Bob Cordell Interview: BJT vs. MOSFET

[Bob Cordell]

Originally posted by john curl
That is a good idea, Wavebourn. I was tempted to use something like that for a PA power amp. It has the problem of a mismatch in impedance, at both input and output, which could cause some real problems. Still, it can be an interesting approach.
Personally, I use FET's when they will allow the power output that I need to achieve. Most vertical FET's won't handle difficult loads or a short circuit, for more than an instant.

John,

I'm afraid I have to disagree with your comment about vertical FETs not handling difficult loads and your implied assertion that they are unable to provide high power. Has that been your experience?

The SOA of most vertical FETs is better than bipolars, owing to the absence of secondary breakdown. Of course, they should be protected against dead shorts where instantaneous current might go to many tens of amps for the time it takes to discharge the large rail capacitors. This is easily done with an electronic crowbar. All one has to do is to keep the instantaneous junction temperature from exceeding about 175C. It's strictly a power dissipation thing combined with die time constant. Ring emitter transistors have much improved SOA, but they still can exhibit secondary breakdown.

Can you elaborate on your experience or on your reason for saying this? If you've popped vertical MOSFETs into anything other than a dead short without any protection at all, I'd guess that they blew because you got them into an HF oscillation.

Bob

[john curl]

Maybe, but I can't do it very well. Good luck to you, and you had better publish your layouts. We discussed this about 20 years ago. I finally gave up.

[Bob Cordell]

Originally posted by traderbam
My editing has gone awry.

My other comment is that comparing "ft" is not really valid as it does not take account of charge. Comparing the speed of bipolars and FETs of equal power rating shows bipolars are nearly as fast; same order of magnitude.

It is a common misconception that ft is not a valid way of describing the relative small-signal speed of a FET. It is valid, but one does have to keep in mind that the ft of a FET is often more a function of current than that for a bipolar, since the gate-source capacitance of the FET tends to be fairly constant with current, while the base-emitter hybrid pi capacitance of a bipolar transistor increases in proprtion to current. In both cases the other important part of the ft equation, transconductance, increases with current. The ft of about 350 MHz I quoted for a MOSFET was at a current of about 1 amp. At a current of only 100 mA, the equivalant ft will be closer to 35 MHz. MOSFET ft does indeed properly take into account gate charge, but it is not the best way of describing large-signal speed.

I encourage you to do some SPICE simulations with a MOSFET power transistor and a bipolar power transistor of the same power rating. Look first at the small signal frequency response of a source-follower (emitter follower) driven by a 50-ohm source and loaded by 8 ohms, and biased at 1 amp. You will see that the 3 dB bandwidth of the MOSFET source follower is much greater than that of the bipolar emitter follower.

Next, do a large signal simulation in which you operate both devices at 1 amp into the 8-ohm load and then reverse the gate (base) drive and pull 10 mA out of the gate (base), and see which one turns off most quickly. You will see that the MOSFET turns off much more quickly.

For both small-signal and large signal situations, vertical MOSFETs are 3 to 10 times faster than a ring emitter transistor. You don't see ring emitter transistors being used in switching power supplies do you?

Bob

[andy_c]

Originally posted by Bob Cordell
I encourage you to do some SPICE simulations with a MOSFET power transistor and a bipolar power transistor of the same power rating.

Please be aware that the SPICE models named NMOS and PMOS, levels 1, 2 and 3 assume CGS, CGD and CDS are constant with VGS, VGD and VDS respectively. These models were not meant for vertical power MOSFETs. International Rectifier provide some subcircuit-based models for the IRFP244 and IRFP9240 that use diodes to approximate the junction capacitance variations with varying voltage. I haven't compared their data with the datasheets carefully yet though. LTSpice has a model specifically for vertical power MOSFETs called VDMOS that models the variations of CGD and CDS with VGD and VDS respectively, and assumes CGS does not vary with VGS. I'm in the process of doing some model extraction for some vertical MOSFETs that uses the Excel solver to do a best-fit match for CGD and CDS vs VGD and VDS respectively. So far, I've done the Fairchild fqp2p25 and fqp3n25 TO-220 devices. I hope to do the International Rectifier IRFP244 and Fairchild fqa12P20 TO-3P devices soon.

[traderbam]

Bob,
My comment was "comparing fts is not valid".
You mis-interpreted this as meaning ft is not a valid speed measure of a FET. Which is not what I said.
The speed of a FET is related to Cgs and gm, which is related to its ft. For a given ft, the speed of a bipolar is related to gm and beta. So you miss a trick if you compare only on ft.
I don't think your emitter/source follower simulation is apt. Perhaps if you compared the output Z of the two followers. It may be necessary to have a 50 ohm gate resistor for a FET but no series resistance is needed for a BJT.
Of course SMPS use power FETs because they make reliable and efficient switches. A linear amplifier has different requirements.
Brian

[Workhorse]

Originally posted by traderbam
Bob,
It may be necessary to have a 50 ohm gate resistor for a FET but no series resistance is needed for a BJT.
Brian

Hi Brian,

There are some designs in which BJT employs 2.2ohm Base resistors to prevent the Oscillations and the present day BJT 's have very high Transition Frequency upto 30MHz......

A linear amplifier has different requirements.

Yes but, if you compare the advantages of Vertical Mosfets with BJT's in Linear applications than Mosfet always win , provided that the designer has required skillset to implement the vertical Mosfets in an excellent manner in his/her designs




K a n w a r

[Workhorse]

Originally posted by Bob Cordell



John,

I'm afraid I have to disagree with your comment about vertical FETs not handling difficult loads and your implied assertion that they are unable to provide high power. Has that been your experience?

The SOA of most vertical FETs is better than bipolars, owing to the absence of secondary breakdown. Of course, they should be protected against dead shorts where instantaneous current might go to many tens of amps for the time it takes to discharge the large rail capacitors. This is easily done with an electronic crowbar. All one has to do is to keep the instantaneous junction temperature from exceeding about 175C. It's strictly a power dissipation thing combined with die time constant. Ring emitter transistors have much improved SOA, but they still can exhibit secondary breakdown.

Can you elaborate on your experience or on your reason for saying this? If you've popped vertical MOSFETs into anything other than a dead short without any protection at all, I'd guess that they blew because you got them into an HF oscillation.

Bob

Hi BOB Cordell,

Nice see another lover of Vertical Mosfets!

Majority of Diyers here on forum are very much inclined towards the good old/new BJT designs and they resist Vertical Mosfet designs because implementing them require much more precised skillset to do the job in excellent manner...
Only some of the few designers have this under their possession....

As far as I know Vertical Mosfet amp designers are
Nelson Pass
Bob Cordell[nice meeting you]
Greg Ball
Anthony Holton
and me too


K a n w a r

[traderbam]

Only some of the few designers have this under their possession....

That is an extremely arrogant remark. I have considerable more respect for the capabilities of the members of this forum than you have. My experience is actually quite the opposite...I've seen many people use FETs because their assessment of the parameters of importance to sound quality are superficial, often due to bad advice from "experts", and they end up with sub-optimal sound.

I am not making any judgements here about which device is superior. This would be presumptuous because it depends on the rest of the circuit and what the designer is trying to achieve. I am just pointing out that the blanket assertion that FETs are much faster and therefore superior has not been properly thought through.

[anatech]

Hi traderbam,
I have to agree with you with regard to your last comments, no offense Kanwar. I find that so far, I have to run mosfets pretty hot to get decent performance out of them, and they still don't sound right. Could be my skills, but I've also redesigned a few commercial mosfet output stages to use bipolar types and have ended up with better sound quality and lower distortion every single time. I am talking about vertical type mosfets here.

Bob, I think your error correction approach works well and is necessary for mosfets. I am trying to adapt the idea to a bipolar design I've been working on. I think it holds great promise. I really ought to read your papers on this.

-Chris

[Workhorse]

Originally posted by traderbam
That is an extremely arrogant remark. I have considerable more respect for the capabilities of the members of this forum than you have. My experience is actually quite the opposite...I've seen many people use FETs because their assessment of the parameters of importance to sound quality are superficial, often due to bad advice from "experts", and they end up with sub-optimal sound.

I am not making any judgements here about which device is superior. This would be presumptuous because it depends on the rest of the circuit and what the designer is trying to achieve. I am just pointing out that the blanket assertion that FETs are much faster and therefore superior has not been properly thought through.

Noway Brian, you simply misunderstood it..........

Tell me , How many successful circuits are there on this Forum which are'nt designed by the designers in the above mentioned posts.....employing VFET's ...
Whenever a Diyer dream of amp, he/she prefer Bipolar config, not the VFET....topology...have you ever dare to wonder why its so much inclination towards BJT's

I am not talking about speed of Mosfets Vs Bipolar
I am talking about the SOA capability, ease of gate Drive, ease of protection circuitry, much more Transient impulse in terms of current.....

Vertical MOSFET clearly excels any of the known audio BJT...Tell me any number in BJT , There is always a VFET ahead of it......