Lab Power Supply Design / Build - Page 12 - 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 8th December 2010, 02:12 PM   #111
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
I'll build this to test transient response:

TRANS_RESP.png


The resistors are 10W (except R3). This will switch between 36 ohm load and 18 ohm load and I'll be able to see the results on my scope.
Pretty much a waste of time but should be fun all the same.

Last edited by MJL21193; 8th December 2010 at 02:16 PM.
  Reply With Quote
Old 8th December 2010, 03:13 PM   #112
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
The rig:

NPX_1008.JPG

With the 220uF cap:

NPX_1004.JPG

Without the cap:

NPX_1006.JPG

I see no signs of oscillation with the cap removed now. I did make one subtle change this morning - I added a 470 ohm resistor in series with pot P_R19 to stop it from hitting 0 volts. This seemed to trigger oscillation before.

As for the transient response - I think that I need to go back 'in' and fine tune the compensation caps that I replaced last night in order to stabilize the supply. The one in the CV loop in particular.
  Reply With Quote
Old 10th December 2010, 04:55 PM   #113
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
Much trial and error later, it seems I have the stability problem sorted out. I found that a cap (ideal value yet to be determined) from the base of Q5 to '0' volts most effectively tamed the oscillation. It seems that it was indeed the 'VAS' (as Mega called it, it actually sinks current from the base of Q1 to regulate the output voltage) that was responsible.

As well:
Curious if it matched the simulation results and in response to this comment (again):
Quote:
If I remember correctly you are feeding the pass transistors from some rudimentary voltage with no regulation
I did some more measuring. I scoped the output of the doubler supply again, before R1:

NPX_01015.JPG

Take note of the V/div scale. (~23mVpp)

I then scoped immediately after R1 (what drives the base of Q1):

NPX_01017.JPG

note the V/div scale again, same time base (less than .2mVpp). Interestingly enough, this is free of ripple. This is (nearly) as clean an a pure DC source.

The same, but with a load at the supply output drawing 1.2A:

NPX_01019.JPG

Virtually unchanged. There is no ripple of any significance.

This result is not obvious at first glance. After all, we expect huge ripple from a relatively unfiltered and unregulated doubler supply, and would think it would make for a pretty rough base supply for a regulators pass transistor. TBH, I was very surprised when I saw this in simulation and I felt that it could not be a true representation of real world operation. It in fact was an accurate prediction of the outcome, as shown in the actual scope shots above.

Last edited by MJL21193; 10th December 2010 at 04:58 PM.
  Reply With Quote
Old 10th December 2010, 05:46 PM   #114
akis is offline akis  United Kingdom
diyAudio Member
 
Join Date: Jul 2008
Sorry my bad choice of words keeps appearing all the time, I do not know where to hide

What I was saying in that statement is that the pass transistors are fed a very impure signal with ripple as you saw on the scope.

The diff amp is a monitor, it always senses the output voltage and tries to keep it constant through Q5. If the output voltage increases for some reason, any reason, it will open Q5 more, and the other way round. This is all reactive.

The transistor Q5 opens and closes all the time in unison with the ripple coming through R1 (and other factors) to keep the base voltage constant.

Actually Q5 is the instrument in the hands of the op-amp (the differential amp) which senses the voltage output and tries to keep it stable. This nullifying of the ripple at R1 is not perfect and some will feed through and show on the scope (on simulation at least, with some very little V/div). Same applies with ripple arriving at the collector and with load changes at the emitter of the pass transistors.

The diff amp / Q5 combo has to fight against all these:
(a) ripple through R1
(b) ripple and voltage changes at the collector of the pass transistors (the more current you draw the greater the ripple)
(c) spontaneous changes at the emitter of the pass transistors (when the load changes, as your test with that square wave showed).

As I tested in simulation, the way the diff amp reacts to changes in the output voltage (which tries to keep constant) is the key, because if it is too slow then you will get waves at the output, and if it is too quick then you will get oscillations (eg if you replace the diff amp with a comparator). I found out, what works best is an op-amp/diff amp with a gain of about 10x-15x and adequate HF compensation and negative feedback.


It seemed to me at the time when I was experimenting with similar circuits (on the simulator) that I could help the diff amp to do its job as much as I could, by trying to eliminate/reduce (a) and (b).

However, you have practically proven that you can almost throw the kitchen sink at the output voltage and the diff amp/Q5 will keep it almost perfectly smooth.
  Reply With Quote
Old 10th December 2010, 06:13 PM   #115
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
Quote:
Originally Posted by akis View Post
However, you have practically proven that you can almost throw the kitchen sink at the output voltage and the diff amp/Q5 will keep it almost perfectly smooth.
And this, IMO, is the absolute beauty of this method. It is simple and elegant.
The first objective to to have a stable output voltage that is minimally affected by load - this has been met.
The second objective is to have minimal ripple at the output, regardless of load conditions - this has been met.
Paramount is stability IMO, and this caused me much concern. Getting the supply to a stable operating condition was something I should have done in the prototype stage. The fact is, I didn't do enough testing, wanting to hastily bull ahead (born in May BTW ). There are lessons to be learned in every project and this one taught me a lot, especially to slow down and 'throw the kitchen sink at it' during the prototyping.

PS: Sorry if I came off a bit harsh earlier. Sometimes my frustration at a setback bleeds through to the things I 'say'.
  Reply With Quote
Old 10th December 2010, 07:01 PM   #116
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
Quote:
Originally Posted by akis View Post
(c) spontaneous changes at the emitter of the pass transistors (when the load changes, as your test with that square wave showed).
This is my transient response test and it puts a switching impedance on the output of the supply - 36 ohms and 18 ohms. The rig (shown in post #111 and 112) does this and IT is driven from a squarewave output from my function generator. This doubles the load at each cycle (if testing at 1khz, it would switch 1000 times per second) and shows how quickly the supply copes with load changes.

To stabilize the supply, I put a cap from the base of Q5 to 'ground'. I went for broke here and put a 1uF cap, just to see if it would halt the oscillation. It did and before I do more testing to see if a smaller value will work, I ran the transient response test again.
This is the result:

NPX_01011.JPG

This is at ~1KHz. As can be seen this response is much improved over the previous runs and I'm tempted to leave things as they are. The response is free of the over-damped compensation effects shown earlier.

Last edited by MJL21193; 10th December 2010 at 07:17 PM.
  Reply With Quote
Old 10th December 2010, 07:40 PM   #117
diyAudio Member
 
megajocke's Avatar
 
Join Date: Jan 2003
You should turn those VAR controls coming out of the middle of the volts/div selectors fully clockwise until they click into the calibrated position. The > symbol to the left of the sensitivity readout means the sensitivity of the oscilloscope is turned down from its calibrated value.

The ripple looks pretty good but the transient response is a little slow compared to what is possible. But if you are content with it then I guess it needs no bettering, of course. I'd use standard miller compensation around Q5, probably with a resistor in series like you have now to get sensible components values for a nice loop crossover frequency. The current R3 does not do much though. It won't have much effect until about two decades above the loop gain crossover frequency.

A large output capacitor is often used because it helps the transient response but if it's too large the protection of the current limit is reduced because a lot of energy is stored in it. My Hewlett Packard 0-50 V, 0-10 A (but limited to 200W too) power supply actually has about 3000 F of output capacitance if I remember correctly!
  Reply With Quote
Old 10th December 2010, 08:14 PM   #118
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
Quote:
Originally Posted by megajocke View Post
You should turn those VAR controls coming out of the middle of the volts/div selectors fully clockwise until they click into the calibrated position. The > symbol to the left of the sensitivity readout means the sensitivity of the oscilloscope is turned down from its calibrated value.
When I bought the scope, I asked the seller if it was recently calibrated. He said that it has a calibration seal and a sticker showing calibration by Agilent in 2006. I figured it couldn't be that far off so I bought it.
I searched online for a manual (it didn't come with one) and I found a calibration guide issued for the US Army. I used that to quickly check things over and saw that it is out of calibration and at the time I calibrated (by use of the VAR knob) the vertical amp to a very close setting (I'm not designing nuclear clocks here, so that would be close enough).
While looking at the ripple at the 2mV scale, I forgot my 'calibration' and turned the VAR knob to get a better look. Ooops.
I've since re-calibrated it back to 'normal'.

Quote:
Originally Posted by megajocke View Post
The ripple looks pretty good but the transient response is a little slow compared to what is possible. But if you are content with it then I guess it needs no bettering, of course. I'd use standard miller compensation around Q5, probably with a resistor in series like you have now to get sensible components values for a nice loop crossover frequency. The current R3 does not do much though. It won't have much effect until about two decades above the loop gain crossover frequency.

A large output capacitor is often used because it helps the transient response but if it's too large the protection of the current limit is reduced because a lot of energy is stored in it. My Hewlett Packard 0-50 V, 0-10 A (but limited to 200W too) power supply actually has about 3000 F of output capacitance if I remember correctly!
Funny, I attributed the poor TR in post #112 to the 220uF cap. I'll try it again.

As for compensation, I'm open to suggestions. At this point, I'm going to get ALL of the bugs worked out before proceeding. When I'm absolutely satisfied that everything is A-OK, I'll make NEW boards (again!), including all of the changes/refinement. I also want to include the digital pots on the supply board - slivers of hard-wired ribbon cable running here, there and everywhere is no good.
  Reply With Quote
Old 10th December 2010, 08:34 PM   #119
akis is offline akis  United Kingdom
diyAudio Member
 
Join Date: Jul 2008
I am not sure about the cap on Q5, I would think Q5 needs to be quick to react when it receives an order from the diff amp. Maybe the diff amp needs a cap here and there in its feedback loop? On my simulation I used 1nF negative feedback on the voltage regulator op-amp and 10X gain.

One more thing, you may want to put a 22uF-47uF cap at the ADJ pin of the LM317, as per manufacturer instructions to eliminate ripple from the reference voltage and from the diff amp.

I attach my almost final schematic so you can see the similarities and differences. I use a separate supply to power the op-amps and again a separate zener to feed the pass transistors' bases and again a separate zener to provide the voltage reference and finally another zener to provide the reference for the current limiter. Lots of zeners
Attached Images
File Type: jpg PSU_DEC_2010.jpg (245.2 KB, 228 views)
  Reply With Quote
Old 10th December 2010, 08:40 PM   #120
Account disabled at member's request
 
MJL21193's Avatar
 
Join Date: Mar 2007
Quote:
Originally Posted by akis View Post
One more thing, you may want to put a 22uF-47uF cap at the ADJ pin of the LM317, as per manufacturer instructions to eliminate ripple from the reference voltage and from the diff amp.
Ahead of you there:

LAB SUPPLY SCH.1.png



You have been busy, I see. I'll take a closer look.

You are using MS. If you don't mind, can you attach it? Just change the extension to .txt

Last edited by MJL21193; 10th December 2010 at 08:46 PM.
  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
Want to build a Pass lab design....help desired jmillerdoc Pass Labs 25 22nd August 2010 11:58 AM
Lab Power Supply Nikon1975 Tubes / Valves 2 11th December 2009 10:52 AM
Regulated Lab DC power supply Sir Trefor Tubes / Valves 6 28th July 2007 06:13 PM
diy dual lab power supply morpheus82 Power Supplies 25 10th July 2006 12:41 AM
Lab Power Supply Current Limiting Q Wynand Equipment & Tools 4 27th September 2005 10:44 PM


New To Site? Need Help?

All times are GMT. The time now is 08:50 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