I am repairing a HP 6023A lab power supply for a friend. The problem was that the power fets had failed. This is what the relevant schematic looks like:

Detail HP6023A PSU

The incoming 120VAC or 240VAC is rectified and the resulting approx 340V (they use a doubler for 120V) feeds the switcher. Here Q3 and Q4 had developed shorts and F1, F2 had blown.

Problem is that IRF440 is a High-Reliability part. MIL std and radiation hardened. The minimum quantity from the distributor is 100 pcs and none of the regular distributors, say Mouser, Newark carry it. It is available from eBay, but I am a bit wary as there are lots of counterfeits around.

The IRF440 is a 500V, 8A N-Channel FET with an Rdson of 0.85ohms. The IRF840A from Vishay has the same parameters or better, with one exception: Eas, maximum pulse avalanche energy. Rest, like diode recovery time is better (5.0 vs 3.5), Qg (38 vs 68) etc and lastly, they are both max 1C/W junction to case, which is surprising as a TO3 part has so much more metal than a TO220. The FETs sit on separate cooling fins and are cooled by a fan.

So, I am tempted to go with the IRF840A as is is cheaper and also readily available. This PSU will be used by a hobbyist and never launched into orbit :)

Is my reasoning sound?

PS. Should F1, F2 be ultrafast, fast or slow-blow? The manual does not say. They are small axial fuses with leads.

The HP6023A has a very endearing feature: A swith on the motherboard isolates the HV supply from the logic supply. You flick that switch and feed mains to two spade connectors. That lets you run the logic without the power part. You then connect a DC lab power supply to the mains input to test in a controlled fashion.

I soldered in two IRF 540 to test and fed in 60VDC. I could verfy that the unit operated properly, ie CC/CV, current limit etc. I am pretty certain that it is just the two FETs and fuses that are faulty.

The service manual also states that they are prone to blowing :)

  • 1
    \$\begingroup\$ The maximum pulse avalanche energy being worse is unlikely to matter, as I'm pretty sure that doesn't apply to anything outside of some very particular uncommon uses of the FET. However, some figures of merit, being better can be a problem (sometimes the Rdson is used to limit current, for instance). This is unlikely to be the case in precision equipment, though. \$\endgroup\$
    – Hearth
    Sep 19, 2019 at 16:23
  • 1
    \$\begingroup\$ The thermal performance being the same is not as surprising as you'd think, also, because the TO-3 package is not exactly what you'd call well-designed. I don't know enough to really get into the details but I believe it was designed without doing any sort of detailed analysis of heat flow (and it definitely predates being able to simulate heat flow, seeing as it predates computers in general). \$\endgroup\$
    – Hearth
    Sep 19, 2019 at 16:25
  • \$\begingroup\$ The Rdson for the two is in fact the same. I thought the same as you initially. Yes, the fets should not see avalanches here, but I once designed an unipolar stepper driver where the kickback from the motor was taken by the body diode. Called for some rather interesting thermal solutions like Keratherm red and using a torque-wrench to tighten down the fets :) \$\endgroup\$
    – AndersG
    Sep 19, 2019 at 16:45

1 Answer 1


This is pretty much the same piece of silicon NTE2393.
Link for Newark

  • 3
    \$\begingroup\$ This is going to need a lot more elaboration to be a suitable answer. Why do you say the NTE2393 is a suitable replacement? Do you have a source for it being "the same piece of silicon"? What does "pretty much" mean? Just giving a link is not considered a suitable answer here on SE. \$\endgroup\$
    – Hearth
    Sep 19, 2019 at 16:44
  • \$\begingroup\$ Agreed and if I go for "much same" then I would rather use the IRF840 that is one tennth the price :) or actually less as there is an extra 25€ surcharge for stuff from the other side of the pond. But you did not know that :) \$\endgroup\$
    – AndersG
    Sep 19, 2019 at 16:50
  • \$\begingroup\$ I said pretty much the same since it has the same characteristics as before mentioned IRF440. You can verify that by looking at datasheets of both components. However, you could also go with IRF840, since the reason for that prone to blowing statement is almost certainly in the supply and not in the IRF440 because they are used in quite a lot of circuits without any problems. You can use Farnell than which is the European website. \$\endgroup\$
    – Jan
    Sep 19, 2019 at 17:15
  • \$\begingroup\$ I think that Hearth objected because of your blanket statement that it had the same specs without being more specific. Yes, I know that Farnell is the same as Newark and yes, they have a surcharge of 25€ for stuff shipped from an US warehouse. \$\endgroup\$
    – AndersG
    Sep 19, 2019 at 17:20
  • 1
    \$\begingroup\$ I did not check their stock sorry about that. However, I would get IRF840 as you originally stated. As far as fuses go I would take fast-blow version since you do not want them to blow super fast in case your power supply does not react instantly, but you also don't want slow-blow fuses in case something goes wrong with the power supply itself. \$\endgroup\$
    – Jan
    Sep 19, 2019 at 18:07

Your Answer

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

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