Thanks for the input. I can do a "normal" Monte-Carlo. My issue is how to do it when you are trying to simulate loop-gain with a middlebrook probe and need to combine results from different steps.
Hi
The link is excellent. I found and read it a few years back and re-read it when reminded by Michael's post. Thanks to him for that because it was helpful to read it second time. So you probably need to read it a few times too
RHP zeros are practically never desirable in an amplifier so it is worth the effort to eliminate them. In a conventional amplifier the only one is likely to be caused by the VAS aka TIS. Usually not a problem because it is above any frequency of interest.
It is possible for a particular combination of reasonable component choices to result in the RHP zero frequency to be sufficiently low to affect the stability a little. This can occur if the LTP emitter resistors are fairly low (to reduce noise probably) compared to the VAS/TIS emitter resistor (used for over-current protection, often).
If your circuit is unconventional then normal rules-of-thumb won't help. But I doubt a RHP at >3GHz will hurt.
Best wishes
David
In this link you find the classical analysis of an amplifier as a single common emitter amplifier.
The pole position diagram is very enlighting and shows the variation of singularities with gm or Cmiller as parameter. It is easy to see that p2 remains constant with C miller increasing and the zero decreases even becoming lower than the pole.
This is what I explained ( apparently badly) in a previous post and was called 'technobable' by some
In case of a two stage amplifier of the Miller opam/OTA style, the same analysis applies but with some interesting différences:
We now have two gm's one for the first transconductance stage and a second for the transimpedance stage . The pole/zero position diagrams remain the same but gm is now the second stage gm2. The GBW on the other hand is = gm1/cmiller. We can play with gm2 to improve pole separation without changing GBW
A very good in depth reference from which the link above is perhaps inspired is: Laker and Sansen design of analog integrated circuits and systems.
This books explains also the concept of OTA ( a full chapter) and how it is complementary and usefull in the description of opamps.
A dipole can always be seen as Thevenin or Norton one ( Kirchof always applies
). Depending on the view taken and the environment, one of the other is more usefull in the understanding of a circuit or subcircuit function or operation.A miller OTA is a good example: it is analyzed as the cascade of a transconductance followed by a transimpedance stage because voltage at the ouput and current at the input is the good approach to describe the function of the second stage. But the output node is a high impedance one at low frequencies and low one at high freq. So, a Miller 2 stages block is an Miller OTA at low frequencies and can be ( is) considered as an opamp ( voltage amplifier) at higher frequencies because the output impedance becomes low.
Mike should read this book, it will make him perhaps less extremist in his opinions and open his mind. This is not a science fiction book
Cheers
JPV
Hmm. Yes, multi- dimensional steps coupled to MC could be problematic.
The LTSpice guru over there is Helmut Sennewald - he should be able to provide you with some pointers.
Mr Lee, I was persuaded a long ago time ago that you know not and know not that you know not, which is why I elected to ignore all your hilariously captious posts addressed to me.
Where exactly in any of my posts did I use the phrase 'rotten designs'?
This is the last time I'll address any of your nonsensical posts.
Last edited:
Mike should read this book, it will make him perhaps less extremist in his opinions and open his mind. This is not a science fiction book
My views are not "extremist" and, No, a Miller compensated amplifier is not an "operational transconductance amplifier" (OTA) as I explained here:
http://www.diyaudio.com/forums/soli...lls-power-amplifier-book-373.html#post3547804
An OTA delivers, by definition, current out for voltage in, while the first two stages of a Miller compensated amplifier deliver voltage out for voltage in.
Read solomon:
http://ece.wpi.edu/~mcneill/524/handouts/solomon.pdf
Last edited:
I have read Solomon of course but follow my advice: it is very interesting to read Laker and Sansen or Sansen ' Analog design essentials'.
Then we could discuss more openly
You can just plot the same Tian probe function but with the iteration step variables &1 and &2 replaced by &3 and &4, and so on. There is probably a more elegant method but this is easy and works.
Best wishes
Davis
Hi Mike,
I understand what you mean. But the fact is that a miller compensated two stage amplifier (basically the first two stages of a “traditional” audio power amplifier) is usually called an OTA.
The reason is the following: the difference between an OpAmp and an OTA is that, ideally, the former has zero open loop output impedance (ability to force the voltage at its output), and the later has infinite open loop output impedance (ability to force the current at its output).
Any real amplifier will never feature either zero or infinite output impedance, and so it will never be a pure OpAmp or OTA. This leaves some freedom for us to use what terminology we find more convenient…
In any way, in OpAmps - an example of which is the “traditional” audio power amplifier - there is always some kind of emitter/source follower stage to ensure the low output impedance, to try to approach the ideal situation of zero output impedance.
If you remove the output stage in the “traditional” audio power amplifier, then the output impedance will be high (in the audio band – frequency range of interest), and this approaches the situation found in an ideal OTA.
I understand your argument that the miller compensated two stage amplifier provides a voltage output. But you can also think about it as presenting a current output, that is “transformed” to a voltage by the output resistance. Furthermore, there is another feature of OpAmps that the miller compensated two stage amplifier cannot fulfill, due to its high output impedance: driving low resistance loads…
Regards
Evil Zeros etc
Good zeros have a rising response AND phase with frequency. If set just above the ULGF, the rising phase can increase PM though some care is needed to ensure the rising response doesn't reduce GM.
What Prof. Ze and JPV point out is that in some cases notably the Common Emitter stage used as the VAS, there is an inherent zero which is evil.
It has rising response BUT falling Excess (did it right this time, Waly) phase with frequency. Hence it will reduce both GM & PM. It introduces non-Minimum Phase behaviour cos the Excess Phase.
I shall call these Evil Zeros and invite Waly & Michael to pontificate on the semantic & pedantic issues vis a vis these, RHP, Minimum Phase, All Pass bla bla .. which will no doubt help us all design better amplifiers.
Guru JPV, with his practical view, is concerned simply with how to move these evils to as high a frequency as possible, where they won't do any harm, by judicious choice of caps & gm of the IPS & VAS.
Guru Zan has observed that that the VAS emitter resistor (that Michael so despises) and its decoupling also bear on this issue ... while investigating a 'real life' design. Da pseudo gurus might like to do likewise as theory & sims which reflect 'real life' are really useful.
I'm not sure of the use of theory & sims which DON'T reflect 'real life' but some people seem to prefer these.
Together Gurus JPV & Zan slowly move those versed in the art to a better understanding ... and hopefully better facility in dealing with these evils.
________________
Monte Carlo
I'm only a SPICE newbie but in the discrete-opamp-open-design, I tried to sim using all the models I had available. Crude but better than nothing.
FWIW, Bob Cordell's models consistently gave the worst THD & stability so I'm inclined to believe they are 'better' than the others
________________
It's difficult to take you seriously if you can't even get decent THD & stability in SPICE world.
You obviously have a comprehensive library but QUOTING VERY LOUDLY from a book sometimes just demonstrates you haven't read it properly. You may like to re-read some of these volumes from cover to cover ... seeking understanding as JPV has advised ... instead of stuff to exercise your undoubted prowess at semantic pedantry
Similarly, re-visit sims more comprehensively as Bob has suggested. A good approach is to ask 'why a succesful design works well' rather than your standard 'it can't be any good cos its got the wrong TLA bla bla'
Excess (sorry Waly) phase shifts introduced by HF roll-offs (poles) in devices & circuits can sometimes be 'cancelled' by introducing zeros (HF lift). A good example of this is the small cap across the feedback resistor. This introduces more feedback at HF (ie lift) and if used judiciously, can help stability.I noticed that the phase never reached 270' only 147' ? There do though "appear" to be some similarities between my phase plot & the ones in the PDF. I don't pretend to understand it all, or anywhere near, i'm just attempting to establish whether there is "something" worthwhile exploring further in RHP ?
An explanation of what i'm seeing, would be welcome
Good zeros have a rising response AND phase with frequency. If set just above the ULGF, the rising phase can increase PM though some care is needed to ensure the rising response doesn't reduce GM.
What Prof. Ze and JPV point out is that in some cases notably the Common Emitter stage used as the VAS, there is an inherent zero which is evil.
It has rising response BUT falling Excess (did it right this time, Waly) phase with frequency. Hence it will reduce both GM & PM. It introduces non-Minimum Phase behaviour cos the Excess Phase.
I shall call these Evil Zeros and invite Waly & Michael to pontificate on the semantic & pedantic issues vis a vis these, RHP, Minimum Phase, All Pass bla bla .. which will no doubt help us all design better amplifiers.
Guru JPV, with his practical view, is concerned simply with how to move these evils to as high a frequency as possible, where they won't do any harm, by judicious choice of caps & gm of the IPS & VAS.
Guru Zan has observed that that the VAS emitter resistor (that Michael so despises) and its decoupling also bear on this issue ... while investigating a 'real life' design. Da pseudo gurus might like to do likewise as theory & sims which reflect 'real life' are really useful.
I'm not sure of the use of theory & sims which DON'T reflect 'real life' but some people seem to prefer these.
Together Gurus JPV & Zan slowly move those versed in the art to a better understanding ... and hopefully better facility in dealing with these evils.
________________
Monte Carlo
I'm only a SPICE newbie but in the discrete-opamp-open-design, I tried to sim using all the models I had available. Crude but better than nothing.
FWIW, Bob Cordell's models consistently gave the worst THD & stability so I'm inclined to believe they are 'better' than the others
________________
Please don't harm yourself. It would be more useful for the rest of us if you addressed the THD & stability of your designs.Where exactly in any of my posts did I use the phrase 'rotten designs'?
This is the last time I'll address any of your nonsensical posts.
It's difficult to take you seriously if you can't even get decent THD & stability in SPICE world.
You obviously have a comprehensive library but QUOTING VERY LOUDLY from a book sometimes just demonstrates you haven't read it properly. You may like to re-read some of these volumes from cover to cover ... seeking understanding as JPV has advised ... instead of stuff to exercise your undoubted prowess at semantic pedantry
Similarly, re-visit sims more comprehensively as Bob has suggested. A good approach is to ask 'why a succesful design works well' rather than your standard 'it can't be any good cos its got the wrong TLA bla bla'
I don't think there is a case for calling a two stage miller compensated amplifier an "OTA" merely because it has a relatively large output impedance at DC and frequencies of a few Hertz.
In any case the output impedance of a miller compensated two-stage amplifier drops rapidly after the dominant pole when shunt derived negative minor loop feedback takes effect.
Last edited:
I know, you poor thing; I KNOW! And from your posts it is abundantly obvious you're a "newbie" at everything else. Really sad for a man of your advanced "Jurassic" years.
kgrlee,
Please could you elaborate on this point a little. It would really help my understanding of other concepts if I can get this one.
The message I tried to pass is: naming something assumes some kind of convention - no one calls dog to a cow, because everyone follows the same convention.
In this case, textbooks define Opamp as an electrical component with low (ideally zero) output impedance, and OTA as an electrical component with high (ideally infinite) output impedance.
In the frequency band of interest (more on this below), the two stage opamp has relatively large output impedance, hence it is usually called an OTA.
Yes, but let us then get back to the basic definitions: fundamentally OTAs and Opamps exist for us to apply negative feedback and minimize (ideally make zero) their input voltage. This happens because of the large gain/transconductance or the Opamp/OTA, and is absolutely central in feedback theory.
As you say, the output impedance drops to higher frequencies in the two stage miller compensated amplifier… but so does the gain.
When the output impedance gets to be low enough for it fall in the definition of Opamp, the gain is also pretty low, negative feedback is not being effective anymore, and so you’re out of the frequency band of interest.
Note that I'm not trying to convince you to start calling it an OTA, but rather explain you why there are many (many, many, many people around the world) that do that. At the end of the day, if you model it properly (transconductance of the output transistor connected to the resistance "seen" at the node and the capacitances connected there) the circuit will work no mater what you call it
And yet you pompously and captiously pontificate about my SPICE simulations as though you are an Authority on the subject...