Regulators, practical differences

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
G

Guest

Guest
I have been reading about voltage regulation recently and have been considering building the sulzer regulator (schem below) as an alternative to the 3 term regulators. I've found very little info. on the practical differences between the two when used in an audio application. Can anyone with any experience with these comment on audible differences? I'm assuming that audible differences would very somewhat with amplifier design... I know the best solution is to use my ears, but I'm just wondering what others have experienced.

http://www.alw.audio.dsl.pipex.com/sulzer_circuit.htm


Thanks
 
Hello Jason,

I built many of them and they do sound better than your standard 3 pin reg.
But I would jump forward in time to Jan Diddens work with Walt Jung in TAA 3/95.

Sorr
 
If you go to Audioxpress.com

http://www.audioxpress.com/magsdirx/backissues/BISAAM.htm

You will find the back issues for 1995, The Audio Amateur

Sorr

edit: Jakob E.
 
One comment about regulators: they are almost always playing catch-up---the load changes, the regulator senses the change and increases or reduces the drive to the pass element or shunt element (or both in a push-pull regulator).

In other instrumentation I've sometimes made the power supply drain constant regardless of system state, so the regulator doesn't have much work to do. And then one can also use an ancillary circuit to sense the signal and adjust the regulator drive accordingly, while allowing the overall regulator loop to still function.

Many are the ways.
 
FWIW, a simple unamplified shunt regulator fulfills the "constant current draw" criterion.

This is an interesting topic to me because I have a circuit which requires 150V regulated (at about 25-30mA) and I can't quite decide how I want to go about it. There are so many options, though my personal tendency is to lean toward simple and hard-to-blow-up...
 
I've used the Sulzer regulator in several applications, including a couple of iterations of my op-amp-based mic preamp and a phono preamp design. It sounded significantly better than LM317/337 regulators, which in turn sounded significanly beter than 78xx/7900 regulators. I haven't tried the Didden/Jung regulator; it was more overkill than I was willing to do, especially considering that the wires from supply to load would render at least some of the advantages moot.

A couple of notes: I've always used the version of the Sulzer regulator with LM317/337 preregulators. I've usually used LM329Z reference voltage sources rather than straight zeners. Ya gotta be careful about using cap sizes different from Sulzer's recommendation, or the thing will latch up on you. And different generations/manufacturers of NE5534s perform differently; mine have never oscillated, but I've always used Signetics chips. Other folks have had bad problems with oscillation, usually from TI chips but sometimes others too. In those circumstances, a couple of small ceramic discs from V+ to V- helped stabilize them. Worst case, a TL071 did the trick, at the cost of higher output impedance for the regulator. Oh, sometimes I've blown up the pass transistors; something slightly heftier might help if you anticipate shorts (like a supply used for breadboarding circuits).

It's a nice design, with quirks.

Peace,
Paul
 
I did not use the remote sensing in the Didden/Jung regulators and they worked very well.
I used an AD797 and did not have any oscillation problems.
However in the1/87 issue of TAA is the Didden Wide band power supply, which is in between the Sultzer and the later Didden/Jung version.
Whilst I spent little time "tunning" the sound of the later cct (I used panasonic FC, 120 mfd 50V caps) I did spend a lot of time on Jan Diddens first one and whilst every cap is important, I found the one on the + input of the op amp as being very critical.
It uses a pre regulator.
In the article Jan mentions that the MJE200 (+ version) type transistors are critical to the high performance of the regulator.
These from memory are discontinued and he uses D44H11 in the 95 version.

In retrospect, I found I prefered the 87 version.
My power supply interests now lie elsewhere, as hinted by bcarso, and I would be very interested in learning more about his approach.
 
I'm glad to see I touched on something that is of interest here. For my purposes I'm thinking of making a small pcb for +/- 15-24 volts for use with mic preamps for the most part. From what I've read, the Jung regulator is very sensitive to layout/grounding, and since this will be my first pcb design I'd like to just keep it simple, do a bit better than the 317/337 regulators and set myself up for success.

Thanks alot paul, for the specific info. I'll keep that in mind when I start breadboarding... more to read/learn...
 
For anyone else, like myself who hadn't seen this before, Walt has alot of his old articles online in pdf. form.

http://waltjung.org/Regs.html

I got LOTS of reading to do...

Can someone add this to the psu meta?
 
[quote author="NewYorkDave"]FWIW, a simple unamplified shunt regulator fulfills the "constant current draw" criterion.

This is an interesting topic to me because I have a circuit which requires 150V regulated (at about 25-30mA) and I can't quite decide how I want to go about it. There are so many options, though my personal tendency is to lean toward simple and hard-to-blow-up...[/quote]

I used to use a simple emitter-follower in shunt for apps that weren't too demanding. For example, in a positive supply a PNP fed from a voltage divider off of the regulated voltage supply would take up the "slack" when the load current varied with signal. It was certainly fast, if inefficient.

For more careful work I've used composite current sources feeding composite shunt regulators. Typically the pass and shunt elements are DMOS, and the error amps are discrete bipolar. I like the access to all the parts so that the transient response can be optimized. The noise is way down below typical intergrated parts, and the line and load regulation absurdly good. The low-frequency noise is limited by what is used as a voltage reference. I used to favor concoctions of JFETs selected for operation at a low-tempco point, and more recently have looked at a roll-your-own bandgap reference made from discrete bipolars. But both of these are a lot of work, and for audio, wretched excess. For other instrumentation it may be a different story---for example, the photoarray system mentioned spent up to an hour soaking up photons, and then reading out, and the before-and-after bias was a part of the signal, so needed to be ultra-stable.

Beyond this, I think the explicit use of the signal to drive the regulators is the next frontier, as it were. I veer between thinking it potentially worthwhile and being cynical that it will just be another marketing gimmick (I can already imagine the Stereophoole review). I'm pretty happy with the state of the audio art as it is, and even listening to vintage tube gear with the loosest of specs can be pretty damn enjoyable. I'm bracing myself for one of my highender friends saying I'm taking all the life out of the music by being so precise.

After this, the electronics will have to become precognitive.
 
facinating post

[quote author="bcarso"]
Beyond this, I think the explicit use of the signal to drive the regulators is the next frontier, as it were. [/quote]

do you have any more info on this? wouldn't each amplifier stage need to have its own regulator? if not, then what is the "signal" that you drive the regulators with?

essentially this would lead to a system where the supply is unregulated and each amplifier has nearly infinite PSRR. I think.

mike
 
Hello bcarso,

I would be very interested in seeing your schematics of your current sources feeding shunt regulators!

Sorr
 
[quote author="Sorr"]Hello bcarso,

I would be very interested in seeing your schematics of your current sources feeding shunt regulators!

Sorr[/quote]

There are some schematics in an AES paper from the 1997 NY convention, that are representative. PM me and in the meantime I'll try to find the exact reference (it's something with "computer-controlled" and "attenuator" in the title).
 
[quote author="mikep"]facinating post

[quote author="bcarso"]
Beyond this, I think the explicit use of the signal to drive the regulators is the next frontier, as it were. [/quote]

do you have any more info on this? wouldn't each amplifier stage need to have its own regulator? if not, then what is the "signal" that you drive the regulators with?

essentially this would lead to a system where the supply is unregulated and each amplifier has nearly infinite PSRR. I think.

mike[/quote]

It's an area of research and speculation mike. But one idea is to have a composite model of the circuit or circuit section that you are interested in regulating, some kind of "blameless" signal pickoff, and then to use the signal and said model to produce the precisely compensatory loading on the supply. Note that the (EDIT: external) loads need to be specified too.

You can get into an infinite regress here, but practically (if any of this is really necessary enough to deserve using that word to begin with!) I suspect you can do pretty well with just the one pickoff/drive method.

You can also design the given stage to have high PSRR to begin with, and there have been some threads mentioning this I recall. The bottleheads seem to be more interested in it than the sandstates.
 
It's an interesting thread for sure but it still brings up the question, when do we *really* need to apply an advanced technique for regulation, and it is really regulation or is it noise/harmonic suppression?

A Vreg usually does an admirable job of voltage regulation but it does very little in the way of suppressing noise on the rails because it tracks so very slowly among other things, thus the reasoning for using caps, inductors and resistors to filter heavily before and after the regulator.

We have discussed various regulation methods but maybe we should be discussing how to make the average Vreg work better?

And yet again the question comes up, when is good enough actually enough?
 
Something I read years ago kind of off and on topic.

1977? audio critic had a writeup with a Mitch Cotter amp. One of the cool things noted was the inverting of the signal on one side of the amp and reversing the speaker wires on that side. This was so there would be more equal loading of the + and - sides of the power supply.

I believe the Dyna 70 was designed to have a more controlled loading of the power supply from what I have read.

I think the current waveform is sometimes overlooked by people.

In the digital world TTL ECL etc day much work was done to min any switching current spikes caused by enabling a bunch of gates to a buss at once ........

I try to balence stuff when I can when one device is sourcing current I try to have one sinking current on the same supply.

So this is nothing new. I have not been able to find a cool article that Intel had on current flow in digital circuits. I think it was from the late 70's or early 80's. Good article anyone know of this?
 
Back
Top