john curl said:
Lighten up, John. I don't think I was second guessing you. I was not even disagreeing with you. I was just asking you to elaborate on your reasons for stating that you could not use MOSFETs in your amplifiers. You seem so annoyed that I ask a simple question, the answer to which will be interesting to everyone reading this thread. Isn't that why we are here? I'm truly interested in understanding your reasoning, and I'm guessing many others here are as well.
OK, so you dissipate a lot of power in your JC-1 amp; I see that the spec is Class A up to 10W at low bias, and Class A up to 25W at high bias. A lot of heat. I like that. With 9 pairs at more than 100 mA with 90V rails, your idle dissipation per channel is over 160 watts. And you already have boosted supply rails. Thanks for sharing that information. This is one fine amplifier, by the way.
That tells me that high idle dissipation and need for boosted supply rails is not your reason for stating that you could not possibly use MOSFETs in such a design. I'm just interested in what the reason is. So what is the reason???
Nowhere in this post am I insisting on anything, much less that MOSFETs are better for your application. In my earlier post, I even acknowledged that perhaps your concern was with idle dissipation or the cost of a boosted supply. These would be reasonable reasons for not wanting to use MOSFETs (maybe that was where I was second-guessing you?)
I just want to know why YOU insist that they are inferior.
I think that one of the very valuable aspects of these threads is that we each test each others' assumptions, and we all learn from that. You are certainly not shy about questioning my assumptions and assertions, and I welcome that.
Bob
Bob Cordell said:
Haven’t got much time now, but a few pertinent points I think you’re avoiding:
Yes, I was referring to lateral MOSFETs. In amplifiers built with these devices, 100-200mA bias per pair is the basically the norm, ragardless of the number of devices. When it comes to devices specifically manufactured and characterised for audio amplifier output stages, with the best linearity and lowest input capacitance, Lateral MOSFETs are at the top of the performance list.
That’s why manufacturers of high-end Lateral MOSFETs can get a way with charging ~$7 per device, Vs ~$1.50 for a comparatively crappy HEXFET with much more input capacitance and vastly inferior linearity.
Another thing worth noting about HEXFET’s is their horrible Vgs Vs temperature characteristics. There are some HEXFETs out there with the zero temp point at many amps of Id (well above any practical bias point for an AB amp), with horrendous temperature dependence of Vgs at drain currents below 1 amp - for a fixed value of Vgs and a junction temperature range of 25decG to 150degC, drain current variation for HEFFET’s typically exceed a ratio to 10:1.
Attempting to keep many paralleled pairs of these things in a high power amp operating stably with very low bias currents, as you suggest, without elaborate thermal compensation techniques would be a nightmare. I’ve seen people attempting to build high power HEXFET amps (and there are quite a few such amps out there) with multiple paralleled pairs resort to inserting source ballast resistors to compensate in exasperation!
Another place where MOSFETs fall down flat in comparison to BJT’s for high power amplification (and why John is quite correct in his assertion that his amplifiers would could not practically deliver the same continuous power output levels if the BJT’s were substituted for MOSFETs) is the fact that MOSFETs throttle back with increasing rds on as temperature rises. Higher bias currents are not the only contributing factor.
As long as they’re operating within their SOA, BJT’s don’t care what the temperature is - they will keep on delivering the goods. A MOSFET amplifier built with output devices of comparable Pdiss/Id/Vds ratings to the Pdiss/Ic/Vce BJT ratings of a similarly rated bipolar amp will always require significantly greater heat sinking.
abc11 said:
Alfred this is an area of great debate. I believe that negative feedback, used properly on an amplifier that is well designed to begin with, is a good thing that will make for better sound and better performance. If negative feedback is just used as a band-aid for a crappy design, the results may be very disappointing.
However, what sounds good and what measures good do not always correlate. Just as there are some very good-sounding vacuum tube amplifiers out there (which usually don't measure as well as mediocre solid state designs), so there are also some very good-sounding no-negative-feedback amplifiers out there.
I'm sorry I can't give you a better answer. The one strong opinion that I have is that some people have given negative feedback a bad rap for no good reason; but even that is just my opinion.
Bob
Bob
Glenn,
Do us all a favor and read section 1 of my paper "A MOSFET Power Amplifier with Error Correction", readily available on my web site at www.cordellaudio.com.
Your association of price of a lateral MOSFET with higher performance is simply ignorant. They are expensive because they are an obsolete technology that is no longer widely used. 300B vacuum tubes are also horrendously expensive, and adhered to by some audiophiles as well. I'm not one of them, and I don't think you are, either.
As I showed with rigor in my paper, the thermal stability of HEXFETs is significantly superior to that of bipolars. Anyone who cannot thermally stabilize a Class AB HEXFET output stage will REALLY not be able to do so with a bipolar output stage. I have never avoided the fact that thermal stability of laterals is better, but that is not all there is in life. As I mentioned in an earlier post, I am NOT talking about Trench FETs, which are a whole 'nother FET technology that is not advocated for power amplifiers. If you want to make comparisons, stick with the IRFP240.
A single HEXFET will put out 10 amps without even breaking a sweat in regard to RdsON. Why don't you take a look at the 20 kHz tone burst in my paper driving 22V peak into a one-ohm load at 20 kHz.
You keep going around and around on the dissipation-heat-sinking thing. While it is certainly true under many conditions that MOSFET amplifiers will have higher idle dissipation, which I have patiently pointed out, you simply cannot make the generalization that it is ALWAYS true that MOSFET amplifiers will need more heat sinking. The need to be able to cover the power dissipation at at least 1/8 power will usually dominate (remember, I'm talking about HEXFETs, not laterals).
Cheers,
Bob
G.Kleinschmidt said:
Glenn,
Do us all a favor and read section 1 of my paper "A MOSFET Power Amplifier with Error Correction", readily available on my web site at www.cordellaudio.com.
Your association of price of a lateral MOSFET with higher performance is simply ignorant. They are expensive because they are an obsolete technology that is no longer widely used. 300B vacuum tubes are also horrendously expensive, and adhered to by some audiophiles as well. I'm not one of them, and I don't think you are, either.
As I showed with rigor in my paper, the thermal stability of HEXFETs is significantly superior to that of bipolars. Anyone who cannot thermally stabilize a Class AB HEXFET output stage will REALLY not be able to do so with a bipolar output stage. I have never avoided the fact that thermal stability of laterals is better, but that is not all there is in life. As I mentioned in an earlier post, I am NOT talking about Trench FETs, which are a whole 'nother FET technology that is not advocated for power amplifiers. If you want to make comparisons, stick with the IRFP240.
A single HEXFET will put out 10 amps without even breaking a sweat in regard to RdsON. Why don't you take a look at the 20 kHz tone burst in my paper driving 22V peak into a one-ohm load at 20 kHz.
You keep going around and around on the dissipation-heat-sinking thing. While it is certainly true under many conditions that MOSFET amplifiers will have higher idle dissipation, which I have patiently pointed out, you simply cannot make the generalization that it is ALWAYS true that MOSFET amplifiers will need more heat sinking. The need to be able to cover the power dissipation at at least 1/8 power will usually dominate (remember, I'm talking about HEXFETs, not laterals).
Cheers,
Bob
Hi Glen,
It seems to me that you have no experience with HEXFET Vertical Mosfets so far, all you have stated above is in context with Lateral Mosfets, while HEXFETs differ alot......
The Minimum Gate Drive Current of HEXFET is define by
The main factor determining the charging current is Total Gate Charge and the Frequency of operation....
Ig=Qg X F
Ig = Gate Drive Charging current
Qg = Total Gate Charge
F = Frequency of operation....
Suppose we want to implement IRFP360
Whose
Qg=230nC
and desired Frequency of operation at F=50KHZ
from above equation the Ig comes out to be 11.5mA only.....
So its evident that the Driver stage needs to be capable of around 50mA pulse drivability to drive the Mosfet efficiently
WHEREAS
Bipolar Transistors need more than 1 A Base Drive to perform at this frequency.......Also the Driver needs to be a big Transistor as well...while in mosfets all it is needed a BC556 to do the job in a wonderful manner....no matter what is the level of rail voltage...
In my N-Channel HEXFET Mosfet amps i usually bias them at 35mA per device...and the Linearity is controlled by Local Feedback loop giving the distortions less than 1% without Global feedback....
I have designed Both Class-H amps with Bipolars as well as HEXFETS and both differ from each other vastly in terms of Sonic Quality and Reliability...HEXFET is a true winner
K a n w a r
It seems to me that you have no experience with HEXFET Vertical Mosfets so far, all you have stated above is in context with Lateral Mosfets, while HEXFETs differ alot......
The Minimum Gate Drive Current of HEXFET is define by
The main factor determining the charging current is Total Gate Charge and the Frequency of operation....
Ig=Qg X F
Ig = Gate Drive Charging current
Qg = Total Gate Charge
F = Frequency of operation....
Suppose we want to implement IRFP360
Whose
Qg=230nC
and desired Frequency of operation at F=50KHZ
from above equation the Ig comes out to be 11.5mA only.....
So its evident that the Driver stage needs to be capable of around 50mA pulse drivability to drive the Mosfet efficiently
WHEREAS
Bipolar Transistors need more than 1 A Base Drive to perform at this frequency.......Also the Driver needs to be a big Transistor as well...while in mosfets all it is needed a BC556 to do the job in a wonderful manner....no matter what is the level of rail voltage...
In my N-Channel HEXFET Mosfet amps i usually bias them at 35mA per device...and the Linearity is controlled by Local Feedback loop giving the distortions less than 1% without Global feedback....
I have designed Both Class-H amps with Bipolars as well as HEXFETS and both differ from each other vastly in terms of Sonic Quality and Reliability...HEXFET is a true winner
K a n w a r
Bob Cordell said:
Your association of price of a lateral MOSFET with higher performance is simply ignorant. They are expensive because they are an obsolete technology that is no longer widely used. 300B vacuum tubes are also horrendously expensive, and adhered to by some audiophiles as well. I'm not one of them, and I don't think you are, either.
I did not associate the price of Lateral MOSFET’s with their performance. Lateral MOSFET’s have always been expensive, even prior to the emergence of new technologies, principally because they are more expensive to manufacture.
Lateral MOSFETs are still being mass produced for audio applications, and as far as I can see people are still buying them for new designs, despite their cost, due to their desirable characteristics. I think lateral MOSFET’s are just as relevant to a discussion on the merits of Bipolar Vs MOSFET transistors as is your beloved IRFP240.
As I showed with rigor in my paper, the thermal stability of HEXFETs is significantly superior to that of bipolars. Anyone who cannot thermally stabilize a Class AB HEXFET output stage will REALLY not be able to do so with a bipolar output stage. I have never avoided the fact that thermal stability of laterals is better, but that is not all there is in life. As I mentioned in an earlier post, I am NOT talking about Trench FETs, which are a whole 'nother FET technology that is not advocated for power amplifiers. If you want to make comparisons, stick with the IRFP240.
I have built bipolar amplifiers rated at the kW level with low idle dissipation levels that happily keep their bias current within +/- 10% over significant temperature variations. I never said that comparable stability could not be attained with HEXFET’s, but looking at the specifications sheets for HEXFETs at low drain currents, it does not appear to me that there is any significant “superiority” of such a device over simple and very easily applied BJT’s biasing topologies.
According to the datasheet, at a Vgs of 4V the IFRP240’s drain current changes from 100mA at Tj=25degC to over 1A at Tj=150degC. The graphs do not extend below 100mA, but following the progression from the zero temp co point at approximately 10A, it is clear that at an idle current of 20-40mA the temperature dependence would be a great deal worse.
A single HEXFET will put out 10 amps without even breaking a sweat in regard to RdsON. Why don't you take a look at the 20 kHz tone burst in my paper driving 22V peak into a one-ohm load at 20 kHz.
Your 20kHz tone burst test does not represent a comparative continuous high power dissipation test between a MOSFET amplifer and a similarly rated BJT amplifier. I'd dont see how it is relevant to the point I made.
Kanwar:
It seems to me that you have no experience with HEXFET Vertical Mosfets so far, all you have stated above is in context with Lateral Mosfets, while HEXFETs differ alot......
Ok. I’ll admit that the majority of my experience is with Lateral devices. I think that all the points that I have made viz MOSFETs Vs BJT’s in terms of Lateral devices stand.
Kanwar
The Minimum Gate Drive Current of HEXFET is define by
The main factor determining the charging current is Total Gate Charge and the Frequency of operation....
Ig=Qg X F
Ig = Gate Drive Charging current
Qg = Total Gate Charge
F = Frequency of operation....
Suppose we want to implement IRFP360
Whose
Qg=230nC
and desired Frequency of operation at F=50KHZ
from above equation the Ig comes out to be 11.5mA only.....
So its evident that the Driver stage needs to be capable of around 50mA pulse drivability to drive the Mosfet efficiently
WHEREAS
Bipolar Transistors need more than 1 A Base Drive to perform at this frequency.......Also the Driver needs to be a big Transistor as well...while in mosfets all it is needed a BC556 to do the job in a wonderful manner....no matter what is the level of rail voltage...
OK, I disagree with you here a bit. Driving a MOSFET with just enough current to ensure an adequate slew rate isn’t sufficient for extracting the best performance from a MOSFET AB amplifier.
If you want to get the best crossover switching performance, the drive currents requirements are a great deal greater. It is for this reason that many high-end MOSFET designs actually incorporate a complementary emitter follower or source follower stage between the VAS and the output devices.
Cheers,
Glen
It seems to me that you have no experience with HEXFET Vertical Mosfets so far, all you have stated above is in context with Lateral Mosfets, while HEXFETs differ alot......
Ok. I’ll admit that the majority of my experience is with Lateral devices. I think that all the points that I have made viz MOSFETs Vs BJT’s in terms of Lateral devices stand.
Kanwar
The Minimum Gate Drive Current of HEXFET is define by
The main factor determining the charging current is Total Gate Charge and the Frequency of operation....
Ig=Qg X F
Ig = Gate Drive Charging current
Qg = Total Gate Charge
F = Frequency of operation....
Suppose we want to implement IRFP360
Whose
Qg=230nC
and desired Frequency of operation at F=50KHZ
from above equation the Ig comes out to be 11.5mA only.....
So its evident that the Driver stage needs to be capable of around 50mA pulse drivability to drive the Mosfet efficiently
WHEREAS
Bipolar Transistors need more than 1 A Base Drive to perform at this frequency.......Also the Driver needs to be a big Transistor as well...while in mosfets all it is needed a BC556 to do the job in a wonderful manner....no matter what is the level of rail voltage...
OK, I disagree with you here a bit. Driving a MOSFET with just enough current to ensure an adequate slew rate isn’t sufficient for extracting the best performance from a MOSFET AB amplifier.
If you want to get the best crossover switching performance, the drive currents requirements are a great deal greater. It is for this reason that many high-end MOSFET designs actually incorporate a complementary emitter follower or source follower stage between the VAS and the output devices.
Cheers,
Glen
G.Kleinschmidt said:
Glen,
The Slewrate I get with these Drivers is 125V/uS.....is it still less to do the justice with full range audio bandwidth.....
My design donot feature regular VAS + EF stage its of another type...
Seperate Drivers for each half....inverse driver topology is used....
Common Base VAS..driven from emitter....
Could you drive your BJT amp with 50KHZ Sinewave with no load connected ..and see how well the Cross-Conduction current increases and destroys the BJTs in no time....
There is NILL Cross-conduction in my design upto 100KHZ OPEN LOAD
K a n w a r
Workhorse said:
OK, cool. Your design is most likely adequate then. It’s just that a lot of people build MOSFET amps with no intermediate driver stage between the VAS and the output devices. They just apply that formula you provided and think that if they sink 11.5 mA from the VAS, that will be adequate.
Cheers,
Glen
Bob:
OK, so you dissipate a lot of power in your JC-1 amp; I see that the spec is Class A up to 10W at low bias, and Class A up to 25W at high bias. A lot of heat. I like that. With 9 pairs at more than 100 mA with 90V rails, your idle dissipation per channel is over 160 watts. And you already have boosted supply rails. Thanks for sharing that information. This is one fine amplifier, by the way.
I agree with you, the JC-1 is a nice amplifier. However, it is also rated at 0.15% total harmonic distortion at full power.
Funny that. In post 134 you pretty bluntly implied that I should consider my designs lousy if they only manage 0.1% THD at 20kHz at near maximum power.
OK, so you dissipate a lot of power in your JC-1 amp; I see that the spec is Class A up to 10W at low bias, and Class A up to 25W at high bias. A lot of heat. I like that. With 9 pairs at more than 100 mA with 90V rails, your idle dissipation per channel is over 160 watts. And you already have boosted supply rails. Thanks for sharing that information. This is one fine amplifier, by the way.
I agree with you, the JC-1 is a nice amplifier. However, it is also rated at 0.15% total harmonic distortion at full power.
Funny that. In post 134 you pretty bluntly implied that I should consider my designs lousy if they only manage 0.1% THD at 20kHz at near maximum power.
Workhorse said:
Just a note about Qg = gate charge, manufacturers state those figures often for a Vgs going from 0 to 10 Volts, eg. full on and off switching, thats where the Qg(tot) is used, for SMPS.
In audio amplifiers the working Vgs is around the platue only, eg. the "linear" region.
Qg(tot) data is for switching purpose, while Qgd is closer to audio amplifier designing.
See attached pictures here and in my followong post, first picture is data for a IRFP240 FET:
Cheers Michael
Ultima Thule said:
Hi Michael,
Thats why I consider Total Gate charge to design gate drivers.....
The figure shows that Qgd is less than Qg and therefore its more accurate to consider Qg as a main parameter in design
In Linear amps, when the output Clip, at the same time the Vgs increases to around 10V at which Qg tends to maximum and then the Qg is more determining factor towards gate driver...... Design
Cheers
K a n w a r