Small problem at power off

[nilrog]

Hi,

I have recently built Claus Byrith's EL34-PP amplifier, which is a modification of the old Mullard 5-20. The amplifier plays beautifully but there is one small thing that disturbs me.

When I power off the amplifier I can see the speakercones making a large movement in/out once. It doesn't hurt the speakers but my soul hurts from looking at it

I have placed capacitors over the power switch but it didn't improve the situation much. A friend of mine who built the same amplifier converted the filament power for the input and phase splitter tubes to DC and he no longer has this problem.

The amplifier is built according to the designers specs as a stereo amplifier (not separate mono blocks), which you can find at:
http://www.lundahl.se/claus_b.html

Unfortunately where I am now I have very limited possibilities to take measurements in order to locate the problem.

Does anyone have any suggestions on how to "fix" this thing?

Regards...
/Roger

[Sch3mat1c]

I don't see anything wrong with it. Usually a low frequency pulse is a sign of two things: a). LF instability (can be checked with a scope and frequency generator variable down to 1Hz) and/or b). weird power supply capacitance. Assuming you went with the full 110uF specified (i.e., 2 x 220uF in series, why? who knows, a single 450V will suffice), it should take at least several seconds to discharge.

Not only that, but this discharging voltage drop is balanced in the output stage, so for this to be the cause, it would have to be seen at the (unbalanced) input stage, which is even less likely because it has an even longer time constant.

Tim

[Tekko]

Like my amp when i turn B+ on. When i turn my amp off it just fades out for like 20 secs if music is playing but it does no pops or plops, it has 340µF in the B+, 100µF for the phase splitter, 100µF for the stage before the pkase splitter and 150µF for the input stage.

I think that Tim is right about LF instability in your case.

[nilrog]

Thanks for your input Tim!

Unfortunately I don't have access to a scope where i'm located at the moment

A swedish magazine who publish a modified version of Byrith's amplifier claimed that the original suffered from LF instability. Therefor they changed the blocking capacitor between the phase splitter and the output tubes to 100nF instead of 470nF. But they also used a different OPT, power supply and added back the compensation networks that was not present in Byrith's design.

The power supply in my amplifier currently looks like this:

110uF -> 10H -> 110uF -> Resistor -> 50uF -> Resistor -> 22uF -> 200V Zener

And the negative bias supply:

220uF -> Resistor -> 220uF

There is some slight differences between the schematic in the appendix and the one in the description when it comes to the value of the capacitors. But what I described above is the actual values used in my amplifier.

What puzzles me is that the same amplifier with DC filament instead of AC doesn't experience this "problem"

/Roger

[Tekko]

Could you fix a schematic ? Easier to see things that way. By the way, where do you live ?

[PRR]

> 2 x 220uF in series, why? who knows, a single 450V will suffice

If we are looking at this amp, with 350V HV winding, the no-load voltage (before the tubes warm) could be 590V (350V+1.414, plus 20% regulation). 590V is pretty hard on a 450V cap.

> When I power off the amplifier I can see the speakercones making a large movement in/out once.

Clearly the driver stages go way off-balance when the power goes away. That's quite common. If it isn't hurting the speakers, I'd ignore it.

> the same amplifier with DC filament instead of AC doesn't experience this "problem"

That could be a clue. Maybe not DC heat, but maybe enough storage in the DC supply to keep the driver tubes warm a little longer than the output tubes. If the output tubes go cold before the drivers do their their off-balance collapse, the driver thump won't be passed to the speaker.

> 110uF -> 10H -> 110uF -> Resistor -> 50uF -> Resistor -> 22uF

What is "Resistor"? When pondering subsonic instability, those decoupling resistor values are critical. Or for a quick-think: are the R-C products in the power supply at all similar to the R-C products in the signal path?

> claimed that the original suffered from LF instability.

0.47uFd and 390K coupling? Yowsa! That's 1Hz. The other pole in this feedback amp is the output tranny. Lundahl does not cite a rolloff, but low-THD 20Hz-30Hz push-pull transformers usually have BIG inductance to swamp core nonlinearity, and the small-signal frequency response often extends to... about 1Hz. Two poles at 1Hz is stable on paper but mighty marginal. A little extra ~1Hz phase shift in the power supply (that 10K-22uFd node is a suspect) can put it right at the edge of oscillation. Maybe not to motorboat, just enough to SLAM a BIG 1Hz pulse as it collapses.

[Sch3mat1c]

Yep, one amplifier I made (Williamson topology, pentode o/p) had an LF rise (like 6 or 10dB?) around 14Hz, the OPT was rated for 30Hz! Probably quite short on inductance... Coupling caps all the way to 0.047uF > 100k grid leak didn't help too much so I modded the topology. It's still got some rise, IIRC.

Tim

[Sch3mat1c]

Originally posted by PRR
> 2 x 220uF in series, why? who knows, a single 450V will suffice

If we are looking at this amp, with 350V HV winding, the no-load voltage (before the tubes warm) could be 590V (350V+1.414, plus 20% regulation). 590V is pretty hard on a 450V cap.

I didn't look closely at the HT winding... that much!? Well then why the heck does it say 400V in use, no SS rectifier I've ever seen is that crappy!

Tim

[nilrog]

Originally posted by Tekko
Could you fix a schematic ? Easier to see things that way. By the way, where do you live ?

The schematics are in the link I posted. The deviations from what you see there is described in my posts. At the moment i'm located in Uppsala, far away from "home"...

Originally posted by PRR
If it isn't hurting the speakers, I'd ignore it.

That's what i'm trying to do but it's not good for my heart
Since the amplifier is sounding and working great besides this minor glitch I would like to try and fix it :-)

Originally posted by PRR
> the same amplifier with DC filament instead of AC doesn't experience this "problem"

That could be a clue. Maybe not DC heat, but maybe enough storage in the DC supply to keep the driver tubes warm a little longer than the output tubes. If the output tubes go cold before the drivers do their their off-balance collapse, the driver thump won't be passed to the speaker.

Sounds like a reasonable explanation. I was thinking in the same direction that the DC caps could be keeping the driver tubes "working" longer. But working with tubes is a new experience and i'm still learning a lot.

Originally posted by PRR
> 110uF -> 10H -> 110uF -> Resistor -> 50uF -> Resistor -> 22uF

What is "Resistor"?

Sorry, for not including the values Here they are, adjusted from the schematics to deliver correct voltage for a stereo amplifier.

110uF -> 10H -> 110uF -> 4k7 -> 50uF -> 47k -> 22uF -> 200V Zener

Originally posted by PRR
0.47uFd and 390K coupling? Yowsa! That's 1Hz. The other pole in this feedback amp is the output tranny. Lundahl does not cite a rolloff, but low-THD 20Hz-30Hz push-pull transformers usually have BIG inductance to swamp core nonlinearity, and the small-signal frequency response often extends to... about 1Hz. Two poles at 1Hz is stable on paper but mighty marginal. A little extra ~1Hz phase shift in the power supply (that 10K-22uFd node is a suspect) can put it right at the edge of oscillation. Maybe not to motorboat, just enough to SLAM a BIG 1Hz pulse as it collapses.

From your thoughts it sounds like the power supply might be the "bad boy" here. The two slightly different schematics that Byrith presented are both calculated for a monoblock. But it said that you only needed to recalculate the resistors to obtain the correct voltages to use it to supply a stereo amplifier.

Besides changing the resistors there is also a slight difference for the two last capacitors in the power supply.

• The 50uF capacitor in the schematic you posted the direct link for stated that it should be 23.5uF (2*47uF) but the schematic in his appendix said 34uF (2*68uF).
• The 22uF capacitor in the schematic you posted the direct link for stated that it should be 47uF but the schematic in his appendix said 22uF.

Since I don't know enough yet about "poles" I cannot tell how this might affect the circuit stability. All your efforts so far is very much appreciated

/Roger