Best low noise regulator?

[analog_sa]

Spent the last week trying to find something which will power an opamp phono (627->637) and sound better than an ALW-Jung regulator.

In short: nothing came close.

What i tried:

1. 317/337 followed by a cap multiplier. Claim to fame: JC uses the topology in the Vendetta and BT. I was a bit worried about starving the four opamps for current and instead of parallel fets used a bipolar multiplier as Teddy Pardo.
Pleasant, "natural" sound but not much detail or particularly wide soundstage. Diffuse sounding pianos. Still, an obvious improvement upon a 317.

2. Simple 3 transistor regulator using a board from the UGS group buy. A differential amplifier and a pass mosfet (IRF9610). Obviously not so low noise. Similar sound to above.

3. Used a fast opamp (AD811 - 45mA max current) with a low noise voltage reference. Opamp output feeds the phono pre directly. 317/337 used to power opamp.
Interesting sound. Larger than life dynamics, extreme detail in high frequencies. Very impressive on some records but something is obviously not well. Slightly peaky midbass and generally nervous sound. Oscillations that i cannot see on a 60MHz scope? Tried the same using more ordinary opamps. Less nervousness but still not great.


These are not the first regulator experiments i've made. Looks like open loop designs don't attract me too much, at least when powering opamps. This is so uncool

Any ideas i've missed? Is the ALW really the best regulator to power an opamp circuit?

Previously i didn't like the sound of simple shunts (CCS feeding a TL431 with a bipolar shunt). Should i re-examine them? Any particular recommendations?

Not surprisingly the ALW is very sensitive to the opamp used. Its sound flavour is very similar to that of the opamp used as an amplifier. I have had best results with AD797 and OPA627 and less pleasant with the AD825.

[syn08]

Originally posted by analog_sa
Any ideas i've missed? Is the ALW really the best regulator to power an opamp circuit?

Previously i didn't like the sound of simple shunts (CCS feeding a TL431 with a bipolar shunt). Should i re-examine them? Any particular recommendations?

Try this one, you may be very pleased with. I'm using it in my phono preamp. You may replace the resistor with a current source, although I did not find this having any significant impact. Force/sense wiring rules still apply for optimum performance.

http://www.diyaudio.com/forums/showt...61#post1735561

Opamp is THS4031, but other high speed low noise models can be used. LED is a red or infrared model (others have larger noise, definitely not blue or violet).

[analog_sa]

Thanks. I had something similar in mind, probably with a lower noise voltage reference in order to keep the output cap small. No idea what the specs on the mosfet are but i guess an IRF610 will work just fine. Will report the result soon.

[mlloyd1]

analog_sa:

did you try the AD817? walt recommended it in his design. i've used it also and found it works well.

mlloyd1

[syn08]

Originally posted by analog_sa
Thanks. I had something similar in mind, probably with a lower noise voltage reference in order to keep the output cap small. No idea what the specs on the mosfet are but i guess an IRF610 will work just fine. Will report the result soon.

It's a high current (60A if memory serves) MOSFET from ST, in TO220. I had these lying around but any other model will work. You may need to adjust the gate resistor.

[analog_sa]

mlloyd1:

no, don't have any



syn08:

Your resistor values don't seem to make sense. Easy to correct.

[aparatusonitus]

Hi analog_sa,

Image below represents Walt Jung (positive and negative) SupeRegulators build by a friend of mine, but PCB design is all mine. Both regs PCBs have full rectification and CRC filtration on board prior to regulation, error opamp used in both regs is an Analog Devices OP27. All measurement are done by him, so I really don't know how well/bed setup was, I have never had a scope, but anyway...


Next image represents residual noise of the scope not connected to anything (top line) and noise at the Jung regulator output with resistive loading (bottom line) measured as stand alone (outside the Marantz CD14 cd player), vertical scale is 5mV on all images.


Jung regs build in Marantz CD14 powering complete analog section after TDA1547 dac.


Noise at Jung regs output when they are build in Marantz CD14.


Output noise at original Marantz CD14 discrete regulator powering HDAM opamp modules.


Output noise at original Marantz CD14 discrete regulator powering TDA1547 DACs


Output noise at original Marantz CD14 LM7805 regulator powering other digital circuits

[analog_sa]

Something's bugging me. There is a large number of opamp driven shunt regulators similar to Syn08's design. No idea who invented the topology but the first implemenation i remember seeing was Allen Wright's SuperReg. All of these bar Manfred Huber's design run the opamp open loop. Is this good? How is overshooting? Performance at high frequencies should obviously suffer. Any comments?

[syn08]

Originally posted by analog_sa
Something's bugging me. There is a large number of opamp driven shunt regulators similar to Syn08's design. No idea who invented the topology but the first implemenation i remember seeing was Allen Wright's SuperReg. All of these bar Manfred Huber's design run the opamp open loop. Is this good? How is overshooting? Performance at high frequencies should obviously suffer. Any comments?

The HF performance suffers mainly because of the output capacitor! This cap may resonate with the inductive output impedance of the regulator, to create a peak in the output impedance. This peak is also highly dependent on the opamp open loop characteristics. For the THS4031 and 100uF (low ESR) at the output, I was able to measure the resonance at about 900KHz (in good concordance with the simulations). Fortunately, the output impedance peak is only up to under 1miliOhm so everything is fine. Ideally, you should not use any output cap, but usually it's a trade.

The same applies to series regulators, however the parallel regulator seem to be more sensitive to this effect.

I'm not familiar with the Manfred Huber design, any link to?

Anyway, how's the thing working for you?

[analog_sa]

This is one version. Another has the gain fixed at 100 and using an OPA655.