Hi John or Bob,
sorry for revisiting an old question regarding V(Re). In the early 80's Dan Agostino launched his KRELL KSA50 which at the time was a very impressive piece of sound equipment, even by today's standards.
He seemed to break all rules regarding the optimum setting of 15 - 20 mV across RE, in fact he was running each at 620 mV acros 0.68 Ohm.
How come that amp had very high aclaim and became very popular for its incredibly revealing sound quality and solid bass. and still remains one of the refereces by todays audiophiles.
Furthermore he used the MJ15003/4 combination that you seem to dicard as rubbish. Motorola on the other hand was under the impression that they created a good complementary pair for high quality audio.
He used those transistors way into the 90's. The KRELL is capable of driving incredibly akward loads loud while remaining in full control of it. What did Dan do wrong to achieve this a successful piece of equipment.
Regards Nico
sorry for revisiting an old question regarding V(Re). In the early 80's Dan Agostino launched his KRELL KSA50 which at the time was a very impressive piece of sound equipment, even by today's standards.
He seemed to break all rules regarding the optimum setting of 15 - 20 mV across RE, in fact he was running each at 620 mV acros 0.68 Ohm.
How come that amp had very high aclaim and became very popular for its incredibly revealing sound quality and solid bass. and still remains one of the refereces by todays audiophiles.
Furthermore he used the MJ15003/4 combination that you seem to dicard as rubbish. Motorola on the other hand was under the impression that they created a good complementary pair for high quality audio.
He used those transistors way into the 90's. The KRELL is capable of driving incredibly akward loads loud while remaining in full control of it. What did Dan do wrong to achieve this a successful piece of equipment.
Regards Nico
Nico Ras said:
Nico,
I think the factors for success of the KSA are pretty complex. It was physically and (partly) conceptually a new kid on the block.
I never owned one, so have no 1st hand experience, but what I hear among my audio friends is that you either love them or hate. I have friends that don't want to touch a Krell with a pole, and some others can't buy enough Krells.
Jan Didden
Nico Ras said:
I'd just like to point something out here. We often hear lots of talk about linearity, and see a lot of detailed analysis by Self for example, but there is much more to bias point than static linearity, which by the way is easily improved with feedback. ft, for example, is highly dependent on Ic, one reason being that we need to have available current to charge junction capacitances. hfe is also dependent on Ic and junction temperature, run an output stage hot and hfe goes up. There are many factors to consider. There is a reason to use high bias in the VAS and even the diff amp which is to have enough class A idle current so that the stage can easily drive the non-linear junction capacitances.
Edit: should have mentioned that Krell is a Class A or near Class A design, the others are speaking of optimal Class AB biasing.
Pete B.
Nico, I am not going to argue with you about it. I know from experience, and the specifications of the devices.
It is just like when I tried to hop up a 54 Chevy six with 2 carbs and a dual exhaust. It was a good attempt, but I would have been much better off to start with a 55 Chevy V8, by a long shot. However, the first Corvettes used a Chevy six engine, so I guess it could have been possible to be successful with it. Of course, the professional mechanics shook their heads at me at the time, about 45 years ago.
It is just like when I tried to hop up a 54 Chevy six with 2 carbs and a dual exhaust. It was a good attempt, but I would have been much better off to start with a 55 Chevy V8, by a long shot. However, the first Corvettes used a Chevy six engine, so I guess it could have been possible to be successful with it. Of course, the professional mechanics shook their heads at me at the time, about 45 years ago.
Nico Ras said:
I think the KSA50 was a 50 watt per channel Class A amplifier. Being Class A, the question of voltage dropped across RE is not relevant in the context of optimal Class-AB biasing. I forget how many pairs of devices he ran in parallel.
Note also that the use of Class A mitigates somewhat the need for speed in the output devices, since there is no abrupt crossover. Therefore, in this application, the speed of the MJ15003/4 may have been OK. Also, if a good T-like driver circuit was used, and lots of pairs were used, beta and beta droop issues would have also been mitigated.
Bob
Hi Bob
John Curl mentioned them already!
Static induction transistors were first made something like 20 years ago maybe longer. At the time they were introduced they were much like JFETS but had diffused channels between elements of a buried grid.
The grid spacing was chosen so that they fully deplete at zero bias, the idea being that they didn't need to be turned off at turn-on, like ordinary JFETS.
On the other hand they were poor in terms of output characteristics - low Early voltage, more like triodes. And, presumably, it would have been possible to force them into bipolar mode if the grid/gate bias were too large.
Thought they might be the best solid state equivalent to triodes, and require less overall feedback in a push pull amp.
They're still being researched - latest devices in SiC operate at 600V or more. 10 kW amp anyone?
But I've never seen a production version from anyone yet. I guess someone might make them somewhere.
Cheers
John
John Curl mentioned them already!
Static induction transistors were first made something like 20 years ago maybe longer. At the time they were introduced they were much like JFETS but had diffused channels between elements of a buried grid.
The grid spacing was chosen so that they fully deplete at zero bias, the idea being that they didn't need to be turned off at turn-on, like ordinary JFETS.
On the other hand they were poor in terms of output characteristics - low Early voltage, more like triodes. And, presumably, it would have been possible to force them into bipolar mode if the grid/gate bias were too large.
Thought they might be the best solid state equivalent to triodes, and require less overall feedback in a push pull amp.
They're still being researched - latest devices in SiC operate at 600V or more. 10 kW amp anyone?
But I've never seen a production version from anyone yet. I guess someone might make them somewhere.
Cheers
John
KSA50:
Used 2 pairs of 15003/4, with (IIRC) ~1.9A total bias.
The amp was only biased for 50w classA into 8ohms, so if the slowness of the outputs was an issue it would/should manifest itself while the amp was driving 2ohm reactive loads, for which it was considered one of the best amps around...not that I've looked at the spectrum of the output while driving 2ohms, so I have no clue what the real change from classA to classB does to the distortion.
Are crossover non-linearities mitigated in their effect or character by their occurence at higher output levels in a heavily biased amp?
Stuart
Used 2 pairs of 15003/4, with (IIRC) ~1.9A total bias.
The amp was only biased for 50w classA into 8ohms, so if the slowness of the outputs was an issue it would/should manifest itself while the amp was driving 2ohm reactive loads, for which it was considered one of the best amps around...not that I've looked at the spectrum of the output while driving 2ohms, so I have no clue what the real change from classA to classB does to the distortion.
Are crossover non-linearities mitigated in their effect or character by their occurence at higher output levels in a heavily biased amp?
Stuart
AndrewT said:
This is hard to answer in the general case, but do keep in mind that the percentage of time that the amplifier is spending at higher power levels is quite small. This is because of the high crest factor of music. So if that is the only time it distorts significantly, it might be perceived as not so bad. Taken to an extreme, realize that it is likely that tube amplifiers, with their limited power, are probably clipping a lot more than solid state amplifiers (but of course they often are clipping in a softer way, and that compressive type of distortion is different than crossover distortion). I apologize for having to be so speculative here; there may be others who have a better answer to this question.
If I had to err on one side or the other of bias, I would always err on the side of an amplifier that was over-biased than under-biased, as long as no thermal stability or reliability problems arose.
Bob
Bob Cordell said:
Never seen that JAES paper, but it probably pretty much the same thing. You are corrrect about the opamp rails, although I have now changed my mind and will derive the opamp rails from the input signal. That will avoid the ripple on the ground driven power supplies.
No, this isn't a cheap design. It's an esoteric amplifier for my loungeroom. Rated at 512W rms per channel into 4ohms in class A. 350W total idle dissipation in each channel, 8A bias for each of the four class A stages.
Each channel heatsink measures 1500X240X90mm. The final unit will also have built in oscilloscopes as signal monitors to give it that 50's RF transmitter look
G.Kleinschmidt said:
Never seen that JAES paper, but it probably pretty much the same thing. You are corrrect about the opamp rails, although I have now changed my mind and will derive the opamp rails from the input signal. That will avoid the ripple on the ground driven power supplies.
No, this isn't a cheap design. It's an esoteric amplifier for my loungeroom. Rated at 512W rms per channel into 4ohms in class A. 350W total idle dissipation in each channel, 8A bias for each of the four class A stages.
Each channel heatsink measures 1500X240X90mm. The final unit will also have built in oscilloscopes as signal monitors to give it that 50's RF transmitter look
Wow!! That should keep you warm on those cold winter nights! Keep us posted on your progress.
Cheers,
Bob
AndrewT said:
The Class A stages are low voltage, each with a dedicated +/-10V power supply which is ground driven by a Class AB stage to follow the output signal.
+/-10V @ 8A = 160W. The amplifier is a bridged design so there are two class A and two class AB stages in each channel. There is a total of 320W idle dissipation in the class A stages of one channel plus another 30W idle dissipation in the class AB stages.