63 Selmer Bassmaster Volume Drop

[Ted]

63 Crocskin Selmer Bassmaster

Sounds fine when clean but as soon as its driven hard the volume drops. On the bench with a signal generator, when signal is applied the B+ drops "bounces" a couple of times from around 500 to just below 400. - its visible on the AC supply to the rectifier.


A bit of circumstantial evidence: It still has its original power transformer. I've never seen one of these that still had the original transformer - they were fragile and have usually been replaced.
These amps are unusual in that the CT of the output transformer is taken from the clean side of the power supply choke.

I'm thinking power transformer. How to confirm. short of putting a new one in?

[Helmholtz]

PT problem is unlikely as you're getting full B+.
How does B+ behave when you gradually increase the signal?
Do you have a schematic?
Scope?

[loudthud]

Any relation to this amp? It has the Animal skin covering.

[trobbins]

Probably this link from the vintage Hofner website. With schematic link: https://www.vintagehofner.co.uk/selm...assmast50.html

[pdf64]

63 Crocskin Selmer Bassmaster

Sounds fine when clean but as soon as its driven hard the volume drops. On the bench with a signal generator, when signal is applied the B+ drops "bounces" a couple of times from around 500 to just below 400. - its visible on the AC supply to the rectifier.
...

I'd expect any competently designed valve rectified supply to behave similarly.
The protecting resistance required by the rectifier type was probably designed into the mains transformer.

[Ted]

Probably this link from the vintage Hofner website. With schematic link: https://www.vintagehofner.co.uk/selm...assmast50.html

That's the one. And look at the pic. The power transformer has been replaced.

[Ted]

I'd expect any competently designed valve rectified supply to behave similarly.
The protecting resistance required by the rectifier type was probably designed into the mains transformer.

I'd expect the voltage to drop on load. Maybe not that much though. What I wouldn't expect is for it to bounce back up to a higher voltage a couple of times and then settle back to the lower voltage.
But I think there's a pretty good chance that the amp was simply not designed to be driven into distortion. I always got the feeling that at that time British amp manufacturers were waiting for rock to go away so they could go back to selling sensible amps to gentleman jazz musicians.

[Helmholtz]

I'd expect the voltage to drop on load. Maybe not that much though. What I wouldn't expect is for it to bounce back up to a higher voltage a couple of times and then settle back to the lower voltage.

That's likely a low frequency LC filter resonance due to the choke being subject to power tube plate current.
A sudden jump in supply current will excite the resonance.
I think that kind of B+ filtering should only be used with class A amps, where supply current essentially stays constant.

Neverthess I would check the filter caps for capacitance, because the problem could also be the amp being at the edge of motor-boating, often caused by filter caps having lost capacitance.

[trobbins]

The effect could also be from output stage grid conduction - blocking distortion. The 220k - 47nF has a 15Hz corner. There are no grid stoppers in this circuit to alleviate that effect.

Wrt the CLC filter - PSUD2 indicates it likely does have a circa 15Hz damped ring to step load changes. Perhaps that could also interact with a bad case of blocking distortion.

Inserting grid stoppers may be a quick way of investigating.

Disconnecting, or adding series R to the feedback 100k may also help identify interaction.

[Ted]

That's likely a low frequency LC filter resonance due to the choke being subject to power tube plate current.
A sudden jump in supply current will excite the resonance.
I think that kind of B+ filtering should only be used with class A amps, where supply current essentially stays constant.

Neverthess I would check the filter caps for capacitance, because the problem could also be the amp being at the edge of motor-boating, often caused by filter caps having lost capacitance.

Actually the opposite problem. The filter caps have recently been changed for 47uF caps - (not by me). Which would slow down that resonance and make it more noticeable.

[Ted]

The effect could also be from output stage grid conduction - blocking distortion. The 220k - 47nF has a 15Hz corner. There are no grid stoppers in this circuit to alleviate that effect.

Wrt the CLC filter - PSUD2 indicates it likely does have a circa 15Hz damped ring to step load changes. Perhaps that could also interact with a bad case of blocking distortion.

Inserting grid stoppers may be a quick way of investigating.

Disconnecting, or adding series R to the feedback 100k may also help identify interaction.

I did try grid stoppers. Not much effect. But as mentioned above, the first two filter caps had been replaced with 47uF from the original 32uF which would have made that ring more noticeable. The voltage drop on B+ on load is also pretty large. (Scope pic shows Speaker output and PI supply from a stepped 1kHz signal. PI volts drops from ~400V to 280V - B+ was too high for my scope probe).


Transformer impedance is clearly part of this (that voltage drop is visible on a meter on AC from transformer), along with ringing from the CLC filter. All Selmers up to the 70s SV series were designed this way (OP CT after the choke) and I've seen it mentioned as the reason for their characteristic sound. Going back to manufacturers values of C in filter is probs a good first step....

[Ted]

Ah. OK. There is also an interaction with the bias. Biasing the EL34s hotter brings down and stabilises the B+ and damps the ringing. - as you'd expect,as running them this hot takes them closer to Class A with relatively constant current.. You have to run them close to 30W before it goes away completely so this is probably not the long term solution. But it probably also explains what happens to the power transformers of these amps.

[Helmholtz]

Ah. OK. There is also an interaction with the bias. Biasing the EL34s hotter brings down and stabilises the B+ and damps the ringing. - as you'd expect,as running them this hot takes them closer to Class A with relatively constant current..

Higher idle current means increased damping of the CLC resonance.

There might be another effect:

As the bias voltage is derived from the HT winding, the bias voltage will reduce when the HT voltage significantly drops, thus increasing tube currents
Not sure, but this might cause some positive feedback effect on B+ sag.

[trobbins]

Just to cover bases -

Did you confirm each EL34's cathode (or anode) idle current for balanced operation?

Have you tube swapped the rectifier (GZ34 ?) in case it has unbalanced anodes, or is starting to get weak or leak ?

GZ34 with 60uF on datasheet is certainly going to sag, depending on the PT winding resistances, and choke L1 will add to that depending on its resistance, and OPT primary half resistances will kick in as well, so maybe not abnormal operation (ie. no 'fault'). You could collect some winding resistance data, given you have the original PT, and I assume choke and OPT.

You could monitor the voltage across R39 to confirm it is fairly stable and nothing before the output stage is being an influence.