using mosfet as current sense resistor

[Simon]

Right, daring thought: I have a power control project that requires up to 6 paralleled soic8 mosfets. I also need to measure current in this project. Well  as I'm using soic8 mosfets so have multiple pins for the same connections (drain and source) I thought could I use the voltage drop on the mosfet to measure current ? the mosfets will be fully turned on. What sort of variation in full on resistance do you get on mosfets over different temperatures ? are they stable ?

Differencies in manufacture would not be a huge issue as I can fine tune the software providing it falls in a range I know about

[jahonen]

Right, daring thought: I have a power control project that requires up to 6 paralleled soic8 mosfets. I also need to measure current in this project. Well  as I'm using soic8 mosfets so have multiple pins for the same connections (drain and source) I thought could I use the voltage drop on the mosfet to measure current ? the mosfets will be fully turned on. What sort of variation in full on resistance do you get on mosfets over different temperatures ? are they stable ?

Differencies in manufacture would not be a huge issue as I can fine tune the software providing it falls in a range I know about

Temperature is one variable affecting Rds(on), at maximum junction temperature (150°C or so), expect twice the value in room temperature.

Regards,
Janne

[Simon]

sounds like that's rules out then

[Neilm]

I do seem to recall a range of MOSFETS that allowed you to do this. I believe that they had internal compensation. I seem to recall looking at them many years ago when I was working on a SMPS design but decided that it was not safe, or was the wrong value or something.

What current are you pulling that you require 6 in parrallel?

Neil

[Simon]

well starting at 20 amps but may push to 30 amps. The mosfets I'm using have only 2.5W dissipation but are 6.5mR full on. I'd probably be fine with 4 but choose 6 for safety. I've already implemented a sense resistor but was considering the shortcut if possible, maybe not a good idea. I could run up my prototype and take some measurements thought

[tecman]

Rds On is not a reliable resistor.  You could look for a Sense FET.  I remember them 10 years ago or more.  It is a power MOSFET with an output pin that was something like 1/100 of the main D-S current.  You could run that through a small dropping R to measure the main current.  I am not sure if they are still made, or who might make them.  I think IR was one of the original mfgs.

paul

[Zero999]

It depends on the accuracy you desire?

[tecman]

Check these out:

http://www.onsemi.com/PowerSolutions/product.do?id=NILMS4501NR2G

Name SenseFET appears to be gone, but not the idea.

paul

[Neilm]

SenseFET was the name I was trying to remember.

Is the 20A  a continuous rating or is it switched? I'm currently looking at this http://www.vishay.com/docs/65162/sir426dp.pdf for a boost converter that will run at about 1MHz. The low gate charge means I will lose less than 1W at full load (40W)

Neil

[Alex]

The only time I have used the RDSon resistance of a MOSFET was to roughly measure the inductor current in a boost converter, several 10s of amps.

I have also seen it in switchmode converters as a means to detect a fault whithout introducing an extra resistor hence achieving high efficiency.

For any reasonable precision I would advise against using RDSon as it varies greatly. You can possibly use a low shunt resistor and an instrumentation amplifier, which you can get in a black box from the usual manufacturers.

Hall effect current transducers could be suitable, Allegro comes to mind. Careful with stray magnetic fields.

[Simon]

well initially I just need to make sure I'm not passing more than 20 amps +/-1amp. Accuracy is not a huge issue it's just to keep the thing under some rough control. I currently have a 2.25mR resistor (or 2 4.5mR in parallel) and a MAX4080, maybe best stick to that

[Zero999]

I say experiment.

Buy 10 MOSFETs and measure the resistance of each of them at different temperatures. Then you should have a good idea of the tolerances.

As I said before, it all depends on the desired accuracy, for example if you only need a basic current limiter and 40% tolerance is good enough you'll probably be fine but if you want to measure the exact current it probably isn't good enough.

How many are you making? If it's going to be mass produced then even fairly rough calibration will be expensive.

[Simon]

well if successful it may be reproduced but on a to order basis, each unit would need testing anyhow and that's when it could be quikly calibrated. I've got a previous prototype to test yet but after that I'll build my current mosfet version and take some measurements. Operating temperatures can be expected to vary from sub freezing to 40+ degrees depending on how much heat the circuit makes