Using the LM317 with higher voltages

I imagine there's no way to, but can we trim the regulated voltage output on the single transistor regulator?
With the zener, there is no clean way: you will ruin some parameter.
An adjustable reference like the TL431 could do the job, except the 431 is limited to ~30V.
If you can find a 431-like circuit withstanding 70V, it would be possible

That would certainly be a good reason to go back to the 3-transistor regulator, because we have a trimpot there.
Or the 2-transistor one

Another thing: how many volts more does the booster zener have to be to consider the regulator's drop?
Apparently, it requires about 10V dropout to begin to regulate.
Add 20% + 1V as a safety margin, and you are at 13V.

Since a higher voltage comes almost for free, use 15V to be on the safe side.

If you use the smooth, non-tracking version, the zener must be 15V higher than the highest possible V+ voltage.

The 2 and 3 transistor regulators are LDO's, and require <1V to regulate properly
 
With the zener, there is no clean way: you will ruin some parameter.
An adjustable reference like the TL431 could do the job, except the 431 is limited to ~30V.
If you can find a 431-like circuit withstanding 70V, it would be possible

So the resulting V+ and V- voltages would always have some difference. Which Luxman didn't seem to care, or they hand picked the zeners to null the differences.

Or the 2-transistor one

Just compared them, adding the same RC input filter to both. The two-transistor reg is still 10dB worst than the single one @ 10Hz.

Increasing to 220 ohms /1000uF decreases that, but is it worth it?


The 2 and 3 transistor regulators are LDO's, and require <1V to regulate properly

Which would be the drop with the 1 transistor reg?

In any case, would it make much of a difference, audio quality wise:

1) To have a slight difference between regulated V+ and V-?

2) To be 10dB worst at 10Hz?

3) None of them matters much.
 
So the resulting V+ and V- voltages would always have some difference. Which Luxman didn't seem to care, or they hand picked the zeners to null the differences.
Both possibilities are plausible. An asymmetry of a few % normally makes no difference



Just compared them, adding the same RC input filter to both. The two-transistor reg is still 10dB worst than the single one @ 10Hz.
That's unfortunate, but it is the price to be paid for the LDO, good DC performance and low output impedance.
Since the limiting factor is also the Early effect, you could use the same tweak as for the 3-transistor, and gaining 10dB without adjustment is probably possible, but if you want more, an individual adjustment will be necessary
Increasing to 220 ohms /1000uF decreases that, but is it worth it?
I don't think so




Which would be the drop with the 1 transistor reg?
Around 10V, as I said

In any case, would it make much of a difference, audio quality wise:

1) To have a slight difference between regulated V+ and V-?

2) To be 10dB worst at 10Hz?

3) None of them matters much.
Normally 3), if the amplifier is designed correctly, but it is extremely difficult to be accurate without examining the actual circuit.

Just for the fun of it, to see what did change, I added the capacitance multiplier used by Goldmund Mimesis 9 to the 2 transistor regulator.

This time I got -100dB at 10 Hz.
You can stack regulators, passive & active filters ad infinitum and get -300dB PSRR, but what's the point?

What are the pros and cons of capacitance multipliers?
You can find all the information on the net or this forum, but remember that cap-mults, in their simple form are just like simple regulators, except they don't regulate, and their PSRR is limited also by the Early effect
 
OK, things are getting to a nice closure over this power amp supply.

Let me tell you what I think it's worth using.

1) LDO, good DC performance and low output impedance are interesting. Point to the two transistor regulator.

2) Asymmetry is not essential, but easy to have with a trimpot. Point to the two transistor regulator.

3) Adding a CM adds one more transistor, but I get better specs than with the three transistor regulator. Point to the two transistor regulator with CM filtering.

The would provide me with an adjustable LDO, good DC performance and low impedance regulator. Am I right?
 
The would provide me with an adjustable LDO, good DC performance and low impedance regulator. Am I right?
Yes, it a good tradeoff for a decent regulator. Being fully discrete, it will be quite robust if you chose a sufficient pass transistor. You can also add a protection transistor if required, and the circuit has not been tweaked in any way: by optimizing the bias currents, E-B resistor etc., you can certainly scrape some dB's of performance here and there.
 
Yes, it a good tradeoff for a decent regulator. Being fully discrete, it will be quite robust if you chose a sufficient pass transistor. You can also add a protection transistor if required, and the circuit has not been tweaked in any way: by optimizing the bias currents, E-B resistor etc., you can certainly scrape some dB's of performance here and there.

First of all, thanks for your help in getting here. To all of you.

Currents will be quite low for me to worry about the pass transistor being "sufficient".

How and where should I add a protection transistor, and why? I will add DC protection circuit, also inspired on the one used by Goldmund, which also protects from oscillations.

Now I will open another thread, as I had mentioned, for the RIAA preamp's power supply. And your Denoiser and NoNoise design will be main contenders.
 
How and where should I add a protection transistor, and why? I will add DC protection circuit, also inspired on the one used by Goldmund, which also protects from oscillations
A protection is always optional, and you could think that it is only needed in bench supplies, and you could be right -most of the times-, but it is inexpensive, easy, and can save the day when you are experimenting with your circuit, or if a cap goes short, etc.


Here is an example of such a protection, dimensioned for ~60mA:

attachment.php


Thanks to a strange LTspice quirk, the first load is 200ohm and corresponds to the 24mA trace.
60mA is for 1K
 

Attachments

  • Carl62f.png
    Carl62f.png
    68.8 KB · Views: 438
  • Carl62c.asc
    2.8 KB · Views: 45
Great. Of course I will add the protection. Thank you.

You had also suggested optimizing bias currents and the E-B resistor. Would that show on the simulation or only with the supply working?

BTW: I did open another thread for the RIAA preamp PS, and both your D-noisator and NoNoise are on the options. If you can go have a look it would be great.

Looking for a good power supply for a RIAA preamp

There's already a sim loaded, from the original supply, but it's not running as it should, probably because some variable is wrong. The regulated output voltage is showing, but now the curves I expected.

I would like to sim those supplies for PSSR, noise and impedance, if possible. Like you did with your NoNoise.

One thing I perceive with your regulators is that they are all positive types, and I am not sure I will have separate secondaries to use two positive regs. AFAIK only 317/337 have pos and neg versions. Is that so?
 
Some further investigation on the High voltage low current regulators we had been working with.

This time the theme is impedance.

Hopefully the simulating setup is correct.

Curve results are not as good as the superregulators and similar that are being analyzed on another thread.

The unsolved question is how a power supply impedance affects the audio quality. Or how PSRR or how noise does affect sound quality, particularly on this case where the signal levels we are dealing on power amps are a lot higher than on sensitive low current preamps.
 

Attachments

  • Luxman 5M21 optimal_impedance.jpg
    Luxman 5M21 optimal_impedance.jpg
    140 KB · Views: 264
  • Carl62c_impedance.asc
    2.7 KB · Views: 39
An strange case for this comparison was between these two versions of the same original two transistor regulator.

One with transistor protection and the other with a capacitance multiplier between input filter and regulator.

As it can be seen, the version with the CM behaved slightly better than the other one in impedance response.

Stranger yet, even if not shown, was that adding the CM to the protected regulator resulted in an even worst impedance response.
 

Attachments

  • Impedance comparison#1.jpg
    Impedance comparison#1.jpg
    282.2 KB · Views: 256
So you have to get to a compromise between output Z degradation and better PSRR from both regulators?

Though I wonder if this would result in audible changes for different combos.

When I first read about output impedance in a power supply in The Audio Amateur, the idea seemed to be that a wideband lower impedance on a power supply would result in a "better" audio quality. Some instead of "better" describe what they think changed, which I think is more objective.
 
You could try something like the circuit below, it's a bit more complex than the 3 transistor circuits but will provide much lower output impedance and 90 dB line rejection in audio band. It does need some more dropout voltage for the bootstrapping to work, otherwise it can be replaced with a CCS. This is the negative version, just invert polarities and flip diodes for positive version.

attachment.php
 

Attachments

  • reg1.PNG
    reg1.PNG
    31 KB · Views: 446
Sure, here you go. Note that C13 includes ESR and ESL specified. The ESL comes from another thread here where people measured actual ESL of real electrolytic caps. Hence the presence of the output Zobel, without it the output could ring when stimulated with output current pulses.

If you get bored, also attached is the original design which is even more complex. I call it the Blameless regulator since it resembles Doug Self's Blameless Amplifier reconfigured for regulator service. This is more of an academic exercise at this point though I would not be surprised someone actually trying to build it. It even has two-pole-compensation! Maybe I'll put it in a new thread if people care.

attachment.php
 

Attachments

  • regulator1.asc
    5.9 KB · Views: 42
  • blameless_regulator.asc
    7.3 KB · Views: 45
  • blameless_reg.PNG
    blameless_reg.PNG
    29.4 KB · Views: 418
AX tech editor
Joined 2002
Paid Member
Some further investigation on the High voltage low current regulators we had been working with.

This time the theme is impedance.

Hopefully the simulating setup is correct.

Curve results are not as good as the superregulators and similar that are being analyzed on another thread.

The unsolved question is how a power supply impedance affects the audio quality. Or how PSRR or how noise does affect sound quality, particularly on this case where the signal levels we are dealing on power amps are a lot higher than on sensitive low current preamps.

If you replace R2 by a zener and D1, D3 by a resistor and make Q1 an NPN, it'll get better.

Jan