Cathode Bypass Capacitor Doing Nothing!?

[jazzcrisis]

I'm experimenting with some push-pull tube circuits and have a Collins 6R built up on the bench. No matter what value cathode bypass capacitor I use, I can neither measure nor hear any effect on gain or frequency response when comparing bypassed versus un-bypassed cathodes. The stock circuit uses 20uf capacitors but I've also tried 470uf and 47uf which are oriented capacitor positive to cathode pin and negative lead to ground. I've tried cap vs. no cap on both tubes, and each one separately. Measurements with an HP 8903B audio analyzer show the exact same frequency response + gain and signal level doesn't have an effect either. I also played some music through the circuit and did a null test with capacitor connected and disconnected--result was nearly a full null.

What gives? Are there operating conditions or circuit topologies that will make cathode bypassing ineffective?

 

[abbey road d enfer]

The output stage is a class A push-pull. If well balanced, the current across the common cathode resistors (R108-R109) is constant so there is no voltage variation to "bypass".
There may be a difference when the output stage is driven hard.

 

[gyraf]

What is the line named "M" for? If left floating I can see cathode-to-cathode crosstalk if no capacitor there

/Jakob E.

 

[ruffrecords]

It would only affect gain at very low frequencies. Possibly other circuit elements dominate in this region and the effect of changing capacitor values is masked.

Cheers

Ian

 

[abbey road d enfer]

What is the line named "M" for? If left floating I can see cathode-to-cathode crosstalk if no capacitor there

/Jakob E.

Yes, I wondered about it. At first I thought it offered a possibility to run the output stage at elevated current for higher output, but seeing it could short the 2.04 ohms resistor, it would do nothing.
In terms of NFB, the polarity of the common-cathode signal is unknown and even erratic, which is not a good omen for NFB. In addition the cathode signal is attenuated 50dB by the voltage divider R108/R109 and furhermore by the cathode capacitor of the 1st tube.
Maybe some way of monitoring the output stage current, but I don't understande the 1k resistor from the 1st tube cathodes then.

 

[emrr]

M is for an external cathode meter. I’m not sure this is drawn correctly. I’ve had one of these, still have a variation of it I can look at.

 

[jazzcrisis]

M is for an external cathode meter. I’m not sure this is drawn correctly. I’ve had one of these, still have a variation of it I can look at.

I don't doubt it! Check out the typo on the second tube designation. I would be interested to see what's different in the real thing.

The output stage is a class A push-pull. If well balanced, the current across the common cathode resistors (R108-R109) is constant so there is no voltage variation to "bypass".
There may be a difference when the output stage is driven hard.

Interesting. Thank you! That makes perfect sense now that I think about it. I'm guessing Collins didn't know this and included their 20UF capacitors just because? I just tried every signal level from whisper to square wave meltdown with and without cap and output level doesn't seem to affect anything. Also checked distortion and FFT spectra... no change.

Here is the reason I was attempting to change bypass caps... there's about a 5dB frequency response droop from 10K to 20K and I wanted to try a very small value bypass capacitor. Without having to add NFB, does anyone know any tricks for flattening that response a bit?

 

[emrr]

Collins knew everything and didn’t use unnecessary risky parts like 1940’s electrolytics! They were bouncing the first signal off the moon not long after this.

 

[Tubetec]

Maybe its on the asymetric peaks the capacitor has its effect and not so noticable on steady state sine waves where the clipping is symetrical and very little out of balance signal exists .

 

[ruffrecords]

Here is the reason I was attempting to change bypass caps... there's about a 5dB frequency response droop from 10K to 20K and I wanted to try a very small value bypass capacitor. Without having to add NFB, does anyone know any tricks for flattening that response a bit?

Typically a Zobel network of the output transformer secondary would achieve this.

Cheers

Ian

 

[Tubetec]

Get a good sample of a snare rim shot , inject it into the pre-amp so clipping occurs ,audition the results with and without the cap , flip the phase of the input and see if that matters .

 

[abbey road d enfer]

Here is the reason I was attempting to change bypass caps... there's about a 5dB frequency response droop from 10K to 20K

I guess this piece of equipment was designed for AM, where the maximum frequency is about 5k (sometimes 10k)

and I wanted to try a very small value bypass capacitor.

That may work more or less on the 1st tube.

Without having to add NFB, does anyone know any tricks for flattening that response a bit?

Following Ian's suggestion, you may want to try a capacitor across the primary of the output xfmr. Not sure if it works because it depends on the xfmr's damping.
Beware there's high voltage there so dimension the capacitor accordingly.

 

[emrr]

Worst case I measured in an original was -1.3 @ 16K for 600 ohm input. On the bridging inputs (main usage) -1 is about 30K for one loading condition, -1 about 40K for another. I don't have the notes with me to say what condition, would assume better with 600 ohm load versus other bering 10K. The cathode balance control makes a big difference in low end response. Best I saw was -1 @ 31Hz.

Quoted is +/-1 30-15K. All that is entirely transformer dependent. Whatever transformers are being used are the culprit. What are they? Solving their problems is separate from the existing circuit.

There's not really much of anything professional after 1934 that's -5 from 10K to 20K.

The manual says the meter should be 100R internal resistance and 1mA full deflection.

Someone last week pointed out an error on another Collins schematic from the same period, someone not great in the drawings dept.

This piece is post-war, in the 1946 catalog, made through the early 1950's.

 

[Walter66]

Collins knew everything and didn’t use unnecessary risky parts like 1940’s electrolytics! They were bouncing the first signal off the moon not long after this.

Yeah, think so. They were the absolute experts on their field of technology. Love their tube gear. At least, a bypassed cathode resistor on the output stage reduces output impedance, if not reducing gain at all.

 

[emrr]

What is the line named "M" for? If left floating I can see cathode-to-cathode crosstalk if no capacitor there

/Jakob E.

Look at the RCA 41-B for an even wilder example of 2 stage shared cathode resistance. It's the bottom of the two in my drawing, simplified from that hard to follow manual drawing.

https://live.staticflickr.com/4585/38441766472_38b5fb8979_h.jpg

https://groupdiy.com/threads/rca-41b-analysis-and-test-results.81131/
There’s a weirder thing in an RCA/NBC compressor where the PP output stage and the side chain gain tube are off one cathode R stack. That one seems to be a remnant from the early 30’s approach of having a series resistance ’tree’ from B+ to ground as a crude method of voltage regulation, back when it was batteries and unpredictable AC supply.

 

[Tubetec]

Theres nothing there by accident , typical of Collin's bomb proof layout .

 

[aurt]

Theres nothing there by accident , typical of Collin's bomb proof layout .
View attachment 103373

Nice.

Could they have put the cap in to lower noise when adjusting the cathode balance?

 

[Mr. Moscode]

What is the line named "M" for? If left floating I can see cathode-to-cathode crosstalk if no capacitor there

/Jakob E.

Bias supply. Puts the cathode voltage above the grids.

Edit - I missed the post about a cathode current meter. I withdraw my answer. It may have helped to have the whole schematic.

 

[Mr. Moscode]

Nice.

Could they have put the cap in to lower noise when adjusting the cathode balance?

Yes, probably hum reduction. Also to decouple the first stage cathodes from the 2nd stage due to a common bias supply.

Edit - now that I know it's a meter circuit, I withdraw hum reduction. Decoupling still stands.

 

[Mr. Moscode]

I will guess it could be a repeater amp for broadcast? Is this for driving 600 ohms or a speaker?

I believe the cathode biasing scheme is to keep DC out of the input transformer. Or it could be a designer's whim or corporate thinking at the time. It's not really that significant which way it's done.

It's not very hi power, a few watts at best.

Are you sure R109 is 2.04 ohms? That's almost a dead short!

BTW: Get a scope!! It's an indispensable tool for everything DIY folks want to do. Without it you're working blind and guessing. My parents got me an EICO 427 scope kit when I was 12 for Xmas and I had it up and running in a day. They are really cheap, too, especially older units.

RE: Treble boost. You will have to take a gain hit of 5db if you want to flatten it out with or without NFB.

One idea I have is to split the input tubes cathodes and change the 1k resistor to (2) 2k resistors, one for each triode. Now bypass those independent cathodes to ground with smaller caps. I would try (2) .01 mfd and see where that takes you. I don't do things by math but someone could chime in with the math. You are making a high pass filter boosting gain at higher frequencies, but really restricting the gain at low frequencies. The bigger the cap the lower the frequency of the boost will start and vice versa.

Bypass the power supply line marked M with 100 mfd or more to quiet it down. It should be a positive voltage of not much, maybe 4 or 5 volts. Do you have a power supply schematic? Send it please. Also put the voltages on the schematics, please.


Edits - missed the meter ckt post. Sorry.
Are you sure R109 is 2.04 ohms? That's almost a dead short!

Everything is balanced coming in and out so I don't think this solution will be a problem.