Negative feedback shielding

[baudouin0]

Well done.E string as a file HaHa.

 

[jhstewart9]

Well I think I've found the problem. One oxidized connector in 12AX7 tube socket. When I put the tubes back after I fixed the wiring somehow that pin began to be intermittent even though I took the tubes out several times in the process. I measured and cleaned all sockets before I've build the amp but looking at this particular socket hole with magnifying glass, it had quiet a layer of black inside for some reason. Perhaps it was arcing in the first place.
I used E string to file the hole out, now it seems ok (so far!). Spend two day nearly rebuilding the amp because of that, who would have thought.
But I want to say thanks to you all for trying to help, I did learn few things along the way so it wasn't wasted and perhaps someone will have same problem one day and will read this before going crazy.

Many years ago as I used a TEK 540 series scope I noticed that intermittently on calibration the gain of the vertical amp appeared greater on the upper part of the CRT than the lower part. But the condition was a real puzzle, only happened occasionally.


Turns out the problem was no connexion on either pin 4 or 5, don't recall now of the heater on one of the vertical amp tubes. The other side was OK, looks like the scope got past TEK QA & our lab got it. All fixed with a short piece of the TEK silver doped solder.

 

[ArcticBrew]

LTP Calculator

Software and I do not work well together.
I could not get it to calculate the correct values, I had to enter the values,
and then adjust them until the gains matched.
There probably is a way to get it to do it automatically, but I should not have to
learn some clunky software interface.

That is how the calculator works. Select a tube and change the values till you get the gains to match.

Doesn't matter if this is a guitar site the calculators are useful.

Another set of calculators that I like.

~ Online Equation Calculators ~

Steve

 

[PCL200]

I'm still curious at what B+ this calculator works. I thought that the gain for a certain tube type gets higher with higher B+. Or is the amount of FB in the tail proportional to B+?

 

[JMFahey]

I took the NFB off and the ghosting was gone. So I've put in shielded wire and ghosting it's back even at different wire position. It's weird. I've got 820 ohm NFB like Deluxe Reverb,

Then NFB Net impedance is less than 800 ohm and signallevel is>1V RMS

NO WAY shielding affects that signal in any way, low impedance and high level ensure that.

You do have an NFB problem for sure, but it´s a very different one.

 

[jhstewart9]

I'm still curious at what B+ this calculator works. I thought that the gain for a certain tube type gets higher with higher B+. Or is the amount of FB in the tail proportional to B+?

One reason of several is that the load impedance can be increased if higher B+ is available. OTOH, if load impedance remains the same the extra B+ will result in higher Ip, so Gm increases & so does gain.

 

[JMFahey]

If the tube rocks at all in the socket - REPLACE IT.
It's not an option, it's Mandatory for reliable operation.


Take the hint from an old seasoned tech.

There´s rocking and rocking

Metal tubular contacts are smaller diameter than actual hole in ceramic (or bakelite) so they allow some "play" and are free to match individual pins, so slight tube rocking while pins are firmly inserted in each contact is fine and actually expected.

Now if metal contacts are SO worn and abused that pins dance inside them, it´s a COMPLETELY DIFFERENT situation.

 

[wiseoldtech]

There´s rocking and rocking

Metal tubular contacts are smaller diameter than actual hole in ceramic (or bakelite) so they allow some "play" and are free to match individual pins, so slight tube rocking while pins are firmly inserted in each contact is fine and actually expected.

Now if metal contacts are SO worn and abused that pins dance inside them, it´s a COMPLETELY DIFFERENT situation.

What I was actually referring to is coming across a tube that easily moves inside a socket that has worn, loose terminals.
A tube that easily comes out of its socket, or for that matter, is easily inserted.


The new noval 9 pin sockets that I installed on that Citation II - I literally have to use pounds of force to insert the tubes - a good thing.

 

[SemperFi]

Lead dress is extremely important for stability, even in stages with low gain. Things you don’t expect to make a difference can. The bias board at the left of the photo was originally located at that second set of holes to the right of the k-followers, with the coupling caps flying over the stage back to the phase splitter. Couldn’t get the stage stable, and it’s unity gain, dammit! Even with no feedback, or even output tubes. Get the signal level above a certain point, and the followers would regenerate and drive straight to clipping. Square wave clipping, and asymmetrical to boot. That would have made some NASTY popping sounds. After a lot of head scratching the only thing it could have been is unintentional coupling. Moving the bias board and putting the coupling caps on it to minimize lengths fixed it immediately.

wg_ski, I see no grid stoppers! Mandatory even on CF with no gain. And daisy chained b+ wiring, shared along the route by all the CFs makes some stopper needed, either at grid or plate. My bet is that would have stopped the oscillations just as well as moving the board. That said, nice build, looks great.

 

[PCL200]

One reason of several is that the load impedance can be increased if higher B+ is available. OTOH, if load impedance remains the same the extra B+ will result in higher Ip, so Gm increases & so does gain.

Thanks.

I just sent an email to "Ampbooks.com" to ask:

"At what B+ does the Long Tail Pair Calculator on your site works? And is B+ the same for all the tube types the calculator contains?"

I hope I get an answer.

 

[wg_ski]

wg_ski, I see no grid stoppers! Mandatory even on CF with no gain. And daisy chained b+ wiring, shared along the route by all the CFs makes some stopper needed, either at grid or plate. My bet is that would have stopped the oscillations just as well as moving the board. That said, nice build, looks great.

Nothing made sense since it was oscillating at 40 Hz. The square wave was the oscillation, you can see the music signal riding on top of it. If it were something a grid stopper would fix, I would have expected high frequency oscillation. Unless it was really oscillating in the MHz (or GHz) range, and gating on and off at a 40 Hz rate. I’ve seen that before in RF circuits. Keeping the input leads short and at right angles to everting else, preventing crossovers where possible, and twisting pairs (including the output) seemed to keep it stable. The white wires going from the bias board to the grids could be replaced by 1k resistors, and then a grid stop would be in place.

Since the OP was complaining of “ghosting” sounds, it would seem his stability issue is low frequency, maybe the 2 Hz stuff I was “warned” about on that build. That amp was stable. I’ve got a 25 watt per channel 7591- based build with a sub 1Hz oscillation I’ve yet to get rid of. I’ve pushed the low frequency poles down as far as they will go without resorting to electrolytic coupling caps and no effect. Make them a lot smaller and I get a traditional motorboat that’s audible. I’m suspecting the output transformers just aren’t good enough for the amount of loop gain and scrapping it. But I’ll look closer for GHz oscillations that may be hiding and really causing the issue. That could also be the OP’s problem. That stuff is b*** to find without a specan.

 

[baudouin0]

Before you condemn the OPT's have a think about the number of high pass filter poles in the feedback. You have the OPT of course and a number of coupling caps. You also have the cathode bias caps you have added. The important thing is that one high pass must dominate so that the loop gain gets to zero before the phase goes through 0 deg. This pole could be the OPT. If that's the case make sure all your other poles are low enough in frequency. A simple test is to make the final caps that drive the grids of the output valves first bigger and smaller. See what makes it worse and better. Also move the LTP plate supply off the first b+ and onto the second (unconnected b+)

 

[baudouin0]

Other things to check. Ground the LTP input to check the stability problem is not getting through the b+ supply. Definitely don't share the B+ for the first stage and the LTP.


The circuit can be modified to cancel a pole if needed. This would consist of replacing the 47R NFB resistor with 47R+220R in series. A cap would then be placed across the 220R (~100uf).

 

[SemperFi]

Nothing made sense since it was oscillating at 40 Hz. The square wave was the oscillation, you can see the music signal riding on top of it. If it were something a grid stopper would fix, I would have expected high frequency oscillation. Unless it was really oscillating in the MHz (or GHz) range, and gating on and off at a 40 Hz rate. I’ve seen that before in RF circuits. Keeping the input leads short and at right angles to everting else, preventing crossovers where possible, and twisting pairs (including the output) seemed to keep it stable. The white wires going from the bias board to the grids could be replaced by 1k resistors, and then a grid stop would be in place.

Since the OP was complaining of “ghosting” sounds, it would seem his stability issue is low frequency, maybe the 2 Hz stuff I was “warned” about on that build. That amp was stable. I’ve got a 25 watt per channel 7591- based build with a sub 1Hz oscillation I’ve yet to get rid of. I’ve pushed the low frequency poles down as far as they will go without resorting to electrolytic coupling caps and no effect. Make them a lot smaller and I get a traditional motorboat that’s audible. I’m suspecting the output transformers just aren’t good enough for the amount of loop gain and scrapping it. But I’ll look closer for GHz oscillations that may be hiding and really causing the issue. That could also be the OP’s problem. That stuff is b*** to find without a specan.

Got you. I agree that stoppers are for HF. I admit I didnt bother look at the scope shot.

 

[PCL200]

Thanks.

I just sent an email to "Ampbooks.com" to ask:

"At what B+ does the Long Tail Pair Calculator on your site works? And is B+ the same for all the tube types the calculator contains?"

I hope I get an answer.

This is their (prompt!) reply:

For audio frequencies, B+ is shorted to ground via an RC ripple filter capacitor, so its value is irrelevant for AC analysis. The calculators use amplification factors and plate resistances that are valid for typical DC operating points. For a 12AX7, for example, the DC plate current is assumed to be at least 0.5mA.

My homework: Study the LTP...