MOSFET Rds(on) minimum

Question

When a vendor specifies an Rds(on) value for a part I have seen typical and max values specified but never minimum values. I understand that Rds(on) is a function of temperature and gate to source voltage (Vgs) but for a given temperature and Vgs is the variance of Rds(on) basically an intrinsic of the MOSFET design? How much does the Rds(on) vary from FET to FET of the same kind at a given operating point?

What motivates this question is current limit sensing of the drain to source voltage (Vds). I can measure Vds and approximate a current if Rds(on) is known. However if Rds(on) varies too much the current limit sensing may erroneously report no fault with a huge current.

Answer 1

It's possible to use Vds to detect a "huge current' such as a short circuit (and that is done) but Rds is not a good way to measure current.

It's variable from unit to unit, and it varies considerably with temperature, perhaps increasing 50% from room temperature to normal operating temperature.

If you like guessing you can take the difference between typical and maximum on the high side and guess it might be similar on the low side. You'll probably hit three sigmas that way and maybe that's good enough if 1 in 1000 is outside.

Answer 2

Sensing Current using mosfet can certainly be done , and i know that a product ( old curtis controller ) that uses it for current limiting , another note is that they installed on the heat sink near the mosfet a NTC temp sensor ( maybe to compensate the calculations at different temperatures .

Also there is an IC that International rectifier ( now infineon ) for this purpose :

IR25750

an-1199

Answer 3

A good N-channel MOSFET can have an Rds in the low milliohm range. For example, http://aosmd.com/res/data_sheets/AOD508.pdf has Rds of 0.0033 to 0.0045 ohm when the gate is driven with 4.5V. I use similar parts frequently.

Figure 3 shows that Rds varies very little over a wide range of Id, up to 30A, when Vgs is constant. Note E further states it is only turned on for < 300uS for the chart, so the device is not heating up much, indeed it is only dissipating P = I^2 * R = 2.97W at 30A and Rds = 3.3mohm. With a good heatsink, the part would not likely even get warm when on for 300uS, so I would expect the Rds to remain quite steady.

Use the charts to find your usage condition and determine how much variability you might see.

Answer 4

The fact that the minimum is not specified means you have to do your own research, and use the results at your own risk.

The fact that there's a difference between typical and maximum values warns you that there is a significant spread of values.

Without doing the research, one option would be to deem that the likely minimum = 50% of typical. This was a rule of thumb we used for propagation delay in logic gates, fairly successfully. Another would be to take the ratio of max/typical, and use that as the typical/min ratio as well.