U310 JFET - anything more modern?

[G6Tanuki]

Historically [that is, for 40 years or so] I have used the Siliconix U310 JFET in a lot of applications - oscillators, mixers, RF-amps - up to and including 144MHz; replacing the front-end 2N3819 transistors in a high-band Pye Westminster with some U310 gives a significant improvement in gain [remember to adjust the bias].

I got wondering - as my supply of genuine U310s is now precisely one, whether there is a modern/better alternative to he 310, given that it must be at least 50 years old as a design.

[Bazz4CQJ]

Does the 2N4416 appeal, at least technically? The price of them, from reputable suppliers, might be off-putting.

B

[Restoration73]

These are inexpensive;

https://cpc.farnell.com/on-semicondu...12?ost=sc15912

[Bazz4CQJ]

Originally Posted by Restoration73

These are inexpensive;

https://cpc.farnell.com/on-semicondu...12?ost=sc15912

Those are indeed well priced. I don't have an account at CPC and RS has discontinued them. Cricklewood has them at £3 ea.

B

[trsomian]

One does not need an account at CPC, order online and pay by card. There may be a postage charge for small orders. I wouldn't call them more more modern; they have been around for many years. None the worse for that though.

[G0HZU_JMR]

The U310 is a classic JFET for sure but it is a process 92 JFET so it is very different to the process 50 JFETs like the 2N4416 and the BF256B. The internal die is very different and the Idss and gm of the U310 is much higher for example.

There is the modern MMBFJ310 and the MMBFU310 in sot-23 and these are also process 92. If you don't mind SMD I think now is the time to stock up on MMBFJ310 and MMBFU310 if you can still find any. I don't think they are obsolete but stocks have plummeted since the effects of the pandemic.

In many databooks the J310 and U310 share the same datasheet but I have typically found the U310 to have slightly higher transconductance. I've still got lots of U310s from the 1990s here. I stocked up on them when they started becoming scarce and expensive. All of mine were from Farnell with part 352-251. I'm now down to the last 4 bags of them and I've used them for all kinds of applications over the last 30 years or so.

There are modern SMD alternatives that are still available like the CPH3910 JFET from On Semi but this device has a very high transconductance. It therefore isn't really a direct replacement as such and it might not work well at higher frequencies. This part typically costs about 80p. It should offer low noise and high gain but it might be a bit twitchy in terms of stability unless are is taken with the PCB layout.

[Terry_VK5TM]

Originally Posted by G0HZU_JMR

There is the modern MMBFJ310 and the MMBFU310 in sot-23 and these are also process 92. If you don't mind SMD I think now is the time to stock up on MMBFJ310 and MMBFU310 if you can still find any. I don't think they are obsolete but stocks have plummeted since the effects of the pandemic.

Other than current supply shortages - the MMBFJ310 is in last shipment but there are other variations (eg MMBTJ310L) still in active production so it will be around for some time to come.

MMBTU310 are still in active production.

Plus, they can always be fitted to little adapter boards to make them pseudo through hole components.

[G0HZU_JMR]

Thanks. The last time I bought any I got them from Mouser. Looking at the bag they sent me 125 when I'm sure I only ordered 100! These are marked as MMBFJ310LT3G and they are made by On Semi. I've also got the (older?) MMBFJ310 version here from Farnell. I think they are still made by On Semi. I'm not sure what the difference is when the L is added but it might be to do with them being suitable for lead free soldering. The labels have faded to white on the Farnell bags as they always seem to do but I have written the details on each bag. I also have some of the even older SST310 parts made by Vishay and I think these are long obsolete.

I've always assumed they are all made to the same process 92 internally but maybe the metal finish on the pads has changed over the years to suit modern soldering processes.

[Terry_VK5TM]

I think the difference is only in the removal of "supposedly hazardous substances", hence the different suffixes.

I know I looked at it but didn't take a lot of notice - fairly sure still the same process.

Just as well they are still around, last time I ordered several months ago I got 500 and it looks like I'm going to have to order more shortly.

Just for reference, if you go on the Onsemi site and put 'N channel RF FET' in the search box (and select Products just above the search results), about the 2nd page of results are a few of the old favourites still in production as smd FETs.

[G0HZU_JMR]

Whilst home in my lunch break I've had a quick look back through the 1986 Siliconix databook for the J310 and U310. Both JFETs are process 92 but they are listed on different pages.

The typical transconductance in common gate for the U310 is given as 17,000umho at 10Vds 10mA Id.
The typical transconductance in common gate for the J310 is given as 12,000umho at 10Vds 10mA Id.

This does agree with what I've seen over the years. All of my U310 parts here are made by Siliconix. Looking back through my library of S2p data I have measured the U310 on my VNA in common gate and at 10V Id and 10mA Id. The real part of the Rp resistance looking into the source in common gate mode was a flat 65 ohms from LF through to about 280MHz. This means gm should be 1/65 = 0.0154mho or 15,400umho. This is very close to the typical performance on the datasheet. I also have VNA derived s2p data for the SMD MMBFJ310(L) part at 10Vd and 10mA Id and this showed 83 ohms Rp at the source. This equates to a gm of 12,000uhmo at this agrees with the datasheet. However I did choose this part after measuring several on a jig. I chose the one closest to the datasheet data so maybe the result was no surprise. However, there wasn't much spread between devices.

I guess the conclusion here is that all of the modern SMD 310 alternatives are likely to produce a slightly lower gm than the classic old U310 in the metal can. So prepare to be a bit disappointed if a typical preamp circuit produces slightly less gain with the modern SMD parts. There won't be that much difference though.

[x72]

These people do some replacements for many JFET parts discontinued by others:
https://www.linearsystems.com/product.html

The UK distributor appears to only deal with bare die and Digikey don't seem to have much stock. However, this company had some in stock https://www.nacsemi.com/products/linear%20systems/

[G0HZU_JMR]

I ordered some CPH3910 JFETs from Farnell and measured the s-parameters on a VNA a few days ago.

I had a go at doing a practical LNA design using the CPH3910 s-parameters today. The aim was to make a low noise common gate preamp for the 88-108MHz band. See below for the simulation results based on the s-parameters at 10Vds and 10mA. The result was quite good in terms of passband gain and port match across 88-108MHz. I'd expect it to achieve a noise figure of about 3dB but with an L match at the input this could probably achieve 2dB noise figure with a bit of input mismatch.

I'd expect the OIP3 to be better than +20dBm so this would be a cheap LNA for the 88-108MHz FM band with 13dB gain and fairly good noise and signal handling performance from just 10mA.

Here's the On Semi datasheet for the CPH3910.

https://www.onsemi.com/pdf/datasheet/ena1965-d.pdf

My s-parameter measurements using the VNA show that it can still produce enough power gain to go unstable at well over 2GHz so some care and understanding is needed with the circuit design and PCB layout.

The performance below is a bit better than I would expect to see from a genuine Siliconix U310 JFET in the metal package. So in this sense it is possible to find parts that can perform at least as good as the old U310. The circuit simulation below would make a good reference design for a VHF FM preamp. It would still perform well at just 5mA although I think the noise figure would degrade slightly. it might be 0.5dB worse for example.

[G0HZU_JMR]

There's no problem with price and availability for the CPH3910 from Farnell. The CPH3910 costs about 70p and they have nearly 2000 of them in stock.

https://uk.farnell.com/on-semiconduc...-23/dp/2724389

This is where I bought mine from so I should have genuine CPH3910 parts here. I'll probably buy a few more the next time I order anything from Farnell.

[G0HZU_JMR]

I just built the VHF FM LNA circuit this evening but I tweaked it for 50 ohm ports rather than the expected 75 ohm. This makes it easier for me to measure as all my test gear here has 50 ohm ports.

See below for the simulation (based on my s-parameter data of the CPH3910 and some simple circuit models for the R, L and C components) and this is compared to a measurement of the built circuit using the VNA. I think if I had measured the inductors on the VNA and I had modelled the circuit strays better in the simulation the agreement would have been closer. Even so, there will be some spread due to component tolerances.

This does at least show that I can measure s-parameters of SMD components quite well!

I'll have a go at measuring the noise figure and the signal handling tomorrow.

This looks to be a useful preamp circuit and is designed to cover the whole 88-108MHz FM band. With 75R terminations I think the noise figure will be close to 2dB across the whole band. I don't think the U310 JFET can match this overall level of performance for gain and noise figure across the whole of 88-108MHz.

[G0HZU_JMR]

I managed to measure the noise figure of the CPH3910 88-108MHz LNA this evening. To ensure a reliable measurement I fit the LNA inside a screened enclosure to minimise noise pickup. This enclosure works really well and is essential for measurements like this.

As predicted the noise figure was about 2.5dB when driven from a 50R source. I also fitted a simple (low loss) L match to step up the source impedance to 75R and the noise figure improved to about 2dB.

The noise figure plots below are taken using an Agilent 346A noise source and the noise figure option fitted to my spectrum analyser. The uncertainty of the measurement below should be quite low so these are realistic measurements in my opinion. The first plot is for a 50R measurement and the second is for 75R although I didn't try an optimise/correct the response for the 75R measurement. A 2dB noise figure is good enough for this band I think.

I measured the signal handling yesterday and the OIP3 (output third order intercept point) was about +20dBm across the whole band.

[Terry_VK5TM]

You mentioned the "classic old U310 in the metal can" further back in this thread and per chance I checked on it at one of my suppliers while looking for something else - they have about 1100 in stock but at $24+ each (about £12), I think I'll pass on those.

[Radio1950]

Thanks Jeremy

I will tuck that away for future use.
And so cheap.
.

[G0HZU_JMR]

There is also the CPH6904 dual JFET and this is effectively two CPH3910 devices in a tiny SMD package. These are probably quite well matched so they could be paralleled together although I'm not sure I'd want to do this for mass production.

https://www.onsemi.com/pdf/datasheet/ena1966-d.pdf

I tried paralleling two CPH3910 devices in the simulation and the gain went up and I'd expect the noise figure to be slightly lower. It's probably possible to get another 1.0 to 1.5dB gain in the LNA circuit with two matched devices in parallel.

[Terry_VK5TM]

If the Drain & Source were interchangeable like the J/U310, that little chip might have made a reasonable replacement for a dual gate Fet or even be used in a cascode amp arrangement.

[G0HZU_JMR]

I've not tried the CPH3910 in a cascode circuit but I do intend to measure it on the VNA in common source at various bias points. So I could simulate the small signal behaviour of the cascode quite accurately as I've already measured it in common gate mode.

The BF999 is a single gate MOSFET and this might be an interesting alternative for some applications. I've had some of these here for years but never got around to playing with them.

https://www.infineon.com/dgdl/Infine...14936ed6171043

I think the BF999 is now obsolete though. Mouser still have stock but this is going to vanish fairly quickly I think. They look to be restocking in the New Year so there may still be time to buy a few.