• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Cathode bypass caps?

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
What is the rule of thumb with cathode bypass caps, both in the preamp and power amp sections of a tube amp? I am talking about the caps which are in parallel with the cathode-to-ground resistor.

I am rebuilding someones homemade amp from the 70s, and it uses 100uF electros for cathode bypass caps everywhere. However, I have read that you dont really need these, and they may just be screwing with the sound. The amp sounds good, but the bass is a bit "rumbly" and there is a little bit of distortion i am trying to hunt down.

Could I just try swapping the caps for smaller values? Or just removing them all together? What would happen if i used 1uF polyester (non-polar) caps?

All answers appreciated. :D
These caps are used to couple the ac signal current to ground while still letting the bias voltage be dropped through the resistor.
If you remove the caps or reduce their value, the audio signal will also partly be dropped by the resistor causing negative local feedback thus reducing the gain a lot. If you remove these caps, the amp will still work but the gain will be much lower and you might get really weak bass response. They should be kept at 100uF to keep the audio gain good. You'll find that if they are too small, the bass response will suffer, definately don't change to 1uF or you'll get no bass with your highs and mids. If you wan't to enforce bass stability, you could even increase the value of hte caps but too much capacitance will make for more needed time to stabilize even after the filaments heat up. Also, make sure your caps are rated for the same or even a higher voltage than that of the originals!
Joined 2001
Paid Member
Re: bypass cap

If you use a cap it should be big and as good as you can get. If you don't you eliminate a cap from the signal path (a good thing) but you get 100% regenerative feedback so less gain.

Sometimes it sounds better w the cap sometimes not. Try it with & without. Use the method one that best suits.

Well, I tried disconnecting the caps altogether, the ones on the power tubes had NO discernable effect on the sound when removed, so i left them off!

The caps in the preamp had a slightly strange effect when removed: the bass seemed to get a bit "boomy" and the sound quality was... changed. Quite what i dont know, but it wasnt entirely nice sounding. So anyway i got some 3.3uF polyester caps (big suckas) and chucked them in: the amp now sounds quite good. I have noticed another set of 100uF electros in the negative feedback loop, so i think these are the next to go!!!

BTW, i did not find that the gain was changed at all. The volume control was left at the exact same position and the amp did not appear quieter!! Also the bass sounded much the same with the caps removed; maybe a bit "tighter" even.

Tweak o rama time!!
Hi ShiFty,

Ya done good for yourself. The name of the game it try it and see. However, one could do some calculations and try to see if one gets as good a result without such a dramatic change.

Here are some hints for either this project or the future:

First off, leave the one in for the power output. Why? It aids in maintaining the class AB function and maintaining the bias at idle. If it is 100 or more bypassing a 100-300 ohm resistor, better to use a smaller value like 22 to 33µF instead. This way one can have a quicker dynamic recovery.

There is something about capacitors called a time constant. That is the time it takes for a given capacitor to charge and discharge in the circuit. A larger time constant means that the capacitor will take longer to charge and discharge. So making this smaller rather than eliminating it will allow the bias to remain constant while giving them the ability to change in step with music demands.

As for the preamp stages, I usually go with making the bypass cap's reasctance 1/10th the value of the cathode resistor, at the desired frequency. So, if I want the bypass cap to have that value at 20 Hz, then I take the resistors value, divide by ten and lug the numbers into the formula for capacitive reactance.

For example, for a cathode resistor of 2.2K, I want 220 ohms. I want the frequency to be 20Hz (that gives a -3dB at 4 Hz).

The formula for capacitive reactance, turned around to get a capacitor value, is:



1/2(3.14159)20(220) = 49µF

A 47µF cap will do fine.

Even smaller, but no less than 10µF (for resistances of 1K to 4.7K), will also do well without degrading the overall sound. In fact, smaller values might improve the high frequency content since the larger capacitors have larger internal inductance. But that is a whole other issue better left for another thread.

Hope this helps.

  • Like
Reactions: 1 user
Joined 2001
Paid Member
Duo said:
If he got 100% regen feedback, he'd have a good oscillator. Actually, if you remove the bypass cap from a class A amp, you introduce degenerative, or negative feedback. Also, it won't be 100%, if you have 100% negative feedback, you have a unity gain amplifier, which makes a pretty useless VAS!!

You are right, for some reason i had cathode follower in my head...

Wow thanks for teh info....

I am going to take the amp to be analysed with a scope and see what i can find... Hopefully I can get a tube guru to give me some advice... I guess any techs who work on guitar amps et al should be able to help me. I want to fine tune this amp so it sounds not just good but great.
I usually bump that cap up to 470uf or larger, parrallel in a nice poly to "clean" the sound and it acts more in the way of a low pass and your bass is improved. And those have been usually 10mF caps I replaced. If you had 100 in there trying a 1,000 would not be out of line. A lot of guys don't like tubes because of the lack of punch in the bass. This seems to help.
Here is a real suggestion; next time you ask a fairly specific question it is very helpfull to the people you ask to tell them just what it is you are playing with and what you are trying to accomplish. If you had been doing this to a SE amp I would suspect that you would now have a very nice looking oscilator. So I am guessing you have a PP, but still only know that is a hombrew from the seventies not even if it is a stereo to listen to or an amp to play a guitar through.:confused:
  • Like
Reactions: 1 user
Well, in case anyone is interested (!) I have now reduced the sum total of bypass caps to 2, the 3.3uF ones in the input. The amp has never sounded better, I definitely think those cheap electrolytics were screwing with the sound a little. Bass response and gain is still ample.

I am going to try putting some different values in there just for fun. However I think after reading this I will not need to use them. (cheers to the usenet posting of Mike Kent of Canada!)

From Ghirardi's Radio Physics Course 2nd Ed. (1932) on page 670:

"If both tubes (push-pull configuration) have the same mutual
conductance and output resistance no condenser is necessary across this
automatic grid bias resistance, for the reason that the plate current
variations will always be equal and 180 degrees out of phase. The sum of
both currents will then always be a constant. Since there is no a-c
component in the plate current, no bypass condenser is necessary."

"If both tubes of a push-pull amplifier are not matched, so that the
mutual conductances are not the same, then the plate current variation
of the two tubes flowing through this grid bias resistance will be
unequal and 180 degrees out of phase. The result will be an a-c
component which can be bypassed with a condenser in the usual way if
desired, but it is really not necessary. If this bypass capacitor is
omitted, the voltage across the grid bias resistance will vary. This
variation will be 180 degrees out of phase with the grid voltage of the
tube having the greater variation of plate current (greater mutual
conductance) and in phase with the grid voltage of the other tube. In
other words, the tube having the greater mutual conductance will receive
degenerative action while the other tube will receive regenerative
action, the tubes tending to divide the load equally. Thus the effect
of the mismatching of tubes is not so serious."

That is all.
Just a further update: I have tried a variety of values and types of caps as cathode bypasses for the preamp section...

I have found that NON-polarised electrolytics of 15-50 uF 50V sound great. Polarised ones seemed to add a little distortion, whatever value you use. A 3.3uF polyester has amazingly clean sound, but lacks bass. I have yet to try paper and oil or polystyrene types.
diyAudio Senior Member
Joined 2002

After reading all these messages I can only say it all depends on the circuit which only the owner can tell.
Playing with bypass caps on cathode resistors will definitely influence the sound.Keeping the values the same as they were originally calculated by the designer would be sound advice in case we're talking about a hi-fi kit here.
More impotantly as a general tweak/update it would be a good idea to replace them with non polarized caps since the take a lot of time to discharge.Use in order of preference use polyprops if you can fit them in if not go for the bi-polar ones as used in speaker filters (keep in mind thes are usually low voltage though).
Making the original values bigger will extend low frequency response,making them too big is risky for stabilty and if your preamp is really wideband you may as well fry the woofer coils.
In an push-pull output stage tubes will try to self balance if they are not matched but the weakest link will go first and will most likely damage the rest of the set in the process.

Yrs sincerely,

Frank De Grove.
ShiFtY said:
From Ghirardi's Radio Physics Course 2nd Ed. (1932) on page 670:

"If both tubes (push-pull configuration) have the same mutual
conductance and output resistance no condenser is necessary across this automatic grid bias resistance, for the reason that the plate current variations will always be equal and 180 degrees out of phase. The sum of both currents will then always be a constant. Since there is no a-c component in the plate current, no bypass condenser is necessary."

Been thinking a lot about long-tailed pairs lately and it seems to me that if the single, shared cathode resistor in the abovementioned class A push pull output stage was replaced with a (solid state?) constant current sink with no bypass cap, this would balance the halves of the output stage virtually perfectly. All you would have to do is set the bias on each tube separately so the zero-signal currents are initially balanced and then you could use virtually *any* pair of non-identical tubes and they should work together perfectly.

Also, how would it be to replace the various other cathode resistors with zener diodes? You could then use a much smaller capacitor because the voltage drop would stay realtively constant even though the cathode current varies during the audio cycle. One disadvantage is that it would not change the bias to compensate for tube aging.

Joined 2002
Paid Member
By pass capacitor value

This is something for which I have some suggestion but need someone to explain the way it functions.
I have read that the bypass capacitor does not seriously improve the gain. So if that was true then the bypass capacitor should not improve the gain by any significant amount. In fact I also found that the frequency response I checked with "Clio" did not show any improvement that would be audible to me. In spite of all this, with the bypass capacitor the amplifier ( power amp ) went audibly deeper in the bass. There is no doubt about this. Measurement did not indicate this ! The resistor was about 1K and the capacitor 470uF which seems to be an over kill. I suggest you try the amp with and without the capacitor. Pick the one that sounds best.
Instrument results may not necessarily show up the difference. I suspect that it affects the output impedance and that is why it sounds different.
Hi guys,

Now we are getting into some fun tube stuff.

The typical triode has cathode resistance. It is generally between 200 and 1K ohms, depending on the transconductance (Gm). This is calculated by a simple formula, 1/Gm. So if your triode has a Gm of 1600 µmho, or µSiemens (the 12AX7 or 6SL7), then the cathode resistance is 1/0.0016 or 625 ohms.

The calculation for gain in a stage of amplification is simply Rp/Rk (some will quote more elaborate and esoteric equations, but it can be simplified into this one). This is total resistance, which means Rp is the plate resistor in parallel with the plate resistance in parallel with the resistance of the following stage. Rk is the cathode resistance in series with the cathode resistor. So for a typical 12AX7 stage with a 100k resistor and a next stage of 470K, with a 1K cathode resistor, the gain would be about 32k/1625 or 19. With a bypass capacitor, the gain now becomes 32K/625 or 51, which seems significant numerically, but audibly amounts to about 3 or 4 decibels difference. Not much.

However, if you have a global NFB in that circuit, you now have some more gain to play with, and some more NFB for the frequency response to add to. IOW, deeper bass and higher highs. And less peaks throughout.

One thing you have to ask yourself: Am I hearing deeper bass, or more bass?

The reason I ask is because when going from one quality of anything audio to another, one of lesser quality may sound more bassy but may be limited in range, whereas the better quality one may sound less bassy, but you can get 10 Hz out of it with equal force as 100Hz.

As for larger caps in the bypass, or interstage for that matter, I have an opinion. If one were to rearrange the triode amplifier such that the plate were the input and the cathode the output, it would look like a rectifier. So the bypass cap will be a filter.

In essence that is what it does. With low frequency sound, the larger caps fill up to the higher voltage of the signals positive peaks. What happens then? For a brief moment the tube goes into cut off, or close to it. In a word, distortion. Also, the tubes characteristics are no longer constant.

With musical instrument amps this is more of a problem, because the pluck of that string introduces a very high amplitude very low frequency signal. So for a moment right after that pluck, you hear nothing, then the sound comes as if someone raised the volume quickly.

For high fidelity it is bad because we want out tube amps to sing properly all the time. It is sort of a testimony for fixed bias. Now that is another thread altogether.

I need to understand something here. How are you measuring the gain of this amplifier? In the over 30 years I have been fiddling around with electronics installing a bypass cap in the cathode circuit of single ended gain stages always increases the gain of the stage. The cathode resister does two things, 1) provides self bias for the tube: 2) provides DC degenrative feedback for the tube. The ratio of the plate load Z to cathode resistor "unbypassed" determines the feedback amount. Bypassing the cathode resistor with a cap removes the feedback above the rolloff frequency. The values of the R and C set the R/C time constant which determines the rolloff frequency of the feedback circuit.

Small cap = more feedback at low frequency = less gain at low frequency

Big cap = less feedback at low frequency = more gain at low frequency.

It always works this way. It has too.

If the gain determination is being made by ear then that explains it for me. Subjective not measured. All bets are off. That doesn't mean you can't experiment. It just means you need to be carefull about how you report your results.

If your measureing this and playing around with cap values. you need to use several frequencies and make sure that the volume control does not change, I set it to max. I always use 20 Hz, 200 Hz, 2kHz and 20kHz. I set the input level so that the output is well out of clipping and then measure the input level. Every time you change frequencies set the input level to the original measured value and then measure the output level. I also use different wave shapes to give me a better idea of amplifer performance.

Yes you are right, but:

Once the calculations have been made and the amplifier assembled, then the simple formula Vout/Vin can be used to verify gain. So for an input of 1 volt, if, without bypass cap, the output is 19 volts, then the calculation is confirmed. Just use a function generator and a scope.

Now, if you look at the difference between a gain of 20 and one of 60, that seems like alot. But, when one calculates this using 20LogX, where X is the gain, then one finds that the decibel gain is small. Audibly, we need to have a voltage gain of 20 decibels in order to have the sound be percieved as double. In terms of voltage gain, that would be from a gain of 20 to a gain of 200. So in these terms, one can see how a small audible difference in gain, as the original poster observed, could be perceived.

This is stuff that has been around longer than both of us together have been fiddling with it.

One will find that these calculations will be correct 99% of the time.

Roll off frequency is another issue. If you want a cap to give you 1/10th (an old timer's rule of thumb) the resistor as its reactance, in the case of the 1K, 100 ohms, for 20Hz, then use this value in the capacitive reactance equation. But use not much more than that value or one runs into the condition I mentioned before.

Hi Gabe

I would guess that the roll off is the part that is heard when changing the cap values not the level so much. In fact the original problem could have been open or dried out electorlytics bypass caps. That's why taking them out made no difference. Now, by replacing them with some new value will change the sound/tonal balance of the amplifer.

Just my take on the subject. Later Bruce
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.