• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

MOSFET Follower Choices

Hi All

I've been experimenting with MOSFET followers as Drivers for Output triodes recently. The benefits are impressive - especially considering the meagre cost involved. ;)

Unfortunately, many of the original threads and references seem somewhat dated now though. Some of the devices mentioned seem out of current production. But more interestingly, there seems to be many new MOSFETS available on the market which might be excellent choices too.

I have been going over the mouser catalogue and found for example this excellent looking MOSFET from CREE - C3M0280090D which is rated up to 900v

datasheet http://www.mouser.com/ds/2/90/3m0280090d-838555.pdf

Here is another great looking FET from Fairchild might be the FCPF290N80 which is rated up to 800v

datasheet http://www.mouser.com/ds/2/149/FCPF290N80-604363.pdf

The Fairchild device claims a reverse transfer capacitance that is seems ideal for high MU triodes.

Unfortunately, I am a no-talent at interpreting all semiconductor specs...

So, here goes: Any opinions on the two MOSFET I listed above? Any other more recent production recommendations?

Best regards
Ian
 

exeric

Member
2008-03-24 6:23 am
soulmerchant,

I was asking the exact same questions you are asking and from the same point of ignorance. Here is a link to the thread:

http://www.diyaudio.com/forums/tubes-valves/285884-infineon-ipa50r140cp-n-channel-mosfet.html

One word of advice. To get the very best information from your fellow commenters it's always best to be completely overconfident of your knowledge and assert that opinion strongly. I do that and it has served me very well in getting the very best information out of people. You are being way too modest.:p
 
soulmerchant,
One word of advice. To get the very best information from your fellow commenters it's always best to be completely overconfident of your knowledge and assert that opinion strongly. I do that and it has served me very well in getting the very best information out of people. You are being way too modest.:p

Sometimes I do that and get interesting answers! ;)

If you look at the Crss plot on the Fairchild device I listed, its not at all flat... so maybe it needs to be carefully implemented?
 
When I look at the specs sheet, it looks like that really nice low Crss is only achieved at a Vds of about 67 Volts..

So here I look at the IXYS IXTU 01N100 put forth in the other thread by smoking-amp:

http://www.mouser.com/ds/2/205/98812-72889.pdf

What I see in this specs sheet is that the Crss drops nicely below 2pF as it approaches 40V. But the graph only goes up to 40v....
 
So, here goes: Any opinions on the two MOSFET I listed above? Any other more recent production recommendations?

Best regards
Ian

When it comes to solid state, the actual type really isn't all that important, given the high gain nature of the device that makes in-circuit performance almost completely independent of device characteristics.

That said, I would avoid both the listed MOSFETs. That Cree device is a SiC device. These typically like high voltage. Look at the spec sheet to see where the reverse transfer capacitance settles down: abiut VDS= 300V. That's pretty high. If you could manage that, let's say, driving 845s, then it would be useful. For more commonly encountered DC rail voltages in audio amps, not so much.

That Fairchild device was intended for switching applications, not linear amplification. As a result, its reverse transfer capacitance characteristic is hideous. In a source follower, Crt becomes your input capacitance, and this one is all over the place. That's bound to compromise sonic performance.

These days, it's unfortunate that all the emphasis is on switching. That's what the gov't and industry are demanding for these "eco-friendly" electric cars (though all this does is move the source of the pollution from the tailpipe to the power plant smoke stack). It's the same insanity that banned tungsten filament light bulbs for those unreliable CFLs that're filled with toxic mercury that'll leach out of landfills filled with busted CFLs.

All the really good transistors are becoming unobtainium. Lateral MOSFETs are only made by a few boutique manufacturers. VFETs are definite unobtainium even though they had some of the best transfer characteristics for audio.

For source follower duty, you want the Crt to be as low as possible, and you want it to settle down at reasonable voltages. The SiC devices don't do that, but good ol' silicon does settle down at more reasonable voltages: 20V -- 50V or so. For Fairchild offerings, what you want are "QFETs". These don't have that squirrelly capacitance characteristic.
 
FWIW, my go to MOSFETs have been, and remain, the ZVN0545A and the IRFBC20. Those 2 exhibit low and "stable" reverse transfer capacitance. Even wimpy types, such as the 12AX7/ECC83 and the EF86, easily drive the ZVN0545A.

Hi Eli, I realize that - thanks! They are old and definitely work. Of course if you are already close to 450V then its perhaps not so wise to use the ZVN0545A..

Also, the ZVN0545A specs sheet does not show Crss as a function of Vds. Not that I don't trust your word on this. :)


IXYS IXTU 01N100 looks also pretty nice...
 
Pardon if this question is like “fârting in a flower show”, but is there some sort of justified prejudice against using NPN Darlingtons for the follower? They're as robust as rocks, offer near perfect “follower” service. hFE of about 100, VBE of about 1.4 volts at 0.10 amps (not that you'll be using that many). Dunno… no MOSFET gate static sensitivities… 1 ma nominal drive, few frequency limitations. What am I missing? GoatGuy
 
piano₃;4625092 said:
current gain rather higher than 100 surely?

Actually no. High voltage NPN transistors offer surprisingly low hFE … due to the fact that to get their high-voltage blocking 'strength', something has to give. Channel-width of course. The 'P' part. Which in turn diminishes current gain substantially.

But here's a different way to look at it (and pretty close to factual, too).

RIN ≈ hFE ROUT

So let's say one is aiming to drive a double-triode (pair of them ganged in parallel), which for other reasons has an grid-shunt resistor of something low, like 120 kΩ. Since RIN is hFE or 100 × large effective input impedance, that's what … 12,000 kΩ. A very light load indeed on the previous stage.

Featherweight.

Proving, yet again, that “anything-followers” do a remarkable job at scaling input impedance … which is what they're asked/supposed to do.

To me I'd think that the NPN Darlington at $0.90 apiece, is kind of an elegant solution which by rights ought never to expire due to something fiddly. A 120 kΩ grid-to-ground resistance is neither very high or low for output triodes. One might argue lower … but it hardly makes a difference to the previous stage with the emitter-follower in place.

GoatGuy
 
From other hand, if the tube requires say 20 mA grid current peaks, that means base currents from very low to 200 μamp! But I agree with you, it is one of available options, as well as a cathode follower.

True… there is that positive grid ≡ real current draw thing. One might look at it as a feature and not a bug! The current multiplier effect also presents a current-draw-during-positive excursion to the previous stage driver. Albeit lowered by hFE.

200 μamp would (at say +50 volts?) look like (E = IR … R = E/I … R = 50 ÷ 0.0002 = 250 kΩ) … or more if the driving voltage is higher.

GoatGuy
 
Hi GoatGuy

Have you actually tried High voltage NPN transistors as follower-driver? If so, what ones did you use, and what kind of circuit?

Don't worry about farting. All mention of sand is farting in the tube board. :)

btw - 90 cents is indeed dirt cheap. But so is the ZVN0545A. Mouser makes me pay for shipping if my order is not big enough, so it really makes no difference...
 

thehoj

Member
2009-08-12 3:23 pm
That said, I would avoid both the listed MOSFETs. That Cree device is a SiC device. These typically like high voltage. Look at the spec sheet to see where the reverse transfer capacitance settles down: abiut VDS= 300V. That's pretty high. If you could manage that, let's say, driving 845s, then it would be useful. For more commonly encountered DC rail voltages in audio amps, not so much.

Is a variation of 10pF really that big a deal?.. @ 20V it looks like it's 8pF, and at 300V it's 2pF
Even the Fairchild one, it stays under 10pF from 40V up to 800V..


In a source follower configuration, with the Vds pretty much fixed, wouldn't one only be concerned about the Crss at that particular voltage anyways?

I'll use my own scenario, because I'm pretty interested in this, and I'm in a position to try different options at this point in time.
My Vds is about 600V.. Drain is at +200V, and Source is at -400V. If I were using that Fairchild linked in the first post the Crss is at 5pF at 600Vds. Does it matter for my scenario where my Vds is ~ 600V that Crss varies between 10pF and 1pF at different Vds points?
 
Last edited: