Super low noise regulator

[Koifarm]

I was in need for a low noise regulator for a mc phono preamp.
Walt Jung/ Didden had a great design for it.

I used a better/newer opamp and lower noise references(LED).
The result was noise below -130db and very low impedance.











Gerber for PCB in attachment.

[magic]

That -130dB doesn't really tell much.
What's the FFT size and sample rate (or simply bin width) and soundcard full scale input voltage?

Generally, you shouldn't be able to see typical opamp noise at unity gain with a soundcard, so if you are seeing noise above soundcard noise floor it likely is the LED, which means that opamp choice may not make much difference.

If the output is shorted, those 10μF and 47μF caps will discharge through opamps's protection diodes, not ideal.

[sorin]

Not only that but the step response will be horrible.
Clearly he didn't know anything about the topic.

[floobydust]

OP's circuit only has a little resemblance to Walt Jung's Super Regulator. He used a few solid-polymer caps, had op-amp (differential) input clamp diodes. No electrolytics for feedback comp!
Also LM317 pre-regulator. Good reads, his Regs esp. how slow the 7815, LM317 can be https://refsnregs.waltjung.org/

Metrology guys might benefit from the circuit, it's very low noise.

[PCB.Wiz]

I used a better/newer opamp and lower noise references(LED).

Not bothered by temperature stability ?

Using LEDs as references is interesting, and these days may even be easier to do.

Modern LEDs have tighter process control, and often come with Vf bin markings per reel.
For thermal coupling there are good looking packages like SMD2835-2P, which can directly couple to a collector TAB of a modern LFPAK or similar.
The temperature coefficient of BJT and LED do not need to match exactly, as you can inject extra correction, based on either sensed voltage (temperature)

The limit here is the bends in the curves, once you have a match at a given temperature, the BJT and LED coefficients do not exactly track over temperature, so a parabola type correction results.

Probably more than good enough for Audio Amplifiers.

[Koifarm]

Thanks for the replies.

The -130db  reverences to 0db at 0,775Vrms.

External clamp diodes not needed because opa1611 has build in clamp diodes.

And it has less noise then LT3045/LT3094 combo.

Yes, It is good for audio like MC phono pre or discrete preamps.

[magic]

The -130db  reverences to 0db at 0,775Vrms.

OK, let's try differently, how much in nV per root of Hertz?
LT3094 claims 2.2nV/rtHz at 10kHz FWIW.

External clamp diodes not needed because opa1611 has build in clamp diodes.

Not sure if they are rated to discharge 10μF through a short circuit.

[AnalogTodd]

One additional question is what output voltage are you running at? The LT3045 and LT3094 have a distinct advantage in that the error amplifier always runs in unity gain, so the noise is the same whether you are getting 1V out or 15V out.

[Koifarm]

12V

[PCB.Wiz]

One additional question is what output voltage are you running at? The LT3045 and LT3094 have a distinct advantage in that the error amplifier always runs in unity gain, so the noise is the same whether you are getting 1V out or 15V out.

I guess one could always design a low noise, reference current source using a LED + Bipolar correction, and create the same structure.
Or you could amplify the voltage reference followed by a noise LPF, and a voltage follower stage.