Shucks.
But seriously, when I find a difference I do like to make sure it's not my misconception, but mainly I like to use it as an alert that there is some subtlety that needs to be examined.
I have learned so much from debates between experts on this forum.
This latest discussion has prompted me to reread Dr P J Hurst's article that compares the Return Ratio with the Loop Gain, "A Comparison of Two Approaches to Feedback Analysis"
http://web.ece.ucdavis.edu/~hurst/papers/Compare2FbApproaches,EDU.pdf
Hard work but educational to make the effort.
He discusses when it is accurate to discount feedforward but I don't understand that bit yet.
Any help welcome
Best wishes
David
Yes, here you go.
I’ve done some further experimenting, looking at differences observed for “isolating caps” (the big 1 kF caps placed at the input nodes to disable the global loop when the inner loop is simulated) being present vs. not present, and two different locations for the tian probe to simulate the inner loop. See attached images.
Notes: top pane = closed loop response, middle pane = global loop, bottom pane = inner loop
green trace = inner loop not compensated, blue trace = inner loop compensated
Enjoy!
Harry.
Attachments
Last edited:
No need to go that far. Just get a 30MHz RET devices triple EF output stage without Zobel and output inductor, with some 100nF capacitive load, and close a global loop around a standard Miller compensated Blameless.
If you consider the EF as the inner loop, then you can easily miss the EF triple instability if you look at the global loop only. Whatever gain and phase margins you'll get, the amp will be unstable on the bench because of local instability in the output stage.
Of course, people don't usually like to think of an EF as a local feedback stage.
Interesting Waly,
Please could you explain how we can check the stability of the 3EF stage? Perhaps with a Tian probe?
Paul
You might also be interested in some 'real life' stability testing of a single stage EF output stage.
#14 stability-analysis-ef-output-stages
It seems I picked up some of synaesthesia's points in Jurassic times too.
Last edited:
Thanks for this Harry.
I've done quite a lot of comparisons of Tian vs zillion Farad cap. My early efforts with LTspice used your zillion Farad cap or zillion Henry inductor.
When I learnt the true Tian way, I did these comparisons to see which of my old sims I had to bin. IMHO, differences are usually well above ULGF of even the 'inner loop' so I was happy to keep my early efforts ... except for some exotic stuff to do with impedances.
_______________
Your Closed Loop responses are sorta what I'd expect. IMLE, the 'real life' amp will be unstable .. perhaps only on part of the cycle depending on load/signal/thermal history bla bla ... and at about the frequency of your 'inner loop' instability. You can probably see this using .TRANS
I/we seem to be jumping between this thread and Toni's http://www.diyaudio.com/forums/solid-state/235194-2stageef-high-performance-class-ab-power-amp-200w8r-400w4r-125.html.
I think you need to check stability of 'inner loops' with the 'main loop' intact. My posts in Toni's thread starting from #1242 show what a powerful influence it has on the 'inner loops'.
_______________
Guru Zan, I'd be interested in your comments on my latest NMP posts on that thread from #1250.
I think the Tian probe IS measuring Hurst/Bode's Return Ratio. It is 'direction independent' so should pick up feedforward too. Am I right?
Last edited:
I recommended that link in response to a similar question a while back, but now I have to say I don't fully understand how it's intended to work.
Maybe it's just because I find his simulator results less readable than LTspice, intended to ask there but can anyone here explain the details?
Best wishes
David
Your results are consistent with my quick lool and my comment to Harry that it doesn't affect the results well above the ULGF of the outer loop.
Yes, the stability difference may be small but it's a simpler technique anyway, take the extra accuracy as a bonus.
I saw the posts but didn't want to comment until I had it clear in my mind.
There is a tricky point here about what the different probes actually mean.
Frank Wiedmann discusses it in 'What about other definitions...' https://sites.google.com/site/frankwiedmann/loopgain
but I don't have this completely clear either.
The results of Tian and Middlebrook GFT seems fairly similar so it probably makes little practical difference.
But I find attempts to understand the conceptual difference are useful to clarify ideas.
Best wishes
David
It depends on what you are trying to do or see.
If its stability, either will probably do.
I you want to show impressive figures in the audio range, the one on the base is superior.
OK. I think the one on the VAS base is the correct one for this type of compensation
In this amp, it cuts ALL the important loops. Not sure about feedforward though
Definitely. I already wrote this in Toni's thread so I don't want to labor the point but, as you noted, we have skipped around between these two threads.
It does not include feed-forward for the inner loop, nor should it, because feed-forward does not contribute to Return Ratio of that loop.
The Hurst PDF discusses this but it's heavy work, read it a few times and understand better but need a few more tries.
I notice no one has offered to explain it simply, or Ovidu either.
It is relevant to our earlier discussion about whether Emitter Followers were Minimum Phase in view of the fact that Common Emitters clearly are Not Min. Phase (eventually
).I now understand this better.
There is feed-forward in a CE and it causes the NMP.
There is feed-forward in EF too. It does not cause obvious NMP so it's less overt.
But it still alters Return Ratio calculations.
I think this is what one of Middlebrook's presentations is all about, I didn't understand the context 'til now.
Best wishes
David
Already posted this link in the other thread, to avoid a hijack of Toni's and Damir's threads I have answered here http://www.diyaudio.com/forums/solid-state/235188-tpc-vs-tmc-vs-pure-cherry-9.html#post4253118
Best wishes
David
I Childproofed The House, but THEY KEEP GETTING IN.
-RNM
Hi Richard,
I would like to send you PM, but it's full and I can't.
BR Damir
The PCB is narrow to get lower price, as the pcb price is per square cm (mm). The mosfet mounting holes are on the alu angle bracket only. With this amp I did not measured the temperature at the full power yet, but my other amp (TT amp) reaches 50 to 60 degree Centigrade on the alu bracket.
Damir
First proto board had to much layout errors and I ordered new boards with corrected errors. With the first board(corrected with some cut and wire) I tested the concept and it worked so now first channel with new board is ready. Next I have to assemble second channel and try it is stereo.
Here are the photos.
Damir
Here are the photos.
Damir
Attachments
Listening 200W CFA VMOSFET amp
I did my first listening with two channel, stereo. The source was DVD player DENON 3910, the preamplifier was my GainWire mk2 in CFA mode, and the loudspeaker my design with long ribbon from NewForm Research and for the woofer 8 inch carbon/paper from Scan-Speak in closed box.
I was listening Patricia Barber on CD and Sara K. on SACD.
Two channels where powered from 500VA transformer and 2x 25000uF, so actually not enough to get full power from the amp.
First impression was total silence when I switched it on, no hum no any noise even with my ear close to the loudspeaker, and all that with wild wiring as could be seen from the photos.
What about music? It was marvel, great dynamic with punchy bas, clear middle in heights, excellent stereo image. It should even improve after electrolytic caps settled.
This is my design and it's not easy to write about it, but still I am impressed with this amp.
It uses cheap vertical mosfet, but different compensation I never found in other power amps. It is very stable with no hint of any oscillation.
Sorry for bad photos, I used my mobile phone quite old.
Damir
I did my first listening with two channel, stereo. The source was DVD player DENON 3910, the preamplifier was my GainWire mk2 in CFA mode, and the loudspeaker my design with long ribbon from NewForm Research and for the woofer 8 inch carbon/paper from Scan-Speak in closed box.
I was listening Patricia Barber on CD and Sara K. on SACD.
Two channels where powered from 500VA transformer and 2x 25000uF, so actually not enough to get full power from the amp.
First impression was total silence when I switched it on, no hum no any noise even with my ear close to the loudspeaker, and all that with wild wiring as could be seen from the photos.
What about music? It was marvel, great dynamic with punchy bas, clear middle in heights, excellent stereo image. It should even improve after electrolytic caps settled.
This is my design and it's not easy to write about it, but still I am impressed with this amp.
It uses cheap vertical mosfet, but different compensation I never found in other power amps. It is very stable with no hint of any oscillation.
Sorry for bad photos, I used my mobile phone quite old.
Damir
