Filter cap tricks, opinions?

[wizard333]

How many folks have experimented with:

1) Diodes between the screen grid and pre-amp filters,

2) bypassing the last pre-amp filter stage with a small non-electrolytic cap.

I've screwed with this on and off for years with mixed results on whether I thought it did anything at all, or whether I thought what it did was good or bad. Sort of like how I've heard or read so many times how more filtering is "good" but I generally hate the sound/feel of it, and haven't found it necessary for a quiet amp nearly as much as good design or lead dress.

Any opinions on feel or sound of either of those? RE 2, for those who have experimented with different values, have you noticed much difference based on the value of cap used?

[überfuzz]

#2. Do you mean a small cap between grid and anode? I'm not sure you really like to bypass with a cap that attenuates audible frequencies.

[wizard333]

No not at all, I mean bypassing the last pre-amp filter cap with a smaller non-electrolytic cap to, at least in theory, improve transient response.

[überfuzz]

No not at all, I mean bypassing the last pre-amp filter cap with a smaller non-electrolytic cap to, at least in theory, improve transient response.

No, still not sure what you mean. It is quite common to bypass the cathode resistor with a cap to maintain the electrostatic attraction across the triode and to maintain the voltage level at the cathode.

[thehoj]

No, still not sure what you mean. It is quite common to bypass the cathode resistor with a cap to maintain the electrostatic attraction across the triode and to maintain the voltage level at the cathode.

I believe he's talking about this - AX84.com - The Cooperative Tube Guitar Amp Project

[eschertron]

I believe he's talking about this - AX84.com - The Cooperative Tube Guitar Amp Project

Nice read! I've also heard about paralleling the first PSU node with a smaller cap to help silence the hash that can be generated by a SS recitfier. Same idea, slightly different symptom.

[exclamationmark]

In theory the minimum impedance a capacitor can have is equal to it's ESR (which is something like like 1 to 3 ohms in the case of a 100u filter cap). However, in practice I'm fairly sure the impedance can decrease past this point slightly, then it will rise back up to the effects of foil inductance. A film capacitor in parallel with the filter cap will help bypass extremely high frequencies, but really only in the MHz ranges, as the electrolytic cap does a fine job in sub MHz regions due to the sheer size of it's capacitance. A similar thing is done in high speed logic design - you can parallel multiple capacitors of different sizes (physical as well as capacitance size) to form a wider decoupling range, but these signals have rise-times in the order of nano or picoseconds which corresponds to frequencies in the GHz range. Personally, I don't think the effect is even audible (I've tried it too).

It's also applying some sort of backward logic... The typical output impedance of a RC filter combo is actually the highest at lowest frequencies and the lowest at high frequencies. ie, the output impedance at DC is equal to the voltage dropping resistor, and the output impedance at high frequencies (say past 5kHz) is roughly equal to the ESR. Thus you're actually trying to bypass something that's already pretty good at high-ish frequency stuff, except when it's practically inaudible. You can certainly try the bypass method if you want, but if you have an oscilloscope you can visually check to see if it actually makes a difference at standby or when you're playing anything - I haven't ever seen anything that indicated that it does.

[wizard333]

I believe he's talking about this - AX84.com - The Cooperative Tube Guitar Amp Project

Yes that is exactly what I meant. Although the only thing I gleened from that discussion that I didn't already know is the preference for going 1/1000 with the poly cap.... although thats a theory. All theory is suspect until proven by ears/fingers. For example the theory that "more filtering' is better and is required for a quiet amp, which my ears and fingers have proven to me is hogwash, or that "pickups that look the same on an oscilloscope when you put a sine wave across the leads must sound the same".....also utter hogwash.

I was more interested to hear from folks who have done it and experimented with different values in the same amp and what their conclusions were about the affect on sound/feel.

Also refering to putting a series diode between the power dropping resistors, inserted between the PI and the pre-amp stages.

[Gingertube]

I use parallel film caps routinely. The rule I generally use is 1/100th the value of the electrolytic. I usually do the main filter bank (power amp feed), the feed to the preamp and the input tube B+.
For "modern" amps I use the in line diode feed for the preamp B+, this prevents the preamp B+ sagging along with the power amp B+. Watch this - it means you need to add a bleed restistor to the preamp B+ supply so you don't get nasty surprises when working on your amp. The bleed resistor(s) across the main B+ does'nt discharge the preamp B+ caps (after switch off) because that diode becomes reverse biased and blocks current flow back from the preamp (that is why it was put in there in the first place).
For "vintage" style amps I don't use that diode, just let the preamp B+ sag along with the power amp.

While the parallel film cap on the main B+ will help reduce SS diode rectifier switching noise, I prefer to treat the disease and not the symptom by using only ultrafast/soft recovery diodes in the power supply. Bog standard 1N4007 have no place in any decent amp. That switching noise will appear not just on the B+ but will couple back through the power tranny into every other winding and can therefore appear on heater supplies, bias supplies etc. Far better to not generate the crap in the first place than to try and get rid of it afterward.

Cheers,
Ian

[deci belle]

hi Wiz! I've tried both and can say that they are kind of old hifi tricks— I think that's what Gingertube said to me in the past. I tried them on a bf Showman and a 5E3-something a couple years ago.

The diode between the power and preamp sections wasn't a bad thing for the little tweed circuit but made the big bf too stiff, it didn't need it and it didn't help anything that I could hear. I removed the diodes from both amps, but if I made another little-ish tweed I'd try it again— the difference was noticeable, making the little amp less sloshy.

I didn't really notice tonal differences when I installed the parallel tube cap at the last filter cap for the preamp. I took it out after trying it out for over a year. I've also tried the parallel cap with the electrolytic at the bias supply. Eh…

[Joe L]

The idea of putting a diode between the power and pre filter stages seems fine on paper. But if you have played with rack preamps much, you will find that there aren't many that really feel like a full amp, even if the preamp is identical to what is in the full amp. The difference of course, is the preamp has a stiff supply which isn't affected by the separate power amp. That is why quite a few racks use the effects send/pre out of a full amp that is driving a dummy load.

[loudthud]

Generally, voltage sag is around 10%. Slightly higher at the screen node in an amp with a choke. A diode inline with a dropping resistor won't do anything unless the upstream voltage drops lower than the downstream voltage. In preamps the voltage sag is fairly slow because of the long time constants in the decoupling string.

[wizard333]

Interesting food for thought, thanks for the replies!

Bog standard 1N4007 have no place in any decent amp.

What are you using?

[Gingertube]

Where 800V PIV is adequate I'm working my way thru' a reel of 200 off BYT11-800 I bought for $5 but I have also used UF4007.

Actually about the worst switching noise I've ever seen in an amp was with 1N5408, I also have a reel of something I can't recall right now B??13-1500 I use in place of 1N5408 but once again at least go for the UF4508.
I also use these when I need more than 800V PIV rating.

Use ultrafast recovery diodes, they have much less junction capacitance and therefore generate much less switching noise (the switching noise comes from a "splat" of current in the reverse direction which sweeps out the junction capacitance charge in order to turn the diode OFF, smaller junction capacitance means smaller junction capacitance charge means smaller reverse current "splat"). The problem arrises because that "splat" has very high di/dt (rate of current change) which means its radiates and couples into damn near everything.

For low voltage rectification (heaters etc) use schottky diodes which are better again (much lower junction capacitance charge).

Cheers,
Ian

[wizard333]

Had a little time to play with this tonight on an amp of mine that is a Fender BF clean channel with a 3 stage low to 80s metal type gain dirty channel. 2x6v6 with merc iron.

The diode thing- not sure it made a difference, but it didn't hurt anything either. I didnt mess with clipping it in and out of the live circuit so no real time A/B, but it either made a subtle bit of good or did nothing at all.

The bypass cap was where it got interesting. I did clip this in and out of the live amp in real time. On the clean channel, really no difference that I could swear to hearing. It would show up on good test equipment I'm sure, but it didn't impress me enough one way or another to give an opinion. On the gain channel (one filter stage farther down the line than the clean channnel for the 1st 2 stages) I could definitely hear it. I started with the 1/100 value and to my ear it made the attack far too hard sounding. Not bright, not brittle, not fuzzy, but like, a hammer to the ear when the pick hit the string. I like a strong attack envelope but it over-emphasized it to the point of obnoxious, it was like someone turning my volume knob up 3 notches for the first half second of the notes. A very strange effect indeed.

Moving through various values and ending at the 1/1000 value, I have to say I liked that the best (.022uf orange drop across a 22uf electrolytic). Vs. not being there at all, it emparted a subtle buttery-ness to to notes aurally; it seemed to cut a tiny bit of highs, make things smoother and "greasier", and gave the impression of very subtle compression. Feel wise, it made the notes really jump off the board. Probably something that you wouldn't notice if your style doesn't include a lot of fast picked runs, but I REALLY felt it when doing some serious double or sweep picking. Hard to describe, but it was like swimming in a heated pool vs. jumping into a cold pool. If you've ever experienced the difference the VHT Valvulator 1 makes, it was similar. This is with low-medium gain, the kind of setting where it can be clean if you back off your attack or fairly gainy when you hit hard

Usually when you cut high end aurally, you start killing the feel of the pick attack, but not with this. It actually smoothed out the highs while making it easier to play.

I wouldn't bother on a clean channel, but I liked it enough that I'm going to experiment more with it on other gain channels and see if it gives the same result.