The Well Regulated Power Supply - Page 6 - diyAudio
Go Back   Home > Forums > Amplifiers > Power Supplies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 10th June 2012, 05:18 PM   #51
diyAudio Member
 
andrea_mori's Avatar
 
Join Date: Jan 2005
Location: Italy
Default Two more options

There are two more options to set operating voltage of the OPA350 around
5V.
The first option uses an AD820 opamp as series regulator to supply directly OPA350 taking voltage reference from AD780.
The second option, the simplest, use one more LT1085 regulator.
In any case for 5V output regulator this section should be omitted, and the OPA350 should be supplied directly form series regulator output.
Attached Images
File Type: jpg HG-SHUNT_REGULATION_4V_OA.JPG (35.7 KB, 506 views)
File Type: jpg HG-SHUNT_REGULATION_4V_LT.JPG (32.8 KB, 491 views)
  Reply With Quote
Old 10th June 2012, 05:25 PM   #52
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally Posted by AndrewT View Post
Gootee,
looks like they have not read your two posts. They are still worried about the opamps.
Good decoupling at the load will reduce the slew rate of the step response on the supply rails and that reduces the need for ultra high speed opamps. They don't seem to see the connection.
AndrewT,

I actually hadn't thought of it that way, either. But you have made what could be a very good point.

I was originally only thinking that a more-realistic simulation scenario should be used, in order to give a more-realistic estimate of performance, to possibly enable better requirements and performance metrics to be defined, which might result in a design that would perform better in the real application.

I assumed that adding a simulated load scenario might expose some deficiencies that could require design improvements. But as you have pointed out, it might actually allow the performance requirements to be relaxed, somewhat.

Cheers,

Tom
  Reply With Quote
Old 10th June 2012, 05:26 PM   #53
diyAudio Member
 
andrea_mori's Avatar
 
Join Date: Jan 2005
Location: Italy
Quote:
Originally Posted by andrea_mori View Post
There are two more options to set operating voltage of the OPA350 around
5V.
The first option uses an AD820 opamp as series regulator to supply directly OPA350 taking voltage reference from AD780.
The second option, the simplest, use one more LT1085 regulator.
In any case for 5V output regulator this section should be omitted, and the OPA350 should be supplied directly form series regulator output.
Ooops.. there is an error in the AD820 schematic in the previous post: AD820 output connected to positive rail.
The following is the right schematic.
Attached Images
File Type: jpg HG-SHUNT_REGULATION_4V_OA.JPG (35.2 KB, 466 views)
  Reply With Quote
Old 11th June 2012, 03:54 PM   #54
diyAudio Member
 
andrea_mori's Avatar
 
Join Date: Jan 2005
Location: Italy
Default Ready to prototyping

Follows the revised schematic ready to build a prototype.
Some changes:
- AD797 op amp in series regulation stage
- AD780 now works in shunt mode
- AD820 supplies fixed 5V for OPA350, so it could be used at any regulator output voltage

Comments are wellcome

Andrea
Attached Images
File Type: jpg TWRPS-HG-POS.JPG (115.9 KB, 446 views)
  Reply With Quote
Old 12th June 2012, 01:28 PM   #55
UnixMan is offline UnixMan  Europe
diyAudio Member
 
UnixMan's Avatar
 
Join Date: Apr 2005
Location: Perugia + L'Aquila, Italy
Send a message via ICQ to UnixMan
Although it's true that high frequencies should be handled by (local) by-pass capacitors, I would rather avoid high feedback designs (thus Op-Amps) and would use simpler, faster (wide-bandwidth), all-discrete designs instead.

A simple integrated pre-regulator shall be more than enough to kill ripple and keep the voltage stable at the source. I don't quite see the need for a second cascaded one. To get better filtering, instead of a second series regulator I would rather use a CCS between the series (pre-)regulator and the output shunt reg (in parallel with plenty of by-pass capacitance).

The CCS very high series impedance followed by the very low shunt impedance of the shunt reg & by-pass caps forms a voltage divider (RC-like filter) which is extremely effective in cutting away any kind of noise coming from the previous stages (as well as blocking any noise from the load trying to travel in the opposite direction). Add RF-chokes and/or ferrite beads in series with the CCS to stop any HF/RF noise too.

N.B.: also for the CCSs you must use simple, wide-bandwidth, all-discrete designs. Do not use op-amp based CCSs such as a CCS-connected integrated regulator!

P.S.: use only Schottky barrier diodes for the rectifier! That's the only kind of diodes (besides vacuum tubes...) which do not suffer from reverse-recovery and associated noise (having only one type of carriers, there's nothing to recombine).

Another often neglected problem is the rectification noise due to the current peaks in the PT/rectifier/filter capacitors circuit. That's a huge source of wide-bandwidth noise which is radiated just about everywhere (and possibly conducted, too).

The solution is to slow down and drastically limit the amplitude of those current peaks. To do so, again I'd use a simple CCS (such as a depletion-mode FET or a couple of BJTs "back-to-back") in series between the diodes and the first cap. That will work sort of like an inductive-input (LC) filter, but without the many problems associated with the "swinging choke" (of course, as for the LC type filters, you'll need a higher voltage on the PT secondary).

It is even possible to use a modified (voltage-controlled) CCS circuit designed so that the "set current" will be reduced as the voltage on the filter caps approaches the desired one. Basically, that's sort of a current-limited, high-impedance series regulator working directly on the unfiltered (pulsing) voltage from the rectifier into the filter caps (but voltage pre-regulation and ripple reduction are just welcome side-effects: the actual goal is to reduce as much as possible the power and bandwidth of the rectification noise!).

Last but not least, you're better off distributing constant currents rather than constant voltages!

Yes, have the PSU output constant current(s). Have these DC currents routed through the circuit. Convert back to constant voltage only where required, that is right at the load(s) that needs it. Use fast, simple and small shunt regulators. One for each "atomic" load!

(use shunt regulators possibly as simple and small as just a Zener diode mounted right across the load pins, right next or even on top of the local by-pass capacitor!).

That way you can avoid interactions between different loads and minimize the area interested by the the load current loops.

That's particularly useful when you have to deal with loads that may produce fast (wide bandwidth) current pulses, such as any digital circuitry.

My 2.
__________________
Quote:
"We should no more let numbers define audio quality than we would let chemical analysis be the arbiter of fine wines." N.P.

Last edited by UnixMan; 12th June 2012 at 01:35 PM.
  Reply With Quote
Old 12th June 2012, 11:47 PM   #56
diyAudio Member
 
andrea_mori's Avatar
 
Join Date: Jan 2005
Location: Italy
Quote:
Originally Posted by UnixMan View Post
Although it's true that high frequencies should be handled by (local) by-pass capacitors, I would rather avoid high feedback designs (thus Op-Amps) and would use simpler, faster (wide-bandwidth), all-discrete designs instead.

A simple integrated pre-regulator shall be more than enough to kill ripple and keep the voltage stable at the source. I don't quite see the need for a second cascaded one. To get better filtering, instead of a second series regulator I would rather use a CCS between the series (pre-)regulator and the output shunt reg (in parallel with plenty of by-pass capacitance).

The CCS very high series impedance followed by the very low shunt impedance of the shunt reg & by-pass caps forms a voltage divider (RC-like filter) which is extremely effective in cutting away any kind of noise coming from the previous stages (as well as blocking any noise from the load trying to travel in the opposite direction). Add RF-chokes and/or ferrite beads in series with the CCS to stop any HF/RF noise too.

N.B.: also for the CCSs you must use simple, wide-bandwidth, all-discrete designs. Do not use op-amp based CCSs such as a CCS-connected integrated regulator!

P.S.: use only Schottky barrier diodes for the rectifier! That's the only kind of diodes (besides vacuum tubes...) which do not suffer from reverse-recovery and associated noise (having only one type of carriers, there's nothing to recombine).

Another often neglected problem is the rectification noise due to the current peaks in the PT/rectifier/filter capacitors circuit. That's a huge source of wide-bandwidth noise which is radiated just about everywhere (and possibly conducted, too).

The solution is to slow down and drastically limit the amplitude of those current peaks. To do so, again I'd use a simple CCS (such as a depletion-mode FET or a couple of BJTs "back-to-back") in series between the diodes and the first cap. That will work sort of like an inductive-input (LC) filter, but without the many problems associated with the "swinging choke" (of course, as for the LC type filters, you'll need a higher voltage on the PT secondary).

It is even possible to use a modified (voltage-controlled) CCS circuit designed so that the "set current" will be reduced as the voltage on the filter caps approaches the desired one. Basically, that's sort of a current-limited, high-impedance series regulator working directly on the unfiltered (pulsing) voltage from the rectifier into the filter caps (but voltage pre-regulation and ripple reduction are just welcome side-effects: the actual goal is to reduce as much as possible the power and bandwidth of the rectification noise!).

Last but not least, you're better off distributing constant currents rather than constant voltages!

Yes, have the PSU output constant current(s). Have these DC currents routed through the circuit. Convert back to constant voltage only where required, that is right at the load(s) that needs it. Use fast, simple and small shunt regulators. One for each "atomic" load!

(use shunt regulators possibly as simple and small as just a Zener diode mounted right across the load pins, right next or even on top of the local by-pass capacitor!).

That way you can avoid interactions between different loads and minimize the area interested by the the load current loops.

That's particularly useful when you have to deal with loads that may produce fast (wide bandwidth) current pulses, such as any digital circuitry.

My 2.
Hi UnixMan,
thanks for comments and suggestions.
BTW I have a different point of view about most topics you treated.

I don't think op amps are so diabolic as you said: I like vacuum tubes, now I'm building a modified Atma Sphere OTL , but I don't disregard what technology today offers. Op amps are very useful in many applications and one of these is just regulated power supply. I'm not the one, not the first and surely not the last who use op amps in this kind of applications, Walt Jung "docet". All his regulator are op amp based, and certainly he's not the last coming in this applications.
Also if you take a look at AD797 datasheet you see that it shows impressive electrical characteristics: GBP 110 MhZ, SR 20 V/us, CMRR 130 dB, PSRR 130 dB, Input Voltage Noise 0.9 nV/sqrt(Hz), THD -120 dB. If you see block diagram of AD797, attached at this post, you can deduce it works just as you ask, so I think it will work as just as well as discrete replication.

I'm not sure that a simple monolithic IC regulator do the same work of a pre-regulation followed by a series voltage regulator rejecting ripple and noise.
I attach two simple simulations of both configurations: as you can see there are at least 120 dB of difference between the two options. Just why I chose to adopt this configuration, followed by a feed forward shunt regulation that improve noise rejection and transient response.

In a future post I'd like to discuss about how fast should be the circuit response to the transient pulse. For now I can say that I prefer series regulation rather than shunt regulation, just because tipically it's slower than shunt. That's why I chose a feed forward shunt at the end of the regulator, only to act a limited noise cancellation, a little more than 1 mV. Most of the work is done by the series regulator.

Andrea
Attached Images
File Type: jpg AD797_BD.jpg (49.4 KB, 394 views)
File Type: jpg pre-reg_op_amp.jpg (36.8 KB, 121 views)
File Type: jpg LT1085_only.jpg (39.5 KB, 71 views)
  Reply With Quote
Old 13th June 2012, 10:36 AM   #57
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Quote:
Originally Posted by andrea_mori View Post
.................... Walt Jung ...........All his regulator are op amp based.................
No.
Many of his example regulators, in his various papers, use discrete active devices and some do not use any IC, they are solely dependent on discrete devices for their exemplary performance.
__________________
regards Andrew T.
  Reply With Quote
Old 13th June 2012, 12:16 PM   #58
diyAudio Member
 
andrea_mori's Avatar
 
Join Date: Jan 2005
Location: Italy
Quote:
Originally Posted by AndrewT View Post
No.
Many of his example regulators, in his various papers, use discrete active devices and some do not use any IC, they are solely dependent on discrete devices for their exemplary performance.
Andrew,
I have not read all Jung's documentation, but his most famous regulator, the 2000's Jung Super Regulator, use an AD825 as circuit feedback control. Also his previous version use an AD797.
In his article "Low noise power for analog circuits" he use an OPA165 and AD780/AD589 as voltage reference.
Do you really think is not possible to build an high performance regulated power supply with op amps and discrete devices?
Do you think you get better performance using a couple of 2SK170 plus a couple of BC550, a mosfet and a zener or an LED as voltage reference?
Don't forget, for example, AD780 works as a zener, as a super zener, with lower drift and lower noise than the zener.

Andrea
  Reply With Quote
Old 13th June 2012, 12:19 PM   #59
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
I don't claim anything for any of your examples.
I quoted Jung's work, which you seem to be unaware of.
__________________
regards Andrew T.
  Reply With Quote
Old 13th June 2012, 12:55 PM   #60
diyAudio Member
 
jackinnj's Avatar
 
Join Date: Apr 2002
Location: Llanddewi Brefi, NJ
Quote:
Originally Posted by jackinnj View Post
I'm going to write some "Cardinal Rules" for regulators:

1) Regulate the clock Vcc/regulator separately from analog circuitry regulation.

2) Load the regulator for as much current as the pass transistor can handle (heat sinked of course.) If you have to, add a ballast resistor to accomplish this.

3) Don't skimp on trace width.

4) where you think you can use a 1/4 watt resistor use a 1/2 watt resistor.

I would take a look at the layout which Jan did for Jung's regulator and go from there.
Add another 2 rules:

5) Use a fast error amplifier.

6) Use high hfe and/or gm pass devices (this last one is right from Jung's "Don't Shun the Shunt Regulator").
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Regulated power supply meanman1964 Everything Else 13 9th April 2007 02:50 AM
Regulated power supply akunec Chip Amps 3 5th November 2006 10:27 AM
Regulated power supply croccodillo Tubes / Valves 2 2nd November 2006 10:47 AM
regulated power supply adason Swap Meet 1 11th December 2005 06:32 PM
regulated power supply yugaaa Solid State 17 28th January 2003 02:39 PM


New To Site? Need Help?

All times are GMT. The time now is 09:37 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