some are better than others, but they rule. always have.
15 responses Add your response
There is no perfect device. Mosfets output stages are:
- more robust than transistors
- have higher bandwidth
- are more tolerant of difficult loads
- less efficient since have lower voltage swing from the same supply voltage
- are more nonlinear than transistors around cutoff region
- require more gain in preceding stages to eliminate crossover distortions
- limited ability to apply local feedback (low gain)
My experience with different types of amps is very limited but judging by reviews and opinions there are good and bad amps in each category.
The only area where Mosfets win hands down are class D amps but this is different type of Mosfets (vertical vs lateral) specifically designed for switching.
To continue and correcting Kijanki, especially nowdays, bipolar transistors are substantially higher quality than a couple of decades ago. Hence most of advantages/disadvantages of bipolar transistors vs. MOSFETs disappear (especially for audio applications) such as
- higher bandwidth(no need to go onto MHz ranges I believe)
- more tolerant of difficult loads (never had been more tolerant than bipolar transistors that have substantially lower output impedance with common emitter connection)
Therefore the preference to MOSFET based power amplifier could only be for older vintage models.
They are tools, nothing more or less, some of my favorite amps have them, some do not. The GamuT amps , which I have been considering, have only one pair of transistors a channel, I don't know of any bipolars that can do this for 200 watts a channel. So circuit simplification remains an advantage for them.
"no need to go onto MHz ranges I believe"
In general, to keep things stable output stage should be as fast as possible (little phase shift)and input stage as slow as possible (large phase margin). Power output transistors of modern amplifiers have hundreds of MHz bandwidth (even though slower devices are available). Also smalls signal bandwidth is different (much higher) from large signal bandwidth. We need to keep total amp's bandwidth as high as possible to avoid phase shift in the pass-band. Jeff Rowland uses 180MHz opamps (OPA1632) in Capri preamp. I have impression, so far, that he knows what he is doing.
Kijanki, some of the material you have written doesn't make sense to me & also appears to be contradictory: in the advantages you wrote that MOSFET have higher bandwidth than BJTs. In the disadv you wrote that they have lower voltage swing from the same supply voltage. This is contradictory to me.
Lower voltage swing from the same voltage supply compared to the BJT implies lower gain. Bandwidth & gain (Gm) are directly related. Lower gain (for the same amount of bias current) compared to the BJT implies lower bandwidth. This seems to be what I observe too.
I also don't know if there is any substantial evidence of BJTs being less robust than MOSFETs. Ditto re. the ability of MOSFETs to handle difficult loads any better than BJTs.
MOSFETs are a square law device hence the distortions are more along the tube-like behaviour. That's why one often reads of MOSFETs being tubey sounding. In comparison the BJT is an exponential device & the distortions are odd-order harmonic based. If the implementation is not correct/good enough this can lead to very discordant sound very quickly.
I agree w/ most of the other members that MOSFETs should sound pretty good if implemented "correctly" (another audio fuzzy term!). I believe that even better than the MOSFET is its cousin the JFET in the output stage - MOSFET electrical characteristics but BJT-like sound. Preferred by the discriminating amp builders. The sad part is that no one is making large quantities of power JFETs like they are power MOSFETs & BJTs. Thus amps seldom use JFETs.
Now, First Watt *seems* to be using JFETs since they output only a few watts.
"Lower voltage swing from the same voltage supply compared to the BJT implies lower gain" - No it has nothing to do with gain but with characteristic.
"Lower gain (for the same amount of bias current) compared to the BJT implies lower bandwidth". - No it doesn't. Even between bipolar transistors current gain has nothing to do with max bandwidth. It is just matter of design. Mosfets are generally much faster.
"I also don't know if there is any substantial evidence of BJTs being less robust than MOSFETs" - sure they are more robust since Mosfets don't suffer secondary breakdown (hot spots of current on the wafer with higher voltages resulting in thermal avalanche - very nasty complication) and are thermally stable (BJT's Vbe drops with temperature increasing collector current hence increasing temperature - thermal runaway). Mosfet drain-source resistance has a positive temperature coefficient, so they tend to be self protective.
As for tubey sound - Mosfets have softer clipping than BJts but are more nonlinear and require much higher gain before feedback to correct it - in result behave pretty much the same as BJts. Higher gain results in design more prone to TIM and therefore higher order odd harmonics.
Just think about it - if one type of devices would have clear advantage nobody would use other devices. Some manufacturers (Krell, Levinson etc) use bipolar designs.
MOSFETs ARE transistors. Field Effect, but still a transistor.
I have been in-fab for 30 years, the last making 'HexFets', a propriatary technology of my company...Bob Carver thought highly enough of them to use our devices some (all?) of his Sunfire amps.
This is not 'secret' info, but our 'discreet' (non IC devices) the entire bottom of the die is the drain. Since there is a conductive layer there(can't elaborate here) the RDON (resistance of the device ON) is lowered and when used as output devices results in a higher damping factor.....
Another device, not yet mentioned is the IGBT, Isolated Gate Bipolar Transistor. I have NO idea if there are audio applications here.
My company also has a line of class 'd' amps, one of which is out on my bench awaiting a power supply. One of these days!
Another device, not yet mentioned is the IGBT, Isolated Gate Bipolar Transistor. I have NO idea if there are audio applications here.Just FYI, the older Thresholds (T200, T100, T400, etc from that era) & several of the down-market Forte amps also built at the same time used IGBTs in their output stages. The sound of these class-A amps & the Forte class-A, class-AB amps is quite good. I find it a bit confined compared to my class-A BJT amp but this is just 1 amp compared against my own.
I am NOT an engineer. I MAKE the darn things in a wafer fab. The HexFet is a pretty neat design, and i guess you'd say it IS vertical. The gate and source regions are on the top while the drain is the bottom. Yes, one of the selling points is apparently RDON, which is Resistance of the Device ON. I suspect this plays directly into damping factor and maybe why Carver used our stuff.
I, too, am a fan of 'd' amps, having a PSAudio GCC series integrated, using the B&O ASP modules. I have an International Rectifier stereo amp out in the garage waiting for me to figure out how to build a power supply. I could take the chickens way out and buy Gel Cells for a brute force DC/Battery supply.
Again, I am not an engineer but I suspect anything that switches quickly enough could be used as an amp's output stage.....RE:: IGBTs. These devices, at least as I see them are usually higher voltages, maybe 600v.