Another Linear Pass Regulator - 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 1st January 2009, 02:40 PM   #1
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
Default Another Linear Pass Regulator

Hello all. For some reason or another, I have learned a lot since I posted the topic below, about 6 months ago.

http://www.diyaudio.com/forums/showt...99#post1551799

The way I approach electronics is becoming more versatile, I am getting more clever and creative.

My learnings have dealt mainly with amplifier design, but there is little difference between a linear regulator and an amplifier. Both are interesting to design.

So my question was at first, "if we can get vanishingly low distortion in amplifiers using simple circuitry, why wouldn't be be able to do the same with a linear regulator?"

I do believe it's possible with discrete circuitry, which is one of the reasons I posted.

I understand that many of you do not usually put the time in to build regulators for your amps. Twenty 2200uF caps usually does the job, doesn't it!?

Sorry, even though most would be happy with just a thousand caps on the rails, with some filtering here and there, I absolutely HAD to do this.

With clever design, we should not need caps (except for after the rectifier), even though they look cool. If you want it to look good, perhaps you should try replacing the zener with an LED. Tests on this forum have indicated that IR LED's are the best, though they don't emit light.

You may think I'm slightly insane, but I wanted to make a good regulator and for this to happen, traditional values must be reexamined for their value. We see this commonly in amplifier designs, especially before the "great awakening" of Doug Self and other such designers.

At any rate, Notice the transistors used. They were all chosen with Cjo in mind. This is essential to HF performance.

Q5 and Q6 form a Rush cascode, which works relatively well with regulators, based on my simulator experience. I am absolutely certain that if I used an ordinary LTP as the voltage sensor here I would be fried within seconds from RF oscillation.

Q9, Q7 and Q8 form a very nice CCS, and with the filtering cap this works quite well.

The MOSFET and attached circuitry is a diabolical testing device to prove the supply's strength. This is very unethical I might add. It generates a VERY FAST negative pulse after climbing to 1A. If C4 goes any lower than 10nF, the circuit begins to oscillate. Some over-voltage protection might help calm oscillations.

It helps reveal the finer details of operation in a supply. This idea was put forth by Mr. Mooly, whom I must give credit.

Regulation is EXTREME for this circuit, I'll just hope it will work in real life.

Apart from the spikes generated, you'll see that the output voltage varies almost imperceptibly from 0-1A. The sim won't let me zoom in any more. If I could guess, somewhere around 20-30uV. And this stays the same across the entire audio band.

Naturally, criticism and suggestions are encouraged as I posted this primarily so that I might learn more through them.

I will post simulation results right after this. Attacxhed is the schematic.

- keantoken
Attached Images
File Type: png discretesuperreg.png (28.6 KB, 642 views)
  Reply With Quote
Old 1st January 2009, 02:44 PM   #2
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
Here is a sim thing.

- keantoken
Attached Images
File Type: png discretesuperregsim.png (13.4 KB, 480 views)
  Reply With Quote
Old 1st January 2009, 03:37 PM   #3
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
Also interesting, "Static" at 40KHz sine 0-1A load:

I'm fairly sure this occurs because the regulator "trips" but can recover before it starts oscillating.

- keantoken
Attached Images
File Type: png discretesuperregsim2.png (24.2 KB, 451 views)
  Reply With Quote
Old 1st January 2009, 09:18 PM   #4
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
Ah, here's something interesting.

See the strangeness? This happens because of R8. When the output current goes lower than about 1mA, Q4 begins to shut off, which reduces the overall gain of the darlington pair, which reduces open-loop gain, increasing error. Regulators are no different than amplifiers.

Putting a 1.2k resistor across the rails improves operation by providing bias current. This is wasteful, however. Removing R8 altogether does not help.

Multiple 2SC's can be paralleled to provide more output power, as one can only handle about 1A. It is important to use a pass transistor with low Cjo, like the 2SC4793 here. If you don't, it will surely oscillate.

- keantoken
Attached Images
File Type: png discretesuperregsim4.png (16.9 KB, 414 views)
  Reply With Quote
Old 1st January 2009, 09:22 PM   #5
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
With a 1.2k resistor across output to provide bias current. Noticeable improvement. The aim is to reduce gain error without wasting output current, the resistor is only a temporary fix.

- keantoken
Attached Images
File Type: png discretesuperregsim3.png (17.5 KB, 414 views)
  Reply With Quote
Old 1st January 2009, 09:28 PM   #6
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
I hope I am not just talking to myself.

- keantoken
  Reply With Quote
Old 2nd January 2009, 11:21 PM   #7
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
Well, whatever. Lowered supply voltage, we certainly don't need 36V.

An interesting tidbit here:

At high frequencies, the Cjc of Q10 became a problem. For those of you who simulate this you will see the cause and effect if you look at Q10's emitter current and the output voltage.

This fix is to use a 100-ohm resistor on Q10's collector. This causes Q10's collector capacitance to not just be a short to ground.

The result, is no funny glitches in the output at 40KHz, and more stability.

I have seen people put diodes in the position the 100-ohm resistor now is, though I don't know if that solution was for the same problem.

Also, temperature compensation. However, due to the diodes' constant-voltage effects, this lowers the ability for regulation. I don't think temperature stability is that important, really, so I don't think it's necessary. Maybe it should be a switch option.

Also used Christer's models for the output transistor, not sure if it's better or worse than the ONSemi model, but the circuit works just as well with both.

I will gladly receive any kind of criticism or advice, even if you tell me that I'm a failure and I should go jump off a cliff. Or if you tell me it doesn't work, etc. It gets scary when no one's around...

- keantoken
Attached Images
File Type: png discretesuperreg2.png (36.2 KB, 368 views)
  Reply With Quote
Old 3rd January 2009, 12:00 AM   #8
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
Quote:
Originally posted by keantoken
I hope I am not just talking to myself.

Attached Images
File Type: jpg i'm listening.jpg (37.4 KB, 374 views)
  Reply With Quote
Old 3rd January 2009, 12:18 AM   #9
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Blog Entries: 2
Thank you, MJL.

So, is it an interesting design? Think it will work as well in real life?

- keantoken
  Reply With Quote
Old 3rd January 2009, 09:33 PM   #10
diyAudio Member
 
Join Date: Apr 2006
Location: Minnesota
Keantoken,

Here's a little feedback.

You could replace all of the transistors and resistors associated with the zener biasing with a single resistor, because the current through the zener will be almost constant anyway.

I don't know what you mean by Cjo. The most important SPICE transistor parameter for frequency response is Tf, the forward transit time. Tf=1/(2*pi*fT). fT is the frequency where the magnitude of the current gain reaches 1. Note that this is an approximation that neglects the base to collector (Miller) capacitance. The SPICE BJT model is a form of the hybrid pi model. Do a little research on this. http://www.intusoft.com/lit/WkwModels.pdf is a excellent reference on modelling.

Your voltage regulation loop is interesting, but I would use a differential amplifier. This will provide better temperature stability, and probably would need fewer transistors.

You should use a higher bias current through the zener. You don't want the zener operating on it's knee. A 5.1 volt zener has the lowest temperature coefficient. This is the crossover point between a negative and positive TC.

Rick
  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
linear regulator question rtate Digital Line Level 6 17th December 2008 03:36 PM
Modular Linear Regulator TheDriver Parts 0 28th July 2008 08:27 PM
Linear Pass transistor regulator - my design keantoken Power Supplies 49 5th July 2008 11:24 PM
Linear regulator eltonseemann Power Supplies 10 1st November 2007 04:44 AM
Linear regulator for power amp mr.duck Power Supplies 3 10th April 2007 11:59 AM


New To Site? Need Help?

All times are GMT. The time now is 10:25 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