2
\$\begingroup\$

A classic way to protect opamps from excessive input current and still preserve a low source resistance (for noise reasons) is to use dual JFETs in combination with rail-clamps internal to the op-amp.

schematic

simulate this circuit – Schematic created using CircuitLab

With JFETs all slowly going EOL, I wonder how to do this without JFETs. Ideally, how do it better than JFETs.

This means:

  1. Negligible resistance in a non-overvoltage case. Say maximum ~100 ohms.
  2. Allowing no more than e.g. ~5mA to pass in either direction.
  3. No leakage to rails, to preserve the low input current rating of FET amps. E.g. maximum rail leakage of 10 pA over temperature.
\$\endgroup\$
1
  • 1
    \$\begingroup\$ I can see that suppliers don't want to produce a zoo of almost obsolete devices. However, some JFETs will remain. For example, MMBFJ113 is widely available. \$\endgroup\$
    – Janka
    May 26, 2021 at 13:10

2 Answers 2

3
\$\begingroup\$

You can probably use depletion-mode MOSFETs, though 5mA may require some added resistance (to have more negative gate voltage for a given current), that's pretty low for Idss.

Edit: A pair of LND150/250 would seem to meet your requirements. Maybe too high resistance.

I'm pretty sure JFETs are not going to go completely out of use, they're still used by the boatload in electret microphones. There are also boutique suppliers of them, if cost isn't a big issue, it's very difficult to get the combination of low noise and high input impedance any other way for some kinds of instrumentation. For whatever reason(s), JFET-input op-amps do not have competitive performance. I snapped up a reel of BF862s for that reason (0.8nV/rt-Hz noise)

\$\endgroup\$
5
  • \$\begingroup\$ JFETs in general are still around, especially for low noise inputs, where they are run with a lot of current. But for the abovementioned protection purpose, one requires a low IDSS part. The 2SK2145 seems to be last Dual N JFET with substantially sub 10mA IDSS that fits this role and has good availability. Of course one could migrate to 2 single JFETs.. but for how long ? \$\endgroup\$
    – tobalt
    May 19, 2021 at 13:25
  • \$\begingroup\$ Usually the MOSFETs pinch off only at 100s of mA, but the LND150 is even on the other extreme (already too resistive at ~1kOhm). You mentioned modifying the above circuit with added resistance to make them pinch off faster. Will that allow trading higher on-resistance for lower saturation current in a useful way? \$\endgroup\$
    – tobalt
    May 19, 2021 at 13:39
  • \$\begingroup\$ Adding 100 ohms (per MOSFET) would give you -500mV to help turn the MOSFET off. There may be some higher voltage ones that are more in the middle Idss-wise. I recall using some with a 300-600V rating. \$\endgroup\$ May 19, 2021 at 13:46
  • 1
    \$\begingroup\$ I have looked up dozens of datasheets of depletion MOSFETs now and simulated for some of them, but unfortunately the product of saturation current and on-resistance is always much worse than for JFET. Using resistors does not improve this metric either. I guess it's time to stock up on 2sk2145's then. \$\endgroup\$
    – tobalt
    May 19, 2021 at 16:23
  • \$\begingroup\$ Good information to have, thanks. \$\endgroup\$ May 19, 2021 at 16:35
3
\$\begingroup\$

Thanks to Spehro I experimented a bit more with MOSFET. While the selection of depletion MOSFET is quite limited, there are of course tons of enhancement MOSFET available. I adopted the two-transistor current limiter circuit to create this:

enter image description here

Looks more complex than it is. It is two tiny logic MOSFET (M2, M4) with very low gate threshold voltage (e.g. IRF7301) and two main blocking MOSFET (M1, M3) which also can be small, but should be able to block the signal fault voltage and handle the power of the clamping current times fault voltage. Just like the original JFET circuit, this limiter is completely serial and thus creates no signal leakage current. Resistance in the unclamped state is basically R1.

The only part that is a little tricky is the floating 3 V source. As this only needs to supply a couple of µA max, a blue photodiode in PV mode would be ideal. There are also typ. 8V output photovoltaic MOSFET drivers to supply this.

Performance of this circuit is indeed even better than the classic JFET limiter. For the same clamp current target, the regular resistance is 2-3 times lower. Another benefit is the much smaller component variations in MOSFET compared to JFET.

\$\endgroup\$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.