andy_c said:
FYI the Middlebrook technique you describe (succesive voltage and current injection), is only gauranteed to give accurate results up to unity loop-gain frequency. Beyond this, the error introduced in the numerator by subtracting a small product from unity begins to dominate....
Exact results may only be obtained for all frequencies by Middlebrook's simultaneous voltage-current injection technique....
Christer said:
Note that what you've described here is called the return-ratio, which is never the same as loop gain.
If your test signal is a voltage presented with an infinite test input impedance for all frequencies, and zero return impedance, then the magnitude of the voltage return ratio will exactly equal the magnitude of the circuit's loop transmission.
However, in this idealised case, the phase relationships between the measured return ratio and loop gain is then strictly 180 degrees.
viz: You have to phase-invert the return ratio to obtain the true loop-gain as a function of frequency.
mikeks said:
Hi Mike,
I'm going to call you the re-animator after the movie of the same name
Where did you find out about this improved technique? I went to Middlebrook's web site and found that he had some nice looking videos. Did you get these? They're pretty pricey though. I thought he was working on a new book but apparently it's not out yet.
I did go through the derivation of the Middlebrook equations on Jim Thompson's site, and it does appear that in the absence of reverse transmission and numerical precision errors, the result is exact. So I assume the errors you're referring to purely numerical precision issues, correct?
Another interesting, related item can be found at the Yahoo LTSpice user's group at http://groups.yahoo.com/group/LTspice/files/Examples/Educational/LoopGain_Probe/. This guy created a loop gain measurement component for LTSpice that uses a technique due to Tian, who I'd never heard of before. This technique is supposed to take into account reverse transmission, which Middlebrook's technique doesn't do. I have yet to try it, but using the component has got to be a lot easier than the nasty subcircuit replication approach I was using before. Discussion of the technique begins here: http://groups.yahoo.com/group/LTspice/messages/2484?viscount=-30 beginning in post number 2482. I'd like to dig up Tian's paper on this somewhere if it's available.
Nelson Pass said:I think the moderators should create Simulated forums
where Simulated DIYers can have Simulated discussions
about Simulated circuits and speakers and so on. And don't
forget: We will need a Synthesized music area also.
This would only be good if communication were required to be in the form of sound files rather than text, and the posters would have to record their sound files by speaking through "robot transformers" so they sound like Darth Vader.
And on a more serious note, it pays for computer-oriented types to remember Hamming's motto - "The purpose of computing is insight, not numbers".
jackinnj said:
when this thread first started I mentioned that I had some boards made up using the HFA3102 transistor array as a comparator -- the device seems to be out of production (although not listed as obsolete) so I am switching to the HFA3101 -- if you can bear to measure phase to a few degrees (i.e. not 0.1 degree to 10 MHz) you can use a regular high-speed comparator.
National Semi has put together a new application note for characterization of low-dropout regulators (which are amplifiers anyway.)
Nelson Pass said:I think the moderators should create Simulated forums
where Simulated DIYers can have Simulated discussions
about Simulated circuits and speakers and so on. And don't
forget: We will need a Synthesized music area also.
Remember Nelson, Simulation is a good tool. And some things are getting build and mostly they work in the first run.
The MIRAND A1 is a simulated amp which worked in the first run. And it behaves like it did in the SIM.
Just a picture of a simulated amp which are working .. just for you Nelson.
An externally hosted image should be here but it was not working when we last tested it.
sonnya said:
Remember Nelson, Simulation is a good tool. And some things are getting build and mostly they work in the first run.
The MIRAND A1 is a simulated amp which worked in the first run. And it behaves like it did in the SIM.
Just a picture of a simulated amp which are working .. just for you Nelson.
An externally hosted image should be here but it was not working when we last tested it.
Correct in all respects...Thanks Sonnya...
Actually I was referring to reverse transmission of the forward path. Matrix notation would be y12, z12, h12 etc. of the forward path. Or what RF/microwave people refer to as "non-unilateral" behavior of the forward path. Whereas the simplified block diagram approach assumes that no signal goes backward through the blocks.
If this is indeed true about Tian's approach, I think this would be a worthwhile enhancement to Middlebrook's original technique.
If this is indeed true about Tian's approach, I think this would be a worthwhile enhancement to Middlebrook's original technique.
andy_c said:
The contribution by such transmission is vanishingly negligible....practically nil........but if you insist:
http://www.ece.ucdavis.edu/~hurst/papers/ExactSimFB,CAS91.pdf
http://www.ece.ucdavis.edu/~hurst/papers/Compare2FbApproaches,EDU.pdf
http://www.ece.ucdavis.edu/~hurst/papers/FullyDiffRR,CAS.pdf
Is the Middlebrook paper available in PDF form somewhere?
Thompson's site is the only free site I've found that actually covers the derivation of Middlebrook's formula. I thought his algebra was convoluted, but the starting point for the calculations, with the simple circuit diagram made a lot of sense to me. I just started where he started and worked out the equations myself. It made a lot of sense once I was done.
Thompson's site is the only free site I've found that actually covers the derivation of Middlebrook's formula. I thought his algebra was convoluted, but the starting point for the calculations, with the simple circuit diagram made a lot of sense to me. I just started where he started and worked out the equations myself. It made a lot of sense once I was done.
Nelson Pass said:I think the moderators should create Simulated forums
where Simulated DIYers can have Simulated discussions
about Simulated circuits and speakers and so on. And don't
forget: We will need a Synthesized music area also.
Actually Nelson, Middlebrook only discussed practical laboratory methods for measuring loop gain....
.....No simulators involved...so....you must be hallucinating......
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.