ECL86/6GW8 Console opinions

[rotaspec]

Attached is a rough schematic of my old console amp that I use for my computer listening. It actually runs quite well after replacing the Cs and Rs and removing the tone control section (which was a convoluted arrangement in a feedback loop). Note there is no direct feedback loop now . I'm not going to push the envelope too much with this, as the tubes are almost unobtainable now.

The main reason for this message is to see if there are any glaring mistakes in the design (or my alterations to it; I don't have an original schematic).
Note that the triode section gets its anode supply from the pentode screen. Would there be any reason not to take them from the B+ point instead? It would make measuring screen and anode currents easier.

The output stage only runs 32mA anode current by my calculations of drop across the o/p trans. Sounds a bit low to me, but these are old tubes not NOS. I think it only gets to about 2W out into 8ohms before it starts to clip.

Thanks, Gary

[DougL]

The only thing I see is you left in the 330 pf cap that rolls of the high end to maintain feedback stability. You do not need it. Just adds to high frequency distortion would be my guess.

Doug

[rotaspec]

Yes, I noticed that today when I was drawing this up. I guess I should have removed it as the response is down 3dB at 10kHz (and 100Hz which I blame more on the small o/p trannys)

Gary

[DougL]

Note that the triode section gets its anode supply from the pentode screen. Would there be any reason not to take them from the B+ point instead? It would make measuring screen and anode currents easier.

The front end and screen grids both benefit from being isolated from the B+.
You could add a separate RC for the front end if you want.

The 1 Meg pot may be contributing to the HF roll off as well. There is no real reason to use anything bigger than 100K. Having said that, the Dinaco ST 70 had a 250k grid leak because the pentode front end shielded the input from the miller capacitance. If I were a betting man the input triode is high mu, and has a high voltage gain. Sorry, too late at night to look it up. That virtual cap probably interacts with the pot to roll off the highs.

Another thought is to replace the RC on the input with an LED string that gives you about the same voltage drop. It will eliminate an RC time constant cheaply may be better than most caps.

You may be right about the LF roll off being due to the transformer. It's that or the cathode bypass caps.

HTH

Doug

[aardvarkash10]

what is the 47k res at the input doing?

[planet10]

Originally Posted by rotaspec

as the tubes are almost unobtainable now.

Every time someone says that i'm surprised... my stash of these keeps growing.

dave

[rotaspec]

Originally Posted by aardvarkash10

what is the 47k res at the input doing?

That's a leftover from a divider network I thought I might need to match the computer output. It really doesn't do anything now.

[rotaspec]

Originally Posted by planet10

Every time someone says that i'm surprised... my stash of these keeps growing.

dave

Well, you can send some my way if you like

Seriously, the best price I have seen for these tubes NOS is about $18 USD each for an unnamed brand. None at any price available here in New Zealand.

If my tubes give out (I still have a couple of spare used ones), I'll rebuild the amp with ECC82/EL84 as there is room on the chassis.

Gary

[rotaspec]

Originally Posted by DougL

The front end and screen grids both benefit from being isolated from the B+.
You could add a separate RC for the front end if you want.

The 1 Meg pot may be contributing to the HF roll off as well. There is no real reason to use anything bigger than 100K. Having said that, the Dinaco ST 70 had a 250k grid leak because the pentode front end shielded the input from the miller capacitance. If I were a betting man the input triode is high mu, and has a high voltage gain. Sorry, too late at night to look it up. That virtual cap probably interacts with the pot to roll off the highs.

Another thought is to replace the RC on the input with an LED string that gives you about the same voltage drop. It will eliminate an RC time constant cheaply may be better than most caps.

You may be right about the LF roll off being due to the transformer. It's that or the cathode bypass caps.

HTH

Doug

The pots were a leftover from the original console, as I didn't have a dual gang one to replace it. Might have to hunt down a lower value one.

I'm still trying to get my head around the LED cathode bias idea, it's something new to me. As the cathode DC point is only about 0.8v on the triode section, can I just stick an LED in there on its own and lose the R & C?

Actually, I measured both channels, and one cathode is 0.8v and one is 1v DC at idle, and the anode voltages are 140v and 113v so I guess one tube is down a bit. Would the LED help correct that?

I trained on radio valve gear for a year, but that was 35 years ago, and haven't done anything since with them until now, so I'm re-learning things from scratch

[DougL]

I'm still trying to get my head around the LED cathode bias idea, it's something new to me. As the cathode DC point is only about 0.8v on the triode section, can I just stick an LED in there on its own and lose the R & C?

Short answer, yes.
Longer answer is that you need to pick a diode with the right voltage drop for your circuit. IIRC, the infer-red LED drop about a volt. you could also use small signal diodes for lower voltage drops. The point is a diode has a low dynamic resistance and a fixed voltage drop and looks much like a RC from AC and DC perspective but much faster overload recovery / less memory and does not have the Dielectric absorption issues of a capacitor. Its a drop in, parts are inexpensive, and its worth a try at some point.

Doug