cathode resistor ?? in p-p fixed bias

[richwalters]

I’ve noticed sometime ago, that the thd in an p-p UL output stage operating in fixed bias AB1 goes down dramatically when separate cathode resistors of around 10R are fitted to each tube without a decoupling cap. The thd reduction is more dramatic than trimming the negative bias volts even when each tube has a separate bias volt control. Some explanations mention the cathode resistor micro-trimming the tube currents with a view of reducing IM thd and altering the screen resistor has the same effect. I don’t do the last for obvious reasons.There does seem to be many circuits around with the cathode resistor fitted and some even mention risk to instability.
Can anyone come up with an explanation to why the thd decreases ?

richj

[aletheian]

My knee-jerk reply would be that it is from the bit of feedback from unbypassed cathode resistor, even if it IS a very small resistor... I don't know if it goes deeper than that.

[HollowState]

Aletheian is correct. Since the output stage is the largest contributor of distortion, any small amount of local feedback will help reduce it. This is why Audio Research runs their output tube cathodes through the secondary of the output transformer. I have been using 10 ohm resistors in the cathodes for many years, but mostly for a different reason. It's a convenient way to monitor tube current.

Victor

[richwalters]

Originally posted by HollowState
I have been using 10 ohm resistors in the cathodes for many years, but mostly for a different reason. It's a convenient way to monitor tube current.

Victor

Thanks all. In my example I'm using a 100W parallel p-p UL amp with 6550s, which was fitted with 1R resistors in each cathode leg. The thd at 1Khz at 100W is typically around 0.1% (which is already good) and with 10R resistors in each cathode leg it drops to around 0.06%. This is quite dramatic and as Vic mentions by also measuring tube current one kills two birds with one stone.
The thd at 10Khz is more problematic, the benefits are less noticeable as the output transformer parasitic losses come into play.
The next question is the optimum value of resistor for a particular circuit.

richj

[DigitalJunkie]

I once got into an e-mail conversation with the late Fred Nachbaur (may he RIP) about using a bit of cathode R with fixed bias.

IIRC,he had seem similar results,and suggested that I try from 10-50ohms in the cathodes,to give just a bit of local feedback,and that it would help the crossover distortion a bit. (I was prototyping a P-P EL-34 amp,using some guidelines from the Duncanamps site.)

I think he mentioned one project where he used about a 50/50 mix of cathode and fixed bias with good results...don't remember which one it was though.

[SY]

Search for "harmonic equalizer."

[ray_moth]

De Palma:

The output cathodes are tied to ground through a 10 ohm wire-wound common cathode resistor. The size of this resistor regulates output stage distortion at high levels. Distortion at high levels is reduced at the expense of a small increase in low level I.M. The 10 ohm value represents a compromise.

See his design for a PPP El34 amp at this link.

[richwalters]

SY...Ah!
..Ray thanks. I've found that with a Sowter made o/p tranny has the greatest effect with 10R res in each cathode leg whilst using a Majestic specimen (same size +power circuit, tubes + freq response etc) I get no improvement at all. So there is something magic in the o/p transformer going on with phasing and capacitive elements.
However, at this stage worth mentioning I get the best overall performance from the Majestic. Someone wrote in another thread that Sowter were the best.....Next.

richj

[kevinkr]

I too have noticed and measured the same effect, originally I used 33 ohm cathode resistors in my modification kit for the ST-70. I have noted the effect with 10 - 47 ohm resistors in PP amplifiers, This should also be the case in SE amplifier designs as well, but I have not measured it.
(My fixed bias SE amplifiers either have this resistor or don't - and in the case where they do I never considered shorting it to see if there was a difference.) I usually use 10 ohms as a matter of convenience for a little cathode degeneration and an easy point to measure the idle current.

Something to be aware of is that this approach does raise the effective plate resistance by rp + r * (mu+1), and carried to an extreme will reduce mu so you shouldn't get too carried away. The implications are obvious with higher mu output triodes, as well as the relationship between rp and primary z in the output transformer. (Reduced LF extension and increased distortion.)

[richwalters]

Kevin, I 've been suspicious about modern days manufactured tubes actually having a lower plate resistance than NOS spec. Swapping Tungsols 6550's for NOS 88's suggest the 88's are going soft or have a slightly higher rp. The ratio of the dummy cathode 10R compared to the plate resistance seems next to nothing but for AC degenerative on harmonics it isn't insignifigant.
Some time ago I modified an old amp design with also degenerative output stage cathode resistors (that design used 12BY7's for front-end and driver and it oscillated). I seem to remember it had alot of inverse plate to plate feedback and in any case was rather flighty design. Harmon C ? Citat ? not sure.
So adding degenerative cathode resistors in output stage tubes on some designs can have it's limitations regarding stabilty.

The old trick I use to check for stability is a clip a decade resistance box across the fixed value global nfb resistor, observing the existing amount of nfb already used and then start switching down high values and see how much extra nfb can be added before instability starts. The DePalma rule as shown in the(ray_moth) link on this thread, works out correct. i.e if a design has 20dB global nfb, one should be able to add another 10-15dB before instability takes off. Test freq 1Khz into res dummy.
I think it is a tough test. With the 10R resistors fitted to conventional non complicated amp circuits I didn't detect any worsening of stability, Hence the motto << keep everything simple >>.

richj