What's even more bizzare, at least I can understand that if one is using awkward TO-3 devices, as it minimizes wiring, but when using TOP-3 plastic pack devices? Lunacy!
For the love of God, it was made precisely so to enable direct mounting on the heat sink! Use what you already have!
Actually, though, the L shaped angle CAN work rather well IF three criteria are met:
1. If the "crossover" L plate covers the WHOLE area of the heat sink proper on which it is mounted,
2. If the said plate is at least 3 mm thick and the heat sink is a big, generous one and if one uses heat compound between the two VERY liberally, and
3. If the said L plate is made of copper, which has a considerably higher heat transfer rate than aluminium.
As you might guess, all of this hullaballoo costs a fair penny or two, and it is my belief that this is the key reason why so few people do it right. And it works up to a point, as in case of home use amps, which don't often need to dissipate serious power on a continuous basis.
I do it the other way, even if I have what are, to my knowledge, the easiest dynamic speakers to drive I have ever seen: effciency 93 dB/2.83V/1m, nominal impedance 8 Ohms, minimum impedance 6.5 Ohms, worst case phase shift -25 degrees. Quite simply, I realize it shoud be done so, and a job well done has its own rewards.
How many people even bother to look at the power derating curve? Even when they are made so easy to read by people like Motorola/ON Semi. Even fewer bother (or know how) to work out their heat sink requirements, and with anything added in between the transistors and the actual heat sink tends to escalate the required heat sink values, while people prefer to invest money in sexy wiring or some such.
It's like buying cream for a cake which is not there.
Quite to the contrary, I think, it shows the author took the time and trouble to research his own topic.
And that he's honest about it. Ridiculous as that may seem in this day and age.
One think i like with MOSFETs power transistors is there is an easy to find polarization region where transconductance is near flat with temperature changes. Ususally around 150mA. Reason why they often sound better in class AB than A when quiescent current is accurately tuned?
The level is enough for Class A operation in normal listening levels, and class B only during very short high transients.
The other thing, with laterals, is they don't need temperature compensation, with all the stability/ distortion problems induced by temperature compensation networks.
Last edited:
Heat and de-rating . It's always 100 % worse than you think . I want to buy the cheapest car radiator and put the blighters in cooking oil and use a pump , one day I will . The angle bracket is the least bad thing I find ( that looks to be dreadful ) . The unexpected one is the insulator . If the heat sink is isolated it is possible to use high performance heat pads ( non isolating ) .
They can be very good at 50 mA , more like a bipolar and no real distortion that can be measured ( extra that is ) . 100 mA is said to be positive temp co tipping point . This man found the same as me .
http://sound.westhost.com/project101.htm
Not necessarily so, Nige, and you know it.
All it really takes is some reasonable care at the very beginning, meaning you need to plan for it at a proper time, no afterthoughs for heat sinks.
This is why I plan mine from the output stage and anticipated requirements from it backwards to the input stage; naybody can do a decent differential stage.
In my experience, a decently designed bipolar amp will ideally be biased between 100 and 130 mA per output device.
Above 130 mA, in my experience, the sound tends to be too rounded off, unnaturally soft, losing bite when and where it shouldn't, even possibly starts to sound a little unfocused. I believe this is due to overbiasing, as the emitter resistors can no longer spread the bias properly.
Also above 130 mA, there is nothing further to be gained, leaving full pure class A operation as the only next possible step.
Do not try 100 mA with emitter resistors of higher value than 0.27 Ohms or less. For emitter esistors of 0.33 Ohms, even 100 mA is overbiased, but isoften still tolerable. 130 mA only with 0.22 Ohms or less.
Again, I underline these is only my own experience, no absolute truth carved in stone.
I disagree. I find that D. Self is scrupulous in giving reference to other work. The only possible exception was his section on the fact that the current mirror in the input stage of the Thompson topology minimises even-order distortion. Here he should have given reference to an article by someone else who wrote about it in electronics world; I cannot remember the chaps name at the moment.
Absolutely correct. In fact, beyond giving credit where credit is due, there are three other reasons that come to mind:
First, references to earlier work afford the reader the ability to reseach the topic more deeply and learn more or make up their own minds.
Second, references are valuable in providing support for what the author is saying.
Third, references can often help the reader to see the other side of the coin from what the author is asserting. This is especially useful in giving the reader a fuller picture on subjects that may be a bit controversial.
In fairness to all authors, while there are many instances in which a citation is obviously called for, there are others where it is a judgment call. We are also not always consistent. For example, I don't reference Widlar when referring to the common current mirror, but I do refer to the Wilson current mirror as such - but do not cite a particular reference where it was first published. Some stuff is just well-known in the art and where it came from is not that important.
Cheers,
Bob
I agree completely. MOSFETs also have a softer crossover region. Moreover, MOSFETs do not get into gm doubling if they are biased at a higher current. They like high bias - for the most part, it increases their transconductance and reduces transconductance droop. MOSFETs are FAR more tolerant of changes in the bias level than are BJTs. A bit of under-bias under dynamic conditions can be death to the sound of a BJT. With bias levels generally higher than the optimum amount for BJTs, the class a region is naturally larger.
Cheers,
Bob
You should set a reminder about for your book's next edition, and expand the references at least to the number and quality included in DS's book.
I disagree. I find that D. Self is scrupulous in giving reference to other work. The only possible exception was his section on the fact that the current mirror in the input stage of the Thompson topology minimises even-order distortion. Here he should have given reference to an article by someone else who wrote about it in electronics world; I cannot remember the chaps name at the moment.
Maybe the chaps name figures in the references of Wilson's "Currents Mirrors, Amplifiers and Dumpers" Wireless World December 1981, pp 47-50.
By the way, I found in a repeated manner that, as load of a differential input stage, the Wilson's 4 transistors current mirror gives a bit more distorsion than the 3 transistors version (which is much better than the 2 transistors version). I recently checked that with the unity gain buffer described in Figure 3.17d of DS's "Small Signal Audio Design".
While I agree with all the reasons you quoted Bob, I still think it's really primarily about how good is the author in what he's writing about. If he's really lord and master of the subject, he will not be bothered by quoting.
But if he's not too sure of himself, he will quote only assertive work, and even that frugally.
In the end, it's never the topic, it's always the man. In just about everything.
But if he's not too sure of himself, he will quote only assertive work, and even that frugally.
In the end, it's never the topic, it's always the man. In just about everything.
While I agree that BJTs have an optimum bias point, "death to the sound" is a little of a stretch. It would be interesting to find out if say 10% under-bias is audible in a rigurous listening test. I certainly doubt it.
If your answer was about my words (not very agreeable for D.S) i suppose you can understand my reaction to a man who publish a thread with "Douglas Self wants your opinions" as a title and answer NO in a very disagreeable and arrogant manner when you suggest to him to consider a chapter about current feedback amplifiers topology.
http://www.diyaudio.com/forums/soli...-self-wants-your-opinions-34.html#post3446004
This means for me the book will not be complete and the author objective enough to be a reference.
As i'm not enough interested and have no need to read a book about some Mister Self's preferences and opinions, i will not buy it.
Specially in the view of the fact that he provided wrong arguments in LTP analyses, confusing symmetrical signals distortion cancellation where it is true with feedback operation where it is not. " This is easily demonstrated mathematically, by simulation, and measurement."
http://www.diyaudio.com/forums/soli...-self-wants-your-opinions-34.html#post3445992
My mistake was to put in concurrence Bob, who share for free a lot of his knowledge, is so modest, nice and technically accurate, and have all my sympathy.
Last edited:
Esperado,
Glad to see you writing in this section, I haven't seen you for awhile. I understand your trepidation about buying a book by someone who obviously leaves out your favorite topology about the current feedback amplifiers, but I will say that Doug does have much information on other subjects that is worth reading. Yes he is opinionated and shows it but that is besides the point as long as he gives his reasons I think anyway. As to the writing style and teaching style I much prefer Bob's style of writing, I know who I would want to have in a classroom! I think this has much to do with the difference between us Yanks and the Brits and that stiff upper lip type of diction that Doug uses, kind of my way or the highway style of writing! But I would gladly have both the authors books on my bookshelf for good information though there will always be disagreements between designers and engineers on specific points of view and different outcomes. I am still waiting to see a complete pcb design by Bob that goes along with his book so we can build along and learn all the things that you learned long ago in your years of building actual products for real world usage. I only wish it wasn't a year away for Bob's next book!
Glad to see you writing in this section, I haven't seen you for awhile. I understand your trepidation about buying a book by someone who obviously leaves out your favorite topology about the current feedback amplifiers, but I will say that Doug does have much information on other subjects that is worth reading. Yes he is opinionated and shows it but that is besides the point as long as he gives his reasons I think anyway. As to the writing style and teaching style I much prefer Bob's style of writing, I know who I would want to have in a classroom! I think this has much to do with the difference between us Yanks and the Brits and that stiff upper lip type of diction that Doug uses, kind of my way or the highway style of writing! But I would gladly have both the authors books on my bookshelf for good information though there will always be disagreements between designers and engineers on specific points of view and different outcomes. I am still waiting to see a complete pcb design by Bob that goes along with his book so we can build along and learn all the things that you learned long ago in your years of building actual products for real world usage. I only wish it wasn't a year away for Bob's next book!
Although I do not agree with everything Doug is claiming, could it be that he simply does not have time for discussion, therefore his answers are more laconic than polite. It is usually so if one is too focused on some work. I have impression that he is writing more books at the same time!
Last edited:
Ivan,
I'm not sure by how he replies in his forum answers? He just seems like one of those professors who won't take a question in class until he is done lecturing, like don't disturb my train of thought with a damned question until I say you may, rather than a back and forth discussion where you try and understand as you go if you know what I am saying. Not to say you wouldn't learn from that type of teacher, I have had many, but it is not as satisfying or as easy to follow along and understand as you go. It requires much more independent researching to understand all that is said.
I'm not sure by how he replies in his forum answers? He just seems like one of those professors who won't take a question in class until he is done lecturing, like don't disturb my train of thought with a damned question until I say you may, rather than a back and forth discussion where you try and understand as you go if you know what I am saying. Not to say you wouldn't learn from that type of teacher, I have had many, but it is not as satisfying or as easy to follow along and understand as you go. It requires much more independent researching to understand all that is said.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.