I don't have time to look at the paper in detail but this would suggest that for a valid analysis we'd actually need to dig into the opamp and place the probe inside its Miller loop. What's the phase margin of your opamp model BTW?
Don't think so. The opamp Miller loop doesn't have a common node at the output with the rest of the network. For including the opamp Miller loop, a more general criteria is required (multiloops not having a common node at the output, based on the number of turns around the -1 point for each loop, it's way off topic here). Or you can consider the opamp dominant pole far away from the rest and leave the opamp Miller loop as it is.
Because there is more local feedback of the opamp--the ULG of the global feedback loop surely decreases. And we know that the opamp is unity-gain stable so this is not much of an argument. As I said earlier in this thread the use of Ahuja compensation (i.e. including the cascode within the Miller compensation loop) opens the possibility for local instability of the compensation loop. But it works well in this particular case--the bandwidth of the cascode must be at least an order above that of the opamp. And it just increases PSR so much...
The 39 degs phase margin reflects exactly the trend for local instability. In a multiloop environment, the global stability margin is actually determined by the least stable loop.
-30dB at audio frequencies. For a closed loop gain of 32dB (as I use it in my HPS5.1), the PSR is -70dB at the same frequencies. Yours is better, though.
Don't think so. The opamp Miller loop doesn't have a common node at the output with the rest of the network.
Many current IC opamps do have inclusive Miller compensation and I don't see how a buffer in between should have a significant effect on loop gain (if we assume its bandwidth to be sufficiently large). But from my side we can stop the discussion here. I've never had (and probably never will have) sufficient confidence in opamp macro models to accept a simulation as proof or disproof for anything. I'd rather prefer to spend my efforts on the original topic. I do have a couple of ideas based on your design but will need some more time to check a few things.
Samuel
But from my side we can stop the discussion here.
Ok. My bottom line is that the Ahuja compensation can be great for ICs (and probably optimal for CMOS linears) but for DIY discrete designs (with inherent unity loop gain frequency limitations) is less than optimal.
Nice talking to you.
My bottom line is that the Ahuja compensation can be great for ICs (and probably optimal for CMOS linears) but for DIY discrete designs (with inherent unity loop gain frequency limitations) is less than optimal.
This is in general my observation as well. I've spent considerable research in other high-PSRR topologies suitable for discrete implementation and came up with at least two excellent alternatives (not yet time to share those though).
Samuel
Samuel, I asked you this before, but the question got lost among others. A low noise measurement preamp like this is probably best powered by batteries; if you agree, then why do you insist on high PSRR?
Somewhat related to this issue, syn08 has a whole series of HPS phono stages, and I recall very clearly how, before HPS 5.0, he pointed out the inferiority of some other designs because of their poor PSRR, and how it's not good engineering to put all that effort in the power supply. It surprised me (pleasantly) when he came out with version 5.0 in which he explores a much simpler head amp design with lower PSRR, moving the complexity to the psu. I always thought it was more a matter of taste whether one gets rid of ripple in the psu, or in the preamp. For a home builder there are all kind of criteria which the designer might want to rather take into consideration, for instance, can the thing be built and tested with a minimum of measurement instrumentation? So perhaps the question is, are the specs well defined, do those specs make sense for the intended use (in reality), and does the design meet those specs?
Somewhat related to this issue, syn08 has a whole series of HPS phono stages, and I recall very clearly how, before HPS 5.0, he pointed out the inferiority of some other designs because of their poor PSRR, and how it's not good engineering to put all that effort in the power supply. It surprised me (pleasantly) when he came out with version 5.0 in which he explores a much simpler head amp design with lower PSRR, moving the complexity to the psu. I always thought it was more a matter of taste whether one gets rid of ripple in the psu, or in the preamp. For a home builder there are all kind of criteria which the designer might want to rather take into consideration, for instance, can the thing be built and tested with a minimum of measurement instrumentation? So perhaps the question is, are the specs well defined, do those specs make sense for the intended use (in reality), and does the design meet those specs?
I should add that I don't mean to be speaking for syn08, he knows better what he has done; just wanted to mention my perspective, that's all. Any errors in the report are mine.
Surely in the end overall performance matters. In general it is however much simpler to design an amplifier with good PSR than a low-noise/low output impedance supply, particularly if we insist on full (e.g. short-circuit and thermal) protection and low extra power consumption. As long as there is no significant disadvantage from optimising the amplifier for high PSR (usually a few simple tricks bring things up an order of magnitude) I see no reason to rely on PSU performance.
I've not yet investigated Battery noise in detail but as far as I know some types ain't that low noise at all; also they show considerably increased noise towards their end of life and there might be substantial differences from specimen to specimen.
Samuel
I've not yet investigated Battery noise in detail but as far as I know some types ain't that low noise at all; also they show considerably increased noise towards their end of life and there might be substantial differences from specimen to specimen.
Samuel
I've not yet investigated Battery noise in detail but as far as I know some types ain't that low noise at all; also they show considerably increased noise towards their end of life and there might be substantial differences from specimen to specimen.
There is a slew of articles in the IEEE library regarding batteries noise, starting with the early 60's; I haven't followed closely the topic, but last time I've checked anything except the lead-acid cells was improper for anything well under 1nV/rtHz. Of course, adding a cap (or whatever R-L-C network) across the battery always helps, but this is, as usual, a tradeoff with the output impedance, etc... The common NiCd cells were among the worse offenders.
The idea to feed a pre from lead acid cells is to me a major turn off, to put it mildly, although I know some eccentric high end audio designers are placing their bets on such a solution.
A bit old, but it does give an idea.
1995 IEEE INTERNATIONAL FREQUENCY CONTROL SYMPOSIUM. MEASUREMENT OF VOLTAGE NOISE IN CHEMICAL BATTERIES. Chadwick K. Boggs, Alan D. Doak, F. L. Walls
http://tf.nist.gov/general/pdf/1133.pdf
For those in a hurry, attached Figure 2.
I would definitely not like batteries in a phono stage for instance, but the idea doesn't bother me for a lab measurement preamp which is only occasionally used. In fact many instruments have a battery option as well, but can't be sure if it's for portability reasons or low noise (or both).
1995 IEEE INTERNATIONAL FREQUENCY CONTROL SYMPOSIUM. MEASUREMENT OF VOLTAGE NOISE IN CHEMICAL BATTERIES. Chadwick K. Boggs, Alan D. Doak, F. L. Walls
http://tf.nist.gov/general/pdf/1133.pdf
For those in a hurry, attached Figure 2.
syn08 said:The idea to feed a pre from lead acid cells is to me a major turn off, to put it mildly, although I know some eccentric high end audio designers are placing their bets on such a solution.
I would definitely not like batteries in a phono stage for instance, but the idea doesn't bother me for a lab measurement preamp which is only occasionally used. In fact many instruments have a battery option as well, but can't be sure if it's for portability reasons or low noise (or both).
Attachments
Thanks. Their technique to measure noise in the pV/sqrt(Hz) range is sweet. I might adapt this for 1/f resistor noise measurements.
Samuel
I agree, nice simple idea, lots of averaging, but then again, it's just a matter of time.
I'll see if I can dig out some results on lead acid batteries, which I was planning to use for this preamp.
I'll see if I can dig out some results on lead acid batteries, which I was planning to use for this preamp.
Thanks. Their technique to measure noise in the pV/sqrt(Hz) range is sweet. I might adapt this for 1/f resistor noise measurements.
Samuel
For low frequencies the averaging time for the cross-corellation technique will get VERY long. Too bad it's an instrument technique and not an amplifier one.
Thanks. Their technique to measure noise in the pV/sqrt(Hz) range is sweet. I might adapt this for 1/f resistor noise measurements.
Samuel
Yes, but... it would have been proper for him to credit Ardelt van der Ziel for the cross-correlation measurement technique.
I wonder if it's possible to implement a cross correlator with an analog multiplier.
For low frequencies the averaging time for the cross-corellation technique will get VERY long. Too bad it's an instrument technique and not an amplifier one.
4nV squared is a very low number!
Yes, but... it would have been proper for him to credit Ardelt van der Ziel for the cross-correlation measurement technique.
I wonder if it's possible to implement a cross correlator with an analog multiplier.
I think there was one in an old Analog Dialog. I think it's Aldert, but anyway he would probably say "it's obvious" 🙂
it would have been proper for him to credit Ardelt van der Ziel for the cross-correlation measurement technique.
That would be like crediting God for the Creation. Almost everything we know about noise in solid state systems is originating from van der Ziel's work. The guy will probably get a Nobel soon.
That would be like crediting God for the Creation. Almost everything we know about noise in solid state systems is originating from van der Ziel's work. The guy will probably get a Nobel soon.
I think he passed away in 1991 -- I actually learned of him from one of your posts!
This thesis by Judith Furlong references van der Ziel -- I wonder if Scott knows the referees?
http://dspace.mit.edu/bitstream/handle/1721.1/26831/25315341.pdf?sequence=1
I think he passed away in 1991 -- I actually learned of him from one of your posts!
This thesis by Judith Furlong references van der Ziel -- I wonder if Scott knows the referees?
http://dspace.mit.edu/bitstream/handle/1721.1/26831/25315341.pdf?sequence=1
Hermann Haus taught me field theory, brilliant and wonderful teacher.
EDIT - Prof. Haus has also passed away http://web.mit.edu/newsoffice/2003/haus.html
Last edited:
I think he passed away in 1991 -- I actually learned of him from one of your posts!
I'm getting old, my memory is short

- Home
- Amplifiers
- Solid State
- Simple 60dB discrete low noise amplifier (lna)