Cap types for large cathode bypass values?

General discussion area for tube amps.

Moderators: pompeiisneaks, Colossal

R.G.
Posts: 1579
Joined: Tue Dec 02, 2014 9:01 pm

Re: Cap types for large cathode bypass values?

Post by R.G. »

thetragichero wrote: Sun Dec 15, 2019 2:34 pm is there really any reason (besides needing to replace every couple of decades) behind using something besides a sub-$1 electro cap for cathode bypass purposes?
Good on you for asking the pertinent question.

Probably not. It's really a matter of objectives - what is one (as the amp builder) trying to accomplish?

If you (the amp builder/hacker in general) are after your amp somehow being "original", electros are the way to go. Actually, you want not-very-good electros, the way new ones were back in the 19x0s when whatever amp is being re-born was made.

If you (the amp builder/hacker in general) are after sonic accuracy (?!) using tubes is the wrong way to go about it. Tubes lost to semiconductors in low distortion on all of the various fronts decades ago.

If you (the amp builder/hacker in general) are after the amp sounding good to you personally, you're wasting time on internet forums when your soldering iron could provide even-more-instant gratification by swapping caps.

If you (the amp builder/hacker in general) are only after bragging rights about having the most internet-politically-correct amp, this forum can't help.

If you (the amp builder/hacker in general) are after understanding, you have to learn what the cap is there for in the first place. It's there for self-biasing, in conjunction with the cathode resistors. The cathode resistors cause a DC offset from the idle/bias current, pulling the cathodes higher than the DC-grounded grids. This is right out of the triode playbook for self biasing. It's intended to provide a constant voltage for biasing the tube. Well, "constant" over signal-significant periods of time. It has to average out the lowest frequency signals to insignificance, so its value must be big enough that its impedance compared to the actual cathode resistance must be very small. It has to be small enough that different tubes can eventually make their own bias point.

Imperfections in the cathode cap - ESR, ESL, leakage, nonlinearities - affect the tube operation to the extent that they affect the tube's open loop voltage and current operations. With a many-K reactive plate load, these are going to be small. How small and how much that affects sound gets into a discussion of infinitesimals.
i understand not placing electro caps in the signal path
Maybe. Not many people really understand this.

A cathode bypass cap IS in the signal path. It's moving signal >current< to ground. This affects the signal voltage by how much signal current is not bypassed to ground. The power supply filters are actually in the signal path. A shunt-to-ground cap is intended to let a frequency dependent current into ground. If an electro is somehow too "dirty", you can make the effect of the electro's imperfections smaller by making its capacitance larger. At some point, the impedance of the electro cap is just too small for the even smaller imperfections to matter to the audio it leaves behind. Power filter caps are a good case in point. Their value are intended to make them dramatically smaller impedances than the series resistors, inductors, and plate loads that use current from them. When this is true, the remaining effect of the cap is an almost-constant voltage, which is where you were trying to get as a power supply designer.
but say i paralleled my 22uf electro cap with a 100nf ceramic or film cap (should help with any esr-related issues, no?),
The whole parallel-the-cap thing comes from RF practice in the early/mid 1900s. RF guys noted that caps have an impedance that is proportional to 1/F, declining as frequency increases... until the cap's impedance dipped below ESR. ESR puts a floor on how low the cap's total impedance can go. Even worse, at some frequency the ESL of the caps' winding and even the parasitic inductance of the leads makes the total impedance rise again. So normal caps have an impedance curve that looks like a bathtub - as frequency increases from near-DC, the impedance goes down as 1/F, hits a floor at ESR, then rises again as F when the ESL impedance gets bigger than ESR. If all of your frequencies are RF, you REALLY want the start of the inductive part to be high in frequency, and you'll happily trade off the low frequency end of the bathtub. If your frequencies are audio, which RF guys think of as DC, you need to be pickier about how low the low end of the bathtub is.

Many caps are made by winding spirals of conductor and insulator until enough surface area has been wound to get the needed capacitance. This necessarily creates some resistance as the signal spreads from the leads' contacts through the cap plates, and some inductance - those spirals! However, modern cap making practice is far better than it was in the bad old days. Cap makers will sometimes make caps with the leads connected to the conductor spirals along the ends of the spiral wind. This means that the incoming current doesn't have to traipse through the whole length of plate to get to the ends, so the ESR and ESL are cut dramatically. Extended foil and low impedance electros can be remarkably low, especially if compared to bad-old-days wound-strip film caps.

Parallel-the-cap schemes can also only help above some frequency where the added/smaller cap's capacitive impedance is less than the ESR/ESL of the cap it's paralleling. Until then, it's just as big an impedance as the ESR/ESL it's trying to shunt. This is all to say - frequency matters. Paralleling caps makes a great deal of sense at high frequencies. But you have to know your caps' ESR and ESL and the frequencies you're trying to pass. Circuit values matter.
besides being able to puff my chest out about spending tens of dollars on mojo-brand film caps? I'm sure we could all come up with good ad copy but actual nuts and bolts technical reasons i am open to hearing
And good on you for asking the Right Question. I've laid out the technical background. NOT listening to the hifi-tweako ad-copy mania is a sign that you're a smart, thinking amp builder.
"It's not what we don't know that gets us in trouble. It's what we know for sure that just ain't so"
Mark Twain
thetragichero
Posts: 478
Joined: Tue Sep 10, 2019 7:46 pm

Re: Cap types for large cathode bypass values?

Post by thetragichero »

i came (back, because the first time i was too lazy to learn and thought i should be able to build great things through intuition and my superior intellectual capacity...) to effects and later ammo building because I'm a cheapskate: why pay a hundred bucks for a big muff pedal when i can make my own using components i salvaged from an old organi found on the side of the road? read about mods folks were charging good money for on the boss metalzone so i socketed every op amp and diode, recorded clips in my daw to test the differences and try to remove as much of my biases as i could. the result being that just about any old op amp sounds the exact same (save for the tl072 which gave unpleasant oscillations) and besides the volume/compression differences based on Vf there was no "magic" using the oddball diodes i had vs your standard 1n914/4148.
I'd love to say i do as much breadboarding with amps as i do low voltage stompboxes but I'm not setup for that so i rely more on wonderful online calculators
plus amongst my reading i took some advice to "use what ya got" to heart, and I'm more likely to have larger than 1uf values in electro caps as they're relatively cheap
your point of paralleling smaller value caps rang a bell with me as I've been repairing a Marshall jcm600 series head and i was wondering why just paralleling the 22uf nonpolar electro coupling caps with lower vague film caps didn't help with my runaway bias issues (took resistance measurements and i was getting on the order of 100k resistance between the two terminals of the caps.... talk about esr!)b but replacing them with 1uf film box caps did

i have a hard enough time listening to live music because my ears are trained to hear when the notes or tooling are off... it must be MISERABLE for those folks who can tell that the amp builder used cheap ceramic caps instead of film caps hand rolled by monks while burning sage.....
PRR wrote: Plotting loadlines is only for the truly desperate, or terminally bored.
LightningPhil
Posts: 15
Joined: Sun Dec 29, 2019 11:54 am
Location: Nottingham

Re: Cap types for large cathode bypass values?

Post by LightningPhil »

Have you checked out this calculator:
https://www.ampbooks.com/mobile/amplifi ... capacitor/

Great for basic sizing of caps.

Regarding electrolitic vs film, there are various types of both which complicate matters. But to simplify the argument - film caps at low voltage are very linear, electrolitic less so. Electrolitics end up having a little hysteresis as the voltage goes up and down. Notably more as the voltage limit is approaching and boat loads if exceeded. This results in a little distortion (not good tube distortion).

But, and an important but - on the preamp tubes the voltage is tiny and the AC signal even smaller. Couple that with the distortion from the amp and it would be difficult to detect, measure or notice.

And another but... - as the capacitance is reduced and only the highs are bypassed and the lows are not, the voltage swing greatly increases. This may cause the lows to distort more as they would suffer more hysteresis with their larger voltage swings.

So, to make a recommendation, for electrolitic caps, higher voltage rating is better than low. And if using small caps (1uF and lower), use film. Also, measure them, as low price ones have wide tolerance so could be all over the place. Ceramic caps are often less accurate than film and potentially more microphonic.

For output tubes, where the bias voltage is much higher, aim even higher in cap voltage rating. More is more good, with diminishing returns. Electrolitic here is the norm, so why rock the boat. Furthermore, if negative feedback is being used, distortion from caps in the output section would be reduced greatly.
LightningPhil
Posts: 15
Joined: Sun Dec 29, 2019 11:54 am
Location: Nottingham

Re: Cap types for large cathode bypass values?

Post by LightningPhil »

Here's some further reading:
http://stephan.win31.de/capdist.htm
Post Reply