IF3602 Very Low Noise Dual JFET

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Samuel Groner

Well-known member
Joined
Aug 19, 2004
Messages
2,940
Location
Zürich, Switzerland
Hi

By accident I run across this very remarkable part--the InterFET databook (Databook_final.pdf) quotes a voltage noise of 0.3 nV/sqrt(Hz) at 100 Hz! This comes not without drawbacks though, the capacity is pretty high.

Anyone already used this part? Thoughts?

Samuel
 
PHAT FET!

No, I haven't seen let alone used that part. Man is Crss large!

It does look as if 10 paralleled SK170s would be competitive, but then here you are getting a dual with some matching as well. And in one small package.
 
[quote author="Samuel Groner"]Hi

By accident I run across this very remarkable part--the InterFET databook (Databook_final.pdf) quotes a voltage noise of 0.3 nV/sqrt(Hz) at 100 Hz! This comes not without drawbacks though, the capacity is pretty high.

Anyone already used this part? Thoughts?

Samuel[/quote]

Classic no free lunch... but remarkable noise... The best I recall in a fet from years ago was .7nV and that was some obscure military part.

I still wonder what kind of bare parts could be made with that new SiGe process?

JR
 
[quote author="JohnRoberts"]
Classic no free lunch... but remarkable noise... The best I recall in a fet from years ago was .7nV and that was some obscure military part.

I still wonder what kind of bare parts could be made with that new SiGe process?

JR[/quote]

And will anyone find a market there of sufficient size to have the incentive to make them?

Teledyne Crystalonics made the C413 many years ago, which was in the <1nV territory and had gm of about 30-50mA/V or more at fairly modest currents. Also high C and low pinchoff. Cute little TO46 package, although with the gate tied to the case. Made it difficult to cool, although there were some dissipators with BeO washers that could be used.

Then evil ion implantation took over, producing defects that couldn't be fully annealed out (according to Ed Oxner). But a few companies continued to use diffusion, like Toshiba.
 
I ordered samples from them a couple of years ago. Can't remember for sure if the IF3602 was that I got. I just called them up and ask for a 2SK170 sub and they sent me 2 or 3 different part #s. In spite of having them all this time, I've yet to try them out. I'll put one in a bloak when I get around to building one.
 
Isn't 20V mentioned? That's probably a CYA spec though.

What one would like to see is the dependence of Ig on Vdg. The 2SK170 gets pretty lousy at high Vdg, despite being still well within the max rating.
 
This part is probably not really practical for an audio DOA. It is clearly intended for instrumentation applications with this performance level and price. One can buy an assembled, factory fresh 990 opamp from John Hardy for about the same money as these cost, per package and there is a high minimum order.
 
[quote author="burdij"]This part is probably not really practical for an audio DOA. It is clearly intended for instrumentation applications with this performance level and price. One can buy an assembled, factory fresh 990 opamp from John Hardy for about the same money as these cost, per package and there is a high minimum order.[/quote]

Their yields are probably pretty low, as well as the demand.

I would just parallel a bunch of 2SK170s. You could run more current to optimize noise and gm that way too.
 
Looks like I'll get some samples in a month or two. Thinking of using them in a 100x/1000x differential amplifier for measurements and simply to see how close to zero noise I get. Not easy, even a few ohms will degrade the 0.43 nV/sqrt(Hz) of the differential pair...

Samuel
 
Everything gets "heroic" pushing these kinds of noise levels. Second-stage contributions that can usually be ignored become important, among other things. B field pickup, power supply noise, all sorts of things will spoil the results. And if you are using resistive feedback it must be awfully low-Z.

Cooling helps gate leakage a lot, and thermal channel noise a little, but of course is cumbersome, since among other things you have to prevent condensation. Also, at low temperatures some interesting physics gets involved and for still-modest drain-gate voltages the gate leakage has an anomalous term due to impact ionization in high-field regions, when the electrons bang into the colder lattice. Things are usually safe below about Vdg of 5V, but this is hardly optimal for reverse-transfer C.

I'm sure Interfet has a pretty small wafer size, and these parts are big, so even if all the parts work they probably don't get that many per.
 
[quote author="bcarso"][quote author="Samuel Groner"]Got a pair! Time to design something cool.

Samuel[/quote]

Or even something cooled.[/quote]
That's almost halfway CoolEditPro...
...but then it's not longer related, apologies....
 
Back
Top