SemiSouth SiC Power JFETs

[audioman54]

Hey Bob,

Could you email me at audioman54@yahoo.com?

Thanks,

Mark

[KSTR]

I'd like to chime in to Bob's questions (especially "What do you think is the biggest reason they have such superior distortion characteristics to vertical MOSFETs?").... albeit some aspects seem to covered already in the datasheets and appnotes, like:

Only N-Channel, thus quasi-comps or circlotrons.

175deg is a package limitation.

The gate diode is a parasitic, the devices are true voltage controlled.

- Klaus

[Wavebourn]

Originally posted by audioman54
Wavebourn,

Are these pure class A amplifiers? Also our LME49811 based amplifier has beaten every tube and solid state unit is has gone up against (even Dennis Had at Cary was stunned when I did a demo of the power amps and D/A box at his place in North Car. He was so stunned he designed the new Exciter Dac around the national LME parts in 4 months!...which is quite a thing for a tube guy to do! Also I turned tube guru Joe Curcio into a believer. Joe is also still an FAE at National!) The LME49811 is class A internally by the way.

Listening is #1 of course but I (and Bob Pease) believe it must be combined with good measurements. If something measures poorly, .1 to 1% THD+N or higher, then the listening tests are suspect! We found in our single blind tests at National that small improvements in THD+N, that were linear with increasing power levels, resulted in audible improvements. Also the New LME parts have a unique process that matches the NPN and PNP transistors in such a way that they have identical dimensions, carrier flow, and ft's! Because of this we were able to run much lower bias currents than normal Class A/B amplifiers and still get great measurements and superb sound quality. That is why the opamps are doing so well in the high end community.

Has anyone done any set of measurements with an AP2 Cascade on one of the Pass amplifiers with these new output FETS like the graphs shown in AN-1850 for the LME49830?

I also attached a THD vs Power graph I made of my mono LME49811 based Amplifier with 50VDC rails into an 8-ohm load.

I would love to try these new FET devices with the LME49830 boards and compare them (listen and test) with the LME49811 based power amps I have. Benchtester and I should have the 811 based amps on display at Burningamp!

Best Regards,

Mark / Audioman54

Hi Mark;

it will be indeed very interesting experience to see and listen results of your work during a Burning Amp!

But currently, we are species of different Universes, despite I've started from the same one; my career started in Tomsk Institute of Semiconductor Devices, it was (and is) one of flagmans of Russian Semiconductor Industry.

At the same time I had a voluntary laboratory of electronic music where I was experimenting a lot deliberately creating distortions designing analog synthesizers and guitar effects. I did not think then that it is so hard to design and build a nice sounding power amp.

I had several design then following fashions, some of them were fully complementary. But I was very surprised when people were saying that my class A+C amp sounded better than all complementary amps, I did not believe that! It had fair class A amp and couple of unbiased complementary transistors working in parallel. Class A amp supplied an output current through resistor, and drove bases of class C output stage. The whole thingy had a complex feedback, so switching from class A to class C was very smooth, and I used as fast as I could find then opamp in an inverting mode then. Anyway, I was thinking that I was cheating approximating a transfer function, and kept believing that properly biased class AB fully complementary amp is the best.

Later I've discovered from Wireless World article that similar approach was patented and called Current Dumping, and the amp that used it was regarded well, even though I did not like that design, it was ugly, according to my preferences.

Switching my career to software design, then to design of networked systems, I still followed audio design as a hobby, and was curious why tube amps are still though out, even though their measurements were worse.

Thanks to Dr. Richard Bandler, the real Master of practical psychology in the field of subconscious perceptions, I finally realized what it is all about. And now I don't believe that complementary topologies can beat asymmetrical tube amps with class A semiconductor output stages, in terms of fooling imagination as if the sound is real.

However, I would like to hear your amps, and get converted back.

See you and your creations on BAF!

Sincerely,
Anatoliy Lisovskiy

[Nelson Pass]

Originally posted by Bob Cordell
So if we wanted to do class AB push-pull, we would need to go quasi complementary, right?

these seem to be sort of like enhancement mode JFETs where the gate junction is actually forward biased to make them turn on. It looks like one might have to shove over 100 mA into the gate in order to get in excess of 10A of drain current.

In the circuits you built, were they all class A no-NFB designs?

What do you think is the biggest reason they have such superior distortion characteristics to vertical MOSFETs?

Could we expect the same degree of superiority in class AB push-pull designs?

I didn't see any gm vs. Id or vs. Vgs curves (and didn't yet calculate it off the slope of the Id vs. Vgs curve), but is it your impression that one gets more gm per mA out of these devices than for vertical MOSFETs?

N channel, that's all there is, and I don't expect a P channel in
my lifetime But of course that still leaves lots of possibilities.

I think enhancement mode JFETs is the accurate description. I
haven't seen significant gate current, but I haven't run them past
10 amps or so.

I have a simple Class AB design with a little feedback, and I was
very pleased with it. I plan to kick that out as a DIY project.

Part of the lower distortion you see there is load line cancellation
on that circuit - you'll notice that the dependence on Vds extends
to a higher voltage range than the Mosfets we play with.

Also there is the high transconductance of the part, which is
on the order of 10S or so at the currents I run (<2A), so it responds
well to a little degeneration.

The temperature coefficient is a bit better than vertical Mosfets,
and they bias up at about 1.4 V Vgs at a couple amps or so.

Worst case improvements are about 1/2 the distortion of their
Mosfet counterparts, and the linearity of the capacitance seems
better, with flatter thd vs frequency curves.

All in all, a significantly better part IMHO.

[KSTR]

Thanks, Nelson, especially for the load line tidbit

[RocketScientist]

Originally posted by Bob Cordell
Could we expect the same degree of superiority in class AB push-pull designs?

Hey Bob. That's my (personal) question. As I posted at the outset, it would have to be a quasi-comp. design. And I haven't ever attempted to design (or even simulate) one of those. I've only fixed a few in my day.

So it would seem one of the key questions is do the SiC JFET advantages outweigh the disandvantage of having to use them in quasi-comp. for push-pull? You (or Nelson) are probably better able to answer that question than I am!

[syn08]

They are obviously not targeting the linear market. For linear applications, limiting these devices at 175 C is absurd, SiC can easily take 250 C junction temperature before even thinking about increasing the leakages.

For that, they need to choose another packaging technology. TO3 would be nice, but unfortunately copper and the SiC chip would not (thermally expansion wise) match. A special mounting preform is required, and that would be very expensive, adding to an already huge cost to process those SiC wafers.

Nice devices, but I wouldn't hold my breath for something optimized for linear applications. Great for DIY enthusiast, but little hope for a slew of new, improved performance, designs. I'll check if I can get some, I can see them in a Pass type class A monster amp. Can you feel my HiFi oven singing and cooking french toast at the same time?

[Wavebourn]

Originally posted by syn08
Can you feel my HiFi oven singing and cooking french toast at the same time?

Not yet, but why should we deny such a great possibility?

[RocketScientist]

Originally posted by syn08
Nice devices, but I wouldn't hold my breath for something optimized for linear applications. Great for DIY enthusiast, but little hope for a slew of new, improved performance, designs. I'll check if I can get some, I can see them in a Pass type class A monster amp.

Given your MOSFET YAP (and other) excellent amp designs, I'd be very interested to hear more of your thoughts about using the currently available SiC devices in a quasi-complimentary push-pull design? As I posted earlier, I'm especially intrigued by what could perhaps be accomplished with a single pair of these avoiding all the multi-pair issues.

[Wavebourn]

Originally posted by KSTR
Son Of Zen, www.passdiy.com/pdf/sonofzen.pdf

Forgive me Papa, did not recognize the abbreviation!