Hi guys. I fully understand the idea of a n-channel JFET (2SK170) constant current source with just the JFET and a source resistor. The resistor provides negative feedback pinning the current to a constant value. I've designed them and used them for years without a problem. I've been selecting on the breadboard resistor values for specific JFETs to give a chosen mildly accurate current. On a whim, it seems like the first time I can remember trying it, I've just tried to set up a p-channel version (2SJ74) as a mirror image of the n-channel and to my surprise it isn't working. I'm getting no Ids current at all. This now has me questioning my understanding of JFETs. Is a p-type constant current source feasible in this role, and if so is it just a mirror image as I had imagined?

 

Please post schematic EXACTLY as you wired it, including the JFET part number.
I have used p-chan JFETs several times in that role and always they work just fine.

 

Ok, here they are. The left hand works fine as is. The right hand has had R2 varied and the supply voltage tried from -6V to -15V without any signs of source current, (measured as voltage across the resistor). The supply is definitely as shown, -ve to the p-type. I have tried a couple of JFETs in the role and all do the same thing.

 

Should work.
What are the measured voltages at the three JFETs terminals?

 

Exactly, it should work as shown.
Let me ask the obvious question; Are you sure the p-chan device is legit? Did you actually acquired it from a reputable vendor?

I am asking this not to be fastidious, I am asking because in another forum a member had exactly the same problem. It turns out it was a re-labeled PNP transistor!!

 

btw, why do you need P fet? current limiter is a 2-wire device, you can use either type. i sure prefer to reduce diversity of parts (fewer to stock in case there is a need to do repair).

 

It's to match an existing PCB Panic Mode. There are a pair of circuits which have been designed to be fully complementary. Of course it could be replaced with a n-channel device, there is already one on the PCB working, but the layout is already there as a p-channel. I can maybe cut and shut the tracks to a n-channel type.

The voltages are as they should be at 0V or -6V except for the only other node Crutschow. The resistor source connection is locked at -6V. There is simply no current at all being passed.

Your point about the reliability of the devices is a good one Schmitt Trigger, thanks for the heads up about that previous problem. I don't know the origin of the devices I am using. I'll check out the idea of it being a BJT. Sounds a bit worrying and I would not have suspected that in a month of Sundays. Devices from dodgy sources out of spec I can understand but a completely different device relabelled? I think I might need to source a supply of these or an equivalent from a reputable source for any final version. I'll report back with what I find.

 

The voltages are as they should be at 0V or -6V except for the only other node Crutschow.

Is the only other node pin 3?
If so, what is its voltage?

 

As I said, "The voltages are as they should be at 0V or -6V except for the only other node Crutschow. The resistor source connection is locked at -6V."

So, well, yes, the only other node is pin 3, and that one is the resistor source connection which is locked at -6V. Those two setups are not part of a common circuit. They are tests done independently. The left one based around the n-channel 2SK170 works perfectly. The right hand one based on the p-channel 2SJ74 is separate to that and doesn't work.

I can now confirm that this is due to the fact that the 2SJ74 is actually a rebranded BJT and not a JFET at all. You were spot on Schmitt Trigger, it is that same problem reported elsewhere. My little tester showed all my examples as BJTs immediately. I really should have checked before! I have now looked for an alternative source of my own, (I did inherit these), but boy are p-channel devices thin on the ground nowadays. Anyone got any suggestions as to a suitable source of them in the UK? At a reasonable price preferably. I am not too worried about the specific 2SJ74 for a simple constant current like this so any decent alternative would do. But a few low noise ones for low level signal circuitry would come in handy.

LSK170s are currently £8.58 each + p&p from Mouser, more from elsewhere, and mostly coming from US. I'm looking at the new n-channel Texas JFE150 devices, (got it right in the end), for crucial noise performance to replace them. I can use my iffy K170s for non-crucial tasks but I won't now assume these are genuine not knowing their true provenance.

 

Relabeling semiconductors, from the lowly transistors all the way to microcontrollers, is quite prevalent from non-franchised vendors, specially coming from the far-east. And it has been so for at least 25+ years, when I found my first counterfeit device.

 

Luckily it has been a cheap lesson for me which I will take into account in future. I've heard of reliability problems in Chinese sellers' semiconductors but hadn't come across this sort of level of dishonesty. Most of the time I'm looking for general purpose devices with which they have never let me down. This time these were pretty specialised as you can see, very low noise JFETs. While my own needs this time round are only for relatively low noise use, in preamp PSUs rather than at the front end of MC head amps for example, to find a totally different device under the part number was a surprise. Mouser do stock a few p-channel types with low noise, though not as low as the original 2SK170/2SJ74 pair. Remember VAT if you are in the UK, they don't add it on until your come to order!

 

If it is any consolation, most of us in this forum have stumbled into this problem.
I’ve seen more insidious examples, for instance microcontrollers with a few defective RAM addresses. Then you scratch your head how come some variables become corrupted.

 

Luckily it has been a cheap lesson for me which I will take into account in future.

If you consider your time troubleshooting problems worth anything, it isn't so cheap.

I've heard of reliability problems in Chinese sellers' semiconductors but hadn't come across this sort of level of dishonesty.

It's widespread. There are places in China that specialize in counterfeit parts. Many things you buy from Amazon, eBay, AliExpress, etc, are suspect. Even legitimate businesses get stung by Chinese manufacturers. The manufacturers start out legit and then start cutting corners to increase their profit margin. If the buyer isn't diligent with quality assurance, they get bit.

I bought a solderless breadboard from a US company that eventually got out of the business because they got hit with substandard product and didn't catch it in time. I bought the board at a closeout store. I contacted the company and the CEO told me the story, but he didn't offer to buy it back...

From SMT Corporation (smtcorp.com):