Normalized On Resistance vs Static On Resistance

[GARL]

Hi everyone,

I have a question about a Power MOSFET that Engineers at my work place used in one of this designs. I am doing a tolerance analysis (STD DEV using Crystal Ball) on this unit, and I am puzzled by which On resistance I have to use in order to model the power MOSFET (IRF5210S).

On the datasheet

http://www.irf.com/product-info/datasheets/data/irf5210s.pdf

it has an on resistance of 0.06 ohms, and if you scroll to page 3 of the data sheet it has a graph of the Normalized On Resistance VS Junction Temperature.

I am uncertain of whether I use the 0.06 ohms or pull a value off the graph provided in the datasheet.

Thanks,

GARL

 

[IRstuff]

Tolerance analysis for what purpose? Variations in device parameters? Over temperature?

Depending on what you are actually doing, you need both, AND you need Figs. 1 and 2 as well, since they describe the drain characteristics as a function of drain voltage. If you're doing parameter variance analysis, then you'll also need the information on Pgs. 1 and 2



TTFN

 

[GARL]

IRStuff,

I am doing tolerance analysis on internal signals, to see whether they fall within the Tolerance Boundaries.

My supervisor only wants to me model the MOSFET as a resistor in Crystal Ball.

So what do they mean by Normalized? I have tried to do an internet search, but can't find anything, may be I am using the wrong key words.

GARL

 

[cbarn24050]

Hi, normalised in this sence means the datasheet value (0.6R). to get the value you need you first decide what temp you are running at, use the graph to find the normalised value (say 1.5) multiply it by 0.6 to give you the value.

 

[IRstuff]

What you'll need to do is to determine the load current, which gives you the Vds and hence the resistance as well as the power dissipation.

Then, using the power dissipation, the ambient temperature and the thermal resistance values, you can determine the junction temperature, which will give you a scaling factor from the initial resistance to the resistance at temperature. If you iterate that process a couple of times, you should be able to zero in on a good value of resistance.

TTFN

 

[logbook]

I don’t think the guys answered your question as to the definition of “Normalised”.

You are probably more familiar with percentages. If you take the nominal value of a quantity as 100% then you can plot what happens to it when you change something like temperature. Maybe the graphs shows 2% change per degree centigrade. A related scheme is “per-unit”. Really per-unit quantities are more natural than percentages. If you divide the percentage values by 100 you get per-unit values.

If a 13K resistor increases by 5% it has changed by 0.05 per unit.

13K +5% is 13K x 1.05 or 13.65K

In order to work it out in percent I had to take the 5%, divide it by 100, then add it to 1.

Now normalisation makes the per unit value 1. This is done by taking some reference condition as the “normal” value.

Taking my 13K resistor, the normal value is 13K at room temperature. The normalised value is 1.00 (per unit).

For the FET in question the normalised value goes all the way up to 2. That means the resistance doubled.