Squeal on shutdown

[famousmockingbird]

So I finished a build on a KT77 PP amp, same circuitry as a Mullard 5-20. The amp is dead quiet with the input shorted and sounds fantastic when given a signal. The only problem is after about 10 seconds of turning the power switch off an annoying squeal comes through the speaker for about 10 seconds or so.

I was thinking that I need to switch the speaker leads around (positive feedback instead of negative feedback) but wouldn't that squeal all the time not just on shut down?

 

[lhgin]

hmmm.... could it be some PSU caps that is not drained properly? maybe you could try a bigger bleeder resistor on your power supply caps? that will cause it to drain faster and avoid the problem altogether.

pls let us know if it works.

 

[famousmockingbird]

I increased filter and decoupling cap values. C15 is two 100uf caps in series with 100k balance resistors. The balance resistors act as a drain too right? C12 is two 50uf in parallel with no drain resistor. C5 and C4 are both 50uf with no drain resistors. Maybe my reservoir cap is draining quicker than the other caps?

 

[lhgin]

try a 50-100k resistor on C12. we would want to EL34 stage to drain faster than the others to prevent the audible noise.

 

[multi]

I think you must have an amp that is oscillating at a high frequency and when you turn of the oscillation is at a lower frequency. i had this problem with a Chinese made amp. was oscillating at 810KHz. When you switched it off had a big squeal. Tune an Am radio near the amp and you will pick up the signal, if its in the Am band.
Phil

 

[famousmockingbird]

I pulled the EF86 and the 12AX7 and the squeal starts immediately once the voltage comes up.

 

[famousmockingbird]

I switched the two output leads to speaker and that DID NOT help! lol

Also added a leak resistor to C12, no help.

I will try the am tuner tomorrow because I have to dig it out from storage.

 

[multi]

Not likely to be oscillation if you still get the problem with the EF86 & 12AX7 out.
Phil

 

[DF96]

Probably oscillation in the output stage. Quite common with UL. Add CR snubbers from g2 to anode.

 

[famousmockingbird]

Probably oscillation in the output stage. Quite common with UL. Add CR snubbers from g2 to anode.

I am new to UL output transformers but after reading up on the subject I believe you hit the nail on the head. Instability in the output causing the oscillation.

So the parts list calls for only half watt 1K screen resistors and I used 5 watt wirewound, would it be worth trying a different screen resistor that isn't inductive? Or is this plainly a transformer issue? I have some 2 watt metal oxide resistors and some half watt carbon film resistors in the 1K value. I only ask this because it seems that the snubbers will effect the high frequency response of the amplifier.

Also it looks like picking the RC snubber values is trial and error to get the least amount of high frequency roll off. What's a good starting point? 10k/500pf?

Thanks again

 

[famousmockingbird]

The CR are in series correct? And I would want to put the capacitor on g2?

I picked 10k/500pf because it had a cut-off frequency of 31,830hz. I need to order parts and will probably order a bunch of values to play with.

 

[richwalters]

Don't worry about the screen/anode dampers effecting the HF audio response...they won't. The issue is damping the leakage parasitics of the UL o/p transformer, i.e usually well above the audio range. In physics terms, the leakage inductance of the UL screen winding sections in relation to the anode winding section is quite high, and there lies the problem.
Rule of thumb, stick to 1nF & 1-2K between each screen tap and anode; both MUST be good HV quality and well rated types.
In an undamped LC circuit, between each push pull pair there is Miller interaction with the grid, high gm tubes will find whatever way to resonate with the leakage parasitics.
The evidence; this is the reason the lower gm of the 6L6 class were so immune to these problems.

Even if a particular circuit doesn't show any oscillatory symptoms, I have always been inclined as an insurance to avoid secret oscillation, to add the screen/anode RC combination. It's cheap and peace of mind.

richy

 

[famousmockingbird]

Don't worry about the screen/anode dampers effecting the HF audio response...they won't. The issue is damping the leakage parasitics of the UL o/p transformer, i.e usually well above the audio range. In physics terms, the leakage inductance of the UL screen winding sections in relation to the anode winding section is quite high, and there lies the problem.
Rule of thumb, stick to 1nF & 1-2K between each screen tap and anode; both MUST be good HV quality and well rated types.
In an undamped LC circuit, between each push pull pair there is Miller interaction with the grid, high gm tubes will find whatever way to resonate with the leakage parasitics.
The evidence; this is the reason the lower gm of the 6L6 class were so immune to these problems.

Even if a particular circuit doesn't show any oscillatory symptoms, I have always been inclined as an insurance to avoid secret oscillation, to add the screen/anode RC combination. It's cheap and peace of mind.

richy

Thanks Richy

So does it matter what order the RC snubber network is in? So for example in series
g2 --> 1nf --> 1k --> anode or g2 --> 1k --> 1nf --> anode?

 

[multi]

Also it looks like picking the RC snubber values is trial and error to get the least amount of high frequency roll off. What's a good starting point? 10k/500pf?
The values depend on the quality of your out put transformer; 1K 2 watt/1000pf. some transformers need 3K/4700 pf. The circuit i looked up had 1K/1000pf for medium quality.
The circuits recommended 1000 volt caps.
Phil

 

[Merlinb]

So does it matter what order the RC snubber network is in? So for example in series
g2 --> 1nf --> 1k --> anode or g2 --> 1k --> 1nf --> anode?

They're in series, so no it doesn't matter what order you put them in.

 

[DF96]

I used 1k+1nF snubbers on PP EL34 with a Sowter transformer. I used 1kV ceramics, IIRC.

Keep them near the output valve, so they don't introduce too much extra capacitive coupling back to an earlier stage.

 

[famousmockingbird]

They're in series, so no it doesn't matter what order you put them in.

Got it, thanks wizard.

I used 1k+1nF snubbers on PP EL34 with a Sowter transformer. I used 1kV ceramics, IIRC.

Keep them near the output valve, so they don't introduce too much extra capacitive coupling back to an earlier stage.

Well this is a PP EL34 amp (edcor iron though) so 1K + 1nF snubbers works for me, and yes I am going to mount them on the socket from the unused pin 6 over to the anode on pin 3. So pin 6 will have the 20% screen tap connected to it and then run the snubber network (1K + 1nF) to pin 3 and also run the 1K screen resistor to pin 4.

Thanks DF96 and also to everyone else's help and suggestions. I will let you all know how I make out.


I had the amp on the bench today on a dummy load and was measuring 1.5vac across the 8 ohm load (input was shorted to ground of course). So this is some inaudible oscillation going on because when I had it hooked up to the speaker it was dead quiet. Hopefully the snubbers do the trick so I don't lose any reserve power getting eaten up by some HF oscillation.

Thanks again to all

 

[Junk Audio]

Probably a lot of people here better at this then me, But I built this particular project before with Acrosound TO-300 output transformers and the amps have been running for years without a peep. When I first built these I had a problem and one of the oldest tricks in the book for building is to have a local friend that builds go over your work. I had two friends go over mine and my one friend spotted that all I did wrong was to forget to bond the output transformer negative. my point is, either have a friend go over it. Or take some good pics and have the guys here go over it. It may be something simple like mine was.

 

[richwalters]

Just in case not sure of the config..see pdf...

 

[famousmockingbird]

Thanks Richy for the PDF, it confirms exactly how I was going to implement the snubber network.


I don't have any friends that are into this type of stuff for a hobby so I don't really have anyone I can ask to take a look and double check my work but here is a couple pics if anyone is interested. Once I am happy with the amp I am going to clean up some of the wires so it isn't quite done yet.