The plate supply is far from clean DC. It probably has around 10V sawtooth riding on it, which gets cancelled in the OT. There may be some ripple in the bias supply too, which is again common mode, and it will be cancelled as well

[Stephen1966]

Here's a hypothetical... What would happen if someone made the schoolboy error of forgetting to install the cap across the Master pot?

I will give you a hint... I forgot to install the cap across the Master pot.

I am getting a sawtooth waveform at the plates with no signal and while the line into the Master is clean as a whistle - I followed the trace all the way from the input jack - the wiper of the Master, leading out to the PI appears jumpy and almost with several waveforms dancing around. Almost like the signal from the power tubes was feeding back.

If you are referring to the bright cap on the master volume pot, I don't think it will make a difference (other than to allow some highs to pass at low volume). I don't think it is there to block anything coming back from the power section. I typically leave that cap off.

That's interesting, I will try it though with the spike suppressing diode and cap on the power tubes I'm already filtering out some of the higher frequencies. Hard to tell, my hearing is not so good in that range... I'll try it in any case. Thank you for clarifying.

[Stephen1966]

Here's a hypothetical... What would happen if someone made the schoolboy error of forgetting to install the cap across the Master pot?

I will give you a hint... I forgot to install the cap across the Master pot.

I am getting a sawtooth waveform at the plates with no signal and while the line into the Master is clean as a whistle - I followed the trace all the way from the input jack - the wiper of the Master, leading out to the PI appears jumpy and almost with several waveforms dancing around. Almost like the signal from the power tubes was feeding back.

The MV cap is a bright cap. You can experiment with the value to see what you like. Some people omit it.

The plate supply is far from clean DC. It probably has around 10V sawtooth riding on it, which gets cancelled in the OT. There may be some ripple in the bias supply too, which is again common mode, and it will be cancelled as well

This is also interesting but this is in AC mode... I will try it in DC and see what happens. It sounds like you are describing what I am seeing though - and hearing because once it's running I don't hear any distortion. Still not getting any meaningful readings from the PI trimmer though, another approach (from reading the signal at the plates) might be called for.

[Stephen1966]

Having swapped the PI tube for a balanced 12AX7, here are the traces I get from its plates (after the coupling caps with probes on the power tube grids)

SAM_7923.JPG

[Stephen1966]

... the traces are very even, and have almost exactly the same deflection factor

SAM_7926.JPG

After inverting CH2 and adding them, I was able to adjust the balance on the PI trimmer to almost zero perturbation...

SAM_7925.JPG

[Stephen1966]

... Oddly, the voltage difference between the two plates is now around 1V and the trimmer has a range from one side to the other of around 3 volts. The A+B trace above is with the trimmer just off centre.

However, this is only the PI tube and it doesn't take account of any imbalance in the power tubes.

I tried taking the signal from the cathodes, with the pairs on each side tied together and probes after the 1ohm resistors. This is what I got...

SAM_7932.JPG

This is off axis and could be corrected but when I add these signals together, I don't get a straight(ish) trace like I found after the PI plates.

I'm aware CH2 could be inverted to bring it into phase with CH1 but the wonky resulting A+B trace doesn't appear to show any change with adjustment to the trimmer.

Calling Mr Manning - Martin, I believe you used the scope at the cathodes for this purpose, is it possible you could outline how you did it? Is there a better way to use the scope to take the power tubes into account?

Again, I tried the Ampeg V9 procedure with the DMM but the results are just floating around -2mV/+2mV - if you saw it, you might conclude as I did that it is not conclusive.

[Richard1001]

Your signal looks correct. Measuring voltage over the cathode resistors, due to the low resistance of the cathode resistors tied to ground you don't see a bottom half of the sinewave.

With the DC voltmeter methode using a low frequency +/- 100Hz) on the signal generator, the problem is that the powersupply ripple (the sawtooth) will add and substract from the amplified signal causing interference with your measurement. It keeps going up and down. Its like the wobble you hear playing two notes close to each other on your guitar. Use a higher frequency between 800 - 1000 Hz on the signal generator and you most likely get a stable readout.

First measure the current at zero signal and set the AC balance to the same value. Make sure your amp puts out 1/2 its power over the load resistor to set the balance correct.

[Stephen1966]

Your signal looks correct. Measuring voltage over the cathode resistors, due to the low resistance of the cathode resistors tied to ground you don't see a bottom half of the sinewave.

With the DC voltmeter methode using a low frequency +/- 100Hz) on the signal generator, the problem is that the powersupply ripple (the sawtooth) will add and substract from the amplified signal causing interference with your measurement. It keeps going up and down. Its like the wobble you hear playing two notes close to each other on your guitar. Use a higher frequency between 800 - 1000 Hz on the signal generator and you most likely get a stable readout.

First measure the current at zero signal and set the AC balance to the same value. Make sure your amp puts out 1/2 its power over the load resistor to set the balance correct.

Many thanks Richard, I appreciate the confirmation. I suspected it was normal to see an asymmetrical trace but didn't fully understand why.

I'm not clear about this step, could you elaborate?

First measure the current at zero signal and set the AC balance to the same value.

The current is about 45mA unloaded, across the PT cathodes, but where would I set the AC balance?

On a seperate, but related topic, I am considering playing with the load resistor in the PSU for the PI. The voltage is nearly even on both sides of the PI and now, about 10V lower on the input side of the PI. I'm also getting very little voltage difference between the sides of the trimmer pot (a 5k), I am guessing a 10k might be better. Naturally, if I increase the voltage to the PI from the PSU, it is going to increase the voltages down the chain to the preamp. In the clean channel the amp is quiet, compared to the OD channel. I understand the OD is going to have more gain but to get an equivalent volume the Volume pot is close to 50% and the OD Level pot is at about 25%. It's really bedroom level stuff! It only really starts rattling the walls above these levels.

[martin manning]

I tried taking the signal from the cathodes, with the pairs on each side tied together and probes after the 1ohm resistors. This is what I got...

This is good. You are looking at the current pulses from each side, and they look to be very nearly balanced.

This is off axis and could be corrected but when I add these signals together, I don't get a straight(ish) trace like I found after the PI plates. I'm aware CH2 could be inverted to bring it into phase with CH1 but the wonky resulting A+B trace doesn't appear to show any change with adjustment to the trimmer.

Adding them will show you the complete sine wave, and you can still observe the imbalance by looking at the alternate peaks (which correspond to the top halves of the PI output signals), and adjust the trimmer to get them all at the same height. You don't want to invert Ch2.

Calling Mr Manning - Martin, I believe you used the scope at the cathodes for this purpose, is it possible you could outline how you did it? Is there a better way to use the scope to take the power tubes into account?

Yes, and this my preferred method, and I use a 400-1kHz signal. A 5k trimmer is pretty small. 10k is shown in the layouts, and if the PI tube is bot well balanced even that won't be enough. With a larger value trimmer you will see more variation. I do not recommend using the plate signals.

[Stephen1966]

I tried taking the signal from the cathodes, with the pairs on each side tied together and probes after the 1ohm resistors. This is what I got...

This is good. You are looking at the current pulses from each side, and they look to be very nearly balanced.

This is off axis and could be corrected but when I add these signals together, I don't get a straight(ish) trace like I found after the PI plates. I'm aware CH2 could be inverted to bring it into phase with CH1 but the wonky resulting A+B trace doesn't appear to show any change with adjustment to the trimmer.

Adding them will show you the complete sine wave, and you can still observe the imbalance by looking at the alternate peaks (which correspond to the top halves of the PI output signals), and adjust the trimmer to get them all at the same height. You don't want to invert Ch2.

Calling Mr Manning - Martin, I believe you used the scope at the cathodes for this purpose, is it possible you could outline how you did it? Is there a better way to use the scope to take the power tubes into account?

Yes, and this my preferred method, and I use a 400-1kHz signal. A 5k trimmer is pretty small. 10k is shown in the layouts, and if the PI tube is bot well balanced even that won't be enough. With a larger value trimmer you will see more variation. I do not recommend using the plate signals.

Very much appreciated, thank you, Martin. It's a relief to know I'm on the right path. I think I was able to get a complete sine wave but I was thinking I should be looking to create a zeroed out trace like I did for the PI plates... I knew that could only take me so far, and it was instructive because I could detect some small variations there when adjusting the trimmer. So it got me in the ball park before I went to measure the signal at the cathodes. The old, (nothing special) 7025 in the PI gave me a 10V difference between the plates but the 5k trimmer wasn't making any difference I could see. I think in any case, I am going to try a 10k trimmer. Anything higher than this could be forcing an imbalance greater than the 10-20% difference in plate resistors though that might be just an arbitrary rule of thumb. The other rule of thumb, to make the plate resistors about double the internal plate (non-linear) resistance, and to increase the resistor value on the trimmer could affect the load line and possibly reduce the higher frequency response of the tube.

I guess it is possible I could replace the plate resistors with smaller values and increase the value of the trimmer so the sum of the external resistance at the plates is about the same as the original, but the increased value of the trimmer would allow for more variation on either side. Say a 25k trimmer, with 100k on the inverting side and 110k on the non-inverting (feedback) side. What do you think?