Improving the filtering.....

[Paultergeist]

Greetings,

I had a couple of questions regarding the power-supply filtering on tube amps:

1. It seems commonplace in most designs to take the B+ for the OT Center Tap (CT) right from the very first filter node. Electronically, this means that the CT is connected to the rectifier output. It seems to me that the rectifier puts out "noisy" (i.e. rippled) DC -- does any of this ripple create noise in the circuit? I realize that the "amplification" is pretty much done by the time the signal is at the OT, but I am just wondering if it ever makes sense to have a filter node upstream of the OT CT?

2. I was reading about some audiophile tube circuits which use an array of capacitors of varying values at the first node. In other words, as opposed to just using -- say, a 47uF cap at the first node -- some audio amps would include a varying array of capacitors (example: a 47uF, but also a 5uF, 1uF, .1uF, .01uF -- all in parallel). The rationale -- as much as my limited understanding will take me -- is that certain cap values more easily convey certain frequencies of noise, and that by having some varied values, it is easier / better to clean up the post-rectifier noise initially?

This idea is just my limited understanding -- I am not suggesting it to others at this time. What I am hoping to learn is if others have done this, and if they can explain the idea better......just hoping to learn.


Thanks for any insights.

Paul

[surfsup]

For (1), for a pushpull amp, common riple rejection cancels any ripply on the PS. Extra filtering won't hurt of course,

For (2), you have it correct, imo.

[Super_Reverb]

You can add a 75 or 100 Ohm resistor after SS rectifiers prior to 1st filter cap: this will give you some sag and will make for a better filter - a larger source impedance working into a filter cap reduces the filter corner frequency. Remember that a resistor here will drop your B+ quite a bit due to large'ish current going to outputs. You could use an inductor, but getting a significant amount of inductance with 150-200mA current capacity makes for a physically large choke.

It's fairly common to place 0.1 or 0.01u in parallel with filter caps. This is standard practice in logic, analog and switching PS circuits. Electrolytics can look inductive at higher frequencies, so a high quality 0.1uF gives you improved HF rejection.

Like surfsup said, since ripple voltage on 1st B+ feeding outputs creates out of phase ripple currents in OT primary, net effect of 120 Hz ripple on secondary signal is minimal (assuming bias currents are reasonably equal).

rob

[gingertube]

Paralleling the filter cap with smaller value caps.

What is this about?
Electrolytic Capacitors are only capacitive up to their self resonant frequency, above which they look inductive. This USED TO BE a significant problem. The way it was fixed was to put a film cap of 1/100th the value across the big electrolytic such that as the electrolytic turned inductive at higher frequencies the film cap took over. The self resonant frequency of the older electrolytics could be quite low - well down into the audio band.

This is not such a problem anymore as electrolytic caps have come a long way in recent years - mostly to cope with the demands of switch mode power supplies. The ESL (Equivalent Series Inductance) value will give you the clue.

I still add a 1/100th value film cap across the main filter electrolytic mostly because it helps shunt any high frequency rectifier switching noise (Solid State rectifiers) which won't go through the electrolytic due to its self inductance.

Hang on I hear the techo types say - if you put a cap across something which looks inductive won't you get a tuned circuit and resonance - generally not but if this is of concern then put a series connection of say 2 Ohms plus the 1/100th value cap across the electrolytic instead. That will damp any resonance (and is what I routinely do).

In the day job I occassionally have to do power supplies for amplifers for up to 2 GHz. In those cases I use 22uF Tantalum + 220nF + 2.2nF +22pF (last 3 all ceramics). For a Git Amp we don't need to go that overboard.

Cheers,
Ian

[Paultergeist]

Friends,

Before this thread becomes buried in obscurity, I just wanted to say a sincere "thanks" to all of those who replied. All three of the responses offered valuable insights -- copy-n-pasted in my ever-growing file of tube-amp-building insights -- and the information provided helps fill gaps in my knowledge base. Thanks again.

Paul