Aargh
Anyhow, TMC is not the same as OIC period.
Please take a look at the underlined characters, and it becomes clear that OIS was just a typo and should read as OIC. Besides, also this post should have made clear that I meant OIC.
This post:
If you refuse to be serious, we better stop this fruitless discussion.
Anyhow, TMC is not the same as OIC period.
Dave Zan, thanks for looking into this matter, probably this is the reason TMC outperforms TPC.
Edmond did you offend D Self or Waly in a previous life or something ??, they are doing their outmost to discredit the work you did with TMC. This is shamefull behavior, jealousy makes one nasty. Just ignore these trivial attempts.
Doesn't matter what you believe, in general TMC does not outperform TPC in the distortion department (that is, the loop gain vs. stability trade is exactly the same). Particular circuit topologies that favour either TMC or TPS are not relevant to the general case.
You are missing two years of discussions about.
There are a few points here.
Quite early on, Bode had a deep realization that only for a simplistic idealized case is distortion reduction necessarily identical to the loop gain.
So while TPC and TMC can have matched loop gains this does not prove they will have equivalent distortion performance.
Edward Cherry applied Bode's maths to the specific case of "blameless" style amplifiers in a JAES paper in 1982, May and confirmed that this inequality was not only mathematically possible but did occur in a typical amp. In fact it was much more common than the simple "loop gain=distortion reduction" case.
So the question is whether the load on the "VAS" is likely to make a difference.
Harry Dymond's influential JAES paper could be considered "canonical" and the value used there is 390 ohms.
So up at at few MHz, where the compensation capacitors start to drop out of the picture, there is an extra load on the VAS in the order of 400 ohms.
That doesn't seem trivial to me, especially when it's close to the unity loop gain frequency.
Other people use more resistance and there are issues about the ratio of the capacitors and so on, but the added load seems to be in the order of 1 kohm.
By the by. I asked previously about your loop gain probe.
It's a floated V source, like the Omicron Labs "Bode 100"?
So no impedance correction?
Best wishes
David
Last edited:
shamefull behavior
Cheers, E.
Yes, of course I did. Proving that someone is wrong is considered a serious insult.
You have mail.
Cheers, E.
Meh, that's nothing. You should've seen how outraged he was when he found out that somebody patented it after he'd been promoting it publicly for several years.
Please have a look at: US Patent 7652531
Those pirates have stolen my baby!
P.S. Turns out it had been previously patented a few decades earlier too.
Several I would say.
1) Independent AC and DC gain
2) DC bias in VAS independent of input stage and vice versa
It's possible to optimize bias point for each stage without changing the other
3) Matched dual transistors easily deployed for excellent tracking between input and VAS
4) Low part count
5) Excellent THD
\\\Jens
Last edited:
PHP:
Cherry already discussed this stuff, and a long time ago; there is nothing to linearize with, beyond a) increasing the bias, so that active devices are in a more linear region (that is, dynamic variations of the bias points are smaller) , b) use distortion cancellation (that is, use circuit symmetry wherever possible, to cancel odd harmonics) and c) use negative feedback.
If you mean the variation of the open loop distortion as a result of TPC loading the VAS, you probably missed the fact that TMC extra loads the input stage (that was another moment when E went ballistic, until he realized this is true). There's no free lunch here as well.
Otherwise, I'm not sure what the Bode deep realization was, and I don't find any of your references of any help. What I do know is that for a minimum phase system, the Bode sensitivity integral ( aka the Waterbed principle) of any system of order > 1 the only way to decrease the sensitivity (that is, increase the loop gain) is by sacrificing the stability margins. In every control theory book, electronic amplifiers are used to exemplify this fundamental result.
P.S. I'm afraid I can't further discuss my loop gain probe details.
Last edited:
>"outraged" (Those pirates have stolen my baby!)Meh, that's nothing. You should've seen how outraged he was when he found out that somebody patented it after he'd been promoting it publicly for several years.
P.S. Turns out it had been previously patented a few decades earlier too.
Huh? You don't understand my sense of humor, do you?
>"outraged" (Those pirates have stolen my baby!)
Huh? You don't understand my sense of humor, do you?
Can't speak for others, but I certainly don't. I have though noticed at least another person that doesn't.
No I probably didn't
It is quite obvious in the Y-Delta transform that I already mentioned (and JCX posted). Also Edmond has a discussion on his SuperTIS site, but it is not yet clear to me, so I left it out despite my interest.There is no reason to believe that the effect on the input load of TMC will be the same as the effect on the VAS load of TPC.
So no reason to assume that it will cancel out and leave them equal in performance.
The IPS is usually analysed as a transadmittance, with a current output.
So my first impression is that lower impedance presented to the IPS by TMC makes little difference.
Whereas the "VAS" is closer to a Transimpedance that ideally works into an open circuit, so the lower impedance presented here by TPC makes more difference
This idea misses a point. There's "no free lunch" in thermodynamics either, but that does not mean that all heat motors are equally efficient.
One can't do better than reversibility, but one can do worse.
My idea is not that TMC can exceed Bode's limits but that TPC underperforms.
His realization was 'The fact that sensitivity and return difference are not necessarily identical is of considerable theoretical interest, however, since the limitations on available "feedback" developed... are actually limitations only on the return difference.' p.102 of "Network Analysis and..." 10th print run.
The distinction seems to have been lost in Control Theory literature.
Hence talk of a "sensitivity" integral when it is actually a return difference integral.
Yes, there's some exceptions for "tricky" systems I haven't really studied.
Here you conflate the two concepts. This is common in Control Theory where it usually doesn't matter but may not be correct here.
Best wishes
David
Last edited:
No I probably didn't
There is no reason to believe that the effect on the input load of TMC will be the same as the effect on the VAS load of TPC.
So no reason to assume that it will cancel out and leave them equal in performance.
The IPS is usually analysed as a transadmittance, with a current output.
So my first impression is that lower impedance presented to the IPS by TMC makes little difference.
Whereas the "VAS" is closer to a Transimpedance that ideally works into an open circuit, so the lower impedance presented here by TPC makes more difference
This idea misses a point. There's "no free lunch" in thermodynamics either, but that does not mean that all heat motors are equally efficient.
One can't do better than reversibility, but one can do worse.
My idea is not that TMC can exceed Bode's limits but that TPC underperforms.
His realization was 'The fact that sensitivity and return difference are not necessarily identical is of considerable theoretical interest, however, since the limitations on available "feedback" developed... are actually limitations only on the return difference.' p.102 of "Network Analysis and..." 10th print run.
The distinction seems to have been lost in Control Theory literature.
Hence talk of a "sensitivity" integral when it is actually a return difference integral.
Yes, there's some exceptions for "tricky" systems I haven't really studied.
Here you conflate the two concepts. This is common in Control Theory where it usually doesn't matter but may not be correct here.
Best wishes
David
Mr. Zan, as soon as you start considering feedback loading effects, you lost the apples in an apple to apple comparison. You are now involving in the comparison the amplifier topology, which is definitely something that your best friend Bode did not consider when developing his sensitivity integrals. You cannot consider (or ignore) real world effects at your convenience (like: consider the effect of a RHP at beyond ULGF, but ignore the TMC input stage loading), that's simply not right. You either chose a topology and do a comparison in that particular case (and don't claim that TMC is in general superior to TPC - or the other way around) or you consider an ideal amplifier, and then you already know the answer, based on Bode sensitivity integral.
Regarding the transconductance-transimpedance amplifier model, I would be every bit as concerned about loading the input stage, as for loading the TIS/VAS. A load at any is shunting away current, leading to a loss in open loop gain. Also these loadings may lead to increased distortions, but this has an easy cure, based on Cherry principles: increase the bias current, so that the devices are deeper in class A. Coincidentally, this may also help with restoring the lost open loop gain.
Which means that even for the simplest Blameless topology, you can flip the TMC-TPC score by simple rearranging the gain distribution between the input stage and the TIS/VAS. It is pretty obvious that if the TIS/VAS gain decreases, there's less feedback to linearize the output stage, and if you add the input stage loading (hence at least lost gain) you will get a pretty bleach picture. In the same topology, TPC has an advantage, because by loading the VAS there's not much gain to lose, anyway, but the IPS gain will be conserved for the global loop to linearize. Always recall, we are talking about max 3-5% differences, which, in my opinion, is not worth even considering. There are things that are worth much more attention in a practical amplifier.
Member
Joined 2009
Paid Member
Well, Bode didn't develop any sensitivity integrals.
A scan of the entire book shows he never used the term, neither is there any trace of such an integral in the sensitivity analysis. It seems to be a later misnomer, which is a pity after he made a particular effort to separate the concept of sensitivity and handle it very precisely.
But in any case, the feedback integrals don't depend on specifics of the amplifier and apply equally well to TPC or TMC.
So we can compare them perfectly fairly.
Nor do I propose to do so. Bode's analysis is not just for an idealized amplifier. It inherently considers load effects.
There seems to be a difference here.
The TPC extra load on the VAS is a dead-loss, just a shunt to earth.
The TMC extra load on the IPS is a result of feedback around the VAS, similar to Miller compensation.
The low impedance presented to the IPS by the typical "blameless" Miller compensated VAS is not usually considered a problem, why is the TMC case different?
Surely in a sensible amplifier the bias current has already been optimised?
Why would you have sub-optimal bias current just so you can increase it?
No, I don't recall. Where was this number arrived at?
Yes, I enjoy these discussions with Waly too. But that one was more of a debater's post - present dubious claims as facts. But well done.
Have you any expertise on the subject to share? I find it helpful to kick ideas around because Bode is hard work to understand and many of the later writers are shallow.
Best wishes
David
Last edited:
Well, I don't (or I no longer). As long as you say I "present dubious claims as facts", I'm afraid I'll stop here and suggest you create your own fan club to find an audience, good luck with that.
If you need to be taken seriously, next time you make an extraordinary claim just come up with some complete and relevant examples and numbers to substantiate them. I'm myself sick of nitpicking around the bush while ignoring the tons of examples (e.g.) regarding the same distortion numbers (within a few percent) that were posted here in in almost all threads debating TPC vs. TMC. Even the inventor of the TMC acronym, and Mr. Cordell, conceded that in a Blameless topology TMC has no significant advantage over TPC, in the distortion reduction department.
C'mon Edmond, let's not take this too seriously. That was one aside in a long technical discussion.
But why is that an advantage?
Their operating currents are completely independent in the standard Blameless circuit.
??? Tracking of what?
Surely having different AC and DC gains is going to require more components.
Are you saying THD would be improved? How ???
Development, project of a diy Class D discrete amplifier
This is something i am searching for...maybe a good thing to your book..despite Class D is not a serious High Fidelity (or was not in the past)...it is something people has interest.
Crazy thing when produce power..the speaker wires can kill you... a lot of high frequency from the oscilatorremains there..unfiltered...at least all i have made here result this way.
Well...they may assemble and give up.
regards,
Carlos
This is something i am searching for...maybe a good thing to your book..despite Class D is not a serious High Fidelity (or was not in the past)...it is something people has interest.
Crazy thing when produce power..the speaker wires can kill you... a lot of high frequency from the oscilatorremains there..unfiltered...at least all i have made here result this way.
Well...they may assemble and give up.
regards,
Carlos
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.