The ultimate DIY? A Stax SRM-T2!

[spritzer]

I've never seen a fake 7 pin package and it would be pretty costly to do.  The fets are just relabeled BJT's which would probably make for excellent input devices anyway... 

[s_r]

Seems like oscillation was indeed the cause of the noise & high offset which has been sporadically happening. Happens now with either tube or solid state input. For reference both channels have 15pf bypass caps across the 100k feedback resistors. What about populating both C12 & C13 with 2.2pf caps? Apparently doing this killed the oscillation in pedefede's T2 a few years ago.

[Craig Sawyers]

Getting a particular T2 stable can be a bit of a game. The problem arises from the fact that the feedback trace is quite long, physically. It was less long in the original Stax T2 because it was more compact (and as a result ran incendiary hot).  Anyhow, the feedback trace has to get all the way from the output tube's anodes/plates all the way back to the input tube's cathodes. Mine used to hoot too - you need to adjust the 15pF value in small increments both upwards and downwards, and possibly add one or both 2.2pF caps, and again be prepared to tweak the value. While looking for oscillation and its frequency. Once it is stable, do a frequency response sweep and look for a response peak - if you find one continue adjusting those caps until it is critically damped.

It is interesting to note that 15pF and 100k gives a dominant pole at ~100kHz whereas the oscillation is at 685kHz, which implies that there is a secondary feedback loop, possibly inductively coupling into the feedback trace.

I have a vague recollection that I cut a trace and re-routed in some way, or changed the position of the upper feedback resistor, but I'd have to tear my T2 apart to find out what I did.

[s_r]

Your board version is probably older than mine too, right (production v1.07/silkscreen v1.03)? Either way I'll see how it responds once I've added a second 2.2pf cap.

On that note, what about soldering in SIP sockets to make swapping bypass caps easier?

Edited May 20, 2017 by s_r

[JoaMat]

The oscillation you have - is it only on one channel or both?

[s_r]

Just the right channel, and it's come and gone up until now.

[Craig Sawyers]

10 hours ago, s_r said:

Your board version is probably older than mine too, right (production v1.07/silkscreen v1.03)? Either way I'll see how it responds once I've added a second 2.2pf cap.

Mine is the original - no idea what the issue number was. Kevin might have that data. There have been various iterations since then

[s_r]

Put in the second 2.2pf cap just now, and the oscillation looks nigh identical as before. Guess I'll try changing the bypass cap value next.

Edit: This is what the oscillation looks like with the bypass cap set to 20pf.

Edited May 20, 2017 by s_r

[Craig Sawyers]

OK - frequency has gone down (not too surprising since C has gone up) and amplitude has gone up a bit.

The issue in my mind is that a gross instability around the feedback loop usually builds up to a rail to rail oscillation, whereas this is almost sinusoidal and well below the peak-peak of what this amp is capable of.

Next line of attack - cascode circuits are notorious for self oscillation, which would probably limit the amplitude. So first look for differences in layout from one channel to the other - particularly (a) the position of R86 and R87 to U1 and U2. The position and routing of R6/C2 and R5/C1/R92.

If there are differences between L and R - and a few cm of track might be enough - they might be significant enough to cause this.

Not saying this is the cause - just working the options.

[s_r]

Well I loaded up the gerbers and took some screencaps of the relevant parts. There's certainly a difference between channels. I've also noticed that R86/R87 are physically higher off the board on the right, if that makes any difference.

Edited May 21, 2017 by s_r

[GeorgeP]

Unless your specific board has a problem, your board version was the later version with the corrections, so the layout shouldn't be an issue. Also your first t2 used the same board version and didn't have any oscillation.

 

[headinclouds]

On 5/8/2017 at 4:05 AM, Kerry said:

Finished...

 

Back plates before I mounted them...

Just returning to catch up with this thread. A terrific project Kerry - a milestone.  Many congratulations.

[Craig Sawyers]

8 hours ago, GeorgeP said:

Unless your specific board has a problem, your board version was the later version with the corrections, so the layout shouldn't be an issue. Also your first t2 used the same board version and didn't have any oscillation.

 

Well something is clearly different between his first T2 to his second. Any idea what he should check to find out?

[Kerry]

On my build, which is much smaller, I did need to adjust the 5pF bypass caps.  I was also seeing an oscillation which caused a similar voltage drop.  I move them up to 15pF and everything is great.

I agree with Craig, you can try lower values as well as higher, but keep checking to see if it's getting better or worse.

It's still good news that you've gotten it closer. 

 

[s_r]

I was thinking of adding resistors to the gates of the K216s in the battery next. If I recall correctly this helped solve some noise in Inu's T2. I'll also try lowering the value of the bypass cap and seeing what that does.

Edited May 22, 2017 by s_r

[Craig Sawyers]

Another possibility (which alas involves a trace cut to try) is to add grid stoppers on the input tubes - about 1k.

[JoaMat]

I just checked my T2 with an oscilloscope at outputs. 750 kHz, amplitude 5 V, nice sine wave. The amplifier is modified - offset servo and output current sources. I've no idea how to kill that oscillation. But that amplifier is the one I like best….

[G600]

I'm pretty sure I have the same problem too.

No devices to test it.

But as far as I remember in this topic, it is always the right channel that causes trouble. Mine too.

There must be something in the layout, no clue till now.

[GeorgeP]

What is yours doing? If I recall you have the shrunk board from the failed lil knight group buy. Interestingly, I cannot recall reading anyone who has had any issues with the corrected boards, until now. Even S_R's first build, using the same board iteration, had no issues.

[Kerry]

8 hours ago, JoaMat said:

I just checked my T2 with an oscilloscope at outputs. 750 kHz, amplitude 5 V, nice sine wave. The amplifier is modified - offset servo and output current sources. I've no idea how to kill that oscillation. But that amplifier is the one I like best….

Focus on the feedback bypass cap.  I bet you can get rid of it there.  I played with adding caps to a bunch of places (can't remember them all) and this was the solution for me.  Do you have it on your smaller build, bigger build or both?

[JoaMat]

Thanks for the advice. It's one of the "original" DIY T2. Modified with something like Kevin's offset opto servo, DC/DC converter current sources and sj79 and sk216 replaced with BJT. After reading about s_r trouble with his T2 I checked my own with the scope and then I found the 750 kHz, 5 V amplitude sine wave.

Despite the 750 kHz wave I'm very happy with how it sounds so  I'm not in an hurry to try get rid of the oscillation…. Kind of work that can be very time consuming and boring.

1 hour ago, GeorgeP said:

What is yours doing? If I recall you have the shrunk board from the failed lil knight group buy. Interestingly, I cannot recall reading anyone who has had any issues with the corrected boards, until now. Even S_R's first build, using the same board iteration, had no issues.

I've built three T2s with the "corrected" boards. Just put the right components it the right position, I don't elevate except power resistors. Power on, trim batteries to 740 volts and done. I don't think there is a board issue. All right, I've had problems but most of them have been builder induced.

[GeorgeP]

2 hours ago, JoaMat said:

 

I've built three T2s with the "corrected" boards. Just put the right components it the right position, I don't elevate except power resistors. Power on, trim batteries to 740 volts and done. I don't think there is a board issue. All right, I've had problems but most of them have been builder

That has been my experience as well.

[G600]

6 hours ago, GeorgeP said:

What is yours doing? If I recall you have the shrunk board from the failed lil knight group buy. Interestingly, I cannot recall reading anyone who has had any issues with the corrected boards, until now. Even S_R's first build, using the same board iteration, had no issues.

Sometimes I hear some random cracks.

They come and go, and at some point the channel goes south.

Voltages are off, with something like 40 V between the two right batteries, and a very large offset at ouput.

It could be cured instantaneously by putting a Voltmeter between battery testpoint and +500V supply entry.

Tried to adjust the batteries in many ways, but it comes back anyway.

I have also uprated the compensation cap to 15 pF. The amp runs cooler afterwards, but the problems is still happening.

Edited May 23, 2017 by G600

[GeorgeP]

You might be the only person who built one using that board - 4oz copper and super thick makes for a tricky build and zero tolerance for error. Did you diode test the whole thing? Also, is the right side the one that was missing the trace?

[JoaMat]

19 hours ago, Kerry said:

Focus on the feedback bypass cap.  I bet you can get rid of it there.  I played with adding caps to a bunch of places (can't remember them all) and this was the solution for me.  Do you have it on your smaller build, bigger build or both?

Thank you , Kerry.

Added capacitance in feedback.. With 15pF the sine wave is replaced with something erratic - amplitude 100 to 200 mV.

 

13 hours ago, G600 said:

I have also uprated the compensation cap to 15 pF. The amp runs cooler afterwards, but the problems is still happening.

What is that compensation cap?