Loop stability clarification . . . - Page 4 - diyAudio
 Loop stability clarification . . .
 User Name Stay logged in? Password
 Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read Search

 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
diyAudio Member

Join Date: Nov 2010
Quote:
 Originally Posted by pete_schumacher ... It was supposed to be a discussion on how one could determine stability margins from the Bode Plots.
I think you have determined correctly - that the amp is unstable. I think that the problem is not with your interpretation of the Bode plots but of your results that show the amp will work. This may be a simulator artefact. The pencil example is actually deeper than a casual metaphor. It will simulate as stable but be unstable in the real world.
I understand your point that you are interested in principles, but the fact the amp is so unrealistic makes it difficult to tell if there is any other problem. Unrealistic values make it harder to spot a suspect result by comparison with previous experience and mean I can't do much more.

Best wishes
David

Last edited by Dave Zan; 10th February 2013 at 01:30 PM.

 10th February 2013, 01:48 PM #32 diyAudio Member     Join Date: Aug 2006 Location: Texas Blog Entries: 2 Maybe you should describe how the pencil metaphor applies, since Pete doesn't have the knowledge to figure this out. That is the reason he asked. How can he fix the circuit if he doesn't know what's wrong? Writing it off as a simulator artifact tells me that you don't know either, you're just guessing that the simulator is wrong because the results don't make any sense. But I think if you can't describe why the simulator is wrong, then you don't really know. OLG@0db is -270 degrees which is actually +90 degrees. I need to review information on stability. At least it will be a learning experience. __________________ The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer! Need something built, repaired or modded? PM me and ask what I can do!
diyAudio Member

Join Date: Aug 2006
Location: Texas
Blog Entries: 2
Okay pete. I knew I was forgetting something. You can't measure actual in-circuit OLG your way because in practice, the total OLG is divided by the closed loop gain and this changes everything. That's why you MUST use the method in /examples/educational/loopgain2.asc from your LTSpice directory (comes default with the program).

So, the circuit IS theoretically stable. The loopgain analysis checks out.

The reason the 1.5k resistors are MANDATORY is that they eliminate the cascoding effect at RF so that the miller capacitance of the VAS causes stability. It is not obvious, but it works.
Attached Images
 Pete_OGLAMP.png (101.3 KB, 105 views)
__________________
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!

diyAudio Member

Join Date: Nov 2010
Location: The City, SanFrancisco
Quote:
 Originally Posted by pete_schumacher To generate loop gain, I put a 1F capacitor at the (-) input to ground to basically short it out without removing the DC portion. I'll check out their suggested method and see if it does anything different.
Although Keantoken has given you the answer, the suggested method can go unappreciated, may I suggest you sneak up on it.

The 1F capacitor you used essentially also removed the miller compensation of the 1pF cap, as it too is now shunted to ground.
If you split R5 into 2 series resistors say 50K from the output, then place your 1F cap to ground and continue with 250K to the - input (to join the 10K and 1p miller cap) you will still have the miller capacitor in place. The resulting "open loop" should be a lot closer to Keantokens plot.
The next progression may be to insert an ac voltage source in series between the "out" and the right side of the R5 (301K) feedback resistor. Use this ac source to excite (set the input source to 0) and plot the ratio of the voltages on each side of the ac voltage source. This should get you even closer to Keantokens.

Thanks
-Antonio

diyAudio Member

Join Date: Oct 2006
Quote:
 Originally Posted by magnoman Although Keantoken has given you the answer, the suggested method can go unappreciated, may I suggest you sneak up on it. The 1F capacitor you used essentially also removed the miller compensation of the 1pF cap, as it too is now shunted to ground. If you split R5 into 2 series resistors say 50K from the output, then place your 1F cap to ground and continue with 250K to the - input (to join the 10K and 1p miller cap) you will still have the miller capacitor in place. The resulting "open loop" should be a lot closer to Keantokens plot. The next progression may be to insert an ac voltage source in series between the "out" and the right side of the R5 (301K) feedback resistor. Use this ac source to excite (set the input source to 0) and plot the ratio of the voltages on each side of the ac voltage source. This should get you even closer to Keantokens. Thanks -Antonio
Ding ding ding!!! I can't believe I missed that. But you are exactly right. Adding the extra resistor in the feedback path before shunting the net with the 1F cap did match up with Keantoken's sim. The open loop picture I was looking at was wrong, wrong, wrong, at least with the FB compensation in place.

At least we know it's not the simulator not oscillating even though it supposedly should have.

Much thanks guys. Mystery solved.

diyAudio Member

Join Date: Nov 2010
Quote:
 Originally Posted by keantoken ... I can't tell from the schematic how you're deriving loop gain, I suggest you might be doing it wrong like djoffe says. Look in the /example/loopgain2.asc file in the LTSpice directory and use that method.
So the problem would have been solved immediately if Keantoken's advice was followed?

Quote:
 Originally Posted by magnoman ...The 1F capacitor you used essentially also removed the miller compensation of the 1pF cap... Antonio
Hi Antonio, neatly explained. The LTspice method seemed a bit complex but this thread shows why. Nice to learn.

Best wishes
David

diyAudio Member

Join Date: Nov 2010
Quote:
 Originally Posted by keantoken ...A phase margin of 90 degrees theoretically would have no peaking and be phase linear...
My previous post was only intended to be a preamble, but I was interrupted so here is the real question.

In a 2nd order filter the phase linear response (or best approximation, to be pedantic) is not the same as the best unpeaked response. How does this relate to phase and slope of closed loop plots? Or do I need to reread Bode and think?

Best wishes
David

 13th February 2013, 06:02 AM #38 diyAudio Member   Join Date: Nov 2007 Location: Dallas I seriously don't understand the analysis method that Kean and most are using. Still throwing a huge simulated inductor at my feedback path, to break the loop for AC, without permitting DC to swerve into lala-land. From there, its easy to compare input to output, and Bode plot the open loop gain and phase shift. Between these methods, not really sure whats supposed to be the difference? Last edited by kenpeter; 13th February 2013 at 06:05 AM.
 13th February 2013, 11:33 AM #39 diyAudio Member     Join Date: Aug 2006 Location: Texas Blog Entries: 2 Kenpeter, the difference is the reason Peter's OLG plots were wrong at the beginning of the thread. The feedback loop modifies OLG; that is the whole reason opamps are either unity stable or not. When you close the feedback loop you subtract CLG from OLG. Poles in the CLG response affect OLG phase and so phase lead compensation is one way to stabilize an amplifier. Middlebrook et al developed the loopgain2.asc method to account for the affect of CLG on OLG. Dave, phase linearity is affected by any poles in the amplifier compensation design. Compensation loading determines differential voltage at high frequencies, and so if compensation loading is phase linear, the amplifier's response will be. This is why simple miller compensation is usually pretty successful at small-signal phase linearity. It is easy to goof up phase linearity by applying feedback and compensation tricks. I'm guessing I understood your question? I'm pretty poor on math and theory. I confess I haven't read about it directly since I don't have readily accessible study materials. I see the optimization of square wave plots as somewhat superficial because a subjectively pleasant square wave does not guarantee phase linearity. But I do think if you can make a circuit phase linear in simulation with known good models then the results should transfer to real life. It's often argued that ultrasonic and RF step response is not important but if we worry about it anyways I think it's phase linearity we should be looking at above all else. __________________ The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer! Need something built, repaired or modded? PM me and ask what I can do!
 13th February 2013, 04:35 PM #40 diyAudio Member     Join Date: Jul 2003 Location: Europe Middlebrooke's technique is fairly foolproof. You can use the Vx/Vy across the stimulus source in the feedback network for a quick check, but that's all. __________________ bonsai Amplifier Design and Construction for MUSIC! http://hifisonix.com/

 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 OffTrackbacks are Off Pingbacks are Off Refbacks are Off Forum Rules

 Similar Threads Thread Thread Starter Forum Replies Last Post TheGimp Tubes / Valves 58 24th December 2011 02:15 AM Archwn Solid State 18 10th April 2011 05:16 AM cbj591 Tubes / Valves 9 17th July 2010 03:17 PM Lumba Ogir Solid State 1 4th March 2009 08:46 PM frank2395 Pass Labs 2 17th May 2005 07:46 PM

 New To Site? Need Help?

All times are GMT. The time now is 06:16 PM.