BJT vs MOSFET - taking the best of both

[sonnya]

Id is dependent on Vgs. You want to have as low gatecharge as possible to reduce the gatecharge before the draincurrent starts to rise. A high impedance driver will raise the Turn off and on time.

That is the reason why the driverstage should be able to sink or source as much current as possible.
I think ... I think some times the problem with mosfet and IGBT outputstage (The sound quality) is addressed to the gatecharge.

If your driverstage cant remove the gatecharge fast enough you will a short shortcut in the outputstage when one mosfet turns off and another turns on wich again can be the reason why MOSFET class A sounds a lot better than MOSFET class AB.

The size off gatecharge is dependent of Vgs and Vds. It would desirable go get as low Vgs and Vds chage as possible.

I dont know if i am right? Ask Nelson ... He does nothing but designing with mosfet... Or ask Elso.

Sonny

[Nelson Pass]

Ok, the manuals only offer Gate Charge versus
Vgs, but from these we infer that the Ids is
a reasonably linear function of Gate Charge
at currents significantly above 0. Below that it
is not so good, and that is why we like lots of
bias on our MOSFETs.

[sonnya]

Have i understand it right?

You draw a "parallel" line between "Vgs versus gatecharge" and "Vgs versus Id"?

Yes i can You are interrested in running the mosfet at a high Vgs as possible. The Drain current will be high. If you make a change in Id in the order of 2%. Then the Vgs will only change a small amount wich again will result in a small change in gatecharge?

The perfect would then be a mosfet with really high "gfs" to get the Vgs change as small as possible?

Sonny

[hugobross]

P.Lacombe,

I've checked it out:
there are igbt's made for audio. There are also igbt-amp designs on the net, like:

http://www.arky.ru/audio/shem/igbt/igbt.htm
http://digilander.iol.it/essentialaudio/unetto_home.htm

I've seen a lot of power-mosfets having worser characteristics than those of the power igbts. But you can find good types of mosfets or igbts for audio performance.

Best regards,

HB.

[sonnya]

The first link is a design from elektor some years ago. They use Gt20D101 and D201 but the amp was in the begining made to work with MOSFET. This Amp do have higher power rating thoug.

The UNETTO amp uses the same IGBT.

One question i have on my mind. Why do they depent their frequency compensation (miller cap) in the UNE TTO on Q7 or Q8 b-c capacitance!?!?!? You will not get two amps with the same Phasemargin (read sound)!!!!!!!!!

Sonny

[P.Lacombe]

Hugo,

Thank you for the two links. The first schematic diagram seems to be correctly designed, and perhaps I will try to build this amplifier for evaluation purpose.

But all IGBT based amplifiers I have heard as yet, sounds not very well...

Regards, P.Lacombe.

[traderbam]

Thanks for looking up the Id/Qgs for me, Nelson. This is interesting - in the "active region" the drain current is approximately linear to the integral of gate current.

Some more thoughts on the use of FETs:
I notice quite often designers treating FETs like substitutes for BJTs, especially in emitter-follower output stages. This really doesn't make much sense to me. Using a FET as a source-follower is undermined because of the transconductance characteristic and the very high Cgs and Cdg. So they don't work as well as BJTs as emitter-follower buffers. However, they could be used as current amps as part of a BJT emitter-follower output. Using power FETs in common-source is more likely to be fruitful. Again, remember the Cdg which is usually very large (>200pF) and varies non-linearly and inversely with Vdg. This can cause some interesting "Miller" effects if not dealt with. And the transconductance is non-linear too. Uncompensated, these characteristics will degrade the sound quality, especially so if loop feedback is used (yes, feedback can make the sound much worse and/or cause instability). Another pointer is to not assume a mfrs complementary device is truely complementary - with some power BJTs they are pretty darned close but power FETs tend to be miles apart on capacitances and transconductance curves, so look for the best complement not just the one with the same serial no preceeded by a "9". Of course, if you choose a single-ended stage you don't have to worry about this - instead you have to worry about heat.

[FEThead]

[QUOTE]Originally posted by Morrist
[B]It has been stated previously in this thread but I don't think that there has been enough emphasis placed on this point: MOSFETs are not linear devices.... Note that not all BJTs are equal, some are more linear than others.

(Morrist's words above, mine below) I edited his post for brevity and it did not come out too well. Sorry about that.

Just as some BJTs are more linear than others, some FETs are more linear than others. MOST BJTs and FETs are made for some kind of switching application where linearity is not wanted. For linear FETs I suggest you go to the following links: http://www.hitachisemiconductor.com/...241_2sk213.pdf

http://www.hitachisemiconductor.com/...44_2sk1056.pdf

Go to the third page and look at the "Typical Output Characteristics".

FWIW, I run 2SK1058s in my version of Nelson's Zen amp and I am using 2SK216s in my version of the Pearl phono preamp. That is still in progress . . . but the prototype works fine. The '216s are quiet and linear even using just one '216 non-cascode for the first device! I would use the '216s in a line amp if I had one of those.

FETs are prone to oscillate at a very high frequency, but a couple hundred ohm resistor (or ferrite bead) in series with the gate ends that problem.

I run the '1058s SE, class A, and my speakers are quiet efficent, (SP!) so I don't need a "lot" of power.

Nelson, if you find this I am curious as to your opinion of the above devices.

To repeat, this is just my opinion, I do not mean to "put down" anybody else including Morrist and especially Nelson Pass.

[Nelson Pass]

I haven't used the 1058's, but looking at the sheet
I see that it's transconductance is quite a bit lower
than the IRF's I favor, and is more similar to the earlier
generation of MOSFETs.

Does your Zen amp have less distortion for using them,
or does it simply sound better to you?

[FEThead]

Originally posted by Nelson Pass

Does your Zen amp have less distortion for using them, or does it simply sound better to you?

Nelson,

Thank you for your reply!

I really don't have any way to measure distortion except to put the amp on an oscilloscope and do an X-Y plot. I built a Zen using an IRF130 a couple of years ago. (because that device was on hand) A few months I built my present amp using the 1058's which is actually two Zen amps bridged to eliminate DC across the speaker with using a blocking capacitor. (and fed push-pull) To me the newer amp sounds better, but there is more than one variable here (lack of output cap - which was a cheapo I admit, less voltage swing per device, etc.) and of course I am not without bias. I still have the single ended Zen and will put the new and old on the 'scope in the next day or two to see if I can see any difference.

Thanks again for your time and all you have done for us DIYers!