diyAudio (
-   Vendor's Bazaar (
-   -   SemiSouth SiC JFET from $16/pc, shipping NOW! (

zhoufang 1st April 2010 07:03 AM

SemiSouth SiC JFET from $16/pc, shipping NOW!
3 Attachment(s)
Well, not really shipping now, as I still have some matching to do.
R100 should be shipping out by end of next week.
R550 will take 2 weeks longer.
All Semisouth JFET will be sold Vgs matched.

The stock for R100 is limited, and I have many pre-order to fulfill, so first come first serve.
Stock for R550 is plenty, so no problem.

SJEP170R550 - new 1700V part with lower capacitance, lower transconductance, lower power dissipation.
Price: $20/ea for the first 4 pcs, $17/ea for the 5th to 8th pcs, $16/ea from the 9th pcs onward

SJEP120R100 - the original part used in FirstWatt J2.
Price: $29/ea for the first 4 pcs, $26/ea for the 5th to 8th pcs, $25/ea from the 9th pcs onward

The above list is just calculated for your convenience, you can purchase in any quantity.

The quantity discount will also be valid if you buy a combination of R100 and R500, i.e. $3/ea off base price for 5th to 8th pcs and $4/ea off base price for 9th pcs onward.

Free shipping for orders >=$50, paypal fee inclusive.
$1 will be donate to the forum for every 2pcs of SS JFET sold.

For the list of other parts that I'm selling, as well as more details about how to order, please view my other post.

zhoufang 1st April 2010 07:27 PM

Payment received from ccliu, thank you!
Invoice sent to all how PMed me.

zhoufang 2nd April 2010 02:13 PM

Payment received from:

thank you!

zhoufang 4th April 2010 04:37 PM

Payment received so far:
ccliu 4*R100 + 4* R550
efreak0314 8*R100
flg 4*R100
stefan5000 2*R100
asnakeon 4*R550

Thank you!

zhoufang 5th April 2010 07:24 PM

Just finished measuring the Vgs for all R100.
Will ship out all R100 only orders on Wednesday 7th Apr.

Vgs is measured with gate tied to drain, and with a CSS of 1.216A.
The Vgs varies between 1.0xxV to 1.5xxV.

R550 will be matched and shipped 2 weeks later.

zhoufang 8th April 2010 05:23 PM

1 Attachment(s)
Matching of TO-247 Semisouth JFET

20 pcs at a time, CCS through all of them (from HP PS below the desk).
Id = 1.2165, ambient 27degC

The precision and short term stability of Vgs reading is quite good, when stabilized it usually fluctuation within a range of +/-0.05mV (really takes 6.5 digit meter to see it). A small swing of hand will increase Vgs by about 1mV (reaction time about 2-3 second), and then you will see it slow drop back.

Repeatability and long term stability needs to be improved. typically I see about 3mV of difference when I retest at different position or time, in worst case I have seen 7+mV changes. I believe this is due to air-movement and the variation in the resistance of the socket (each socket to pin connection cause about 7-10mV drop at 1.2A). If I test at smaller Id, I expect the socket problem to be reduced.

The first 2 decimals (0.x V and 0.0x V) are pretty reliable, the 3rd decimal place is useful but less accurate, the 4th decimal, although recorded on the JFETs I sent you, can be neglected.
For current sharing purpose, I think this accuracy is sufficient.
For matching of Gm to decrease of THD (if applicable to your design), this is also good enough as Vgs is not a indicator of Gm anyway.

I will try to improve it when I match R550 10 days later.

EUVL 8th April 2010 05:33 PM

If you wish to match them like that, 20 a time, you need to bolt them all onto ONE big lump of aluminium for temperature stabilisation. Whether the block is at room temperature or at 50C in an oven is up to you.


zhoufang 10th April 2010 03:15 PM

Dear Patrick,

I am not sure if bolting them all onto a big aluminum block will get more accurate result.

When there is a common aluminum block, the temperature Tj of each FET is not only determined by its own heat dissipation (which is directly proportional to its Vgs with single CSS). Instead, the Tj is mostly determined by the temperature of the aluminum block, which is determined by the heat dissipation of all the FETs on the block, which is ultimately determined by the average Vgs of all the FETs.

The problem is, for each group of 20 FETs, their average Vgs can be quite different. The same piece of FET, when measured with another 19 FETs of average Vgs=1.0V, verses with a different group of 19 FET with average Vgs=1.2V, will give you very different result, becase the delta between the aluminum and ambient temperature differs by about 20%.

I did my test in this way, so that all heat is dissipated by natural convection + radiation of the TO-247 package, which is highly uniform, and the temperature of each FET is only minimally disturbed by the neighbouring FET (unlike in the all bolt together case, the Tj is mostly DECIDED by the Vgs of other FETs)

In my hand swing test, it has shown that the mass of the TO-247 device has reasonably sufficient thermal mass to reject fluctuation

One obvious limitation I see in my method is that it limits the maximum power that each FET can handle. With RthJA of about 50degC/W, up to 2W is possible, but I will keep more safety margin unless it is SiC (semisouth) or has larger than TO-247 package (2SK1530/J201).

Zhou Fang

EUVL 12th April 2010 04:32 AM

By large aluminium block, I meant something like 200x300x80mm. You can easily calculate how much mass there is and how much energy you need to heat this up by 0.1C. On top of that you can easily monitor the aluminium block temperature with a thermal couple from some multimeters.

You are speaking from theory, I speak from years of experience.


zhoufang 18th April 2010 06:59 PM

R550 are matched, all R550 orders will be ship out later today (19th Apr Mon), very sorry for the long wait.

R550 are match in a similiar way as the R100 except:
- They are tested with Id = 0.5043A, instead of 1.2+A for R100. This should be closer to the Id that most will be using the R550 at, since it has higher RthJC.
- Vgs is read out by directly probing the legs of the JFET, instead of from the pink extension wire. This remove the inconsistant voltage drop due to resistance of the pin(leg) socket connection.

The repetability of this R550 testing setup is generally below 1mV, with very few exceptions that goes slighly above 1mV.

All times are GMT. The time now is 08:46 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio