Semisouth SiC JFETs: SJEP120R100 or SJEP170R550 better for building amp?
Thank ijomojo for point out to us that Semisouth have a new 1700V SiC JFET SJEP170R550!
With a fair amount of effort, I managed to get through their automatic answering machine system, and got a quote.
The cost of this new 1700V Jfet is about 75% of that of SJEP120R100, and the SJEP120R063 being much more expensive.
Since the SJEP120R125 is obsolete, and SJEP120R063 very expensive, I sees the SJEP120R100 (the subject of Magura's GB) and this SJEP170R550 as the best candidate for DIYers.
The SJEP170R550 has thermal resistance RTHjc of 2.6degC/W, while both SJEP120R100 and R125 are rated 1.1degC/W.
If the insulation mounting of their TO-247 to heatsink (plus the heatsink local spreadout resistance) has a RTH of 0.4degC/W, then they will have a total RTH(junc to heatsink) of 3.0 and 1.5.
In this case the R100 and R125 has double the power capability of the new part.
A more relistic figure of the mounting will be 1degC/W, is that case (1.1+1)/(2.6+1)=58%, which means the SJEP170R550 should be able to handle about 58% of the R100/R125 power.
The semisouth datasheet only has graph of ID vs VGS (Figure 6) but no conductance vs ID curve.
So I spend some effort to manually measured the conductance (under extreme enlargement in photoshop) at ID of our interest.
The transconductance are found to be:
1.4A 7.85Papa's J2 is rated for 30W at 8ohm and 15W at 4ohm, from this it is not hard to calculate that the bias current through each JFET is about 1.4A.
If we use SJEP120R100, we will just keep the 1.4A, and with SJEP170R550 we can go to 0.8A without increasing juction temp.
We can see that a single SJEP170R550 has 80+% of the transconductance of a single SJEP120R100.
Two parallel SJEP170R550 will have much higher transconductance than a single SJEP120R100, and reach about 85-90% of that of J2's single SJEP120R125.
SJEP120R100's ID leakage at low VGS?
From the ID vs VGS graph of SJEP120R100, the ID does not totally shut off at VGS<1V. between 0.3V<VGS<0.8V, the ID is flat at 0.78A.
And the slope (=transconductance) is seems abnormally gentle until VGS pass 0.9V or so.
This is only seen in the SJEP120R100 and no where else in Semisouth's other JFET.
I guess this will means the first 0.75A bias current through the SJEP120R100 is wasted. Anyone has a better idea?
It is very difficult to read off the graph due to the voltage scale.
We can roughly see that, at the very low end of the VDS:
SJEP170R550 shows Ciss of up to 400pF
SJEP120R100 shows Ciss of up to 1200pF
The datasheet gives Ciss:
SJEP170R550 170pF@300V (from graph: <180pF@100V)
If their gate capacitance curve are of the same shape (a linear multiple of one another), then at low VDS SJEP170R550 will still keep a 1:4 Ciss advantage over SJEP120R100
Conlusion: the SJEP170R550 has roughly 1/3 or SJEP120R100's Ciss, and the linearity of the gate capacitance is too hard to tell.
It is such a pity that SJEP120R125 is no longer produced (and guess who has stockpile of it?), as SJEP120R100 does not seems to be as good and it takes 2x SJEP170R550 to get near its performance.
Between the SJEP170R550 and SJEP120R100, from what I can see the SJEP170R550 seems to be more useful.
The only obvious problem I can see in the SJEP170R550 is that its ID vs VDS curve for VGS=1.5V seems to be more flat, suggesting it is less capable of taking advantage of loadline cancellation.
I'm sure I missed out a lot of point, so pls help me to bring them out.
In practice there is very little difference between the R100 and
R125 versions for our applications. I get just about the same
numbers in the J2's with either part.
Your words are more worthy than their datasheet!
I'm really looking forward to see the J2 Circuit being released. Hopefully before Magura spreads out the parts... ;-)
Maybe a circlotron J2 would be very interesting. (I know, there's a version out there, but I would like to see a comparision of the specs between the Nelson optimised version, and the semisouth drop in approach.
Looking at the specs of the drop in semisouth version, it looks like a stronger driver required...
P.S.: Your passdiy website is not reachable... ;-(
A circlotron would be something else.
With regard to JFETs, there are more parts coming out in
time, so patience will be rewarded.
considering results of your photoshoping , and counting also on what I raped my eyes in last two days ,I'm pretty sure that we must use 2 paralleled 550's in every position , if we want nice characteristics and safe operation in same time , at least in voltage/current ranges interesting to common Papa's follower/spoiled kid ;
so - counting on words written by greatest kid of all kids prisoned in mature envelope :
you know familiar feeling - there is always better tube just around the corner :clown:
Take a look at the last curve in "De-Lite"
Now that knowing r100 is also an official J2 part, the option looks clear.
Until Papa said:
SiC JFET in new packaging and withstand 400degC?
Or much cheaper alternatives?
Higher and more linear Gm?
Please enlight me a little more!
( will it drive F4 ? :rofl: )
next jfet :
|All times are GMT. The time now is 10:34 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright ©1999-2015 diyAudio