Low noise N-JFET ?

The BF862 is very low noise.

Look at the Linear Technology site to use it in conjunction with low noise fast Op-Amps.

The BF682 (IDSSS = 10 .. 25 mA, Yfs = 45 mS typ) seems to come in a surface-mount package only. Some Japanese through-hole counterparts (all Toshiba) are:

A higher transfer admittance Yfs means a changed channel geometry corresponding to a paralleling of FETs with lower Yfs. It translates into smaller noise, as the dominant source of white noise is resistive noise from the channel. On the other hand, this means higher FET capacitances.

Similar to bipolar transistors, FET noise also involves a 1/f component below some corner frequency. However, while the designers of the bipolar MAT02, MAT03, LT1028, and AD797 managed to push this corner down to the 1 .. 10 Hz range, the noise of the above FETs already starts to rise below the 1 .. 10 kHz range. Compare the Toshiba datasheets.

Also note that the LT1028 has a noise peak around 400kHz (see page 9 of the Linear Technology datasheet).

Martin.

I was just going to suggest that one, it ought to be the lowest noise JFET around. However, as Martin pointed out one may have to consider the sub-audio region. NXP's data sheet is, ahem, rather skimpy. Do you (or Martin) know about any study of the BF862 below, say, 20Hz?

Also, does anyone know what happened to the old 2N4417? Can't find it anywhere these days.

I am a bit partial to the Texas THS series. Mostly because many can be used at +/-10V and more, resulting in lots of dynamic range.

My problem is that I not only need the noise from 0 - 1MHz, but also the low-frequency noise from 0 - 10 Hz. And it is unclear to me what is does there.

I tried to simulate it in Multisim, but that gave very strange noise results when I changed the frequency range (went from -110 dB to +33 dB).

Pieter

IME Spice is not very helpful here. I am afraid someone has to actually measure it because Philips probably didn't. If you decide to do that please post.

Hey, aren't you in the Netherlands? I've lived there many years ago and most people I knew had some relative or acquaintance who worked at Philips, or now NXP. Maybe ask someone?

No. The transfer function of low-IDSS BF862's closely corresponds to that of high-IDSS 2SK170's, so at the same drain current the white noise will be the same. The 2SK369 has a higher Yfs and its white noise will be lower.

It is remarkable, though, that the BF862 achieves the same Yfs at a third of the capacitances of the 2SK170.

That's why I referred to the Toshiba datasheets. The BF861 too seems to be a close relative to the BF862. It's noise is specified down to

Was this one much different from the 2N4416 = PN4416 ?

(Correction to my first post: the 2SK370 is nothing but a 2SK170 in a miniturised through-hole package, electrically they are the same. And please read IDSS for IDSSS.)

Martin.

Uwe Bonnes wrote: B.t.w.: Congratulations to the NXP web site developper :-(

Their search doesn't find the BF862.

Argh...

-- Uwe Bonnes snipped-for-privacy@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt

--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Hi,

Yes, I live in the Netherlands. I have my own company now, and have in the past also worked for several companies including Philips in Eindhoven. NXP is former Philips Semiconductors in Nijmegen.

I did not find information on noise from 0 - 10 Hz yet. Seems that there are only a few specialists.

Pieter

Ah, finally another one with entrepreneurial spirit in Europe. But I wouldn't use "dit werkt ook niet" in the email address :-)

Randstad area? Or out in the country? I used to live in Zuid Limburg, by many Dutch people not considered to be part of NL because they couldn't understand us. And they also didn't get most of the Urbanus jokes while we were rolling on the floor.

NXP's marketing is a mystery to me. Sometimes I wonder whether they still have one. I found the BF862 by pure coincidence. Never got any email announcement about it, just stuff like all those boring protector diodes they make. Yawn. Yet this FET could really take off in the marketplace if they'd advertise it _and_ provide a little more data or at least cram some more info into an app note. And they should get a better web design time but that's a whole 'nother story.

Fig 14:

Yuck, that's pretty bad. Puzzling: The BF861 is spec'd at 1.5nv/rtHz at

So, is it "aggressive marketing" or is it really that much better? The graph in the BF861 datasheet somehow tells another story.

I'll stay with the BCX70K for now where it doesn't have to be really hi-Z. That's really low noise and much cheaper.

Nope, IIRC it is (was?) a small JFET.

Well, but those are all not mainstream in the US.

That's what I am saying all along. IMHO they ought to fire the web team and maybe some managers. I've tried to help (seriously) but it seems they don't want outside help. Too bad, Philips used to be such a good company. Heck, maybe I should write Mr.Kravis at some point. But only because I have some more emotional ties to Philips. Their EE20 kit is what brought me to electronics.

Helps perfectly against spam.

South, between Tilburg and eindhoven. Near the Belgian border, so we do understand Urbanus.

I know, worked for Philips a lot. And know how it works. Great technical knowledge. Also many people in marketing. But somewhere things are missed. Maybe that webdesign changed when they left Philips. Though the same people still work there. They are oriented towards companies, not endusers.

I will check the BF862 datasheet.

My problem lies mostly in the 0-10Hz range, parts do get get specified there.

Pieter

That would be south enough :-)

Orientation towards companies only is IMHO a big mistake. It misses the next generation of sales. At least in the US it's the little guys who will be the high rollers 10-20 years from now. Many start in a garage just like Hewlett and Packard did. Imagine someone would have refused to sell them stuff.

Looks like web design has been outsourced to some extent. It doesn't work.

Got to go through Digikey because NXP's web site current can't find it via the search box.

Unfortunatley many parts such as the BF862 aren't spec'd for noise down there.

Le Thu, 01 Nov 2007 22:40:48 +0000, Joerg a écrit:

I never understood your rants about philips/nxp website. It's not that bad. Just plug BF862 in the top right search box and it comes out immediately.

You can have a grasp of what it does by looking at the BF861 fig. 14 Not the best for low frequency.

It is bad and I'm definitely not the only one saying that. Old saying here in the US: A product is only as good as potential customers think it is.

Well, today it does, maybe somebody reads these posts ;-)

Yesterday it clearly did not, and I've tried multiple times. So has Uwe Bonnes here in this thread. That NXP site isn't reliable IMHO.

From the NF figure in the table the parts seemt o be much different though. But yes, the trend down there does not look good at all.

I had the same problem. Google helped and found it on the website. Notice that the datahseet still says "Philips".

So a temporary problem.

Pieter

Things is, they've got a lot of those. And some not so temporary ones.

For example, right now when I go onto their homepage the whole center is blank. I cannot get into product categories anymore. Brand new Firefox browser, so that's not the issue. Not good.

Hi,

It works over here. So I checked their code. I think it is a bad idea to make code system dependent. This is a part of NXP system detection (I changed formatting so I could read it), it checks for M$, opera, mac:

var bCompatible = false; if ( document.all && !navigator.userAgent.toLowerCase().match(/opera/) && !navigator.appVersion.match(/mac/i) ) { try { bCompatible = true && new ActiveXObject('Microsoft.XMLHTTP'); } catch(e){} } else { bCompatible = window.controllers && (parseInt(navigator.productSub)>20031001) && new XMLHttpRequest(); } if (!bCompatible) { location.href = "/indexnc.html"; //INDEX Non Compatible }

I wonder what system the developer tests it on? Computersystems that are protected may not give out enough browser information. Do you use firewll, anonymizer, privacy protector etc?

Pieter

Yes, a hardware firewall. And that's not going to come down just because the web designers at NXP lack competence. It wasn't much better in the Philips days. Back then they did not understand what a latency is. It's almost 200msec from Europe to the US, mostly it hung up when they did their stupid stock quote lookup. But with the change to NXP it became a whole lot worse. With older browsers the NXP site doesn't work at all, you get a blank screen all the time. And I can't possibly tell my clients "Oh, let's just install another browser...". We just move on to TI, National, ON, ST or other vendors.

Anyhow, I just pretty much quit using their parts. So have lots of others out here. There are enough competitors with more competence in the dissemination of info about their products.