I'm trying to determine the minimum requirements for a MOSFET that will be used in a op-amp based current source. I've hit Google and this site and looked at multiple pages, but almost always, they are dealing with high voltage and/or current, so I'm not real sure they apply here.

This will be used in the output stage of an impedance analyzer. While I've got op-amps that will easily source the 30 mA(or less) that are required, I'm trying to in essence buffer the op-amp from any capacitance on its output.

The input to the MOSFET is a 3Vpp(or smaller), biased at 2.5V 100Khz(or less) sine wave. The application is a AFE for the AD5933 chip.

Please note, I'm not asking for a specific MOSFET, but for the things I should look for in choosing one. Yes, I do know about the so called linear MOSFETs, but I really think that they would be a massive overkill(and a budget buster, to say the least) in this application.

Just as an example, the SI2300(and I have 30 or 40 of them handy) has my requirements firmly within its SOA chart(which has a decent DC area). Is there anything else I should be looking at on the datasheet that would disqualify it for use here?

In any case, I'm planning on heat sinking the MOSFET however I can. In the above MOSFETs case, I was going to increase the area of the copper pads it is soldered to, and give it copper right below it. If it's a TO-220 or 253, it gets a real heat sink. Etc.

The power supply I'll be using has 5V(10A), 12V(3A) and -12V(800mA) available if that would make a difference.

  • \$\begingroup\$ What is an AFE? \$\endgroup\$
    – The Photon
    Oct 12, 2019 at 15:55
  • \$\begingroup\$ Analog Front End -- AKA A fix up for bad silicon. The AD5933 has a multitude of problems, all 4 of the excitement voltages have different bias, none of which match the input bias, each of them also has a different, but high output impedance(which makes low impedance measurements impossible); without an external freq source it is unable to measure below 1Khz;you really need to have an external transimpedance amp, as the built in one has quirks(you end up running it as a buffer amp). That said, the external interface is not that hard to make and gives it a massive boost in accuracy. \$\endgroup\$
    – GB - AE7OO
    Oct 12, 2019 at 20:37

1 Answer 1


I think I would be looking at low capacitance small MOSFETs like the 2N7000/2N7002 - you don’t want your buffer to add more capacitance than what you are driving. There are others in the BSS series.

Look for low Id parts and fairly high voltage rating (like 30-80V).

A lot of the physically small parts are low voltage high current designs that can switch amperes and have metric tons of capacitance.

  • 1
    \$\begingroup\$ First off I don't have any 2N700X on hand, whereas I do have 20+ AO3400, SI2300, SI2302 available to me. Second is the addition of 50 to 60 pF to the output that major? Or are you talking about the input capacitance, which has a 200pF addition compared to the 2N700X? The op amp I was going to use (AD8692) can drive up to 10nF without the need for a snubber. Even though the AD8692 can supply up to 250mA, I wanted to be able to measure uF caps, hence the need for a buffer of some type. \$\endgroup\$
    – GB - AE7OO
    Oct 12, 2019 at 8:24
  • \$\begingroup\$ It's easier to respond to particulars if a schematic is presented. \$\endgroup\$ Oct 13, 2019 at 0:44
  • \$\begingroup\$ I'm setting it up as a standard opamp mosfet constant current source. I've found a even better MOSFET for this, the STD17N(I had pulled 4 of them from an old motherboard). It has better specs than the above three and it is in the DPAK package. While some of the specs are not as good as the 2N700Xs, it is a lot better than those SOT-23 ones above. And unlike any of them, it is in a package with a 20W dissipation rating that is easily heat sinked. I've ordered a batch of the 2N700Xs;They'll fit the boards I had cut for the SOT23's. I had to make new boards anyway for the DPAK. \$\endgroup\$
    – GB - AE7OO
    Oct 13, 2019 at 3:38

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.