2
\$\begingroup\$

If I have a setup where a N MOSFET drives a heavy inductive load (say, 25A peak), can I use the MOSFET's internal resistance to measure said current the same way I would use a shunt resistor?

The value of the equivalent series resistance when the MOSFET is active (i.e. Rds-on) is usually very low and easy to find in the MOSFET's datasheet. I know it is not ideal, and will strongly depend on the FET's temperature a well as the current (so I have a loop there). Still, is there any serious impediment to this approach?

The reason I want to do this is because I have a system where I need to minimize component count as well as avoid any extra losses (i.e. shunts) when driving the inductive load, but can compute some corrections/linearization on a temperature-sensing micro-controller if needed.

I am almost 100% I saw a LiPo battery manager that appeared to do something similar, but I am unable to find it. As I recall, this IC estimated the charging current using something similar to what I just described. But maybe I am just mistaken.

\$\endgroup\$
  • \$\begingroup\$ You can but the resistance will be low tolerance, it will differ between mosfets and it will drift with tempature. \$\endgroup\$ – user160063 May 23 '18 at 9:45
2
\$\begingroup\$
equivalent series resistance ... easy to find in the MOSFET's datasheet

No. This seems to be your main misconception.

The datasheet tells you the guaranteed maximum, but not what it will actually be in any one device. Sometimes datasheets show typical specs, which are usually significantly less than the maximum. And of course any one device might be lower than typical too, but you don't know how much.

Then as others have said, RDSON has a strong dependence on temperature.

With calibration to the particular device, and maybe some correction for measured or assumed temperature, you might be able to detect very basic current thresholds, like "too high, shut down now". But anything you'd call a "current measurement" isn't really going to work.

\$\endgroup\$
  • \$\begingroup\$ Thank you for your answer. I have found that many datasheet contain a neat graph showing RDSon vs temperature. Are they too innacurate to do some sort of correction with software? \$\endgroup\$ – andresgongora May 25 '18 at 18:50
  • \$\begingroup\$ @and: Those graphs are showing you the nominal relative changes to Rdson as a function of temperature. They don't tell you the absolute resistance. These graphs either show only the typical values, or sometimes just relative values normalized to 1.0 at 20 or 25 degC. Also, determining the actual temperature of the die isn't easy. So even if you did know absolute Rdson as a function of temperature, you're far from done. \$\endgroup\$ – Olin Lathrop May 25 '18 at 20:44
2
\$\begingroup\$

MOSFET's on resistance is not very accurate and not stable over temperature. So if you calibrate and compensate- go for it. Or maybe it's just good enough. LiPo itself probably like narrow temperature range, maybe in application you are limited to 25-35 degrees or something.

Keep in mind, that in two years your purchasing people will ask to replace the MOSFET with something else- and you can't, because there is no same MOSFET in the world.

\$\endgroup\$
2
\$\begingroup\$

RDSon is mostly dependent on temperature and Gate-Source voltage. The Gate-Source voltage part is not really a Problem, the temperature is though. It is not uncommon that on resistance doubles (or even more) at the upper specified temperature limit (compared to 25°C). Take a look at some MOSFET datasheets as reference.

There are some current sense ICs that specialize in current sensing using MOSFET RDSon (eg. IR25750L). There are app notes available for temperature compensation (for example this).

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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