I asked, even begged, not to be asked to answer simple, easy to find the answer elsewhere, questions, as I am disadvantaged medically, at this time.
In truth, the 1166 is an interesting chip, with many problems and compromises. I am not afraid of thermal runaway with my conventional circuits, so why change them for a chip that typically adds more high frequency distortion to the mix?
Paralleling multiple pairs of transistors is another engineering problem entirely, and can be fixed with thermal and beta matching, and experience in placing the temp sensors. As I said, any experienced engineer can do this. I don't have the strength to do so, to answer elementary questions, without doing potential harm to myself. Please keep that in mind, fellow engineers.
In truth, the 1166 is an interesting chip, with many problems and compromises. I am not afraid of thermal runaway with my conventional circuits, so why change them for a chip that typically adds more high frequency distortion to the mix?
Paralleling multiple pairs of transistors is another engineering problem entirely, and can be fixed with thermal and beta matching, and experience in placing the temp sensors. As I said, any experienced engineer can do this. I don't have the strength to do so, to answer elementary questions, without doing potential harm to myself. Please keep that in mind, fellow engineers.
.
Hi John,
You'll get a lot more sympathy if you refrain from comments and innuendo that insults others. Even here, you are demeaning the questions that I and others have asked.
I am sincerely sorry about your medical condition, but it does not give you license to be rude to others without a response.
Most experienced engineers are taught to design circuits that are tolerant of beta and beta mismatch. It is certainly a good thing to match betas in an output stage to achieve lower distortion. But having to match betas to keep your amplifier from going into thermal runaway is not a good thing.
Our philosophies are at times different. If I was designing a big amplifier with 90V rails and lots of output transistors in parallel, with very small RE and running fairly hot, I would be extremely reluctant to put 10-ohm base-stopper resistors in the circuit. That just invites beta to the thermal party.
Bob
john curl said:
Hi John,
You'll get a lot more sympathy if you refrain from comments and innuendo that insults others. Even here, you are demeaning the questions that I and others have asked.
I am sincerely sorry about your medical condition, but it does not give you license to be rude to others without a response.
Most experienced engineers are taught to design circuits that are tolerant of beta and beta mismatch. It is certainly a good thing to match betas in an output stage to achieve lower distortion. But having to match betas to keep your amplifier from going into thermal runaway is not a good thing.
Our philosophies are at times different. If I was designing a big amplifier with 90V rails and lots of output transistors in parallel, with very small RE and running fairly hot, I would be extremely reluctant to put 10-ohm base-stopper resistors in the circuit. That just invites beta to the thermal party.
Bob
Mr.Curl,
To be fair to Bob Cordell. he has asked a fair question, and I for one would like to know the answer. There can be a something useful to be learned here, so if his comments are BS, we would like to know why?
I suspect the the best answer could lie somewhere in the middle, but if we don't know your reasons, there is nothing to be learned.
Which brings up the question...........what is the point of this discussion?
Your humble servant.
Jam
To be fair to Bob Cordell. he has asked a fair question, and I for one would like to know the answer. There can be a something useful to be learned here, so if his comments are BS, we would like to know why?
I suspect the the best answer could lie somewhere in the middle, but if we don't know your reasons, there is nothing to be learned.
Which brings up the question...........what is the point of this discussion?
Your humble servant.
Jam
Stereophile cover
On another note John’s JC-1 amp is listed on the cover of the latest Stereophile. There is text about the amp in the equipment review section of this magazine. Congratulations John for once again having one of your designs make the magazine. Not many other designers on this site can claim multiple front page covers of Stereophile.
On another note John’s JC-1 amp is listed on the cover of the latest Stereophile. There is text about the amp in the equipment review section of this magazine. Congratulations John for once again having one of your designs make the magazine. Not many other designers on this site can claim multiple front page covers of Stereophile.
Thanks Jam, for stirring the pot. Just wait till you need a question answered. 
Now, for what I was researching as Jam made his challenge:
About 1980, separated from Otala and still competitive in my own way, I was assigned a project to develop an all out power amp using the new 'ring emitter' power transistors, just released.
I used a proven complementary differential fet balanced bridge topology, that I had originally used with the GALE power amp, except with much faster driver and output transistors.
In final testing, I ran into problems. Nothing that I did in my lab made any difference, but the customer kept blowing it up. I asked the advice of John Iverson, a colleague, and excellent engineer, and he recommended 10 ohm resistors in each base to negate any -R that may appear at the base of the output transistor. I took his advice and the problem was fixed.
After that, all of the power amps that I developed with bipolar power transistors at the output used a 10 ohm resistor in each base lead. 10 ohms worked, it was a convenient value, and we had no obvious problem with it for at least 8 different amplifier configurations from 100W to 350@/ channel for about 10 years.
Finally, with the JC-1, we found a problem with some units.
We found that IF the betas of the output devices were not reasonably well matched, one pair of devices could go into thermal runaway. I fixed one of the JC-1's myself by just best matching the output devices from my limited personal stock. It is here, within arms reach, and it works OK.
However, we found the most SERIOUS problem to be a defect in the heatsink assembly which is not as 'seamless' as it first appears. We fixed that by redesigning parts of the heatsink.
Now, what about the 10 ohm resistor and what does it do wrong? It isolates each stage from the others at DC, and IF the betas are significantly different, and/or the individual thermal resistance of each output device and its heatsink area are not identical, then current hogging, and even thermal runaway can occur.
What to do? Eliminating the R is dangerous, but maybe, just maybe, a lower R could work and make matching less critical. Well, Toshiba recommended 5.6 ohms. Next time I will try that.
Now are you happy, Jam?
Now, for what I was researching as Jam made his challenge:
About 1980, separated from Otala and still competitive in my own way, I was assigned a project to develop an all out power amp using the new 'ring emitter' power transistors, just released.
I used a proven complementary differential fet balanced bridge topology, that I had originally used with the GALE power amp, except with much faster driver and output transistors.
In final testing, I ran into problems. Nothing that I did in my lab made any difference, but the customer kept blowing it up. I asked the advice of John Iverson, a colleague, and excellent engineer, and he recommended 10 ohm resistors in each base to negate any -R that may appear at the base of the output transistor. I took his advice and the problem was fixed.
After that, all of the power amps that I developed with bipolar power transistors at the output used a 10 ohm resistor in each base lead. 10 ohms worked, it was a convenient value, and we had no obvious problem with it for at least 8 different amplifier configurations from 100W to 350@/ channel for about 10 years.
Finally, with the JC-1, we found a problem with some units.
We found that IF the betas of the output devices were not reasonably well matched, one pair of devices could go into thermal runaway. I fixed one of the JC-1's myself by just best matching the output devices from my limited personal stock. It is here, within arms reach, and it works OK.
However, we found the most SERIOUS problem to be a defect in the heatsink assembly which is not as 'seamless' as it first appears. We fixed that by redesigning parts of the heatsink.
Now, what about the 10 ohm resistor and what does it do wrong? It isolates each stage from the others at DC, and IF the betas are significantly different, and/or the individual thermal resistance of each output device and its heatsink area are not identical, then current hogging, and even thermal runaway can occur.
What to do? Eliminating the R is dangerous, but maybe, just maybe, a lower R could work and make matching less critical. Well, Toshiba recommended 5.6 ohms. Next time I will try that.
Now are you happy, Jam?
john curl said:Thanks Jam, for stirring the pot. Just wait till you need a question answered.
Now, for what I was researching as Jam made his challenge:
About 1980, separated from Otala and still competitive in my own way, I was assigned a project to develop an all out power amp using the new 'ring emitter' power transistors, just released.
I used a proven complementary differential fet balanced bridge topology, that I had originally used with the GALE power amp, except with much faster driver and output transistors.
In final testing, I ran into problems. Nothing that I did in my lab made any difference, but the customer kept blowing it up. I asked the advice of John Iverson, a colleague, and excellent engineer, and he recommended 10 ohm resistors in each base to negate any -R that may appear at the base of the output transistor. I took his advice and the problem was fixed.
After that, all of the power amps that I developed with bipolar power transistors at the output used a 10 ohm resistor in each base lead. 10 ohms worked, it was a convenient value, and we had no obvious problem with it for at least 8 different amplifier configurations from 100W to 350@/ channel for about 10 years.
Finally, with the JC-1, we found a problem with some units.
We found that IF the betas of the output devices were not reasonably well matched, one pair of devices could go into thermal runaway. I fixed one of the JC-1's myself by just best matching the output devices from my limited personal stock. It is here, within arms reach, and it works OK.
However, we found the most SERIOUS problem to be a defect in the heatsink assembly which is not as 'seamless' as it first appears. We fixed that by redesigning parts of the heatsink.
Now, what about the 10 ohm resistor and what does it do wrong? It isolates each stage from the others at DC, and IF the betas are significantly different, and/or the individual thermal resistance of each output device and its heatsink area are not identical, then current hogging, and even thermal runaway can occur.
What to do? Eliminating the R is dangerous, but maybe, just maybe, a lower R could work and make matching less critical. Well, Toshiba recommended 5.6 ohms. Next time I will try that.
Now are you happy, Jam?
Hi John,
What you described about the effect of base stopper resistors on tendency for current hogging is exactly what my point was. I guess you agree that what I said was not BS, at least in regard to my assertion about what base stopper resistors can do to thermal stability.
I still don't know your answer about my assertion about good engineering practice and the need to match beta to avoid thermal runaway, but I'll assume we will agree to disagree there.
Returning to the base stopper resistors, you are right: they are sometimes a necessary evil, and to do without them can be dangerous. We would like them to be as small as possible. This can be a delicate tradeoff. There may have been ways to stabilize the output stage at high frequencies with smaller base stopper resistors, but we may never know. Your choice to increase RE from 0.1 to 0.15 ohms was probably helpful.
At one point I had suspected that you went to the high value of base stopper resistor in order to be able to be stable without a coil and into a capacitive load.
At another point I wondered if the MOSFET source follower driving the capacitance of the large number of output transistors was the reason you needed the isolation of the base stoppers. As you know, MOSFETs like to oscillate, especially if they are capacitively loaded. Sometimes a small resistor in series and a shunting Zobel network in the gate of the MOSFET can help that.
Another approach that I favor is the use of distributed Zobel networks on the output. As you know, the output Zobel is there to provide a resistive local load to the emitter followers out to high frequencies. When you have a large number of output transistors spanning a significant distance, it may be hard for one Zobel to do a good job of providing a good high-frequency return for all of the output transistors (in light of wiring inductance, for example). This can be a bigger problem when using very fast output transistors. Four smaller Zobels might have helped.
Cheers,
Bob
Hi all,
this thread and and it's predecessor are great. Just what DIYaudio should be about.
Look at how many Members read the posts and how many contribute, even regularly, to gauge how popular this topic is.
Can I appeal to all that we try to bury our hatchets and follow the cause. Improve our knowledge (technically). Please.
this thread and and it's predecessor are great. Just what DIYaudio should be about.
Look at how many Members read the posts and how many contribute, even regularly, to gauge how popular this topic is.
Can I appeal to all that we try to bury our hatchets and follow the cause. Improve our knowledge (technically). Please.
AndrewT said:
Hi Andrew,
I agree that this is a very good thread and that we should try to move forward with less dischord. I appreciate the encouragement of you and others.
Sometimes a lively discussion on a controversial topic can get a bit rough and direct, but we should always try to keep it technical. As long as we can keep it from getting personal, sometimes those rough-and-tumble discussions turn over the most rocks and help all of us to learn. I've probably personally learned more from John and my disagreements with him than from those with whom I more often agree. He challenges me and makes me think.
Rather than one participant in a discussion or the other be a "winner", we should try to think in terms of the overall readership of the thread being the "winners".
Cheers,
Bob
Here I thought this was the hockey masked guy, but geezerjohn curl said:(snippage)
(snippage)
instead, oh well.
This is what I've found the best people to do. They try a bunch of
stuff and tweaks...think about it. Then, let is simmer for a while.
Finally, ask a trusted colleague or friend.
It is one of those process things.
and gets you out of the rut.
kinda what the board should
be about too. Helping everyone
learn and get better.
Thanks guys.
The insertion of a base resistor is well known in electronics engineering. It is the exact VALUE that is difficult to pin down. I learned about it from Dr. Don Pederson at UC, Berkeley in class back in 1971 or so. He is the one who explained the -R on the blackboard, from a transistor follower. You should do some research too, if you don't understand the concept.
john curl said:
John is right. I learned it from Jim Meindl at Stanford.
We always have to watch out for the law of unintended consequences. In this case, the base stopper in an output stage can lead to bias instability if its resistance is too high. That is one of the reasons we want to do as much as we can to minimize it.
The mechanism goes something like this: the rich get richer and the hot get hotter.
If you have a 10 ohm base stopper and are running the transistor at 125 mA, and beta is 50, you have 2.5 mA of base current, causing a 25 mV drop across the base stopper (and perhaps another 10-12 mV across the internal base spreading resistance in the transistor).
The transistor was originally biased with this drop taken into account. As the transistor gets hotter, Vbe goes down and beta goes up. The drop across the base stopper resistor is smaller if beta is higher. Transistor current goes up, transistor power dissipation goes up, the transistor gets hotter, and all of these effects then get reinforced. It is like a positive feedback effect, and if the phenomenon gets too strong, it can lead to thermal runaway.
Notice that a part of this process depends on the DC drop that was present across the base stopper resistor, here 25 mV. That drop will be smaller for a high-beta transistor, and will decrease some more as beta increases with temperature. This is a highly over-simplified explanation, but gives a sense of how this unintended consequence can happen, or be made worse by large-value base stopper resistors.
Cheers,
Bob
- Status
- Not open for further replies.