Oh now come on Bob, you contrived to write your book with only five references to my twenty-plus years of work on power amplifiers, but I don't go on about it.
These are exactly the conditions I explained was solved by the HK engineers
for both T-trak and standard versions of their EF3's.
How the standard version works is "elegant".
-Main thermal feedback is based on the thermal equilibrium of the main extrusion. (Average)
-The dynamic thermal feedback "pivots" on the main and is determined
by the "faster" thermals -current demands of the driver stage.
The t-track does exactly what Bob says with and P and N device diode.
Both also use the sensor circuit to eliminate LF modulation as mentioned.
Making the response too fast would actually be bad as the Die and
slug have a defined thermal response determined by the resistance
between them.
So we are actually measuring the response of the interfaces (thermal
resistances) - the time it takes to reach equilibrium. I can test
this on my dynamic driver heatsink ..
I have many interfaces -
- driver die to slug , slug through insulator to small heatsink.
- small heatsink through to-126 epoxy to it's die.
Takes a whole <1 second for this to happen.
T-track eliminates most interfaces , but not all.
OS
for both T-trak and standard versions of their EF3's.
How the standard version works is "elegant".
-Main thermal feedback is based on the thermal equilibrium of the main extrusion. (Average)
-The dynamic thermal feedback "pivots" on the main and is determined
by the "faster" thermals -current demands of the driver stage.
The t-track does exactly what Bob says with and P and N device diode.
Both also use the sensor circuit to eliminate LF modulation as mentioned.
Making the response too fast would actually be bad as the Die and
slug have a defined thermal response determined by the resistance
between them.
So we are actually measuring the response of the interfaces (thermal
resistances) - the time it takes to reach equilibrium. I can test
this on my dynamic driver heatsink ..
I have many interfaces -
- driver die to slug , slug through insulator to small heatsink.
- small heatsink through to-126 epoxy to it's die.
Takes a whole <1 second for this to happen.
T-track eliminates most interfaces , but not all.
OS
Hi Richard,
My short answer is that MOSFETs, especially laterals, have much less issues with bias stability, including vertical MOSFETs of the IRFP240 type. I covered this in my MOSFET power amplifier with EC paper, where I compared thermal stability measurements of a vertical MOSFET amplifier with that of an equivalent BJT amplifier.
That having been said, however, MOSFETs usually create more crossover distortion due to transconductance droop in the crossover region. In general, the MOSFET output stages do not have an optimum bias where crossover distortion is minimized. To first order, the more idle bias the better. The transition toward crossover into the transconductance droop region takes place over a larger signal current range, and so it may be a bit smoother and create relatively less upper harmonics.
But the bottom line remains that as far as raw THD of the output stage goes, all else remaining equal including bias current, MOSFET output stages will generally have higher distortion. My MOSFET amplifiers without error correction and with straightforward Miller Compensation, with a single IRFP240/9240 pair biased at 150mA and a ULGF of 1MHz tend to come in at about 0.01 to 0.02% THD at 20kHz into 8 ohms.
If you want to build an amplifier with a high idle current and fairly large class A region without running into gm doubling, vertical MOSFETs are a good approach.
Finally, matching of the vertical MOSFETs among multiple output pairs can be more important than for BJTs, but my experience with modern vertical MOSFETs is that if they come out of the same tube it is pretty easy to get adequately matched sets for Vgs at the planned bias current.
Cheers,
Bob
Yes, like you made up a new term for TMC. At least I strongly defended the term VAS.
Cheers,
Bob
Hi Mr. Self
I am going to offensive. I have both your book and Mr. Cordell's book. I admit I studied Mr. Cordell's book a lot more in detail. I read through about 200 pages of your book. I found your book gives result and not much of the theory and explanation behind your finding. Pretty much take your words. What I want is explanation to back up the statements, not just take one's words. Particular I have a few things that I don't even agree with your assertion.
It might well be that you are right, but even when I read all your post in the last few days, you pretty much just said this is my experience, and this is my finding. It won't be an issue if Nelson Pass say the same thing as he is a famous designer, not a book writer. This is not comforting by any stretch for people since you are a writer of one of the two most popular book in this field. AND you expect people to learn from you.
Case in point, you said you use 0.1ohm resistor for Re and in your book, you stress running 215mA per stage that you did not give credit to Oliver's condition. You insist it is safe without putting in the limitation of power dissipation. I did a lot of calculation as I was trying to use 0.12ohm and 200mA in my amp that has 5 stages. The idle power dissipation becomes a BIG problem even I use one of the more expensive chassis with 12" X 6" X2" heatsink. And I am not even getting into hogging potential yet. I can't even use 40V rail safely.
Also, please fix the typos in your book. It is very hard to read as so many wrong reference to Figures, referred to the wrong thing etc. I had to go over and over to "GUESS" this is what you refer to. I know this happens from doing revisions, but please proof read first. This is the sixth edition!!!
Maybe I am bias towards Mr. Cordell's book. I do prove all the formulas, I read enough text books not to trust what the book is written. When I read Mr. Cordell's book, I had the Grey & Meyer book side by side to verify most of the formulas before I take his word. Also maybe I designed bipolar IC before, so we speak a lot of the same language. You make assertion, you really need to explain.
Sorry.
Last edited:
Doug, I did not "contrive" to reference you only 5 times (if that is the right count) and I certainly should have been more generous, but quite frankly most of your stuff was out there long ago; your greatest contribution was putting it all together in one place in a very readable and digestible fashion - and good, diligent, common-sense engineering, for which you deserve mountains of credit.
Secondly, two wrongs don't make a right. You specifically asked me if you could use some of my ThermalTrak material and assured me that you would reference that material in my book. So much for that.
I'll try to reference you more in my second edition, and I hope you will be fair in referencing me in your inevitable 7th edition. We disagree on a lot of things, but fair and appropriate referencing to each other should be a no-brainer.
Outright stingy referencing does not do anybody any good, especially the reader who would like to delve deeper or double-check assertions that are made. You made plainly wrong assertions in your section on ThermalTraks that any reader could have discovered by looking in my book if it had been referenced.
Cheers,
Bob
Sliding bias
Hi Alan,
You are quite right. If one of legs of the output stage conducts no current, you cannot control the bias, simply because of lack of information. But if you don't switch it off completely, only a few tens of mA is sufficient, you still have the opportunity to control the bias.
Cheers, E.
Hi Alan,
You are quite right. If one of legs of the output stage conducts no current, you cannot control the bias, simply because of lack of information. But if you don't switch it off completely, only a few tens of mA is sufficient, you still have the opportunity to control the bias.
Using a uP is insane, way too complex. Use an analog computer instead, only four matches trannies will do the job.
Don't give it up that fast and have a look at: Auto Bias part II
Cheers, E.
Ahem. I think you should know that I have been designing power amplifiers for production in serious quantities since 1980, for companies like Cambridge Audio, TAG-McLaren, Soundcraft etc etc. I am still doing it. I would like to think this carries a bit more conviction than being just a 'book writer'. But your standards may be higher than mine.
It is also very hard to write a 700-page book with no errors at all. I hope you are aware that corrections are published on my website as soon as I become aware of them.
Eh? The relevant reference to your TT work is Ref 11 at the end of Chapter 22 in Audio Power Amplifier Design, 6th edition, p542.
Oh , come on ... you both have great books !
Another angle ... the hundreds of builders I've seen have read neither of them. (Unfortunately).
They build a Badger , slewmaster , Apex audio , or DX amp ... it works -
they are happy.
With some designs , they can explore a multitude of "exotic's". This usually
sparks an interest in how these circuits actually work. Then they buy a
book , make their own IPS's.
To discount these "exotics" (symmetrical or CFA) makes for a boring hobby.
(they will move on and NOT buy books).
There are two "camps" -
- build it first ... interest is "sparked" , learn more - buy books.
- Buy books , sift out the subjective and technical differences between them ... then build.
Alan falls into this category. Most fall into the former.
PS - I speak the truth in these matters - builders think compensation and
ULGF/PM are "black arts" , and have a hard time optimizing (or calculating) static operating currents on basic designs.
OS
OS
Another angle ... the hundreds of builders I've seen have read neither of them. (Unfortunately).
They build a Badger , slewmaster , Apex audio , or DX amp ... it works -
they are happy.
With some designs , they can explore a multitude of "exotic's". This usually
sparks an interest in how these circuits actually work. Then they buy a
book , make their own IPS's.
To discount these "exotics" (symmetrical or CFA) makes for a boring hobby.
(they will move on and NOT buy books).
There are two "camps" -
- build it first ... interest is "sparked" , learn more - buy books.
- Buy books , sift out the subjective and technical differences between them ... then build.
Alan falls into this category. Most fall into the former.
PS - I speak the truth in these matters - builders think compensation and
ULGF/PM are "black arts" , and have a hard time optimizing (or calculating) static operating currents on basic designs.
OS
OS
Hi Harry,
I don't think it can be TOO fast - you can always slow it down with some RC time in the bias spreader, if that was required.
As Bob mentioned after your post, putting the diode on the die would not allow it to float - I hadn't thought about that. OTOH, connecting the diode one side to the power device emitter would certainly allow smart designers to come up with a regulation circuit referenced to Vout - after all, it's very close to the Vbe multiplier circuit anyway. I think putting the diode on the die even when not be floating would be a Good Thing.
Jan
Electrical wise, it's not out of the question to put the diode on the same die as the transistor. The bias spreader always swing with the output transistor, it's not as if you put a lot of voltage between the diode and the transistor.
The concern is if they are on the same die, that they are on the same substrate, you have much higher capacitance between the transistor and the diode, you create a feedback path from the output back to the bias spreader. This may or may not be a problem, we need to look into that a little more.
No, that's not correct. All of it, output device, diode, bias spreader all move in unison with Vout, so the voltage across the various caps doesn't change. The caps are effectively 'invisible' and can't do any harm (nor any good either).
Jan
One more diode to collector capacitance. I really don't think that's a problem as the diode on the metal backing will create the same capacitance.
Bottom line, I don't think it hurts to have the diode on the die that I can think of. It has advantage that the diode and transistor are doped by the same process at the same time, tracking got to be better and Vbe must be closer.
Write to On semi!!!
I decided against using it for three reasons, first it's a sole source, I like to have option to switch transistors. Second, this only compensate one single transistor, it's not as if it's the cure all as the others still running open loop thermal wise. Third and less important, the pin out is too closed together, I don't want to have copper plane break between pins on the pcb. forming the lead so it becomes two rolls is more troublesome.
Last edited:
I wouldn't think that's an issue as other measures are taken to share current (and thus dissipation and temperature rise) among the // pairs. So I would think that with a single diode in one device you can meaningful keep in check all devices.
Jan
Last edited:
Which is exactly what I believe about your book, Mr. Cordell, with the addition of lots of pompous hand waving, and at least 25% treated just because your book was supposed to cover everything (class D, for example). And trying to make happy everybody, including as much as possible the Golden Ear Brigade. I believe it's also a new world record of a 600 pages technical book with less than 100 references. And please don't ask for more precise pointers, I am not paid to be your book's professional reviewer (and no, I don't want the job).
But then we reached here the crux of the matter: two famous audio book writers disliking each other, each with his own, ego based, fan club. Cool
.I am not so sure each does not like the other....... but both are rather competitive and since both books are discussing the same subject.... competitiveness is 'normal'. The two books certainly show the authors' thoughts,pretty well. I like both books and both authors..... even when they seem to disagree with my own views and experience..... but then that's because I'm competitive also. We all like to share and learn from one another. I have learned from both books... So I am happy with the differences between the books in style and content. [more details on CMA, please, by both].
THx-RNMarsh
THx-RNMarsh
Last edited:
Pompous is the last word that would come to my mind in regards to Bob. Someone else though, maybe…
Do you have to be so rude? Where’s your book on audio power amplifiers? Or on any subject, for that matter?
Bob’s section about class-D is a damn sight better than Doug’s. It could still do with a bit of work, and I have provided some hopefully useful feedback in that regard. Class-D is capable of exceptional performance and deserves a text book to cover it in detail.
Fewer than.
Being harshly critical without then backing it up with constructive comments is, IMHO, extremely rude. If you can’t make the corresponding constructive comments, please keep your opinions to yourself.
Look into the mirror lately?
Don't care to help Bob, then educate the rest of us then.
And there are a few( say one) that prod both sides to get a kick out of it.But then we reached here the crux of the matter: two famous audio book writers disliking each other, each with his own, ego based, fan club. Cool