Knowledge and understanding are the basis for creativity in engineering (and many other fields too). Problems can occur when someone tries to be 'creative' without them.
It has probably been said already on this thread, but it is surely the case that the "ideal" bypass capacitor will be such that there is zero AC voltage across it, and if this ideal is achieved then it cannot, by definition, be giving any "colouration" to the sound. This ideal should, of course, be approached if one uses a very high value of capacitance. It would also be very easy to measure the AC signal on the real-life capacitor, and see to what extent the ideal is being approached.
Although one might not be able to predict reliably what perceived musical nuance would correlate with a particular AC signal across the bypass capacitor, I think one could say with absolute certainty that if the AC signal is vanishingly small, then there cannot be, and will not be, any audible colouration. So just using a very high value of capacitance should surely be the way to go. If a high-value electrolytic still leaves some measurable deficiency, such as not filtering out the high frequencies well, then a suitable lower-value non-electrolytic in parallel should complete the job. In any case, any additional capacitance in parallel can surely only ever improve the situation, never make it worse.
If someone claims that such a strategy of minimising the AC voltage on the bypass capacitor doesn't produce the most musically pleasing sound, then they would surely have to admit that what they really want is a sound-effects box and not a faithful amplifier whose output voltage is a magnified copy of its input voltage.
(By the way, I'm not necessarily saying there is anything wrong with preferring a coloured sound; I'm just saying that in that case one should be clear about saying that the goal is not to build an amplifier that faithfully magnifies the input voltage.)
Chris
Last edited:
Not necessarily. A parallel cap can create a resonance (high impedance) with the inductance of the bigger cap, thus destroying the bypassing over a limited frequency range. The lower the ESR of the electrolytic, the worse this might be. Things are never quite as simple as they seem!cnpope said:
Well, i have done lots of tests with different types of cathode bypass capacitors, and i had the conclusion that the ones you dont hear are the ones wich are not there...so ,a little bit of NFB in the tube is the best way to do in my opinion... or if you want to put there a capacitor, use a perfect one, and dont forget to tell me where you bought it, because i want to buy some as well...
Best regards
Best regards
Well, yes, you are right, I guess. But anyway, it can all be measured and tested for; it is engineering, not black art.
Chris
Last edited:
Chris, you're making far too much sense. Please desist.
(Yes, the question of signal voltage across the cap was indeed dealt with early on in the thread, as was fidelity versus effects boxes)
(Yes, the question of signal voltage across the cap was indeed dealt with early on in the thread, as was fidelity versus effects boxes)
I remember a now fairly famous tube designer who did just this, always spec'd a 1000uf cathode bypass cap, people would ask why such a large cap? He would never answer.
Are we are going to say that fixed bias sounds the same as a cathode bypass cap or that led's sound the same? The differences can't be measured on the output so its just psychology. Heck a TL431 in the cathode is cheaper than a cap lets use one of those 🙁.
One thing to keep in mind is that unless one is using a shunt regulator B+ the signal current flow goes right thru the last cap in the powersupply, so there is not much gained by getting rid of the cathode bypass cap without a shunt reg from a technical perspective. Maybe this is the cause for confusion
What I like a about SPUD amplifiers is one can be built that blows up a lot of pessimism about DIY fanatism. It can be built with zero eletrolytics in the signal loop. One active device, the most linear known to man. Distortion measuring well below audible levels. A real conundrum for the objectivists.
All in this thread should build one listen to it. I recommend the EML 20B if you have the dough or an E55L if you don't. design it for a pair of monitoring headphones or nearfield monitors speakers. Throw a big ugly electrolytic in the cathode vs LED/filament bias. Tell us they measure the same after the speaker (We should care less about measurements of resistor loads.) In other words measure the system with a microphone.
Or more practically listen with your ears.
Its a very good question to ask for means to eliminate the cathode bypass cap, the OP shouldn't be drug the mud for wanting to do so IMHO.
"Its a very good question to ask for means to eliminate the cathode bypass cap, the OP shouldn't be drug the mud for wanting to do so IMHO. "
I didn't think anyone was dragging the OP through the mud. And wanting to eliminate the the cathode bypass cap is a fine thing to do. All that I would say is that if one is going to use a cathode bypass cap, then there is no deep mystery or subtlety about how to optimise it for minimal colouration of the sound; it can all be settled by measurement of the AC signal across it; choose the capacitor or paralleled capacitors that reduce the AC signal to a minimum (across the audio frequency range).
I absolutely agree with you about the the final capacitor in the power supply; that is in the signal path too. So attention should be paid to optimising things there also, by minimising the AC signal voltage across it. Again, measurements would be the surest way to settle such matters.
Chris
I didn't think anyone was dragging the OP through the mud. And wanting to eliminate the the cathode bypass cap is a fine thing to do. All that I would say is that if one is going to use a cathode bypass cap, then there is no deep mystery or subtlety about how to optimise it for minimal colouration of the sound; it can all be settled by measurement of the AC signal across it; choose the capacitor or paralleled capacitors that reduce the AC signal to a minimum (across the audio frequency range).
I absolutely agree with you about the the final capacitor in the power supply; that is in the signal path too. So attention should be paid to optimising things there also, by minimising the AC signal voltage across it. Again, measurements would be the surest way to settle such matters.
Chris
Ok what you are saying is most capacitance lowest ESR, correct?
Lets name same that work at the cathode of a popuar amplifier, a DC coupled 2A3 SET.
We need 1000uf, low DCR and 300+V rating, got any suggestions?
I'm not quite sure what your point is. It is, I think, a truism that the primary purpose of a bypass capacitor is to reduce the AC signal across it to as low a value as possible, and therefore one should choose the capacitor(s) appropriately. And the extent to which one has succeeded in this aim can be measured by straightforward means. That is all I am saying.
Chris
I'm not being argumentative, just pointing out that such electrolytics don't really exist commercially as far as I can tell. Well they do for an input or driver tube but not an output. But then there is also the zero crossing issue where an electrolytic's performance degrades inversely to the voltage across it, so use LED's for small signal IDHT tubes, filament bias where possible with DHT's. Also there was a good paper that showed the lowest distortion electrolytics for audio are the bipolar type but these days they are available in only low voltage.
I just don't think its a big stretch to say, avoid electrolytic's and to ask for alternatives.
The real gap we have is the cathode of the output stage and the last cap of the B+ as we agreed ( or ultrapath cap which does both.)
How about a Salas Shunt reg under the cathode? Practically no impedance, just a nice 1uF film cap. Could it be used in a stacked arrangement?
Is it worth it? I don't think enough have tried, it certainly adds a lot of expense and complexity but I've seen a few folks intent on here trying it. I think Moglia's 4-6A build and the "mid life crisis amp" are two examples but they are still in the design phase.
I want to try this myself with a 2A3/300B but not sure I can afford the time/expense. An electrolytic is easy and convenient.
Cheers.
I don't get it.😕
A) Why do you want caps rated at at 300+ Volts?
B) They are readily available anyway. e.g. RS online offers a choice of 30 different 1000uF caps rated at over 300V DC.
Well, this has been exactly my experience in practice - I never got electrolytic bypasses to sound anything other than awful, so switched to filament bias with DHTs, which I'm very happy with.
But then again, I didn't measure the AC voltage across the bypass capacitors when I spent a whole afternoon clipping together a vast array of capacitors of all kinds. It sounded to me that polypropylenes were better than electrolytics, but I could make a case for no capacitor sounding better in some ways though it was clearly not "correct". I didn't try very large values of electrolytics, so the theory that I was hearing the effects of AC voltage across the capacitors is plausible.
But if we are saying that the capacitor has no effect (no AC) simply because of its large value, then how can different types make any difference at all? What am I not getting here?
A DC-coupled SET maximises any problems with cathode bypass electrolytics because the cathode voltage is so large compared with signal voltages. Not a good architecture IMHO. By avoiding a (well-behaved) coupling cap you are forced into using a (less well-behaved) cathode electrolytic. Amplifier architectures should be designed around the known limitations of components.
A polar electrolytic should not be used without a bias voltage so zero crossing of voltage should never occur. Zero crossing of current occurs all the time with any cap carrying a signal so I'm not sure what it is you are concerned about.regal said:
If we start with a well designed circuit minimizing AC across the cap and then look at measurable differences between caps.
And yes caps are different.
They have different internal structure depending on use.
ESR is something discussed but comparing a pulse cap against a backup being extreme shows some other differences.
The main question is if theese differences is audible in the circuit ??
And yes caps are different.
They have different internal structure depending on use.
ESR is something discussed but comparing a pulse cap against a backup being extreme shows some other differences.
The main question is if theese differences is audible in the circuit ??
Hi flathead!
Good to hear from you! You're talking about "minimising" AC across the bypass where previous posts seem to be talking about "eliminating" it altogether. Where there is some voltage I could see different caps making a difference.
Good to hear from you! You're talking about "minimising" AC across the bypass where previous posts seem to be talking about "eliminating" it altogether. Where there is some voltage I could see different caps making a difference.
There are more ways to skin a cat than the topology you describe. DC coupled with grounded output stage cathode is perfectly possible and avoids all capacitors apart from those in the power supply where they do least harm.
I never could understand anyone who did a DC coupled amp by burning hundreds of volts in the cathode - that certainly does qualify as very bad design in my book.
Shoog
The bypass caps goal is to provide a AC direct link on that we all agree ?
No cap is perfect electrical they all have compromises depending on use.
The question is as always when is things unmeasurable, non audible and not relevant for the performance.
The search for the perfect cap is long and not very rewarding 😉.
No cap is perfect electrical they all have compromises depending on use.
The question is as always when is things unmeasurable, non audible and not relevant for the performance.
The search for the perfect cap is long and not very rewarding 😉.
- Status
- Not open for further replies.