Power amp output - BJT or MosFET?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This is just my opinion, for what it is worth.

Short answer:

Class B: For low to medium power output (<100 watt RMS), BJTs are the best tradeoff for most applications. For higher output, use MOSFETs -- primarily because they will have higher reliability at high power outputs. MOSFETS will cost more, and require a more complex circuit with more front end gain to achieve similar levels of output distortion.

Class A: I'd always use MOSFETs, simply because the heat so compromises BJT SOA that they are not reliable for significant power outputs (say above 50 watts, absent heroic cooling).

Notice I said tradeoff. There are no perfect devices, and you may value certain device characteristics more or less than I do. For instance, I value reliability most highly; cost is secondary since I am not trying to make a profit. Of course cost is the primary factor in most consumer audio equipment, and they almost never use MOSFETs because the bipolar solution is cheaper.

I have some fairly old BJT amps that still work perfectly (including a 1967 JBL SE400), but I also own several high power BJT amps that failed catastrophically (Phase Linear 400), either catching fire or exploding capacitors or both. The MOSFET amps that have failed were higher power output (Haflers), but only blew fuses and were far simpler to repair; I blame the failure on clogged cooling fans - poor housekeeping.

No doubt many will take exception, but remember these are based on my own personal evaluation, assigning my own weights to different characteristics to make the conclusion. You may assign different weights to these parameters and come up with a different and perfectly valid conclusion for yourself.
 
Bill Fitzpatrick said:
Assuming good class AB design practices do BJT output stages offer more or less quality than MosFET output stages?

On one criteria, output stage crossover distortion its no contest is favour of BJT's.

As c/o distortion is the thorniest problem in aB design using BJT's is a good idea.

:) sreten.

Note :
in my book class aB = class B + optimal bias, class
AB = enriched to work in class A up to a few watts.
 
Re: Re: Re: Power amp output - BJT or MosFET?

Bill Fitzpatrick said:


OK. All other things being equal - which of course they never are - is the difference audible?


IMO absolutely. I can certainly hear the difference between a
class A design (does not have c/o distortion) and an aB design,
and I believe I know what a poorly biased amplifier sounds like,
which is very similar to Mosfets in a standard aB output stage.

(I had a tuned to the nines aB Mosfet amplifier for ten years)

The subject is covered in great detail in D.Selfs books, but his
concentration on THD rather than weighted harmonic distortion
does not lead him to the "optimum" solution IMO.

:) sreten.
 
Disabled Account
Joined 2003
I cannot hear mosfets from bjts but I can hear a poorly biased class b to a well biased class b. I cannot hear class a from class b tho.

Nelson Pass said:
Actually, I think it's Charles who teaches the use of laterals.


that's correct. But I am not sure if I am convinced about Charles' notion that laterals sound better than vertical mosfets.
 
From personal experience, a properly designed BJT amp (with good thermal feedback) will bench better than a properly designed MOSFET amp on the Audio Precision gear.

This is no insult to MOSFET amps though - their performance is very close to that of BJTs, and both can pull off outstanding numbers that'd make me wonder what substance you're on if you can hear a difference between the two... Of course, if people can hear a difference between speaker cables, they can probably hear a difference between output device types... well, I won't get into that ;)

For reliability, you can't beat MOSFETs.

As for vertical vs lateral FETs, I've only had experience designing amps with laterals, and I've always associated vertical MOS with hard switching power supply type applications instead of audio applications.
 
By email, I received an excellent technical rebuttal of my recommendations on this subject. I believe we can all learn from this, so I am posting it here. Real world experience is always a necessary complement to theory.

I suggest you pay particular attention to the recommendations for usage of cascoded bipolar out put stages and to the conservative output stage design for voltages below 65v. I for one will take this information to heart in future design work.
--------------------------------------------------------------------------------
Dear Slowhands,

I disagree with some of your thoughts concerning BJTs. We have used
them successfully in very high powered class A/B amps with no failures.
The biggest problem is that a lot of hobbyists and manufacturers do not
look carefully at the SOA of the devices used. Mosfets are expensive and
need lots of idling current to keep THD down. They require a tert.
supply to fully turn them on. They do have SOA problems since they have
limitations on their dissipation but they do not suffer from secondary
breakdown problems.

We have used both TO-3 and TO-3 plastics in our designs. We do not use
the TO-3s anymore as their mounting is a pain. We use exclusively
Toshiba 2SC5200/2SA1949 plastics. We temp. derate them to 80 deg C and then
calculate SOA data at these temps. Typically temp derating is about 62%
at 80 de g C.

One large problem with high powered amps is that the secondary
breakdown region on the SOA curve is violated due to the high rail voltages
used. When we go above +/- 75 volt rails we use series connected output
stages with no problems. Yes there is more saturation problems with a
series stage but then we simply add a few volts of rail to overcome this.

We also use glass G which basically does the same thing as the series
in terms of SOA. If one examines the SOA curve of high power bipolars
we see that current capacity is good below the 60 to 70 volt region
(assuming similar types to the Toshibas) and above that the current
capacity drops.

We used a +/- 130 volt rail in a 1.5Kw /ch 4 ohm amp we did but we used
a 3 stage class G supply so [the most] each output tranny ever saw was +/-65 v on
peaks (the 3 stage was a dumper type with only 2 devices ever in
series).

The big Toshibas at +/- 65 v had no problem with SOA. Used 6+6 (SOA
tables read 3+3) but we required absolute reliability. Have never had an
amplifier failure with that product todate.

Stephen Mantz
Zed Audio Corporation
Los Angeles
 
I think, it depends on what kind of quality you aim...
If you absolutely want low crossoverdistortions in AB, always
use BJTs, if you are looking for an amp that gives much current
with ease and want a simple design, than mosfets might be a better
choice.
In addition, the mosfets are very robust, it seems to me that the
only way to kill these is to overheat them (>170deg). I only once
killed a mosfet by accidently using a quiscent current of ~10A,
after 5 seconds it burned up.
(powerdissipation 200Watt, TO220AB on a small sink)
That was a IRF540n, costing 80 cents at reichelt...

I had a hard lesson with 2nd breakdown on BJTs, i once tried to
build an amp with mje15030/31 in the outputstage, these exploded
2 seconds after powering up, adjusted to qc of 100ma...

Mike
 
MikeB said:
, if you are looking for an amp that gives much current with ease and want a simple design, than mosfets might be a better choice.
Mike

Just point the extra rail votages for the drivers stage needed
for an efficient MosFet output stage driving low impedance
loads IMO does not not qualify as a simple design, though I
admit adding the extra rails is not complicated.

Without the extra voltage rails MosFet performance into low
impedance loads in terms of power output is very mundane
compared to BJT's.

IMO most MosFet "simple" designs are far too simple and BJT's
would in most of these cases provide better overall performance.

I have a hard time seeing the justification for MosFets as a good
solution for "standard" class aB amplifiers, i.e. most amplifiers.

:) sreten.
 
On the amps I've built, I've seen MOSFETs show greater distortion into low loads than BJTs, in roughly equivalent circuits. This isn't particularly surprising, given the shape of the Vgs/Id curve.

Also, in most (non-bootstrapped) configurations, the MOSFET amp will clip sooner than a BJT amplifier, given the same rail voltages.

All in all, this points at bipolars for driving 'awkward' loads. I'd agree with previous comments about (some) MOSFETs having reliability advantages at higher temperatures, though.

Cheers
IH
 
What about the previous stages?

What is the most suitable device for differential and VAS? J.Curl uses Fets for differential, many uses TO92 bipolars, Mr.Pass uses TO-220 mosfets. What is really good for differential?

As for VAS, almost all uses bipolar. But I see it because the ease to use in mass product (dont need to match or using different R). If we are talking about DIY, what is the most suitable device for VAS?
 
Disabled Account
Joined 2003
I have used mosfets for differential pairs. typically I use medium power mosfets (~1amp Ids).

I have always loved to use mosfets for VAS as well, especially driving mosfets output devices directly (no driver utlized). I tend to use to220 mosfets running 10-40ma (usually 20-40ma), with heatsink. Those are usually 1amp Ids to 5amp Ids devices.
 
I would say, definitely small BJTs in the VAS, at least in spice-simus,
the results with BJTs are much better, as the FETs have much less
gain. Of course, if you want a lowfeedbackdesign, jFETs might
be interesting.
For differentialamps, i prefer jfets, because of their ultrahigh
inputimpedance. This high impedance gives a much more exact
feedback, and a better DC-stability. But the FETs for DAs needs
to be matched, or you use a dual-jfet.

Mike
 
About vertical and lateral mosfet

I have built amp with K1058 pairs, it is lateral mosfet. In the manufacturer's datasheet it is definitely for audio use. But why I feel that with vertical mosfet (IRF) the sound just seems tobe have more character?
There is also controvercy about a certain device is good or bad for audio, like 4558 opamp.
Is it if a device is said for audio it is not automaticly that it will sound good?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.