Simple Low Noise Voltage Regulator?

[lxmyers]

Hello All,
A couple of days ago I asked a question about an Audio Buffer for a 125ASX2, all the replies were very helpful and I have decided to try and build one of the recommended Buffers. I am having trouble figuring out the Power Supply for the OP-AMP, the 125ASX2 has an unregulated aux supply that outputs ±24VDC and the Audio Buffers need ±15VDC.

I've spent the last day looking through the forums and found many designs all with vary levels of complexity. Unfortunately space is an issue in this case so a smaller circuit would be better.

It seems that the Super-Reg is considered one of the best however, it is a rather complicated circuit with many components and takes up quite a bit of space.

On the other hand, the LM317/LM337 has five components, takes up very little space but is rather noisy by itself.

There are circuits like Fred Dieckmann's "one transistor addition to the three terminal regulator" and this "Clean-up Shunt" that could be space efficient but maybe not suitable...

The Question
What would you recommend as a good, simple (LM317/LM337 based?) Low Noise Voltage Regulator for a simple OP-AMP Buffer?

Any help would be greatly appreciated.
Thank you in advance,
Alex

[jitter]

The standard LM317/337 circuit as is is already excellent for low noise. Trying to clean the output even more might just be overkill when opamps are fed.
Opamps themselves have a very good power supply rejection ratio (PSRR). That PSR suppresses the very little noise that might still be there after the regs even quite a bit more.

In fact, when space is really an issue, I would simply use 7815/7915 combo with HF bypassing caps as close to the regs as possible. Dissipation permitting, you might even use the smaller 100 mA version (78L15 and 79L15).

[lxmyers]

Thanks Jitter, that makes things a lot easier!

Would it be a good idea to match the voltages of the LM317 and LM337 using matched resistors or a potentiometer? Would the Opamps care if the positive rail was (extreme example) say 15.24v and the negative rail was 14.91v?

Thanks again.
Kind regards,
Alex

[jackinnj]

If you bypass the adjust pin of the LM317/337 you knock down the noise tremendously. There's a Bob Pease ap note on this on National's site. The reduction in noise comes at the expense of impulse response, however.

[jitter]

Originally Posted by lxmyers

Thanks Jitter, that makes things a lot easier!

Would it be a good idea to match the voltages of the LM317 and LM337 using matched resistors or a potentiometer? Would the Opamps care if the positive rail was (extreme example) say 15.24v and the negative rail was 14.91v?

Thanks again.
Kind regards,
Alex

No, that's not necessary. PSRR also reduces the effect of out of balance power supply. An opamp still works fine on (e.g.) a +/- 15 V power supply with +/- 0.5 V tolerance on both rails.
Besides, even if power supplies were perfectly symmetrical, the + and - input are never 100% identical leading to DC offset on the output anyway.

In critical DC-coupled applications output offset needs to be nulled and that allows only for the use of opamps with offset adjustment capability. In audio DC is usually dealt with by DC-blocking caps to circumvent the whole offset adjustment procedure (you just don't want that time consuming hassle in mass production). The popular dual opamps in 8 pin cases don't have offset adjustment available.

[AndrewT]

the Manufacturers of single amp 8pin opamps, that have DC offset adjustability, seem to remind us that the temp co changes much for the worse if one invokes that adjustment.

[martin clark]

An aside, but don't use the DC offset adjustment if you don't absolutely need it on any opamp. It is an easy and excellent way to utterly ruin PSRR. ( - and if you do need it, ensure the supplies are filtered before the bias adjust pot biasing supply is heavily pre-filtered)

Similarly teh adjustment should only be used to null the opamps output to 0v - not 'system output' (eg at the output jacks). That often ruins performance by excessively unbalancing the LTP pairs internally!

[ChristianThomas]

My advice would be to use the 317/337 with 10uF bypass. If you put a cap on the output as well, use something like a tantalum which will have 4 to 8 ohms of ESR and will damp any resonances. Don't go above 10uF as you will bring the resonance of the o/p impedance into the audio band.

IIRC you shouldn't need a buffer. The input Z is 8k I think and anything should be able to drive that. If you put an op amp in, then take the opportunity to put in a little filtering. 80 - 1000kHz, single capacitor around 220pF.

CT

[Piersma]

For Your Information:

Simple Voltage Regulators Part 2: Output Impedance

Using 3-pin regulators off-piste: part 1

[QSerraTico_Tico]

Noise or output impedance do not control sound quality...