Miller Input Compensation (MIC) Design Example
Wanted to be able to find this from a thread title rather than dig for it:
Bob C. on how to design with MIC:
From: http://www.diyaudio.com/forums/soli...l-interview-bjt-vs-mosfet-78.html#post1207962
Originally posted by G.Kleinschmidt
Could you pretty please articulate your design procedure for this type of compensation? Using your accessible 50W EC amp design as a reference would be good.
Cheers,
Glen
Hi Glen,
Here's how I do it. Let's assume I've chosen a closed loop gain of 20 and a gain crossover frequency of 2 MHz. This means that I want the forward gain of the combined input stage and VAS to be 20 at 1 MHz, assuming that the output stage has unity gain (if the output stage had a gain of 2, the target would be 10 instead). Having chosen the feedback shunt resistor to be 215 ohms, I now know that the reactance of C4 must be about 19 times 215 ohms at 1 MHz, or about 4k. Thus c4 becomes 20 pF.
I then want to put in a stabilizing zero at a frequency above the gain crossover frequency. This is done with R13 at 680 ohms. This gives me a zero at about 11 MHz.
Next I analyze and stabilize the inner loop that is closed by C4 and R13. This can be a high-bandwidth loop with a gain crossover beyond 10 MHz, since only small-signal fast transistors are in the loop. This fast loop is stabilized by C3 and R14. This is very light lag compensation.
Note that there is not a lot of interaction at the base of Q2 because the 215 ohm resistor is small compared to the impedances of C4 and R12.
The advantage of the scheme is that it rolls off the high frequency forward path gain by applying feedback around the input stage, increasing its dynamic range, reducing its distortion, and not making it work harder at high frequencies (the way Miller compensation does). It only has to work harder at very high frequencies where C3 comes in, but this is pretty far out.
This compensation approach is what allows me to get 300 V/us slew rate. It is like input compensation in a way, but is more elegant because it does not use the brute force of shunt lag compensation across the input gates.
Cheers,
Bob
Wanted to be able to find this from a thread title rather than dig for it:
Bob C. on how to design with MIC:
From: http://www.diyaudio.com/forums/soli...l-interview-bjt-vs-mosfet-78.html#post1207962
Originally posted by G.Kleinschmidt
Could you pretty please articulate your design procedure for this type of compensation? Using your accessible 50W EC amp design as a reference would be good.
Cheers,
Glen
Hi Glen,
Here's how I do it. Let's assume I've chosen a closed loop gain of 20 and a gain crossover frequency of 2 MHz. This means that I want the forward gain of the combined input stage and VAS to be 20 at 1 MHz, assuming that the output stage has unity gain (if the output stage had a gain of 2, the target would be 10 instead). Having chosen the feedback shunt resistor to be 215 ohms, I now know that the reactance of C4 must be about 19 times 215 ohms at 1 MHz, or about 4k. Thus c4 becomes 20 pF.
I then want to put in a stabilizing zero at a frequency above the gain crossover frequency. This is done with R13 at 680 ohms. This gives me a zero at about 11 MHz.
Next I analyze and stabilize the inner loop that is closed by C4 and R13. This can be a high-bandwidth loop with a gain crossover beyond 10 MHz, since only small-signal fast transistors are in the loop. This fast loop is stabilized by C3 and R14. This is very light lag compensation.
Note that there is not a lot of interaction at the base of Q2 because the 215 ohm resistor is small compared to the impedances of C4 and R12.
The advantage of the scheme is that it rolls off the high frequency forward path gain by applying feedback around the input stage, increasing its dynamic range, reducing its distortion, and not making it work harder at high frequencies (the way Miller compensation does). It only has to work harder at very high frequencies where C3 comes in, but this is pretty far out.
This compensation approach is what allows me to get 300 V/us slew rate. It is like input compensation in a way, but is more elegant because it does not use the brute force of shunt lag compensation across the input gates.
Cheers,
Bob
Last edited:
The MIC approach feeds back high-frequency signals from "VAS" output to the inverting input. Some people, Bob included iirc, do not promote strong lead compensation in the global feedback loop as this might impair input-stage performance due to high-frequency stress.
The signal at VAS output contains all the "pre-distortions" the feedback loop adds in order to produce a good amp output signal, and these pre-distortions contain a lot of high-frequency components.
Is MIC a good idea in case of BJT input stages, perhaps with only little degeneration?
Matthias
Edit: I'm aware that the C3/R14 lag compensation only is loading input stage at high frequencies. Nevertheless, more hf signals are reaching the input stage junctions. Do they harm?
The signal at VAS output contains all the "pre-distortions" the feedback loop adds in order to produce a good amp output signal, and these pre-distortions contain a lot of high-frequency components.
Is MIC a good idea in case of BJT input stages, perhaps with only little degeneration?
Matthias
Edit: I'm aware that the C3/R14 lag compensation only is loading input stage at high frequencies. Nevertheless, more hf signals are reaching the input stage junctions. Do they harm?
Last edited:
Correction: This is Miller Input Compensation according to Bob C. I got the MIC right but the I should stand for Input and not Inclusive. This has been bothering me since I posted and I looked it up in Bob's book pg. 181.
I've pointed out before that the SWTPC Tigersaurus (Dec. 1973) amp used this form of compensation and also the Adcom GFA-555, it seems to be fairly common.
Tigersaurus, C2 here is MIC: http://www.swtpc.com/mholley/RadioElectronics/Dec1973/RE_Dec_1973_pg44.jpg
Pass A-40, C4 here is MIC: https://www.passdiy.com/projects/images/content/a_6.png
I've pointed out before that the SWTPC Tigersaurus (Dec. 1973) amp used this form of compensation and also the Adcom GFA-555, it seems to be fairly common.
Tigersaurus, C2 here is MIC: http://www.swtpc.com/mholley/RadioElectronics/Dec1973/RE_Dec_1973_pg44.jpg
Pass A-40, C4 here is MIC: https://www.passdiy.com/projects/images/content/a_6.png
Last edited:
I'm not sure if I understand your question but perhaps I can shed some light on it.
Typically we design an amp to have various gains in the different stages at audio
frequencies and then we wrap feedback around it where a side effect is that the finite
bandwidth of the components adds poles that lead to instability if not treated in someway.
I believe that MIC around the VAS and diff input linearizes them, the VAS output is
distorted because the global loop is working correctly and that is the signal required
to correct the distortions in the output stage.
If MIC is done correctly and it provides lower distortion than Cdom compensation for
example, then it has made a better diff-VAS section that does a better job of correcting
the output stage distortions.
I've built amps with MIC and simulated them but never paid close attention to it so I'm
not sure if I'm answering your questions or not, perhaps Bob or others can add to it.
One question that I've not explored in any detail is what impact MIC has on the negative
diff pair input impedance and how does that impact the global loop?
Hi Pete,
fully agree with that.
Did you use JFET or BJT input stage? Can you comment if you experienced any problems with "grainy sound"?
Some people believe that BJT are more prone to e.g. hf rectification effects, possibly leading to subjectively bad treble performance. Avoiding strong lead compensation towards the inverting input, one can hope to reduce hf stress of input stage at least a bit.
With MIC, one deliberately feeds back hf parts of distorted signals. At least two arguments would point out that this is not a problem. Firstly, the input stage has a lighter load than the traditional Miller capacitor. So stage linearity will become better, perhaps even at hf. Secondly, the hf signal does not stem directly from the amp output. So hf ingress from speaker might be a smaller problem.
Nevertheless, if one is concerned about hf susceptibility of the input stage, one should ask: Does MIC introduce an issue here? It seems to be difficult to evaluate the effects in question by "standard methods", since origin and characteristics of the assumed hf signals are unclear. For instance, also power-supply noise will be present at VAS output due to capacities and Early effects in the output stage and corresponding correction actions by the loop.
What are the listening experiences with amplifiers using MIC?
Kind regards,
Matthias
PS. I fully understand and respect if people recognized for excellent no-nonsense engineering do not comment on a question like the last one. From an engineering point of view, MIC proved to be a nice idea, since the measured results are good. But backing this up with subjective impressions would make the picture even more complete.
I've always preferred phase-lead compensation over Miller as some diy'ers will note. It works well if there are a limited number of active devices (two, preferably) in the loop. Otherwise there are too many phase shifts e.g. current mirror for a single capacitor to correct. The circuits Pete posted do not use current mirrors, and simple resistor loading for the VAS base junction. This was Bailey's preference too, (30W Wireless World design 1968).
Using phase lead compensation around the input stage has the great advantage of reducing the differential voltage across the input bases, rather than increasing it (and hence distortion) as with classical Miller compensation. It can be used with current mirrors and Darlington VAS stages but additional phase lag correction is needed, and needs to be implemented carefully so as not to offset the phase lead feedback.
Using phase lead compensation around the input stage has the great advantage of reducing the differential voltage across the input bases, rather than increasing it (and hence distortion) as with classical Miller compensation. It can be used with current mirrors and Darlington VAS stages but additional phase lag correction is needed, and needs to be implemented carefully so as not to offset the phase lead feedback.
Hi sure it is Bob's EC MOSFET amp from the 1980s, .pdf here from his site:
http://www.cordellaudio.com/papers/MOSFET_Power_Amp.pdf
Thanks, Pete.
Bob's circuit configuration is still one of the best IMO. All the phase lead compensation circuits using current mirrors (and differential VAS stages) require some form of RC phase lag, but as Bob pointed out the frequency can be set high enough not to cause problems in the audio band.
Bob's circuit configuration is still one of the best IMO. All the phase lead compensation circuits using current mirrors (and differential VAS stages) require some form of RC phase lag, but as Bob pointed out the frequency can be set high enough not to cause problems in the audio band.
The amp that I built in the 1970s was the Tigersaurus with a comp BJT diff pair, another
amp I use with MIC is the Adcom GFA-555 also BJT input. They both sound fine to me.
I am a believer that "good" amps run below clipping sound the same. We discuss null tests
here: Thoughts Concerning Cordell, Otala, and Gilbert papers
Actually, Bob removes the error correction on the output stage, changes the cascode
to a driven type and uses a hybrid of MIC/TMC for compensation mentioned in this post:
http://www.diyaudio.com/forums/soli...ells-power-amplifier-book-86.html#post2391212
I like this version better.
