MOSFET high temperature leakage current

[jduncan]

I'm trying to switch between several analog signals using pmosfets, and the lowest leakage current spec I can find is 150uA max Idss at 175C.

Are there parts that do better, or some better approach?

[Benta]

That's pretty much typical for MOSFETs.
Tell us about signal voltages and currents, otherwise we're flying blind.

[jduncan]

Vds ~1V, on-state current about 1mA.

I'm unable to find anything but power MOSFETs, I guess there's not much need for small signal parts with that temp rating.

Another thing I'm unclear about is how the leakage scales with Vds - the datasheets all spec it around max (usually tens of volts), but give the same value whether it's Vdsmax of 12V or 40V

[Benta]

Then why not just use an analog switch? Much easier. (Vishay is a good source)

[jduncan]

I can only find one rated 175C and it's very low stock (and costs $200!).

I figured a mosfet was the poor man's alternative. BTW I don't actually need qualification at 175C, just a reasonable expectation/extrapolation that performance will be acceptable at that temp.

[exe]

Are there any relays for such temps?

[jduncan]

I doubt it, and I'm pretty constrained on space as is.

SO linked me to this guide, which has quite a lot of details: https://assets.nexperia.com/documents/user-manual/Nexperia_document_book_MOSFETGaNFETApplicationHandbook_2020.pdf

Apparently decreasing Vds changes the leakage very little unless it's well below 1V. However their comments imply the datasheet value is very much a worst case number, so I might just need to do some real world testing.

I did see another comment that a positive Vgs of a few hundred mV can drastically reduce the leakage, but haven't been able to corroborate that.

[voltsandjolts]

I've worked in high temperature electronics for quite some time now. There are 125degC rated analog switches in ceramic packages which work at 177degC at least, although whether they would meet your leakage spec I can't say. We used them extensively for synchronous detection.

If your application is short term high temperature, or if you can tolerate a some failures after a few months, then try plastic SOIC versions. Many IC's work way past manufacturer ratings, with varying reliability.

[Zero999]

Vds ~1V, on-state current about 1mA.

I'm unable to find anything but power MOSFETs, I guess there's not much need for small signal parts with that temp rating.

Another thing I'm unclear about is how the leakage scales with Vds - the datasheets all spec it around max (usually tens of volts), but give the same value whether it's Vdsmax of 12V or 40V

The leakage is specified at the maximum voltage rating. It will be  much lower, as you're running it at a tiny fraction of that. My advice is to test it.

[jduncan]

I've worked in high temperature electronics for quite some time now. There are 125degC rated analog switches in ceramic packages which work at 177degC at least, although whether they would meet your leakage spec I can't say. We used them extensively for synchronous detection.

If your application is short term high temperature, or if you can tolerate a some failures after a few months, then try plastic SOIC versions. Many IC's work way past manufacturer ratings, with varying reliability.

Thank you, that is good to know. The application is indeed short term high temperature, and will be running several other parts well outside their rated range, although I've tried to stick to 150C ratings. May need to widen the search.

Do you happen to have any experience using 125C ADCs at elevated temperatures?
I am currently using one with pretty poor specs solely because it's rated 150C, and I'm wondering if I should evaluate others.

[voltsandjolts]

Do you happen to have any experience using 125C ADCs at elevated temperatures?
I am currently using one with pretty poor specs solely because it's rated 150C, and I'm wondering if I should evaluate others.

Well, it depends on your requirements for dynamic range, short & long term drift but yes, there are low temp rated ADCs which work at 150C and beyond. If you have modest requirements the ADC in your microcontroller might do the job. Try to use ratiometric measurements if you can, rather than voltage references.

[exe]

Can you switch the signals outside of the hot zone?

[jduncan]

Unfortunately not, no cold zone available.

However I may have been getting worked up over nothing - I wired up an SOT23 FET I had lying around and even at 175C leakage was too low for my multimeter (below 1uA). I have a feeling Vishay just templated Idssmax @ 175C = 150uA across the datasheets of their entire line.

[voltsandjolts]

I think a lot of these datasheet figures are 'guaranteed' by design/characterisation, and are actually way much higher than true typical values, just to cover-their-ass.

[voltsandjolts]

Can you switch the signals outside of the hot zone?

Sometimes the cold zone is a few kilometres away 

[Siwastaja]

I have a feeling Vishay just templated Idssmax @ 175C = 150uA across the datasheets of their entire line.

And not just Vishay. I once looked at MOSFET leakage currents for some dozens of parts and it's basically the same number for all parts from all manufacturers. Since the real-world leakage must obviously vary quite a bit, this must mean the rated number is at least an order of magnitude on the safe side for the manufacturer, if not two.

[David Hess]

Apparently decreasing Vds changes the leakage very little unless it's well below 1V. However their comments imply the datasheet value is very much a worst case number, so I might just need to do some real world testing.

The datasheet value is what the manufacturer was willing to test to.  Lower values take longer on the automatic test equipment, and are unlikely to be relevant to the intended application anyway.

Unless you want to pay a lot more money to have the parts tested to your specification, you may want to qualify the parts for your application yourself, meaning measure the leakage at high temperatures in a representative circuit.