Negative rail - Decoupling needed?

[exclamationmark]

Just wondering... I have a negative rail used to supply a bias voltage to 4 or so stages. I'm getting some weird interaction between notes (especially bass notes) which sort of leads to me suspect something within the power supply chain (and not my flawless design ). Is it necessary to decouple the bias supply rail as well or would I be better off looking for the problem elsewhere?.

[J M Fahey]

You answered yourself: you *do* need to decouple the negative bias for those stages, *specially* if they are cascaded.
See it yourself: scope the negative bias rail: you must have *no* audio visible there.

[exclamationmark]

Decoupling didn't seem to help. Turns out it was some weird intermodulation effects because I was passing too much mud (or low end 'growl' if you wanna be nice) between 250 - 500 Hz, turning any chords into mush when they clipped. I stuck a graphic EQ in the front and now I can actually play chords by making an artificial shelf filter. I guess I need to cathode bias the input stages to get that mid-shelf going (I LED biased the first two stages). On the plus side, the roughness of the bass gives a nice texture for lead playing... Just as long as you don't play more than two notes at a time

[jimboyogi]

Instead of attenuating low freq. with a small cathode bypass cap., can't you just use smaller coupling caps and keep your LED biasing?
Or are you saying that it's the mids that are causing the mush, and not the bass? If so, can you move your tone controls upstream?

[exclamationmark]

It seems to be the low-mids not playing nice with my circuit. The problem with the coupling caps is that they have a static -6db/octave slope, which could be bad if I set the rolloff freq too high. Most of my existing coupling caps are already have a -3db point of 150Hz. The bass is actually very controlled sounding, but it has this sort of 'tearing' sound to it, which just turns into mush when you play chords.

This was also kind of a slapped together design, testing/modifying individual stages as I went along. So the first couple of stages were just plain old LED bias and I went straight to the overdriven stages without thinking. Then it kind of hit me. Realizing this (and pondering the 6505/SLO100 designs), I put a TON of shelving filters between the overdriven stages, but I think it may have been too late. An easy fix, anyway (I hope, haha).

[Chuck H]

Well, cathode bias and smaller bypass caps act very much like the RC parallel circuit. Though there are some more interesting local NFB things happening. The average 1.5k cathode R / 100k plate load on a 12ax7 gives the bypass cap a 6db shelf @ frequency. Increasing resistance in the signal path will increase hiss a little (use metal film 1 watt resistors to minimize the effect) but you could just increase the resistor value, adjust the cap value for impedance on your existing shelf filters and keep the fixed bias.

[tubeswell]

@exclamationmark Try a smaller bypass cap on V1 (like 1 or 2.2uF) which should cut down 'boominess' without robbing low end. Then, if blocking distortion is a problem on bass notes, try lowering the coupling caps after the PI. This will keep a nice amount of bass in the mix (whilst not resulting in a too 'nasally' sound) but should alleviate a teency bit of blocking distortion. JM2CW

[J M Fahey]

Remember that if you are cathode biasing a triode, with a led/diode_string/zener/TL431 (you didn't think about the last one, did you? ) , the parallel/decoupling capacitors you use will have little or no influence either on gain or frequency response, because the dynamic impedance of the active element is much lower than anything else you put there.
Capacitors are only used to kill junction-generated hiss.

[exclamationmark]

Well, cathode bias and smaller bypass caps act very much like the RC parallel circuit. Though there are some more interesting local NFB things happening. The average 1.5k cathode R / 100k plate load on a 12ax7 gives the bypass cap a 6db shelf @ frequency. Increasing resistance in the signal path will increase hiss a little (use metal film 1 watt resistors to minimize the effect) but you could just increase the resistor value, adjust the cap value for impedance on your existing shelf filters and keep the fixed bias.

I just realized this. I was using 470k + 470k to ground with a .0022uF cap across one that used to shelf to 6db, and changed it to 1 Meg/470k so now it shelves to something like 10db, which helped a tiny bit. Gonna sub in a new main board and just stick sockets literally everywhere, so I can plug in random values. I think I may have been shelving things too late and ended up with a weird raspy mid heavy bass. For stupidly saturated vai-esque leads it works great though! Just stick a delay in the loop and wank away...

[Chuck H]

IMHE it's possible to get that oversaturated sound and still retain some note definition if you actually CUT some low mids and raise the high mids. Adjust the bass level to where it begins to "flub" and then back it down a little just to where it doesn't. The upper mids are where it's at for that uber gain thing. Too much low mid (which is really what typical tone stacks adjust) just turns everything from the D string down into atonal mush. If your amp is a do all type design then you need the low mids for good clean and blues/hard rock tones. I usually add an upper mid boost switch to the front panel and when I want to wank I just flip the switch and roll down the mids (and a little bass) at the tone stack. Crank up the gain and wail.

[loudthud]

Remember that if you are cathode biasing a triode, with a led/diode_string/zener/TL431 (you didn't think about the last one, did you? ) , the parallel/decoupling capacitors you use will have little or no influence either on gain or frequency response, because the dynamic impedance of the active element is much lower than anything else you put there.
Capacitors are only used to kill junction-generated hiss.

The problem with the TL431 is it doesn't go below 2.5V, and like most of those active zener type parts, impedance starts to rise at some frequency usually between 100Hz and 1kHz. I was thinking of using one in a no-electrolytics allowed amp. There are newer ICs that go below 2.5V I was looking at, you still need something like 10uF in parallel to keep the impedance low. The LEDs I looked at had an impedance of 30 ohms @ 1mA.

Then there is the problem of what happends when you drive the stage really hard and there is no keep-alive current.

[J M Fahey]

Yes, it's true.
And 2.5V *is* a little high for the typical 12AX7 , although not for a 12AU7.
And Red LEDs provide 1.9V, perfect for 12AX7.

[exclamationmark]

The problem with the TL431 is it doesn't go below 2.5V, and like most of those active zener type parts, impedance starts to rise at some frequency usually between 100Hz and 1kHz. I was thinking of using one in a no-electrolytics allowed amp. There are newer ICs that go below 2.5V I was looking at, you still need something like 10uF in parallel to keep the impedance low. The LEDs I looked at had an impedance of 30 ohms @ 1mA.

Then there is the problem of what happends when you drive the stage really hard and there is no keep-alive current.

I basically made a lm317 ccs powered from a lowered b+ supply and fed it to the LED at tube cathode/led junction to alleviate the cutoff problem. Literally costs 50 cents to do.

[J M Fahey]

I see we are throwing all kinds of "fancy SS parts" at our tubes.
I guess we'll be burned at the stake by the "purist" crowd.
Oh well.
At least the WW2 has just finished and we won.

[loudthud]

I see we are throwing all kinds of "fancy SS parts" at our tubes.
I guess we'll be burned at the stake by the "purist" crowd.
Oh well.
At least the WW2 has just finished and we won.

How about batteries? Would that keep them away? They are cylindrical and have a plus sign on one end.