Luix Dual Channel EL34 50W Power Amplifier

[TUBEDUDE]

Forgot to mention, some of the parts replaced were KT-66, 5881, 6L6GC, 6550 and finally KT-88. All tested at 4k or 8k and I found i really prefer the sound of the KT tubes over all my NOS 6L6GC pairs. The KT's share a fuller richer sound than their American designed brethren. I wish I'd known earlier, when they were cheaper.

[luix]

Forgot to mention, some of the parts replaced were KT-66, 5881, 6L6GC, 6550 and finally KT-88. All tested at 4k or 8k and I found i really prefer the sound of the KT tubes over all my NOS 6L6GC pairs. The KT's share a fuller richer sound than their American designed brethren. I wish I'd known earlier, when they were cheaper.

I think that NOS tubes nowadays are warehouse fund, or worst, old production waste... this opinion can hurt somebody so take it as my personal opinion, I don't want to lit a flame with NOS lovers.
This is why I prefer to spend money for new production tubes, I know that they are far away from the golden age perfection but they are better than an old production waste.

So, even with your iteration method you can't get rid of the commutation spikes, this really makes me sad!

I'd buildt PCB for everything since I was in high school (AD 2000), i love to design PCB even if it is difficult to modify in case something is not working properly.
This time I really don't know what's wrong, PCB I've used are the same used for other perfectly working design.
Next week I'm gonna try two things:
1) I'll test only the PSU with the JCM800 preamp, I'll let you know if spikes appear.
2) I'll try to load the PSU, maybe with two tungsten lamp in series to simulate the EL34 absorbtion and of course I'll report here.

Thank you all guys!

[TUBEDUDE]

Sadly true, l acquired my NOS tubes from my time in the Army, 73-85, and working for a school district with lots of deserted tubes, as their PA, intercom, and other systems went sandy state.
In the early '90's I got several matched quads of Telefunken EL84's from Germany, but that has dried up.
I know there are still pockets of gold out there, but fools gold is prevalent for sure.

My spikes are about 100 mV, substantially smaller than yours. They are annoying at full tilt boogie, but with the master volume less than wide open, they are not noticeable. Still a reasonable signal to noise ratio. I just never had these in my all tube builds. The pulse is clean, no ringing, just rectifier generated. Further investigation indicated. Post 'rona.

P.S. as an aside, I used new KT-66 and KT-88 pairs. Russian tubes marketed as Gold Lions. As I never owned any of these NOS. Also, I've been using Chinese 5AR4's for the last 6 builds as I am down to just one Amperex 5AR4 in my personal amp. I know they were firecrackers years ago but they have figured it out.
Agedhorse was right, new production tubes are proving to be durable and toneworthy. The KT-88's sound as good and get as loud as my GE 6550A's.

[luix]

OK guys, commutation spikes are present in the JCM800 preamp, at the catode follower output, with the input grounded and the power supply loaded only with the mentioned preamp.
I'll play with the PCB to see if moving some ground connection will solve the problem.

Here are some scope screenshot:

DS1Z_QuickPrint6.png

DS1Z_QuickPrint8.png

[TUBEDUDE]

My spikes are normal switching. At 50mV the fall time is 100mS. Yours appear to be spike generated ringing.

20201221_123454.jpg

[luix]

My spikes are normal switching. At 40mV the fall time is 100mS. Yours appear to be spike generated ringing.20201221_123454.jpg

Yes, for this reason I was thinking that those spikes comes from the transformer hi leakage inductance coupling the current pulses of solid state rectifier to filament or to bias winding... this theory seems to be proved by the pulse disappearing using a second transformer for separating the high voltage from the other voltages.

BUT! Is it possible that two different transformers manifacturers have done the same errors???

I don't think so, I'll modify the PCB grounding scheme... I've used this PCB at least two more time without any problem!
The only way to attenuate this ringing is with a generous snubber, 1uF+100ohm that at 50Hz means about 3.3kohm in parallel with the HV secondary!

I'm really getting crazy!

[luix]

Ollè! I Won!

The solution is, 470 ohm trimmer between the filament supply, center pin of the trimmer to 60V given by the classic resistive voltage partition.
The trimmer is setted really unbalanced towards one of the heater wire, this cannot be possible using the transformer center tap.
A small snubber on the HV secondary, 10nF+100ohm, is the cherry in the Martini, the spikes are untriggerable and the noise is much lower.

DS1Z_QuickPrint9.png

DS1Z_QuickPrint10.png

Now I'll modify the PCB, update all and get them from JLC.

[TUBEDUDE]

Congrats Luix,
Good creative troubleshooting. I always pump D.C. into the center tap to get rid of noise, but this balance and D.C. combo was some inspired thinking.

[luix]

I've never used the trimmer solution, this is the first time of my life!

I swear I'll use it for every amp I'll build in the future.

Also the listening test is passed, with the trimmer setted by oscilloscope there is only white noise, the tube thermal noise.
Moving the trimmer the horrible buzz can be heard again.

[brewdude]

To clarify, the noise was from the heater supply, and you fixed it with a “humdinger” balance trimmer/pot as an artificial center tap and elevated the center reference from chassis ground with dc voltage. (?)

[luix]

To clarify, the noise was from the heater supply, and you fixed it with a “humdinger” balance trimmer/pot as an artificial center tap and elevated the center reference from chassis ground with dc voltage. (?)

Yes, I should assume that this is correct.
The noise is from the solid state rectifier as the pulses are spaced by 10ms (100Hz) and get coupled with the filament in some way.
With the heater center tap to 60V the noise is still there, using the trimmer (or humdinger) the noise is almost gone, the final cut is the reasonable RC snubber on the HV secondary, 10nF+100ohm means a load of about 320kohm at 50Hz.

[TUBEDUDE]

Did this also solve the issues with the recorded tone?

[luix]

Did this also solve the issues with the recorded tone?

Nope, because the recorded tone was compressed at the source.

The recorded tone was taken using a recorded clean guitar since I'm really crappy in guitar playing.
The soundcard used to record the clean guitar has some kind of compression even using the instrument channel, probably the guitar signal is too strong.

The chain was: [clean recorded guitar played from a PC] -> [cable] -> [Amplifier] -> [Speaker] -> [SM57] ->[lexicon alpha] -> [PC]

You'll have to wait to have some other samples

[luix]

Good afternoon and good new year to all!

I'm gonna buy a PCB set for this amplifier from JLC, is any european guy interested in one set of this?

JLC only print 5 items for each PCB so the other 4 are useless for me (at least in short period), if anyone needs a PCB I'll be glad to sell them without charge, only PCB cost + shipment to your country.

The complete PCB set is composed of 6 PCB, JCM800 preamp, EF86 preamp, gain recovery & PI+final stage, power supply, bias supply with humdinger , relay switch and cost about 7 euro.

For italian shipment the price can be around 10 euro, even less if the weight is not over 250gr.
For european shipment the cost can be 20 euro, I'm not sure, I'll verify.

Is anyone interested??