Mullard 7 watt clone with oscillation problems

[Baronkarza]

hi everybody,

I've build a 7 watt stereo mullard clone but i changed the first stage for less gain.
A 6s2s tube instead of 12Ax7.
Naturally I changed to 27k the anode resistor,and to 1k the cathode resistor, still bypassed by 22uF.
The global feedback has a 3k3 feedback resistor paralleled with 600 pF capacitor, 330R tail resistor.

Here the link to the schematic, not up to date with some values:
https://1drv.ms/b/s!AlelTeAIElpOhrU_IscJg3ELHqFE2A
Hope it works.

(Now the 1000pF is in parallel with 27k anode resistor, 8k2 instead of 3k3 feedback resistor, 1k screen resistor. A 1M grid leak resistor before the grid stopper.)

The amp do not behave, so I'm trying to debug with a friend of mine.
We had a wrong grounding problem in the power supply, that caused a variable ripple on the supply line, but we solved that. Or we think so.

Now the amp seems to oscillate in the preamp stage. If we inject a signal at the phase splitter input (without preamp tube and obviously, global feedback), all is good. The amp amplify without strange noises or instability.

With the preamp tubes in, the damn thing have a weird response.
Injecting 2Vpp in (1kHz), the output, scoped with a Tek 2246, showed a bizarre phenomenon of "resonance".
Multiples waveform over the main one. We suppose this is an oscillation in the audio frequencies and over.
Trying with a disposable speaker, crackle,and sweeping to high frequence sound.
Modding the feedback resistor value ( first was 3k3 , now 8k2) and not changing the capacitor (600pF), will change the symptoms. At first seem to work, just at low volume. When reaching 2/5 of volume, the right channel waveform disrupt, and after few seconds, the left channel collapse too.

We plan to do more tests this weekend, and take some screenshots of the scope. Do You have any suggestions on what things check, or any other tips? We suspect oscillation due to layout or change of first stage from the original.

We have some experience in building tube amps, have DMMs, scope, bulb limiter and (cheap) signal generator.
We also know the basics to not electrocute ourselves

Thanks in advance.
Emiliano

[nickb]

Can you post the original Mullard schematic? Edit: Is this it?




In the meantime remove the feedback resistor (8.2k) and see if it's stable.

Edit: If that fixes it replace it and try change the 1nF across the 27K to 5.6nF (corrected as I had not noticed the original has 2.2nF rather than 1nF). That will move that pole back to the right place to compensate the loop.

[Baronkarza]

Hi,
thank you for the answer.
Yes, the original mullard schematic is that one.
With a 3.3nF , and a Zout of 9600 ohm, I think i would have a cut-off of 5kHz.
I'll try in the weekend.
If you need scope traces, photos, layout, i'll post in the next days.
I'll post the update schematic and layout too.
Thank you again
Emiliano

[nickb]

Hi,
thank you for the answer.
Yes, the original mullard schematic is that one.
With a 3.3nF , and a Zout of 9600 ohm, I think i would have a cut-off of 5kHz.
I'll try in the weekend.
If you need scope traces, photos, layout, i'll post in the next days.
I'll post the update schematic and layout too.
Thank you again
Emiliano

When I posted I did not notice the original uses a 2.2nF. That means you'll need about 5.6nF with the new stage. Yes the pole (about 2.6kHz) is well inside the passband but that is how the compensation works. The NFB will increase the bandwidth.

[Baronkarza]

This is the link to the update schematic ( gen_feb_2019)

https://1drv.ms/b/s!AlelTeAIElpOhrVB_7vuBNecXF7blw

[Fazer]

Hello everyone, I'm the other guy in this project...

Here some pics of the inside, to get a look at the layout:

https://1drv.ms/u/s!AlelTeAIElpOhrVKnUrzJPwnIuSphg
https://1drv.ms/u/s!AlelTeAIElpOhrVJlxyWZveigjDwqA
https://1drv.ms/u/s!AlelTeAIElpOhrVIJesnAwjjoLbPZg
https://1drv.ms/u/s!AlelTeAIElpOhrVFoO8BxywSqsq3gw
https://1drv.ms/u/s!AlelTeAIElpOhrVHnckNl1K15OfcIA
https://1drv.ms/u/s!AlelTeAIElpOhrVG1aeOQsSjXo3RUg
https://1drv.ms/u/s!AlelTeAIElpOhrVDLzAciqWShi1FNA
https://1drv.ms/u/s!AlelTeAIElpOhrVCvrvPK_wKYHPcpg

And one top view:

https://1drv.ms/u/s!AlelTeAIElpOhrVEk0eOw1xEYZKvyQ

The Layout:

https://1drv.ms/u/s!AlelTeAIElpOhrVL02Q2y-AYuWAABQ

Seems I can't visualize previews, but the links should work.

Regards,
Giampaolo.

[Baronkarza]

Hi,
Result of the weekend tests.
Scope probe were 10:1. Input signals 2Vpp at pre input, 1Vpp at phase inverter.

Phase 1 without feedback (disconnected at the bottom of cathode resistors. Or we should have only removed the 8k2 resistors, leaving the 600pF caps connected?)

This is a video of the amp without FB, with all tubes in, Vin= 2 Vpp at the input in both the channels:

https://1drv.ms/v/s!AlelTeAIElpOhrVWuy5lfDLhJF9XWg

So we tried to remove both 6s2s.
Right and Left Channel with a 1Vpp input at phase inverter were stable. We sweeped the signal at audio frequencies, no troubles.

Phase 2 with feedback connected and 5.7nF across 27k anode resistors. All valves in.

Right and Left Channel weren't stable, raising the volume made the waveform distort more. See photos



This with right Channel only



This with Left Channel only

https://1drv.ms/u/s!AlelTeAIElpOhrVly59okGf2eJY_HQ

With Right Channel only and 6s2s see photo 6 raising volume at photo 7

https://1drv.ms/u/s!AlelTeAIElpOhrVmt9d2dFoeJdmM5A
https://1drv.ms/u/s!AlelTeAIElpOhrVnbK4MZjsygSwyyQ

With Right Channel only no 6s2s, Vin at phase inverter, lowering the signal at frequency generator, the waveform deteriorated. ( see photo 8, 9)

https://1drv.ms/u/s!AlelTeAIElpOhrVoRUXmdNmh_bsliA
https://1drv.ms/u/s!AlelTeAIElpOhrVpthkmIyq5UM22yw

This with Left Channel only, same as Right channel, no problem at all

https://1drv.ms/u/s!AlelTeAIElpOhrVqr0F2k_pK1SMa8A

So, there is a problem with right channel, probably oscillation, but maybe something different. We tried to move wires around with a chopstick, but without any changes. Maybe a fault ground link? We are running out of ideas, any suggestion?

We will have time to make more test next week (middle), have a nice weekend, thanks for now.

Emiliano.

[nickb]

You need to get it stable without feedback for starters. To be honest, the fragments of the layouts posted make it very hard to see what the layout is really like. We just need one clear hi res image of the entire underside.

Looking at the layout sketch you seem to have the output transformer at the opposite end from the ECL82 and so run the primary wires overthe tag board. That seems to be asking for trouble. I would have put the OPT near the ECL82 and kept the primary wires well away from anything else. You can try twisting the primary wires and re-routing them away form anything else, perhaps consider screening them.

If that brings no joy, start adding grid stoppers everywhere.

Only once you have it stable can you tackle operation with feedback.

[Baronkarza]

Hi nickb,
The primaries are ( the opt have UL tap 10k:8R):

black ( B+) : goes to power board supply.
red ( screen) , yellow( anode) on the first 6F3p ( at left side of left tag board)
blu( screen),green (anode) on the second ( at right side of left tag board)
blu, green on the third ( at left side of right tag board)
red, yellow on the fourth (at right side of right tag board)

The secondaries ( White, Brown) are twisted and goes to bananas ouput jack.

The other colors are :
Anode : red
Cathode: Yellow
Grid: Blue
Ground : Black
Feedback: Blue
Grid input 6s2s ( 10k grid stopper): red, blue

I've tried to run the wires 90 degree and on different height in the problematic areas.

I think the black wires ( B+), that goes directly to power supply, i can left where are .

I think to to put all the other wires ( anode, screen) together inside an elevated ( with some stand off) aluminium bar, (elec. shielding), left side of left tag board , right side of the right tag board. (With some terminal solder lugs, if the wires are too short from the 6F3P )
Do i have to twist them all together ? Or by couple ? Anode-anode ? Or anode-screen ?

Thanks for the answer , i'll try the next week.

Bye
Emiliano

[nickb]

Twist opposite phase signals together i.e anode/anode and screen/screen (or all of them).

The center tap is not an issue as it does not have a large voltage swing.

[Fazer]

Full inside picture, for layout reference:

https://1drv.ms/u/s!AlelTeAIElpOhrVtNRmgneA2yGP5MQ

The feedback wires are disconnected for now, and terminated (blue wires, top of pics).

I also noticed the blue wire to the screen resistor of the right channel is about 1 cm near to the cathode of the 6S2S:

https://1drv.ms/u/s!AlelTeAIElpOhrVuVRiZGWFWSU0qnw

Could be this part of the issue? At the left channel, the wires are far away.

Thanks,
Giampaolo.

[nickb]

The cathode is fairly low impedance so I doubt that 1cm is much of a problem. Certainly take care of the primary wires. I suspect you may have grounding issues as I con't really see what philosophy you've used. Are you aware of star grounding? Is the wire to grid of the 6S2S screened?

There are many eagle-eyed people on this forum who are much better than me at spotting things now that we have a clearer layout view in post #11.

Here for example the path for the output current is shared for the input signal. This could well be causing problems. I really don't like the layout at all. The OPT is at the opposite end from the ECL82's. The components for the ELC82's are also at the other end of the board from the ELC82s. It almost as though the layout was chosen to maximize wire lengths rather than minimize them. There seems to be just a single ground for the tag board, that's not optimal.

[Baronkarza]

Hi,

yes, moving the cathode wire was irrelevant. Moving around the PI anode wires, messed the waveform a lot. So, next step is twisting and shielding the OT primary wires and increase separation of the anode wires.

Then if we see improvement, we will look at the grounds issue. The idea is a mixed bus-star ground.

Regards,
Emiliano.

[Fazer]

We are making progress.

After twisting and shielding the OT primary wires and moving the PI anode wires in a different spot, both channels are stable without feedback.



Right channel has some less gain, but that is a problem for later. Probes are 10:1.
Connected a pair of disposable speakers, no crakles, pops or strange noises. There is some 100Hz hum, so we will rewire the grounds to a better scheme.

Next step is reconnecting feedback and adjust if needed for stability/gain.

Regards,
Giampaolo.

[nickb]

We are making progress.

After twisting and shielding the OT primary wires and moving the PI anode wires in a different spot, both channels are stable without feedback.

[ATTACH=CONFIG]52945[/ATTACH]

Right channel has some less gain, but that is a problem for later. Probes are 10:1.
Connected a pair of disposable speakers, no crakles, pops or strange noises. There is some 100Hz hum, so we will rewire the grounds to a better scheme.

Next step is reconnecting feedback and adjust if needed for stability/gain.

Regards,
Giampaolo.

That's progress. I see there is a gain AND a phase difference between channels.