JFETs and active crossovers

[nic]

Most of my OnSemi BF256B come in at around 9mA . Minimum 6mA . None over 10mA . CPC has them on back order . 7p a pop when buying 100

cheers

Nic

[simon]

Thanks nic, good to know. Tighter than J113s. I should be able to do without bias resistors then, though I read last night that a low value resistor, 47R say, below the lower JFETs source was prudent for current sharing. Not sure if that's really needed though if they're well matched.

Baccy tins look to be £6 a pop, so might use some antistatic bags instead, like these - 100 for the same price.

https://www.staticsafe.co.uk/products/e ... ack-100-2/

[Morgan Jones]

You need to search carefully for baccy tins as the prices are very variable. It's probably five years since I bought any, but I paid about 50p each for 1oz and £1 each for 2oz, buying a hundred at a time.

Despite what is now claimed, J113 was designed as a switching transistor. Measuring J112 (same series), 3rd harmonic distortion was 10dB higher than BF256B at all levels below clipping.

[simon]

This is what a bag of 1000 BF256B looks like. 2SK209 and 2SK2145 below. I am VERY pleased I didn't buy a 1000 of either of them.
.

20260413_111032.jpg (172.63 KiB) Viewed 5212 times

[Morgan Jones]

The nice thing about having 1000 of a device is that you get an awful lot of close pairs. 1000 2SK2145 doesn't look very impressive at all, and when coiled fits easily in a 2oz baccy tin.

[simon]

I'm very interested to see how close the matches are compared to J113s, should be much closer I'd think.

Goodness knows how long it's going to take me to measure them all. A couple of years I should think. The J113s took quite a while for the readings to settle down as each one warmed up.

[Nick]

The J113s took quite a while for the readings to settle down as each one warmed up.

Every time I build one of the phono stages with jfets at the front, I have to remind myself there isn't a fault with the DC output wandering around a few 10's of millivolts while I am testing it until I put the top on the box or stop waving my hands over the PCB while checking voltages. Drafts in the air are enough to change the LSK170's current. So good luck with trying to match them. I found the only sane way I could match them is with a populated and working phono board, just with sockets in for the jfet, and test them in situ. I find that the anode voltage of the ecc88 with the jfet in its cathode and a stabilized B+ voltage is the best way of categorizing them WRT DC current which seems to translate very closely to gain.

[simon]

With the J113s I applied voltage and watched the Idss reduce, quickly at first then more slowly, until it was 2 or 3 seconds between mA change. I figured this was as accurate as I'll ever need to be, or indeed could be.

Interestingly I tried holding the JFET body with my fingers thinking that would speed up the warming process and thus the time taken per reading, but it actually slowed the change down. Removing my fingers speeded the change up again. Which suggested my fingers acted as a heatsink not a blanket.

[Morgan Jones]

Goodness knows how long it's going to take me to measure them all. A couple of years I should think. The J113s took quite a while for the readings to settle down as each one warmed up.

It took me about a week to measure 1000 BF256B. I think I did a couple of two hour stints each day. It's not so much a question of waiting until they've warmed up as measuring a constant time after applying power. My V3 FET VTO tester (almost finished, a few wires to be fitted) will measure 100ms after application of power, triggering the DMM for a single measurement.

Quiet JFETs can be matched to 1mV with no trouble at all. My leaded VTO tester has an earthed screening can that goes over the DUTs, but its main benefit is that it prevents air draughts from upsetting measurements. Hot fingers are the other issue, so you need to allow about a minute for the DUT to stabilise temperature after putting it in the ZIF socket.

Sixty LND150N3 (500V 30mA) have just arrived from CPC in an ordinary polythene bag. (Should have been antistatic for DMOS!) Average VGS for IDS = 20nA was -1.687V with an impressively low standard deviation of 12mV. Technically, they're depletion FETs, but you have to operate them in enhancement mode if you want any current, so they're mostly useful as input protection for instruments (see H&H Ed. 3).

[simon]

So, I'm not sure what to do for the best. With the J113s I simply recorded the value of Idss to 3 decimal places and stuck the JFET on lengths of masking tape, so I could make matches to 3DP. This of course assumes that my measurements were accurate enough for 4 significant figures. Which is probably a stretch, as Nick suggests above.

I'm not sure how practical that is with 1000 of them to measure though. Even if I bin to 0.01mA that could be 400 containers if my BF256Bs are similar to nic's. If I bin to 0.1mA that's 40. But it seems a waste to only record to 1 DP if I've measured to 3. And I wouldn't want to have to measure a bin again to get the best matches.

Then again, I think I've read somewhere Nelson Pass saying no JFET needs to be better matched than 0.1mA. Better see if I can find that again to be sure I'm not on the wrong path...

Edit - nope, according to Google search it's 1mA. Bum.

[Morgan Jones]

If there's one thing I hate it's sticky tape (especially insulating tape) and electronics. It always leaves a gungy residue...

I found that there's quite a spread in VGS across BF256B, so even a not-terribly-accurate measurement is useful. Practical storage after measurement is an issue, though. If you don't like the baccy tin approach, then lobbing them in a square of kitchen foil will do fine. I bought forty "morn, noon, tea, night" pill containers also labelled with days of the week, giving 28 individual (not anti-static) containers per tray. That wasn't a good solution (apart from costing £40, all up) and binning into ranges is much better. If necessary, you can always measure from a bin to find a tighter match. But it's very rarely necessary. I consider VGS to within 100mV to be perfectly close enough to be considered matched unless I'm being really fussy. In short, rough binning is fast and can always be improved later. It's the way to go.

It sounds as if you're measuring IDSS rather than VGS at an imposed IDS. Either method allows similar devices to be binned, but measuring VGS gives a result that is more easily translated to circuit design, although it's harder to make the measurement.

Is there any possibility that the forum software could allow for subscript and superscript?

[simon]

I know! Even resistors and caps on tapes have gunk at the ends. At least they can usually be cut off.

I haven't done a lot of looking but even 1oz tins were a couple of quid each. So I'm still mulling.

I'm using a Stompville JFET tester, so measuring Idss and Vgs (apologies for the lower case, my brain works better in a spatial way). But every reference I've seen refers to matching Idss, so I don't really know when I'd use Vgs

[jack]

I used old business card box I collected at a previous job. The PAs had 100s of these translucent plastic boxes and they are simply ideal for small stuff, especially when dismantling things and you need to keep small bits carefully...

[Morgan Jones]

But every reference I've seen refers to matching Idss, so I don't really know when I'd use Vgs

Philips binned their devices by IDSS (BF256A, BF256B, BF256C). Really, it should be IDS(S) with all but the I as subscript and the brackets should be there because it's the current between drain and source (at saturation). But the brackets get lost. Similarly, because it's not a small signal parameter, it shouldn't be lower case. You'd be amazed how difficult it is to be consistent about that one.

Go for the kitchen foil solution if you can't find baccy tins at a sensible price. To be honest a JFET in a country as damp as the UK isn't exposed to much of a static hazard. Unless, that is, you have the wrong sort of carpet. Many, many years ago there was a works canteen with the queuing lines before the counter roped off by those plush maroon ropes with brass hooks that went into upturned trumpets. We soon learned not to touch the trumpets as you would get a shock.

[simon]

The only time I get a belt these days is when getting out of the car. Doesn't happen often so always takes me by surprise when I do.