Strange distortion on Music Angel Amp.

[yestertech]

This is a 'modern' stereo valve amplifier using EL34s in Push Pull.
I can't quite get my head around the distortion visible on the output of both channels.
(scope trace attached )
The trace is clean to about 8 watts and then slowly takes on the shape seen here as the level is increased. These were taken at around 15 watts. (4 ohm resistive test load )
It affects both channels so my first thought was PSU voltages amiss. I have no schematic but anode and screen volts seem OK on the output stage. The driver and phase splitter voltages also seem reasonable and are pretty similar on both channels. Pre amp and driver valves test OK.
The EL34s run in fixed bias mode. I have no details for the Iq value, but they measure around 37-39mA across the 4 EL34's (measured across the small cathode resistor)
At around 15 watts this increases to c. 45ma for each valve.

Any thoughts on what to check next ?

Andy

[John_BS]

Andy: do you think the topology is similar to this?

The waveform is indicative of gross second harmonic distortion, i.e one valve's output much lower.

If your amp uses the same trick for phase-splitting it may be that pot which sets up the drive balance. Or one valve on one side.....

John

[PJL]

This is one of the anodes?
Have you put the scope on the phase splitter?

Is it this one? https://www.diyaudio.com/forums/tube...l-el34-pp.html

[yestertech]

Originally Posted by John_BS

Andy: do you think the topology is similar to this?

The waveform is indicative of gross second harmonic distortion, i.e one valve's output much lower.

If your amp uses the same trick for phase-splitting it may be that pot which sets up the drive balance. Or one valve on one side.....

John

Thanks for that - the topology is different ( 1/2 ECC85 input valve followed by 6SL7 each channel ) I think its a split load phase splitter - still trying to trace it out but I can't easily see a balance pot. EL34s are JJ but I may swap them out just to make sure. I'm not above suspecting the O/P transformers at this stage, but both channels failing seems unlikely.

A.

[yestertech]

Originally Posted by PJL

This is one of the anodes?
Have you put the scope on the phase splitter?

Is it this one? https://www.diyaudio.com/forums/tube...l-el34-pp.html

Not scoped phase splitter yet ( I need to trace out the circuitry ) I will have to undo the feedback loop to avoid confusing results. Just seems odd that both channels present exactly the same fault ( what's the probability of that if voltages look OK ? )
Unfortunately I can't see the image in that link as I've lost my sign on details.
The amp is a Music Angel XD-SE signature and the only schematic I can find contains lots of errors.

A.

[PJL]

This one?

[PJL]

Maybe someone has 'rolled' the valves as the ECC85 sounds very wrong. It is also very possible it has been 'modded' and you may need to check bias conditions are correct.

[yestertech]

Originally Posted by PJL

This one?

yes thank you, that looks pretty near give or take a few resistor values.
I can see from the layout that its a long tail pair phase splitter ( 39K tail resistor )
The resistor linking the 2 halves of the phase splitter is only 560K but I can't trace a connection from the 'far' end to the second grid of the 6SL7.
As to tube rolling, ECC85 is printed on the chassis, but I agree it sounds somewhat unsuitable.
Attached is the trace of the drive signal to one pair of EL34s, feedback removed and then EL34s removed just to be sure. Scale is 2 v /division. The other channel is producing around 10 watts at this point (with distortion )

A.

[PJL]

Go back a stage further and look at the anode of the ECC85 (without feedback) and...have you have checked the signal generator?

[yestertech]

Originally Posted by yestertech

I can see from the layout that its a long tail pair phase splitter ( 39K tail resistor )
The resistor linking the 2 halves of the phase splitter is only 560K but I can't trace a connection from the 'far' end to the second grid of the 6SL7.

A.

The connection was actually there, they seem to have interposed a 10k grid stopper between the ECC85 anode and the g1 pin of the phase splitter.
The sig. gen output is constant and faultless as I'm using the volume control on the amp itself. Interestingly, although the chassis is stamped ECC85 the PCB states '12AT7' Fitting this valve drops the gain somewhat but the problem remains. I will check the output of the ECC85 stage with the 'scope. I can't help feeling there's something very odd going on here...and yet everything appears to be 'functioning'

[PJL]

The ecc85 is DC coupled to the long-tail pair and that spells trouble for swapping valve types. What voltages do you get on the 6SL7 anodes, are they roughly the same?

[GrimJosef]

Originally Posted by PJL

The ecc85 is DC coupled to the long-tail pair and that spells trouble for swapping valve types ...

I was worried about this too. The wrong first-stage valve, or even a sick correct valve, could easily shift the operating point of the phase-splitter pair a very long way from what's optimal.

Cheers,

GJ

[Diabolical Artificer]

That looks like classic core saturation usually present at low frequency's. What frequency was the IP - 1khz?

If I had this and it's not core saturation, I'd check my sig gen and all terminations to/from the amp. If they're all ok I'd check the OPT.

Andy.

[yestertech]

That looks like classic core saturation usually present at low frequency's. What frequency was the IP - 1khz?

Yes 1KHz
I initially thought about OPT problems, but as the drive to the EL34s is also distorted I've discounted this for the moment. ALso what are the chances of them both failing ?
At around 2 W output the OPT response is exceptionally good, being virtually flat from 10 Hz to over 40KHz !!

A.

[yestertech]

Originally Posted by PJL

The ecc85 is DC coupled to the long-tail pair and that spells trouble for swapping valve types. What voltages do you get on the 6SL7 anodes, are they roughly the same?

They were approximately the same on the initial check. I will report back with an annotated version of the schematic in post 6

A.

[TonyDuell]

Unless I am missing something very obvious, there's no way a 12AT7 (= ECC81) and an ECC85 can work in the same socket. The heater voltages are different.

The ECC85 has a 6.3V heater between pins 4 and 5. Pin 9 is an internal screening plate between the 2 triodes, normally grounded.

The 12AT7 has a 12.6V heater between pins 4 and 5, with the centre tap to pin 9. Conventionally you either run it at 12.6V between 4 and 5 (and leave pin 9 open) or you strap pins 4 and 5 together and run the heater from 6.3V between there and pin 9.

So :

If the unit is designed for an ECC85, then, if the heater supply is is floating wrt ground or centre-tapped wrt ground than a 12AT7 will run at 3.15V across each heater

If the unit s desgned for an ECC85 and one side of the heater supply is grounded then, assuming pin 9 is also grounded, one heater will get 6,3V, the other will be effectively shorted out.

If the unit is designed for a 12AT7 with the heaters in parallel for 6.3V then an ECC85 won't light at all.

If the unit is designed for a 12AT7 with the heaters in series for 12.6V then an ECC85 will be run at double the correct heater voltage.

[snowman_al]

As Tony says, the ECC85 and ECC81 cannot be directly swopped for the reasons given...

Then looking at the schematic the phase splitter is given as a 6N8p which is a Russian 6SN7GT equivalent not a 6SL7GT as mentioned in the text. A typo perhaps?

[PJL]

Lack of headroom on the long-tail pair? The 6SL7 anode voltages need to be around midway between the cathode and supply voltage which is in this case determined by some Zeners.

Have you checked the Zeners are OK? The original circuit shows a 12AU7 (ECC82) and they have very different characteristics compared to the ECC85. What are the values of the anode and two cathode resistors of the ECC85?

PS: The circuit on post #6 says 12AU7.

[yestertech]

Here is the schematic from post 6 with some annotations showing voltages and actual resistor values. Note that this has quite a few differences. The voltages measured on the other channel are within about 5%
The 6SL7 anode resistors are 75K paralleled with 100K ( 42K approx)
The output stage is operating in pentode mode.
Nothing looks to be unoriginal or modified. The valve types are all stamped on the chassis.


Seeing the schematic shows a 12AU7 input stage, I fitted one of these just to compare. (NOS CV491) The output waveform is much cleaner and more uniform, but is still quite 'rounded' once it passes around 10W.
More shocking is the measured distortion - THD at 1KHz and 5W is showing around 3% increasingly quickly as the power output approaches 20W.

I'm now wondering how this has ever worked 'properly' with the valves and components fitted, given the performance.

A

[Aub]

Should there be some decoupling where the 1.5k resistor meets the two 42k resistors?

Aub