low noise regulator with reference and buffer?

[t52]

has anyone tried this: use a good zener or voltage reference, noise as low as possible and feed the generated reference voltage into a buffer like the buf634... wouldn't that be a good low noise regulator for low current demands?

 

[Garbz]

The combination as far as I understand it would be no better then building a shunt regulator out of a zener. Infact the buffer could be a limiting factor. In essence though what you suggest is actually how many regulators work. Fixed regulators and even generic LM317 adjustable regulators generate a voltage and using an intricate feedback network ensure the output is quiet and noise free.

Suggesting something though would depend on it's application. For a digital circuit I've used a shunt made from an orange led once, noisy as it may have been it didn't matter. If you're looking for an excellent performing regulator with very low inband noise www.waltjung.org the jung super regulator is hard to pass up. For general audio excellent wide band performance can be achieved with a simple $3 TL431 programmable shunt, and for low inband noise even an LM317 correctly implimented can achieve >80db noise reduction.

 

[tangent]

The BUF634 doesn't spec output impedance, but you can see resistors on the output of the simplified circuit diagram, and we know it has no feedback. Significant output impedance and no feedback means that any change in current demand by the load will change the voltage drop across the output. That makes for a pretty lousy regulator.

 

[rickcr42]

the option would be to use the zener/TL431 shunt the follow it with a feedback buffer to lower the noise and get a lower output Z.

something like this

http://www.maxim-ic.com/appnotes.cfm/appnote_number/899

Just watch botht e idle current and the current drain of the entire circuit and adjust all the elements of the V-reg accordingly.Look at the TL431 data sheet from Fairchild for deas on how to work it out

 

[Garbz]

AHHHH IT'S RICK

Btw that maxim diagram is basically a simplified schematic of the jung super regulator, with the exceptions that the Vout, GND, Vsens, and GNDsense lines aren't seperated.

 

[t52]

thanks for all the input. i should have forced my brain into thinking before posting this

obviously a buffer like the buf634 is not very well suited here because of missing feedback. just to bring the thought experiment to an end, if i'd use a opa551 or any other high power opamp here in unity gain and feed the Vref into the noninverting input, this would probably be no different from any series regulator, right? the output impedance of the opamp should be quite low because of the feedback applied. possibly this could even be better than a series regulator because the opamp also has a second output device hooked up to the other supply rail and so it can also sink current (whats that then - a series AND shunt regulator?).

i'm just a lazy guy and like to use integrated stuff wherever possible

btw. thanks for all the links. i'm drowning in ee related stuff to read lately, which is a good thing...

 

[robzy]

Have you read http://tangentsoft.net/elec/opamp-linreg.html ? You might find it interesting.

Rob.

 

[rickcr42]

Quote:

AHHHH IT'S RICK

wandered in by accident (damn google searches keep turning up headfi

) before I realised it then could not help responding.sent myself back to exile until one of those damn "responded to post" emails popped up that I haven't had in months and well,again could not resist the response so will do it this one time then back to exile

Quote:

Btw that maxim diagram is basically a simplified schematic of the jung super regulator, with the exceptions that the Vout, GND, Vsens, and GNDsense lines aren't seperated.

yup

Kinda the simplified version,one easy to layout and use.The sense lines are nice if you plan on having any real distance between the ps and the amp but otherwise you tie them together anyway so can be tossed for most apications where amp and power supply are either in the same box or on the same shelf

Quote:

i'm just a lazy guy and like to use integrated stuff wherever possible

a couple of the LDO chips are damn near identical to the Jung regulator as are almost all low voltage/low current types so you need to use a pass transistor for increased current and a low noise voltage reference on the control pin (can even be a zener just add the zener voltage to the drop out voltage for the final figure) to raise the output voltage .

there is no one size fits all general duty power supply even though all will get you a stable voltage.Very situational and circuit dependant so for instance with a high gain circuit I would not look past a Jung Reg or similiar ultra low-noise type,for typical opamp circuits where current draw will fluctuate maybe an optimised three pin type and with a Class-a stage where current does not change a shunt regulator

 

[tangent]

Quote:

Originally Posted by t52 if i'd use a opa551 or any other high power opamp here in unity gain and feed the Vref into the noninverting input, this would probably be no different from any series regulator, right?

Oh, there are still a few differences:

1. The output structure of an op-amp isn't really made for passing constant DC current. It can certainly do it, but if you knew that's what the design goal was, you would design it differently....like in the article robzy pointed you to.

2. Although the OP551 does indeed have higher output current than the average op-amp, it's still nowhere near what you'd get if you used a regular op-amp as an error amp to control a pass transistor...again, as in that article. You could step up to a true power op-amp like the LT1210, but that's using a sledgehammer to hunt butterflies.

 

[peranders]

t52, what currents and voltages did you have in mind? This is rather important if you want to get possible solutions.