Rodent said:
I disagree. Designers develop a preference for a particular device for a PARTICULAR reason, be it a good valid judgment call or not. That reason may be as simple as a bad experience with a device type in the past. Or, more likely, it is based on their perceived balance of the goods and evils of a particular device. Picking your poison, as it were. All devices have non-idealities that must be overcome. Some designers are more comfortable and confident dealing with one set of idealities than another. This doesn't mean that one designer's choice is necessarily better than the other designer's choice.
I do agree, however, that there is a lot of human nature (and subjective judgment) at work here.
Bob
Rodent said:
I don't think that it is always true.
On DHE seminar with Richard Bandler and John LaValle we among other things studied calibration of perceptions. One exercise was to discover prices of coins by sound. Another excercise was to discover prices of paper money by touch. When students tried reasoning they were loosing, when reasoning was switched off intuition helped a lot. The core of exercises was, stop reasoning and calibrate own perception, and you will be right when well trained, no matter how it works in reality.
The same may happen with MOSFET amps, when the designer hears well differences, but can't name them, and tries to find some way to measure differences he hears. It reminds me measurements of how well are grand piano strings are aligned horizontally, in order to discover why one grand piano sounds more in tune. Imagine, people did not discover yet frequency measurements, but know well how to measure horizontal alignment of strings... Or they don't believe that frequencies may be measured... Like don't believe that tiny amount of crossover distortions means much more than huge amount of low order distortions...
Absolutely, it use human nature to make a choice based on your perceiption. You probably find that you can design as good an amplifier with either component.
I recall an earlier comment that both sounded good, but not the same. I don't think many of us had the opportunity to be able to switch between identical topologies using the two different devices and yet he was unable state irrevocably which is better.
That brings me to the point that either may be successfully deployed as an audio amplifier and if the user does not know what was inside the box he would happily agree that what he was listening to sounds the best.
I recall an earlier comment that both sounded good, but not the same. I don't think many of us had the opportunity to be able to switch between identical topologies using the two different devices and yet he was unable state irrevocably which is better.
That brings me to the point that either may be successfully deployed as an audio amplifier and if the user does not know what was inside the box he would happily agree that what he was listening to sounds the best.
Rodent said:
I would say it depends on a musical matherial. If he/she listens to something modern, highly compressed, AB class amp of traditional opamp type may be the best while single ended class A may sound bad because of more possible intermodulations, but for live sounds such as vocal, acoustic guitar, symphony orchestra, single ended class A amp most probably will be shosen. Again, if that live sound was made using compressors and noise gates, AB opamp may be preferred.
Waveborn,
I would absolutely agree with you. Different topologies will sound nicer with different material as Nico pointed out. Percussion sounded more natural with mosfets than BJT but BJT sounded more natural on voice and string. I could probably agree with this, but if I owned either on of the amps, I would probably never know the difference.
We are the chosen few with the ability to tweak an amp until we are satisfied with the results no matter what devices we use.
People who purchase their amps can hardly make a decent judgement when listening to the salesman, strange speakers playing unknown material that was selected to make the system sound impressive. The listening rooms are usually so dampened that there will be no boom from standing waves.
When the poor guy takes his newly purchased amp home and playing is material on his speakers in a room with many standing waves, the experience may not be that satisfying, regardless if it was MOSFET or BJT.
The average man's listening room is usually integrated with his living room and his wife will indicate where certain equipment my be place. It is always not optimal, because optimal would look rediculous.
I would absolutely agree with you. Different topologies will sound nicer with different material as Nico pointed out. Percussion sounded more natural with mosfets than BJT but BJT sounded more natural on voice and string. I could probably agree with this, but if I owned either on of the amps, I would probably never know the difference.
We are the chosen few with the ability to tweak an amp until we are satisfied with the results no matter what devices we use.
People who purchase their amps can hardly make a decent judgement when listening to the salesman, strange speakers playing unknown material that was selected to make the system sound impressive. The listening rooms are usually so dampened that there will be no boom from standing waves.
When the poor guy takes his newly purchased amp home and playing is material on his speakers in a room with many standing waves, the experience may not be that satisfying, regardless if it was MOSFET or BJT.
The average man's listening room is usually integrated with his living room and his wife will indicate where certain equipment my be place. It is always not optimal, because optimal would look rediculous.
Wavebourn said:
I think that as long as an amplifier has low THD, is reasonably fast so it has low intermods, is quiet, has a nice low output impedance, and is able to drive a decent amount of power into a speaker, then it matters not a jot what devices it uses.
Indeed, I very much doubt whether anyone here could tell the difference between any two amplifiers with reasonable paper specs, as long as the testing was done using proper double blind methodology.
It's other things like cost or thermal runaway that dictate the choice of one device over another.
Regards,
Suzy
AndrewT said:
Secondary breakdown in BJTs can have an effect on sound quality if the amplifier design employs intrusive protection circuits to protect the output devices from destruction as a result of secondary breakdown. As John Curl has pointed out, this is no longer a problem with conservatively and well-designed amplifiers employing quality modern output BJT's. It was a problem in the 70's and 80's, and could still be a problem with amplifiers designed by corner-cutting manufacturers.
Bob
I agree with Bob Cordell in general. We can make 'good' amps with with either bipolar or FET output devices. If I can, I prefer fets, but I personally found in the past that I could not make high powered vertical FET amps, without additional active protection. Usually, for the same output power, I used balanced bridged outputs (250W Gale 1976, 100W Symmetry 1978, 250W VMPS 1981) in order to get around the second breakdown problem with bipolar outputs.
When I tried to make a single sided output 100W all FET power amp (Lineage 1986), I ran into a short circuit reliability problem that forced the inclusion some sort of active protection, where only passive protection was necessary in previous designs. Had I stuck to a fully balanced bridge design, I would not have had any serious problems, just like Charles Hansen found to be effective, independently.
Bob's initial design from 1984 is a relatively low power design. In fact, I have not made a commercial power amp since the Levinson JC-3 (ML-1 is almost the same) with that low of power. Vertical fets have no problem at +/- 35V, but if you increase the voltage to +/- 50V, you better have some active output protection.
When I started to make power amps with Parasound in 1989, they had already developed a relatively unobtrusive and effective protection circuit to use with bipolar outputs, so long as we used a significant number of output devices in parallel. They pioneered (for me at least) the implementation of a complementary power fet driver stage in order to VOLTAGE DRIVE the output devices. This linearizes the output devices significantly, because the BETA curve is not followed, and the output itself is very low impedance (even open loop) because of this.
Amps made this way can be very fast and relatively linear, without extra internal feedback loops.
It would be conceptually possible to make an all vertical power fet output stage, and even eliminate the fet driver stage. Whether this approach would be as reliable without active protection is a serious question.
When I tried to make a single sided output 100W all FET power amp (Lineage 1986), I ran into a short circuit reliability problem that forced the inclusion some sort of active protection, where only passive protection was necessary in previous designs. Had I stuck to a fully balanced bridge design, I would not have had any serious problems, just like Charles Hansen found to be effective, independently.
Bob's initial design from 1984 is a relatively low power design. In fact, I have not made a commercial power amp since the Levinson JC-3 (ML-1 is almost the same) with that low of power. Vertical fets have no problem at +/- 35V, but if you increase the voltage to +/- 50V, you better have some active output protection.
When I started to make power amps with Parasound in 1989, they had already developed a relatively unobtrusive and effective protection circuit to use with bipolar outputs, so long as we used a significant number of output devices in parallel. They pioneered (for me at least) the implementation of a complementary power fet driver stage in order to VOLTAGE DRIVE the output devices. This linearizes the output devices significantly, because the BETA curve is not followed, and the output itself is very low impedance (even open loop) because of this.
Amps made this way can be very fast and relatively linear, without extra internal feedback loops.
It would be conceptually possible to make an all vertical power fet output stage, and even eliminate the fet driver stage. Whether this approach would be as reliable without active protection is a serious question.
john curl said:
Thanks for the summary, John. I agree. In general, MOSFET outputs should have some form of simple active protection, although I suppose if you use enough of them you can get away with just a rail fuse. The protection needed, though, can be very unobtrusive protection that just comes into play in the event of a short circuit. My preference is for a type of crowbar circuit that is triggered only under those conditions, and which, when triggered, removes gate drive. It is completely out of the circuit in normal operation.
I'd love to know generally what kind of protection is used in your Parasound designs, but understand that you might not want to discuss it, even in general terms. I'd be tempted to guess that if it is a relatively unobtrusive form of active protection, that it would probably also adequately protect MOSFETs.
Cheers,
Bob
mikeks said:
Rail fuses are always a good idea – in my mind, mandatory. They may not fully protect the output devices from failure, but that shouldn’t be their intended purpose. They are there to prevent the output devices and associated components (not to mention your speakers) from burning up if there ever is a catastrophic output stage failure.
G.Kleinschmidt said:
I thought the idea of rail fuses was to put lots of DC on your speaker when one side blows and the other doesn't, so that as well as replacing output transistors, you have to replace your woofer, too.
I don't use them. My only fuse is on the transformer primary. I figure that if an output transistor goes, it's going to be a mess regardless of what protection you have.
Regards,
Suzy