I was trying to compare the dissipation of an electromechanical relay with a MOSFET switch. I thought I knew how to do this but I don't believe my results.

Here is a data sheet for a relay: https://www.azettler.com/pdfs/az2850.pdf
This relay is rated for 30 or 40 amps. My load is 20 amps.
The data sheet says the contact resistance is "< 50 milliohms initially."
So 20 amps times 50 milliohms is 1 volt and 20 amps times 1 volt is 20 watts. And that's just one pole. I actually wanted to switch both poles. I can't really believe that a relay would be dissipating 40 watts!!! I can't really believe a 1 volt drop across a pair of relay contacts.

So, I thought, maybe they mean micro-ohms. But I looked up a P&B datasheet
https://www.te.com/commerce/Documen...&DocNm=1308242_T92&DocType=DS&DocLang=English
and they're saying 100 milliohms.

What am I missing here?

 

In the real world the resistance will be much much less.

 

Yeah, it must be lower. But where would I find a realistic value?

 

Close the relay and use a Megger Ohm meter to measure.
Max.

 

It seems you're not wrong.

https://na.industrial.panasonic.com/blog/how-measure-relay-contact-resistance

I never realized how much resistance relay contacts can be "allowed" to add to a circuit. Note that the 50/100 mOhm value is a (max) value / worst case. I suspect that one whose resistance is actually that high (without any defects) is unusual.

 

Well, I guess I'm going to have to measure it. I'll report back ...

 

OP's specs for contact resistance seem out to lunch like a copy pasta mistake but another example Fujitsu rated 32A 250VAC FTR-K3-PS spec is "Resistance (initial) Max. 100mΩ at 1A, 6VDC" which clashes with their graph below.

Little known is that relay contact resistance increases with age. It is common to dissipate a couple watts of heat just across the contacts at >25A. But after say 50,000 cycles contact resistance can go over 3X the intial spec.
The relay is physically too small to actually meet those wonderful specs (with pcb pins) and have a long life. 1/4" spades crap out above around 16-20A as their resistance is too high, as connectors.
So at say 20A, you have 0.8W of coil heat, 0.3-1W of contact heat, and double that to include the space connectors. The wires act as a heatsink, or your pc board large pads and traces. With a thermal imaging camera you can see that high current relays make a lot of heat, hence why they get so large.