As title.
The ideal transistor for amplifying signal should be high input impedance and able to output enough current with least distortion to drive speaker. MOSFETs have high input impedance and should be capable to drive in high power. But why people keep saying BJTs are best for output transistor in class AB?
The ideal transistor for amplifying signal should be high input impedance and able to output enough current with least distortion to drive speaker. MOSFETs have high input impedance and should be capable to drive in high power. But why people keep saying BJTs are best for output transistor in class AB?
Input impedance can be designed-around.
BJTs have the highest transconductance of any practical device. The most gain. At loudspeaker current levels, that's perhaps more important than anything else.
BJTs have the highest transconductance of any practical device. The most gain. At loudspeaker current levels, that's perhaps more important than anything else.
There are plenty of large mosfets whose transconductance matches bipolars (up around 50S and higher), and do not have SOAR limitations. However, they have very high Ciss though used as source followers this is not too much of a problem. Bipolars require heavy base current which increases heavily at higher collector currents; mosfets have capacitive gates with no current requirement.
At the same current?
"No current" is a DC thing? The capacitances are substantial even in the audio band.
PRR,
At 5A, an IXFH94N30 mosfet has gfs of about 18S, but at less than one amp its down to about 4-5S. However, the effective gm of a source follower (the job it is used for) the gm is largely set by the source resistor and this situation applies to bipolars too. With a 0.22R emitter resistor, the gm would be little more than 1/0.22, say around 4S with a high gm device, so this is probably not the issue it might seem. Mosfets make excellent output stages.
Yes, driving the capacitive gate can be difficult, but the miller capacitor (Coss) is the big issue, and it's only only 965pF on the IXFH compared to the Ciss which is bootstrapped in a source follower and it is 5.5nF.
The large mosfets do not require multiple pairs, and they are easy to drive with bipolar drivers. Morever, their SOAR and thermal robustness makes them good choices.
Different folks, different strokes......
At 5A, an IXFH94N30 mosfet has gfs of about 18S, but at less than one amp its down to about 4-5S. However, the effective gm of a source follower (the job it is used for) the gm is largely set by the source resistor and this situation applies to bipolars too. With a 0.22R emitter resistor, the gm would be little more than 1/0.22, say around 4S with a high gm device, so this is probably not the issue it might seem. Mosfets make excellent output stages.
Yes, driving the capacitive gate can be difficult, but the miller capacitor (Coss) is the big issue, and it's only only 965pF on the IXFH compared to the Ciss which is bootstrapped in a source follower and it is 5.5nF.
The large mosfets do not require multiple pairs, and they are easy to drive with bipolar drivers. Morever, their SOAR and thermal robustness makes them good choices.
Different folks, different strokes......
Here on the forums it is a choice between IRFP240, IRFP9240 and 2SC5200 , 2SA1943. Both are excellent.
Heh, good question!
Let's try to be more comrehensive than previous speakers.
1. The choice about output device type are fully designer's headashe. And the main issue here is how deeply he understand pro's and con's.
2. All of four worlds, BJTs, MOSFETs, JFETs and LatFETs have its own mighty and weaky properties as output devices.
3. BJTs are easy to use and there are plenty os sample designs in typical configuration. As not too much designers wants to search the cat in a dark room - than there were also many designs with BJTs.
4. As not too much designers knows exactly good workarounds for MOSFET gate capacitance drive we do have ugly MOSFET designs with a poor performance.
You must need to know this about BJTs.
1. They are good in a comparatively low unity gain frequency designs because fast of them (like 2SC5200, MJL3281, 2SC2922) have too high dependance of transition frequency versus flowing current while slow of them (MJL21194, BD911) have too low widebandness for modern designs.
2. BJTs needs a more active drive to be fast, this asks for more powerful driver and usually predriver, so 3 stage emitter follower. But it have intrinsic instability which asks to proper understanding by designer and needs to be well-maintained by design.
BJT have been around longer so many will prefer them.
Plenty of established designs and overall familiarity.
True, for high powered designs Mosfets have advantages.
Then again, audio and music is full of art and romance.
Visually some people want to see something different.
Tubes have many disadvantages as well.
But there is a visual romance and higher cost people are willing
to spend.
Far as BJT, I find older TO3 packages, and even M200 packages rather beautiful.
And feel large finned heatsinks add to the romance and appearance. As well as function.
Far as performance. There is plenty of BJT and Mosfet designs. Which bring distortion
levels down. Way beyond our threshold of hearing.
Topology preference comes down to the designers / builders
Neither technology limits the ability for performance.
Eventually over time, prior arts become more romantic and cool like tubes.
BJT has similar status now, and will increase more over time.
Plenty of established designs and overall familiarity.
True, for high powered designs Mosfets have advantages.
Then again, audio and music is full of art and romance.
Visually some people want to see something different.
Tubes have many disadvantages as well.
But there is a visual romance and higher cost people are willing
to spend.
Far as BJT, I find older TO3 packages, and even M200 packages rather beautiful.
And feel large finned heatsinks add to the romance and appearance. As well as function.
Far as performance. There is plenty of BJT and Mosfet designs. Which bring distortion
levels down. Way beyond our threshold of hearing.
Topology preference comes down to the designers / builders
Neither technology limits the ability for performance.
Eventually over time, prior arts become more romantic and cool like tubes.
BJT has similar status now, and will increase more over time.
Oww, tubes
As you mentioned about tubes. This remind me of tube preamp. Many people have tube preamp with solid state power amp setup. Is that a proven setup for good sound or it just a personal taste with euphonic distortion?
Cause I also heard people keep saying that tubes are good (produce very low distortion) when they work as signal amplifier instead of driver.
As you mentioned about tubes. This remind me of tube preamp. Many people have tube preamp with solid state power amp setup. Is that a proven setup for good sound or it just a personal taste with euphonic distortion?
Cause I also heard people keep saying that tubes are good (produce very low distortion) when they work as signal amplifier instead of driver.
Last edited:
It is depend of the goal of the design. If you want lowest distortion in all audio frequency and highest bandwidth, it is easier for BJT. But if you want simple schematic with good enough sound, it is easier for lateral mosfet.
Overall linear Gm makes good outputs, however be it done. Push-pull can present difficulties to this end. Fet OPS is particularly vulnerable in P-P.
Everybody knows I'm an Old School type.
BJT's were used since they were invented, and look at how many products they're used in over the decades.
It's only been kind of recently that other "transistor types" have popped up for outputs.
And at the shop I was getting more blown mosfet/fet messes, rarely something with BJT's.
Bottom line, the "older stuff'" held up better.
BJT's were used since they were invented, and look at how many products they're used in over the decades.
It's only been kind of recently that other "transistor types" have popped up for outputs.
And at the shop I was getting more blown mosfet/fet messes, rarely something with BJT's.
Bottom line, the "older stuff'" held up better.
Ruggedness has gotten better over the years for BJT, and worse for MOSFET. Earlier high voltage BJTs had horrible problems with second breakdown, but they started getting better in the 80’s and 90’s with stuff from Sanken and Toshiba. Motorola quickly caught up. Some early RCA types were exceptionally rugged, but had a host of other problems. They were gone by 1985. Lateral mosfets and early verticals were advertised as a having “no second breakdown”. While not quite true, the Spirito effect on this generation of parts was not apparent at audio frequencies (even subwoofer frequencies) so the ruggedness was much better than BJT for a given dollars worth of silicon. You don’t want to use modern switching mosfet types for amplifiers at all - ruggedness above 10 volts on some of these switching types is worse than the old XPL909 they put in first Phase Linears. If you do use mosfets, stick with the older generation like IRF240/9240 or laterals that have a proven track record.
Ruggedness on BJTs is all over the place. You can buy crappy ones and you can buy good ones. Best ones are the audio-specific ones from ON and Sanken. Toshiba only makes one pair now, the C5200 - and it’s good enough for most typical applications. I think they screwed the pooch when they discontinued the C3281 which was better in every way.
Ruggedness on BJTs is all over the place. You can buy crappy ones and you can buy good ones. Best ones are the audio-specific ones from ON and Sanken. Toshiba only makes one pair now, the C5200 - and it’s good enough for most typical applications. I think they screwed the pooch when they discontinued the C3281 which was better in every way.
Yeah, I agree on that.
However, I've used the NTE 2328/2329 pair with good results in some of the repairs at the shop.
And the NTE 36/37 as well. - those were workhorses.
NTE NEVER made their own stuff. They buy industry standard parts and stick their own label on them. 2328/2329 = 2SC3281/A1302, 36/37 = 2SD1047/B817. When those types were being made, that’s what you actually GOT. Closest equivalents may be coming from Moto/ON, Fairchild, or ISC now. They may also be buying TTC5200 and 2SC5200N (and the PNPs) from Toshiba and calling them close enough.
Regarding big capacitances, the base-emitter capacitance of a bipolar transistor is gm/(2 pi fT) - ccb, which is also huge when you have a 10 MHz fT transistor running at 1 A.
Whether bipolar transistors or MOSFETs are more suitable for a design really depends on the circuit. In classical class-B complementary emitter/source follower configurations with emitter/source resistors, apparently bipolars have a nicer distortion minimum at the optimal quiescent current. That doesn't matter much when you don't want to use a complementary follower anyway.
Whether bipolar transistors or MOSFETs are more suitable for a design really depends on the circuit. In classical class-B complementary emitter/source follower configurations with emitter/source resistors, apparently bipolars have a nicer distortion minimum at the optimal quiescent current. That doesn't matter much when you don't want to use a complementary follower anyway.
Last edited: