Bob Cordell's Power amplifier book - Page 227 - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

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 11th November 2011, 09:50 PM   #2261
diyAudio Member
 
Join Date: Dec 2009
Location: Brussels
Hi Sidly,
As Edmond states, removing Q15 and Q16 is plain suicide. I tried that so many times.
Q25 and Q26 are just there to limit the VAS current by collapsing the VAS.
As I keep seeking I rearanged my grounding in another way and all oscillations were gone except the one from the BLUE node (30MHz 50mVpp) it increased to 100mVpp but all others were 0 and even the noise on all the nodes seriously dropped to under 5mVpp.
I already changed many things on the grounding now I cannot find my best situation again. But as I said maybe it was just temporary luck so it's not just a matter of placing back the probes and clamps on the same wires and nodes ...
Sigh

Hi Edmond,
As you see the struggle continous to make that amp work.
Do you think it has to do with the CMCL roll off point?

To all,
By the way what is a good noise measure on all nodes? I suppose it is not possible to have a perfect 0 line on the scope while on 1mV/division?
For instance is 10mV of noise on the nodes normal or not?
In other words, is me having 50 to 100 mV just quite a bit too high or are my values just awful and should it be 100 or 500 times lower?

Cheers

Olivier
  Reply With Quote
Old 11th November 2011, 10:06 PM   #2262
Waly is offline Waly  United Kingdom
diyAudio Member
 
Waly's Avatar
 
Join Date: Jan 2011
Location: London
Quote:
Originally Posted by Olivier73 View Post
A breakthrough would really make my day !
Other than wiring/ground loops, PCB layout issues and RFI ingress, here are my bets (in the top-down likelihood order). Almost always very high frequency instabilities (5MHz and up) are due to local oscillations and not necessary related to the global feedback loop.

1. Local oscillations in the VAS cascode. At a minimum, you should add a small resistor in series to the Q21/Q24 collectors, 10-47ohm would do.

2. Miller local loop instability. The VAS transimpedance gain is huge (although C12/R25 and C13/R52 are trying to lower the HF gain), therefore the Miller loop (CM1/CM2) may require some frequency compensation. Add 47p in series with 47-100ohm from the Q22/Q23 collectors to the ground. Such a lead-lag compensation network usually helps.

3. The CMCL control loop seems to be compensated, not sure if C15/R33 and C16/R34 are good enough. If you determined the CMCL loop unity loop gain frequency and the phase margin, please post the method and the results.

4. Lack of decouplings, 10ohm/47nF for the entire board is definitely not enough. You should add local decouplings for each stage (input, VAS, output).

Last edited by Waly; 11th November 2011 at 10:32 PM. Reason: True, it is not a Hawksford cascode
  Reply With Quote
Old 11th November 2011, 10:18 PM   #2263
diyAudio Member
 
Edmond Stuart's Avatar
 
Join Date: Nov 2003
Location: Amsterdam
>1. Local oscillations in the Hawksford cascode.
It's NOT a Hawksford cascode.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en
goed verliezen dan dooft het licht…(H.M. van Randwijk)
  Reply With Quote
Old 11th November 2011, 11:10 PM   #2264
diyAudio Member
 
Join Date: Dec 2009
Location: Brussels
Default oscillations

Hi Waly,

Interesting cures you post here.

As now I refound my best grounding my noise is down to +/- 5mVpp on many nodes. Is this good enough?

In the supposition this is good enough I only have one local oscillation situated around R28. On both sides. One is now 20MHz and on the other side it's 30MHz (can this be???). One is 200mVpp and the other is 100mVpp.

I would tend to say my problem lies within the CMCL loop. Since my oscillation are always around that piece of circuitry (also when I had the oscillations indicated as RED in one of my previous postings). So I would not search for origin in my VAS circuit ... I would rather opt for cure 3 in your posting.

So coming back to cure 3 : No I did not determine UGFL for the CMCL (I am simply not capable of doing that ... Edmond helped me with this matter some time ago). How you do that?

About n°4 : there are some Caps foreseen on the actual PCB. But i did not solder all of them because it's not always sure they do good or bad. I send hereby the complete schematic. The one you saw posted before is the one as it is built for now in front of me.

Thanks in advance !

Olivier
Attached Files
File Type: pdf Full Schematich.pdf (25.9 KB, 78 views)
  Reply With Quote
Old 12th November 2011, 01:12 AM   #2265
diyAudio Member
 
EnvisionAudio's Avatar
 
Join Date: Apr 2002
Location: Dayton, OH
Quote:
Originally Posted by Olivier73 View Post
Hard to tell what is the real offender here ... !?

You guy know more now? A breakthrough would really make my day !

Cheers

Olivier
Honestly, I think you are seeing a phantom signal that is really there, but only because of the loading by your standard, but high capacitance, oscilloscope probe. At 6.75MHz, you're looking at exactly 9x the -3dB point of 750kHz which is highly suspicious. In basic terms, I think you're measuring the wrong thing with the wrong equipment and presuming it's significant.
__________________
DCX2496/DEQ2496 Car Audio Conversions
www.envisionelec.net

Last edited by EnvisionAudio; 12th November 2011 at 01:17 AM.
  Reply With Quote
Old 12th November 2011, 01:29 PM   #2266
diyAudio Member
 
Edmond Stuart's Avatar
 
Join Date: Nov 2003
Location: Amsterdam
Hi Olivier,

You are driving me nuts. I just analyzed (simmed) the schematic in post 2258, and in the meantime you pop up with another one in post 2264. Shooting at moving targets makes things not easier.

Regarding the first schematic, I have plotted the gain/phase response of the CMCL (which looks okay) respectively the Miller loop (with disabled global NFB loop).
It's the latter that predicts trouble, as the ULGF is way too high: 65MHz.
Notice that this is a rough estimation, as parasitics were ignored. A real amp might give quite different results and/or might even oscillate.

If you don't mind, I'm not gonna investigate this any further, because you changed the circuit every time and it deviates too much from the starting point as depicted in this post: HEEEELLLPPP : M. Randy Slone Mirror Image Topology Construction - Troubles

Cheers,
E.

PS1: The plots show the loop gain, thus not the return value, which is - (minus) loop gain.

PS2: @Walter cs. "simulation [...] used to AC analyze the loop gain - that's the type of analysis where simulators are shining." Clearly NOT !
Attached Images
File Type: png cmcl.png (15.2 KB, 207 views)
File Type: png miller.png (15.4 KB, 206 views)
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en
goed verliezen dan dooft het licht…(H.M. van Randwijk)
  Reply With Quote
Old 12th November 2011, 02:51 PM   #2267
diyAudio Member
 
Join Date: Dec 2009
Location: Brussels
Hi Edmond,
I understand the trouble of me changing the circuit, but the one you analysed (2258) is the one on the test bench. So that's perfect. This PCB is foreseen for additional components. So in fact the PCB corresponds to the schematic in 2264 but is only equiped with the components of 2258.
The TMC part is just replaced with one 1 cap on each side 33pF for a simple dom. pole compensation (for now).
Some other part may differ like C12/C13 and C15/C16 but the real ones plugged in are according to 2258 (and not 2264).
About the deviation from the starting point, indeed it looks more to the circuit you posted just before the version you linked to.
Thanks for the info, I know now I should rather look into the Miller loop instead of the CMCL loop even if the affected nodes are around the CMCL and ML... But I guess this is normal...
I wll post my findngs
Cheers
Olivier
  Reply With Quote
Old 12th November 2011, 05:01 PM   #2268
diyAudio Member
 
Join Date: Dec 2009
Location: Brussels
Hi Envisionaudio,
Changing my probe from a 1/1 ratio to a 10/1 ratio removed the oscilation around R28 completely. It is expected to reduce 10x on the trace (assuming equal scale) but it went down to nill. Even raising the scale 10 to 1 (should be same amplitude as with 1to1 probe) showed no oscillation.

However Edmonds findings about the Millerloop ULGF to 65MHz worries me.

Only I don't know how to simulate it...

Edmond, could you explain how to measure the millerloop AC analyses (& cmcl loop - will be the same logic i assume)

Cheers
  Reply With Quote
Old 12th November 2011, 06:04 PM   #2269
diyAudio Member
 
Join Date: Dec 2009
Location: Brussels
Hi Everyone,
I hope the situation at this moment is stable and not just a combination of hazards that come out well today but not tomorrow.
At the moment the amplifier behaves very well.
I even lowered the miller caps from 47pF to 22pF to see if any shakiness starts to sneak in.
The amplifier has a BW (-3dB) of 1Hz up to 1MHz.
It's slew rate is around 200V/us.
It's phase lag at 1MHz from the input signal is 36°
Checked scopetrace for peaking (sinuswave) from 1KHz to 5MHz (result = no peaking)
No overshoot on square wave (checked = OK)
Loads were never connected
Should still check if other nodes are oscillation free going through the bandwidth.

One strange thing I noticed however. As for sinewave the upper -3dB is @ 1MHz it seems 1,5MHz if square waves are used ? Why is that ?

Put simply : the output at say 1,2MHz is 20 to 30% higher if SQUAREWAVE compared to SINEWAVE ... any clues? This is not the case at lower frequencies of say 10KHz...

I'll continue trsting

Cheers

Olivier
  Reply With Quote
Old 13th November 2011, 02:11 PM   #2270
diyAudio Member
 
Edmond Stuart's Avatar
 
Join Date: Nov 2003
Location: Amsterdam
Quote:
Originally Posted by Olivier73 View Post
Hi Envisionaudio,
Changing my probe from a 1/1 ratio to a 10/1 ratio removed the oscilation around R28 completely. It is expected to reduce 10x on the trace (assuming equal scale) but it went down to nill. Even raising the scale 10 to 1 (should be same amplitude as with 1to1 probe) showed no oscillation.
Great!

Quote:
However Edmond's findings about the Miller loop ULGF to 65MHz worries me.
Only I don't know how to simulate it...
Edmond, could you explain how to measure the millerloop AC analyses (& cmcl loop - will be the same logic i assume)
Cheers
Yes I can, but first I need to know what version of MC do you have?

Cheers,
E.
__________________
Een volk dat voor tirannen zwicht, zal meer dan lijf en
goed verliezen dan dooft het licht…(H.M. van Randwijk)
  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
Another realization of Bob Cordell's THD Analyzer giulianodes Equipment & Tools 37 9th October 2013 03:48 AM
best audio amplifier book?? Bouvett Everything Else 30 13th August 2012 03:43 AM
Amplifier Design Book pixie Everything Else 27 11th June 2010 09:36 PM
Project 11.1 from Slone "High-Power Amplifier" Book Karl71 Solid State 46 6th October 2008 04:47 AM
book-audio power amplifier by Douglas Self mikee12345 Solid State 8 16th November 2003 02:16 AM


New To Site? Need Help?

All times are GMT. The time now is 05:21 PM.


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