Low audio noise JFETs?

What's wrong with good old 2N4393,92,91? They are well below 10 nv/ rtHz at 20 Hz, specified etc. (siliconix datasheets). I used to have also an alternative (same process, another device), but seem to have forgotten which it was... got it, was easy to find - 2N4856 (I used the lowest R ones - variable R design - there are relatives good for you as far as I can remember :-).

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

Thanks! I looked at that but ON only markets it as a switch:

But then Vishay (they bought Siliconix) indeed mentions noise:

Pretty darn good, I just wonder why it levels off below 30Hz or so, Anyhow, seems there is hope here ...

--
Regards, Joerg

http://www.analogconsultants.com/

It gets better than that. They also make the 2N5566, which is a dual

2N4393 or so - have used them as well. At low frequencies - and being reasonably fast for you - there is nothing better to find. The 2N4416 is a known low-noise FET, but at higher frequencies, not where you need it to be (it is the choice of Ge detector manufacturers, works cooled down to apr. -90 C to achieve lowest noise... :-) - but at much higher speeds of interest).

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

switch:

noise:

They still make those in the TO-71 can? Haven't seen them in a long, long time.

Yep, the 4416 ain't too bad but seems higher:

Just wondering why the 2N4391 levels off for lower frequencies (somehow doesn't make sense) and the others don't. I bet Winfield Hill would know the answer. Win?

--
Regards, Joerg

http://www.analogconsultants.com/

Looks pretty much like because the 4391 can work at higher Id - the graph just is not that steep for the 10 mA Id case. Now why these FETs have lower noise at higher current I don't know, I am only a FET user, not a manufacturer :-).

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

higher:

On a sunny day (Tue, 13 Nov 2007 17:49:50 GMT) it happened Joerg wrote in :

BF245B, one I use for many things, is 1.5dB at 100Hz, it is in the datasheet.

2dB at 2MHz Vds=15V, RG=1kOhm, Vgs=0, T=25C.

What, me? I dunno, something to do with a draftsman?

Might be. Specsmanship has gone down the drain quite a bit from when I started out into serious electronics design.

Do you know any JFET that's really low noise? So far I have:

SST4391 thru SST4393, 2SK3426, 2SK2593, 2SK2751. Specs for the 2SK parts are the skimpiest. Some are claimed "low noise figure" and then there isbn't even a noise figure in the data.

Thing is, we may have to abandon the really low noise BJT stuff because RF from cell phones and the like gets into the first b-e junction. It's a design where shielding is rather difficult.

--
Regards, Joerg

http://www.analogconsultants.com/

That's not 100Hz but 100MHz:

The graph doesn't go below 1MHz. Also, this part is non-stock at Digikey, only TO92, might be going away some day. But it was my staple for RF amps when I was a kid. I've squirreled away a few so I can repair the amps I built ;-)

--
Regards, Joerg

http://www.analogconsultants.com/

On a sunny day (Tue, 13 Nov 2007 19:57:51 GMT) it happened Joerg wrote in :

You are right, I was wrong. Is it wrong to suppose it extends down to 100Hz ?

Try it perhaps. The datasheet says: RF, LF and also DC amps.

It usually start going up again below a kHz.

They often say that. However, I have a feeling that it might not be around for much longer, isn't available in SMT and from my wild days I remember that BJTs were better at audio so it couldn't have been very good. I only used the BF245 for scope pre-amps and such where I needed bandwidth.

--
Regards, Joerg

http://www.analogconsultants.com/

How low do you want? How much do you plan to use? The lowest spec'd are the Interfet IF3601/3602 but I guess you wouldn't want to buy them.

Some Jfets are claimed low noise, 1db or so NF. Then you look the fine prints: hey wait, 3dB with a 100K source resistance!

Oh, uh. I have a real very nice trick for this but unfortunately I can't... blah... Maybe you can figure that out.

--
Thanks,
Fred.

Didi a écrit :

Higher gm: theoretical equivalent noise resistance is 2/(3.gm)

This is only true above the 1/f corner frequency.

--
Thanks,
Fred.

That word doesn't mean what you think it means:

5nV/rtHz at 20Hz would be nice.

Looks nice, although not much of a datasheet. In case anyone else want to take a peek (their server did not find the part...):

A mainstream manufacturer would be nice ;-)

Yep. At the end of the day you just have to buy a bunch of FETs and try it. Just like you had done with BJT.

I've built a few mast preamps with nice (linearly behaving) protection and so on. But if a cell phone is only a foot away and transmitting at full blast that's a challenge.

--
Regards, Joerg

http://www.analogconsultants.com/

datasheet.

Hi,

I have thousands in stock, TO92. Write me if you need one, I have enough. Only mail costs is enough if you only need one or two (for hobbyists, not for businesses: they will actually have to pay..).

Pieter

Hi,

I asked the same two weeks ago. And Jon Elson replied: InterFet IF9030. The datasheet looks very good.

What are you building?

Pieter

Hmm, interesting. At most companies I dealt with it is used that way though (mostly central/western US, no idea how it is elsewhere).

--
Regards, Joerg

http://www.analogconsultants.com/

Which package are you ready for? If you can go discrete, look at Toshiba. I think they still make some of 2SK170/367/369/370/371

--
Thanks,
Fred.