D-Noizator: a magic active noise canceller to retrofit & upgrade any 317-based V.Reg.

[Elvee]

Originally Posted by Mark Johnson

Summary: low noise voltage regulator achieved using twelve electronic components, one of which is an integrated circuit.

This is just an add-on, but I would love to see a superreg in its own right using <12 components, of which none is an IC.
The current season is certainly the right time for such a development....

Is someone up to the task?

v

v

v

v

[Mark Johnson]

I would love to see a low noise voltage regulator using twelve or fewer electronic components, and I don't care how many or how few of them are ICs. If one IC is allowed, seven are also allowed.

[Zung]

If low frequency steady state power supply rejection is your primary concern, you can cascade regulators. This is one of the options I considered (and rejected) for the heaters of the preamp I'm currently working on: a cap multiplier as a pre-filter for the regulator. Sim yields 1.3uVpp@100Hz, but it doesn't account for the random noise of the regulator itself.

[Mark Johnson]

Why no Zener diode at the base of Q1? It will increase mains ripple rejection ("PSRR at low frequencies") by many decibels.

[stormsonic]

@Mark Johnson
Are you familiar with Maxim app notes 3656 and 3657 ?

[Zung]

Originally Posted by Mark Johnson

Why no Zener diode at the base of Q1? It will increase mains ripple rejection ("PSRR at low frequencies") by many decibels.

For no particular reason, it's just a proof of concept. But you got me curious, so here it goes: no change, even after I force double precision.

The explanation: the 1000uF cap is worth about 1.6 ohms at 100Hz, whereas the Zener typically is 7-15 ohms, so it's not very effective in this case.

[gpapag]

Thanks Elvee.
The "Finesse" circuit works OK (and it can be adjusted)

Performance measurements are here (page 2, 3)
https://www.edn.com/electronics-blog...or-noise-floor

And a strong enough noise source (for testing)
https://www.edn.com/design/analog/44...-1Hz-to-100kHz

George

[Mark Johnson]

Yes Maxim application note 3657 looks like it contains some good ideas. Pair it with a 36V supply, RRIO opamp like LT1498 and you can use the Walt Jung trick of connecting the opamp's power supply input pin, to the final regulator's output. With a RRIO opamp you don't need Walt's level shifter. Be sure to filter the holy heck out of the reference voltage (Maxim R1 and C1), I've used 3 terminal "feedthru" capacitors for this purpose. Such as YFF31AH2A105MT0Y0N.

[gerhard]

OMG. A LT3042 with an external transistor to carry the dissipation is
> 40 dB better than any LM317, right from the start.
Circuit is in the data sheet.

There is no point to mess with the LM317's loop gain.

Oh, yes, it's tested:
< Lt3042_DH44 | Low noise regulator consisting of LT3042 and D… | Flickr >
< spectrum_lt3042_DH44 | Noise spectrum of the LT3042+ D44VH10… | Flickr >

And Wenzel's finesse thing cannot be tuned, it NEEDS to be tuned
for optimum results.

cheers & happy winter solstice

Gerhard

[jackinnj]

Originally Posted by gerhard

cheers & happy winter solstice

Gerhard

The days are getting longer.

The easy to characterize parameters for a super-reg:
1) PSRR
2) Noise
3) Z out

The (somewhat) harder to characterize parameter:
4) THD

The error amplifier can't be damped as it will impinge back on the rails part of the problem it just saw. Yet, the error amplifier has to be stable. * This is what one of the respondents to the Linear Audio article commented, and the bit of analysis I did, correlating FFT with the perceived results, seemed to bear it out. Unfortunately, or fortunately, Jan has better things to do with his time and there doesn't seem to be an interest in further research.

I know a bunch of you guys do this for a living, for me it's an avocation.

* -- as I've mentioned before, I could get one of the Jung regulators to play WABC-770kc a.m., the 50kW broadcast blowtorch of New York City, when used with an AD797. No such luck with the AD825AR.