Ah right. I thought maybe it was #960
Either way, you make a good point about the diagnostic use of TMC versus conventional Miller to distinguish between distortion from the output stage and that from the input stage and VAS.
In the tests I did, it seemed pretty clear that what distortion remained with TMC switched in was still basically crossover distortion from the output stage. The Blameless degenerated-input-pair plus beta-enhanced-VAS configuration does seem to be capable of remarkably low distortion for such a simple circuit. I know many more complicated arrangements have been put forward, but that seems to me a bit redundant until we finally get the output stage sorted out.
You are the king of amplifiers...our main source of inspiration
You are the best...also Blameless is the best...i gave my personnal touch to enter the ship of glory,...i have include a bootstrap in the place of the CCS that was feeding VAS...i found the sound more interesting...but cannot say it is better...maybe better to me.
Even with the Bootstrap (my personal taste) it is unbeatable...i could not listen any better amplifier in 50 years building more than 6000 amplifiers..i have never listened anything so good.... also i have searched your other designs and i have tested them..no one has beated Blameless circuit in my ears, mind and heart.
This is the last circuit i will build, i may release higher power versions or soemthing alike..just small touches... because it is the best one.
My deep congratulations Doctor Self...i feel honored with your attention and also Bob Cordell attention, as i am expert to make things work together, i am not a designer....i tune things listening only.
I have searched internet and i could not see your face...this was a pity.
Merry Christmas!
Carlos
You are the best...also Blameless is the best...i gave my personnal touch to enter the ship of glory,...i have include a bootstrap in the place of the CCS that was feeding VAS...i found the sound more interesting...but cannot say it is better...maybe better to me.
Even with the Bootstrap (my personal taste) it is unbeatable...i could not listen any better amplifier in 50 years building more than 6000 amplifiers..i have never listened anything so good.... also i have searched your other designs and i have tested them..no one has beated Blameless circuit in my ears, mind and heart.
This is the last circuit i will build, i may release higher power versions or soemthing alike..just small touches... because it is the best one.
My deep congratulations Doctor Self...i feel honored with your attention and also Bob Cordell attention, as i am expert to make things work together, i am not a designer....i tune things listening only.
I have searched internet and i could not see your face...this was a pity.
Merry Christmas!
Carlos
Last edited:
TPC increases loop gain visible at the input, reducing diff input V, hence input stage distortion
and also modifies ("eliminates"?) the Miller component of the slew rate limit at audio frequencies
doing so could allow a lower noise front end with large area bjt without degeneration - which may be a legitimate design decision with low source Z - rather than "a bad job on the ips"
Last edited:
@ Bob Cordell RE 4 ohm speakers --
What you say is true Bob. And after years in the consumer electronics and audio business at every level from repair bench to engineering lab to advertising, here's what I think I know: The numbers on the equipement have almost nothing to do with some objective reality and almost everything to do with warranty and liability obligations, hence amp makers who rate their units for "8 ohms" do so less to specify an exact load for optimum sound (and what load would that be, anyway?) and more that they can restrict their responsibility for repair costs.
Speaker makers have tended to be a little more honest overall, at least US and Brit builders, usually representing their products' impedances in relation to the woofer's voice coil DC resistance, which in many (but not all) cases is the lowest Z the amp will see. Will a mass-market receiver with an 8 ohm load rating on the back satisfactorily drive a 4 ohm speaker? In most cases, yes, particularly for those who are not commited to the hunt for "audiophile" grade performance. I only meant to point out that an awful lot of customers in the past have related to me that the difference in those numbers caused them a great deal of worry about potential harm, when no real harm would likely ever ensue.
Wouldn't it be great if amp makers provided VI plots vs frequency and Z, and speaker makers provided amplitude and phase plots vs frequency? Do any of them do that?
Sorry, this kinda turned into a rant and that's not where I wanted to end up; as we all know, the gear we have designed and built was not primarily made for use by engineers, but by ordinary mortals, with all that implies for use and misuse.
What you say is true Bob. And after years in the consumer electronics and audio business at every level from repair bench to engineering lab to advertising, here's what I think I know: The numbers on the equipement have almost nothing to do with some objective reality and almost everything to do with warranty and liability obligations, hence amp makers who rate their units for "8 ohms" do so less to specify an exact load for optimum sound (and what load would that be, anyway?) and more that they can restrict their responsibility for repair costs.
Speaker makers have tended to be a little more honest overall, at least US and Brit builders, usually representing their products' impedances in relation to the woofer's voice coil DC resistance, which in many (but not all) cases is the lowest Z the amp will see. Will a mass-market receiver with an 8 ohm load rating on the back satisfactorily drive a 4 ohm speaker? In most cases, yes, particularly for those who are not commited to the hunt for "audiophile" grade performance. I only meant to point out that an awful lot of customers in the past have related to me that the difference in those numbers caused them a great deal of worry about potential harm, when no real harm would likely ever ensue.
Wouldn't it be great if amp makers provided VI plots vs frequency and Z, and speaker makers provided amplitude and phase plots vs frequency? Do any of them do that?
Sorry, this kinda turned into a rant and that's not where I wanted to end up; as we all know, the gear we have designed and built was not primarily made for use by engineers, but by ordinary mortals, with all that implies for use and misuse.
TMC TPC
I fully support D. Self suggestion to open a thread where all posts on TMC vs TPC can be found .
I have read the AES paper on TPC and found it very well written. I have only one caveat: it is the authors opinion that '' an analysis is not provided" in the litterature.
I am recommending the reading of Roberge book from MIT ( 1975 !) ' Operational amplifiers theory and practice' which gives an in depth investigation of compensation and TPC in particular both in theory and with practical hw tests and scope views. One of the difficullty with TPC is the sensitivity to capacitive loading and the bad overload recovery.
He develops a method based on minor loop analysis that is very simple and avoids the huge algebra needed to solve the node equations using the brute force approach .
One of his collaborators at MIT ( Kent H Lundberg) has written a nice paper on the approach: Internal and External OP-Amp compensation: A control - centric tutorial.
JPV
I fully support D. Self suggestion to open a thread where all posts on TMC vs TPC can be found .
I have read the AES paper on TPC and found it very well written. I have only one caveat: it is the authors opinion that '' an analysis is not provided" in the litterature.
I am recommending the reading of Roberge book from MIT ( 1975 !) ' Operational amplifiers theory and practice' which gives an in depth investigation of compensation and TPC in particular both in theory and with practical hw tests and scope views. One of the difficullty with TPC is the sensitivity to capacitive loading and the bad overload recovery.
He develops a method based on minor loop analysis that is very simple and avoids the huge algebra needed to solve the node equations using the brute force approach .
One of his collaborators at MIT ( Kent H Lundberg) has written a nice paper on the approach: Internal and External OP-Amp compensation: A control - centric tutorial.
JPV
Hi Doug,
I agree. The output stage is the main nemesis and is usually the most expensive one to improve (in terms of dollars, heat and size). Making an IPS-VAS that is superb is much less expensive by comparison. That is why I'm inclined to favor TMC over TPC. However, implicit in my leaning is the sense that TMC does more to improve the output stage than TPC for equal margin in the amplifier against any sort of instability. Whether my sense of that is correct in light of a rigorous exploration is unclear.
Cheers,
Bob
You're right about the reduced average error signal at the input stage for TPC vs TMC, and I was a bit harsh in my statement about the IPS-VAS design; I probably should have said one should try to improve further the IPS-VAS. What I meant was that if the IPS-VAS contribution to distortion is not less than that of the output stage, there is probably more low-hanging fruit for improvement in the IPS-VAS than in the output stage.
I have not investigated by how much TPC increases slew rate compared to conventional Miller compensation, and that is worth a look. I usually use Miller Input Compensation, which largely eliminates the slew rate connection to compensation.
In respect to TPC's possible improvement of the slew rate, I would not recommend eliminating degeneration of the input stage. I think that is really asking for trouble. I have not usually found input-referred noise to be a problem in properly-designed power amplifiers. I believe it is far more important to attack output stage nonlinearity than further drive down already-low noise.
Cheers,
Bob
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.151.7235&rep=rep1&type=pdf
Also, Dennis Feucht did the ultimate "brute force" two pole compensation analysis. Can't imagine anything more complete than his work:
http://www.analogzone.com/col_0628.pdf
http://www.analogzone.com/col_0719.pdf
Last edited:
Hi Bob
Just adding, their input compensation is used in this topology :
http://www.diyaudio.com/forums/soli...-complementary-input-stage-6.html#post1298324
Is an elegant circuit, I have not tried to make a simulation, but I have a simple circuit (LTP more VAS Balanced), conventional Miller compensation does not work this topology , because hinders roll-off of the two outputs of LTP. In my case, Miller short-circuit the VAS, I getting 90V/us, PM 80° and crossover in 1.2MHz.
The circuit of Hitachi that uses conventional Miller, Slew Rate is limited to 20V/us due to roll-off asymmetric, if the Slew Rate is not limited (RC in NFB), you will have oscillation on high frequency.
BTW: I ordered your book, must reach me in a few days
Thanks
Hi, Bob
You will notice that i did mention "generally", wich is different
from "all the time"...
Anyway, our views are not antinomic, rather differents things
are highlighted that are not directly related..
What i was pointing is that TPC will have a lower OLG to
start with before compensation is connected, that is the
result of the heavier VAS loading than with TMC.
While it wont appear as significant in a relatively low gain vas,
due to already existing heavy loading by the OS stage,
it is manifest whith high OLG vas, and TPC has about 15db
less OLG than TMC in a wide region where the OPS is still
functionnal and largely encompassing the audio band..
This advantage of higher total OLG for TMC, and thus higher
total available NFB, is surely one among many of the reasons
why it reduce THD so efficiently.
The drawbacks pointed by JCX can easily be overcomed
by increasing the vas OLG.
cheers,
w
Kudos to Bob, excellent book, i'm 1/3 through it and will finish during Xmas holidays.
I really like your IPS with floating jfets (fig. 7.12 in the text)
I really like your IPS with floating jfets (fig. 7.12 in the text)
Howdy,
the most deleterious capacitance in amplification devices is the Miller capacitance whose effect in the VAS is devastating and should be kept small at all cost. Frequency compensation methods extending the Miller capacitance externally should be banned from audio. You end up with distortion with any values because capacitances prolong rise time while increasing damping (by shortening settling time depending on zero frequency and implementation accuracy). A faster response time means more ringing and overshoot and ringing is nothing else but damped oscillation, clearly a distortion (much worse than THD).
the most deleterious capacitance in amplification devices is the Miller capacitance whose effect in the VAS is devastating and should be kept small at all cost. Frequency compensation methods extending the Miller capacitance externally should be banned from audio. You end up with distortion with any values because capacitances prolong rise time while increasing damping (by shortening settling time depending on zero frequency and implementation accuracy). A faster response time means more ringing and overshoot and ringing is nothing else but damped oscillation, clearly a distortion (much worse than THD).
Many thanks, but I have a copy of it now.
Good to see the stuff by Dennis Feucht. His book is very good.
Hang on, there' s only so many hours in the day. I haven't studied the issue much but my recollection is that the phase angle of the load (assuming the modulus is constant) alters the phase relationship of the distortion products to the main waveform, but does not significantly alter their character.