u73 cathode bias

[ioaudio]

i was wondering about the "ac" - cathode bias of the vari mu pentodes, marked red.

i´m also curious about the effects of tying the two screens (G2) together fed with 100k instead of feeding them through 5 - 10k with an additional balance trimmer, as seen in many other schematics.
(probably service reasons? - cathode and screen trimmer let alone all the other trimmers for aligning the compression curve)

big
manual pdf

also interesting: rohde&schwartz u23 pdf

 

[Michael Tibes]

[quote author="ioaudio"]i´m also curious about the effects of tying the two screens (G2) together fed with 100k instead of feeding them through 5 - 10k with an additional balance trimmer, as seen in many other schematics.
(probably service reasons? - cathode and screen trimmer let alone all the other trimmers for aligning the compression curve)[/quote]

I was wondering exactly the same, because I recently tried both and figured that on my units I could reduce the thumping a lot better with the U23 than with my U73b. This might be due to a better loend response of the U73b though, I didn't have time to investigate...

Michael

 

[gyraf]

The AC part of the cathode-to-ground return path is simply for canceling out hum by trimming the 5K pot - same as in some guitar amplifiers.

 

[ioaudio]

thank you jakob.
i´m familiar with the hum cancelling pot, but was wondering why it´s referenced to the cathode of the vari mu stage only. suddenly it seems clear, because it is the only push pull stage in the circuit.

 

[PRR]

> effects of tying the two screens (G2) together fed with 100k

Well, you know the effect. Vari-Mu happens by reducing cathode current in T1 T2. Cathode current splits into plate and screen current, and this ratio is fairly fixed. Screens sit at 40V from a 170V rail, so there is large drop in the 100K. Screen current falls, drop is less, screen voltage rises.

So as gain goes down, screen voltage goes up. Why?

The maximum undistorted input on a pentode is roughly Vs/Mu. Speaking over-simple, as we get big signals, screen voltage rises to handle them. It is a very real effect. You find a similar condition on many remote-cutoff tubes for IF and RF service.

Offhand, E99F Mug2g1 is 27, so we expect 40V/27= 1.5V to just-distort at idle, say 150V/27= 5V to just-distort in deep GR.

This is also true on triodes, and many classic triode vari-Mus have large plate resistors and sit at low idle plate voltage.

It is not that simple, and you can also get good results with fairly fixed plate or screen voltage.

> "ac" - cathode bias of the vari mu pentodes

The vari-Mu stage has the lowest signal, as low as a mike-amp (I mean a classic mike-amp, not a rock-studio input). Here we need two tubes with low heater-hum, low hiss, and good matching. By force-nulling the hum with +/-8mVAC hum from pot 66, we may relax the requirement for low heater-hum, and just try to find two matched low-hiss tubes.

It isn't exactly the same as gitar amps. The usual plan is you twist your heater wires 99% symmetrical, so their radiated hum field is canceled at nearby plate/grid parts, and then tune your "center" tap a few percent off-center to match your few-percent asymmetric twisting. The limiter is a simpler case.

There is more to the tale. Note that T1 T2 are fed one-side-grounded 6.2VAC. In most low-level work, that would be wrong. I bet these Germans laid-out the T1 T2 heater wiring so exACTly that hum in the two tubes cancels 99% perfect. Then pot 66 trims that last dab.

Pot 61 adjusts thump. It will affect hum, but that's not what you should adjust it for. You probably have to go back and forth from 66 to 61 to 66 a few times to get both balances good.

 

[ioaudio]

thanks prr :thumb:

also interesting is the use of the extra heater winding, which shows 6,2 instead of the 6,5 for the rest of the tubes. seems they cheated the max. grid resistance 7 release with lower emmision, similar to underheated tube mic circuits.

i think its a very interesting circuit. 4 tubes for a vari mu seems fair. the input, interstage and control output transformer seems to be the same ei48 model, and the output is known from the V72.

 

[ioaudio]

i measured the transformers t188
it´s 400 ohm dcr each side (very exactly balanced), 30 Henry, no gap.

so for the sidechain that would mean 400R * 0,1µF = 40 µsec for the attack time?

 

[PRR]

> 400R * 0,1µF = 40 µsec for the attack time?

Add the reflected resistance of primary and of cathode follower. I don't have data but I'd guess another 400 in winding, 1K for tube. So more like 0.2 milli seconds. Still quite fast.

My German ist poor, but doesn't page 10 of that big PDF say 0,5ms? (The R-C attack time is only the control voltage risetime. The tube gain is not directly proportional to voltage so the audio attack time is different.)

 

[ioaudio]

yes, it says <0,5ms as upper limit.
i totally forgot about the reflected source impedance on the primary of the transformer - thanks! primary is hooked up parallel, so it´s a probably bit lower.