diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Power Supplies (http://www.diyaudio.com/forums/power-supplies/)
-   -   Improving on LCaudio LowNoise regulator (http://www.diyaudio.com/forums/power-supplies/45451-improving-lcaudio-lownoise-regulator.html)

ABO 11th November 2004 09:00 PM

Improving on LCaudio LowNoise regulator
 
I just ordered one of these regs.

http://www.lcaudio.dk/com/dcreg1.jpg

Looking at the schematic I see some possibilities for further improvement of the noise performance.

http://www.lcaudio.dk/com/dcreg.gif


As I am merely an armchair designer, I am looking your opinions on my ideas:

- Bypassing D20 (led) with 47 uF in order to filter out led noise;

- Adding a cap across the lower leg of the potmeter ( from opamp in+ to ground) to create some lowpass filtering, in order to filter out zenernoise;

- Decreasing noise gain of both opamps by bypassing R39 and R36 with appropriate cap, some 22 uF;

- Adding power supply bypasses (10-22 nF) for the opamp;

- ??

What do you think? Will this work and improve the performance of this circuit?

thanks,

Abo

By the way: Does anybody know what that thing is across the higher leg of R35?

ABO 16th November 2004 07:54 PM

Surely someone must have ideas?

Lars Clausen 16th November 2004 08:26 PM

Hi

That thing on R35 is a bubble to enable the regulator to be used at full voltage with no trimmer.

I personally don't subscribe to filtering out Zener or LED noise with caps, as they will not remove noise, merely move the noise to lower frequency zones. Better is to use a low noise reference, that don't need filtering.

Cheers

LC

jean-paul 16th November 2004 08:43 PM

Quote:

Originally posted by Lars Clausen
Better is to use a low noise reference, that don't need filtering.

Please send that design improvement to the designer ;) It could have been implemented in the design.

At least the decoupling of the opamp supply lines won't hurt.

Quote:

Originally posted by Lars Clausen
I personally don't subscribe to filtering out Zener or LED noise with caps, as they will not remove noise, merely move the noise to lower frequency zones.
??

Elso Kwak 19th November 2004 02:57 PM

Questions
 
Quote:

Originally posted by ABO
Surely someone must have ideas?

Hi Abo,
I would be more inclined to ask:
Why MOSFETs as the pass-element?.
Why a constant current source for the reference comprised of two transistors, three resistors and a blue LED?.
Why the antiquated NE5532 as the opamp? Then.....
Just some questions.:confused:

Lars Clausen 20th November 2004 08:29 PM

Elso: Sorry i didn't see your questions until now. Here are some answers for you:

1..MOSFET's are perfect a pass elements, as they require no drive current, and will stand high loads. I see no drawbacks of using MOSFET's, the highest pole of the regulator loop is handled by capacitors, so a particularly high loop speed is not required.

2..The constant current source, for which the base current is pre regulated using the LED as a sort of Zener shunt, gives the Zener perfect working conditions. Free of the load regulations found in other types of low noise regulators taking the Zener current from the output of the positive regulator, and thus allows the reference to fluctuate when the regulator loop reacts to changing loads. (If present).

3..Well the NE5532 may be old, but as i often tell my employees, just because something (or someone) is old, doesn't mean it's useless. :D The NE5532 is very low noise (definitely not the bottleneck in any regulator, the Zener will most probably have 100 times higher noise than the op-amp) In this case the Zener has typ 7 uV RMS noise, (Max 100 uV) while the NE5532 has just around 30 nV RMS input related noise.
So it makes no sense to use a lower noise op-amp.
And it has one great advantage over any ultra low noise opamp: it can operate on +/- 22V (Philips brand).

I hope this answers your questions.

All the best from

Lars

Elso Kwak 21st November 2004 04:47 PM

Improving on LCaudio LowNoise regulator
 
Quote:

Originally posted by Lars Clausen
Elso: Sorry i didn't see your questions until now. Here are some answers for you:

1..MOSFET's are perfect a pass elements, as they require no drive current, and will stand high loads. I see no drawbacks of using MOSFET's, the highest pole of the regulator loop is handled by capacitors, so a particularly high loop speed is not required.

2..The constant current source, for which the base current is pre regulated using the LED as a sort of Zener shunt, gives the Zener perfect working conditions. Free of the load regulations found in other types of low noise regulators taking the Zener current from the output of the positive regulator, and thus allows the reference to fluctuate when the regulator loop reacts to changing loads. (If present).

3..Well the NE5532 may be old, but as i often tell my employees, just because something (or someone) is old, doesn't mean it's useless. :D The NE5532 is very low noise (definitely not the bottleneck in any regulator, the Zener will most probably have 100 times higher noise than the op-amp) In this case the Zener has typ 7 V RMS noise, (Max 100 V) while the NE5532 has just around 30 nV RMS input related noise.
So it makes no sense to use a lower noise op-amp.
And it has one great advantage over any ultra low noise opamp: it can operate on +/- 22V (Philips brand).

I hope this answers your questions.

All the best from

Lars
Quote:

Originally by MalichiConstant
I think most of the choices are for less $$. The choice of high current mosfets is a mystery to me since at currents of a few 10s of milliamps the transconductance is very low and the output impedance VERY high. Figure 4 volts turn on for the mosfets and a couple of volts overhead from the op amp output and the minimum supply voltage for the op amp is about Vout plus 6 volts or about 20 volts. This is about the highest recommended supply voltage for the NE5532 (and higher than most op amp supply limits). Anyone wanting +/15 volts out of this design topology is out of luck. Using a faster op amp or lower current mosfets without gate resistors is asking for trouble.

IMO this a very poor implementation for a low current op amp based regulator for a commercial regulator. The lack of filtering for the LM 329 defies logic other than cost savings run amuck. Spend another 10 cents for a cap and resistor if you are going to call it a low noise regulator.:smash:
Lars,
I am well aware that a constant current source improves the performance of the LM329, as I am using it all the time in my supply for the TDA1543. I was merely asking why this particular current source for the reference. Why not a red LED and a transistor or two FETs or simply a J505?
Your claim that the extended supply range of the NE5532 (+/-22V) is an advantage is invalid as you are using the opamp on a + or - 20V single supply to ground. 20V total span. So any low noise opamp designed for a regular +/- 15V supply will do.
It is intriguing that you are using the suggested noise filtering by MalichiConstant in you XO-3 circuit. (R14 and C7). I am afraid you are confusing noise filtering with noise shifting by a Delta-Sigma DAC. I never experienced an increase in low frequency noise when applying low-pass noise filtering.
Last question why not using a dual-tracking regulator? (Horowitz page 351)

MalichiConstant,
As Lars is running the opamps single supply he could use a higher supply voltage for more drive. This will increase the dropout voltage. But the dropout voltage of the regulator is already a hefty 7V.
:cool:

QSerraTico_Tico 29th November 2004 10:42 AM

Re: Improving on LCaudio LowNoise regulator
 
Quote:

Originally posted by Elso Kwak



Lars,
I am well aware that a constant current source improves the performance of the LM329, as I am using it all the time in my supply for the TDA1543. I was merely asking why this particular current source for the reference. Why not a red LED and a transistor or two FETs or simply a J505?
Your claim that the extended supply range of the NE5532 (+/-22V) is an advantage is invalid as you are using the opamp on a + or - 20V single supply to ground. 20V total span. So any low noise opamp designed for a regular +/- 15V supply will do.
It is intriguing that you are using the suggested noise filtering by MalichiConstant in you XO-3 circuit. (R14 and C7). I am afraid you are confusing noise filtering with noise shifting by a Delta-Sigma DAC. I never experienced an increase in low frequency noise when applying low-pass noise filtering.
Last question why not using a dual-tracking regulator? (Horowitz page 351)

MalichiConstant,
As Lars is running the opamps single supply he could use a higher supply voltage for more drive. This will increase the dropout voltage. But the dropout voltage of the regulator is already a hefty 7V.
:cool:


Hey Kwak, wake up!
The NE5532 is a dual opamp so Lars is using it on a dual supply. The OPA2604 could be used and this one can accomodate higher supply rails till +/-25V. Lars' regulator is not a dual tracking regulator as claimed on www.lcaudio.dk but could easily be modified to be one.
Here is an example of a dual tracking regulator:
(figure 3)
http://headwize.com/projects/showfil...lmore3_prj.htm
:att'n:

peranders 29th November 2004 10:55 AM

Re: Re: Improving on LCaudio LowNoise regulator
 
Quote:

Originally posted by QSerraTico_Tico
Lars' regulator is not a dual tracking regulator as claimed on www.lcaudio.dk but could easily be modified to be one.
Here is an example of a dual tracking regulator:
(figure 3)
http://headwize.com/projects/showfil...lmore3_prj.htm
:att'n:

Yes it is! Check carefully, once again.

QSerraTico_Tico 29th November 2004 11:06 AM

Re: Re: Re: Improving on LCaudio LowNoise regulator
 
Quote:

Originally posted by peranders


Yes it is! Check carefully, once again.


No, I a dual tracking regulator the negative output is just the inverted positive supply. Or in other words the regulated positive supply is used as a reference for the negative supply. In Lars' design the negative is obtained from the same reference, not the positive output.
Any disturbance on the positive line is not tracked by the negative line in Lars' design.:cool:


All times are GMT. The time now is 07:42 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2