Discrete Junction-FETs

[schmitt trigger]

This is more a nostalgia question:

How long has it been where you have seen a recent-design electronic circuit, which still uses discrete JFETs?

I've been noticing a dwindling supply of JFETs, and it must be because of fewer and fewer circuit applications.

I remember in the early 1970s the JFET were poised to be a replacement for bipolars where high input impedance, improved frequency response and its depletion characteristics were a plus.
However, certain critical parameters like IDss and VGSoff had vey wide limits, and DC biasing wasn't that straightforward.

That an the rise of the discrete Mosfet, essentially limited its usefulness. More so that some companies are offering depletion-mode Mosfets.

What do you think? Will they eventually become obsolete?

[Kleinstein]

In a few application discrete JFETs (and JFET pairs) are still used. So if you need low noise (especially low frequency) at high impedance JFETs are still about the best way to go. So most scopes will use discrete JFETs at there input (a few may used special chips).  As AM radio is loosing ground, there may be lower demand for those FETs made for high quality AM radios.

There are even new JFET type coming out, not just the SOT23 replacements for old TO92 types.
It is only the old TO92 case types get rare.

It is only a few cases (e.g. current limiting) that depletion mode MOSFETs can replace a JFET. They may be similar at first glance, but there strength is high voltage (e.g. 600 V), AFAIK never seen with silicon JFETs.
JFETs used for analog switching tend to be replaced with CMOS chips however.

Another competing part are increasingly better OPs with JFET inputs. It takes some effort to beat an OPA140 or OPA827 with discrete parts.

[David Hess]

It is only a few cases (e.g. current limiting) that depletion mode MOSFETs can replace a JFET. They may be similar at first glance, but there strength is high voltage (e.g. 600 V), AFAIK never seen with silicon JFETs.

JFETs were available up to at least 300 volts as shown in the attachment below.  Teledyn used them in their FETRON solid state vacuum tube replacements and I gather they were popular in military applications.  Some power JFETs were also produced for the commercial market before vertical MOSFETs became available but they were never very practical.

I would have no complaint about depletion mode MOSFETs except that they are very expensive compared to JFETs and small signal devices are basically unavailable; I would like to see some with Idss in the milliamp range.  Of course MOSFETs still have less precision and more noise than JFETs.

Another competing part are increasingly better OPs with JFET inputs. It takes some effort to beat an OPA140 or OPA827 with discrete parts.

Beating them usually just means using a discrete JFET input stage.  Analog Devices published die shots of their equivalent low noise JFET input operational amplifiers showing just how large the input JFETs were to achieve that kind (4nV/SqrtHz) of noise level; they take up almost half of the silicon die.

[T3sl4co1l]

HV JFETs are once again* available, this time in SiC flavor, for high power switching application.

They're on the boutique side, expensive.  Some are mil-spec, covered in gold, $100 and up.

*Or they were.  I think they're going away, enough to support legacy only.  I think the timeline is, back when SiC was just getting good enough to make transistors out of, they went for the easier junction types first: BJTs and JFETs.  BJTs of course stink for speed -- the carrier lifetime is even longer than in silicon, so you get BJT t_stg and diode t_rr much longer than comparable Si devices.  (Or, well, Si body diodes are (necessarily!) pretty crummy too, and the SiC parts are probably subject to the same limitations.)  These were also cascoded with low voltage parts for speed, or left as-is for direct drive.  Later, MOSFETs were perfected, with easy drive requirements (if at odd voltages).  I think IGBTs are here now too, although I still only see research articles using 10kV SiC IGBTs, nothing actually for sale anywhere.

Tim

[JPortici]

synth guys have been using JFETs as buffer for some time...
as switches to short and discharge the integrating capacitors of relaxation oscillators (to produce a sawtooth wave)
as buffer after OTAs, instead of using (for example) the LM13700 integrated darlington pair

[Alex Nikitin]

JFETs and DFETs (Depletion Mode MOSFETs) are my favorite discrete active components, especially for audio applications. I've designed many circuits using exclusively (or almost exclusively) these FETs (and some MOSFETs for power amplifiers). My last phono stage design (Kora 3T, 2008) used DFETs (with an added JFET for the MC version) . I hope discrete JFETs and DFETs are here to stay  . With some JFETs one can build a very decent voltage reference as well.

Cheers

Alex

[schmitt trigger]

I appreciate all your answers....if I understand correctly, JFETs are the best for certain specialized applications.

I personally have used JFET in designs, in which one can create a simple and elegant circuit function.

But my comment (rant?) is that these applications do not create enough volume for Semiconductor houses to keep producing them. Will they become scarce?

[Kleinstein]

There are small depl. mode MOSFETs available and some are not that expensive (e.g. BSS126 at some 15 to 20 cents). I would guess they have enough of a market niche (input protection, current limiting for LEDs and startup of flyback converters)  to stay.

For the silicon JFETs they still get the best noise figure in some applications. So they are to stay unless there is a better replacement to come up. The TO92 case and similar versions are getting rare but usually there are SOT23 replacements.  The duals in a metal case are already rare and rather expensive, but as they are the best choice in come cases they are likely to stay though maybe with specialized, low volume producers like linearsystems.