Capacitance multiplier question

[wavley]

So in my never ending quest to make my space echo quieter from it's inherent hum and hiss, which I've gotten down considerably.  Looking at the input circuitry (which I built a mini booster in front of to fix the miserably low input impedance) it's all powered by a capacitance multiplier which I don't have a lot of experience with.  It's easily replaced in the same holes in the pcb with a modern voltage regulator OR I can just build a daughter board with a voltage regulator on it and leave the capacitance multiplier.

Is there a benefit to either of these arrangements?  RG and PRR, I'm probably looking at you guys on this and welcome the learning experience.

[bool]

I doubt that a voltage reg. will yield quieter rails.

Massaging a capacitance multiplier could bring the rail noise down a couple dB's.

I'm not familiar with the particular circuit, but generally a capacitance multiplier (usually always) benefits from small tweaks like: replacing the pass transistor with the highest gain possible, a darlington or a mosfet (where the voltage drop due to Vgs-off isn't of such concern), second increasing the RC constant to the larget possible value (space constraints etc). With a darlington (or a mosfet), this increase can be accomplished by pushing the resistance, while with a "naked" BJT, it is usually accomplished by pushing the capacitance.

So there, pick your ... lollipop.

[amptramp]

A capacitance multiplier does just what it says - the capacitance on the base of a transistor is multiplied by the hfe so that it looks like the power rail has a much larger capacitor across it.  It is not a voltage regulator and it must be preceded by a large capacitor just to ensure there is enough stored charge to draw down on.  A regulator has a problem - it usually has a zener as its regulation standard and even the buried / subsurface zeners may be quite noisy.  Some use a bandgap reference which is quieter but not by much since it is only 1.23 volts and this has to be multiplied (along with its noise) to provide the output voltage.  In fact, the LM399 discrete 6.95 volt zener is quieter than a bandgap for just this reason.  A capacitance multiplier does not regulate voltage or generate much noise.  I have seen discrete regulators where the regulation standard is a forward-biased LED which is very quiet - if you need a quiet regulated supply, this may be a good idea.

[R.G.]

All of the above makes sense.

Add to that a subtle noise source - transistor degradation. Bipolar devices get noisier if their base-emitter junction is reverse broken even once. Many older circuits have capacitors arranged so that this can happen with power cycling, so the devices just seem to get noisier over time. Protecting the base emitter junction with a normally reverse biased diode can stop much of this, but at the cost of putting the diode leakage and capacitance right on the base emitter.

If I were you, I'd simply replace the transistor in the cap multiplier with a new, high gain device. If it's a TO-92, here we go with Keen's Second Law again; when in doubt, whip in a 2N5088. This may do a lot to clean up the power supply.

It may well be the case that the other transistors in your copy of the Space Echo have had the same issues, and it might profit from some judicious replacements, especially in the front end. Got a schematic you can point to?

[wavley]

Thanks guys, it's funny because I got to thinking about it and realized that it looks like it's mostly just power for amps that regulation wasn't much of a concern, just low ripple.  I spent some time on teh webz looking at them and it's not such a bad idea but could benefit from cleaning up due to age.

While I've worked on a lot of things with discrete regulators and capacitance multipliers they were almost never broken so it's not a subject I've researched much given that I haven't needed to touch them and anything new I do it's easier to just use a modern regulator.  In school we had a night or so on these, but it was more of a "I'm going to teach this a little but you don't need it" kind of thing.

But today I was just kind of musing and thought about the fact that my added on input stage was also powered from this rail and maybe it all could benefit from cleaning up.

So here's the service manual, I have the 101 (and a 501, but it's a different beast inside and isn't noisy, but did benefit from a jfet input) http://manuals.fdiskc.com/flat/Roland%20RE-101%20&%20RE-201%20Service%20Manual.pdf

[PRR]

Do you mean this part?



First: "hiss" is *usually* not power supply. Hiss is however universal so system levels are important. In tape systems tape-hiss "should" dominate over all others.

That capacitance multiplier looks odd to me. 10uFd*hFE is about 1,000uFd, and they already have two 1,000u physical caps. The 2.2K against 10uFd does not look like a lot of hum filtering (maybe -24dB). Perhaps it works better against the 60KHz bias-screamer tapped on the raw 21V.

Being a simple man, I would consider brute-force. Split the 180(?) Ohm as three 60 (56) Ohm and hang new small modern 1,000uFd caps on all nodes.

[wavley]

I've actually gotten rid of most of the hiss to the point that the echoes don't have the same tone when it oscillates, but we've discussed this before.  I'm on a quest to find the rest of the hum and this was the obvious place to start.

Good thing is about these things is that there is plenty of room to add whatever I want in there.  I'm planning on cracking it open this weekend for some tinker time.