Mosfet RDSON test

[dietert1]

Recently i got some IRFB3207 and IRF1405 mosfets. As i have some doubts about fakes, i wanted to test their Rdson and used a 0.5 A test current at Ugs= 12 V, measuring Uds=11.9 mV. That gives about Rdson = 23.8 mOhm, much more than the datasheet spec of 4.6 - 5.3 mOhm.

Except the datasheet spec is at a much larger current = 101 A (that i can't test). The DUT does not get hot during this test. Are there any rules about the current dependence of Rdson?

Regards, Dieter

[eugene]

I have no specific experience with this, but it seems to me ^TM that you should be able to infer RDSON from any of the graphs of I_D vs V_DS in the datasheet. The slope of any of the curves under any particular conditions has units of amps/volt and is thus resistance.

[wraper]

Rdson will rise when you push too much current. But you won't get higher value than specified due to lower current. So the part you tested is fake.

[mawyatt]

Rdson has a +TC, so at lower drain currents one would expect a lower Rdson due to a lower channel temperature.

If your setup is per the data sheet (specifically Vgs), either you have a setup measurement issue or counterfeit devices.

Best

[Avelino Sampaio]

You may be introducing an error in the way you are performing the test. See in the first image that the connection of the GND of the source to the lead of the Mosfet source is being made via protoboard. And as a result I am reading 17mv instead of 3.5 ~ 4mv (test current of 1A). Note in the second image that I am connecting the GND of the source directly to the lead of the Mosfet, without going through the protoboard. The result is now correct, 3.6mv.

[Avelino Sampaio]

I designed a milliohmmeter, with the main objective of measuring RDS and identifying fake MOSFETs. See the project link where I tested several MOSFETs. It has been very useful to me ever since. See the image below of the device measuring the RDS of the IRF1404, at the correct value of 3.5mR, using only 50mA.

https://www.eevblog.com/forum/projects/diy-milliohmmeter-how-to-solve-grounding-issue/msg4258852/#msg4258852

[dietert1]

I have no specific experience with this, but it seems to me ^TM that you should be able to infer RDSON from any of the graphs of I_D vs V_DS in the datasheet. The slope of any of the curves under any particular conditions has units of amps/volt and is thus resistance.

Yes, if i look at the Id vs. Uds graphs in Fig. 1 or 2 in the IRF1405 datasheet i see a linear relationship, at least at reasonable currents, e.g. between 10 and 100 A. They don't show data below 1 A, but certainly the slope will be the same. When i read the slope, it is about 100 A at 1 V, that is 10 mOhm. So one may wonder how they specify Rdson = 4.6 to 5.3 mOhm in the same datasheet. Maybe there is an extra 5 mOhm package resistance involved.
My own measurement was directly on the package between source pin and cooling tab for drain.

Regards, Dieter

Edit: Meanwhile i understood that Fig. 1 is for room temperature and Fig. 2 for 175 °C. The factor two i was missing is due to difference in chip temperature. In Fig. 1 (at room temperature) i read 200 A at 1 V, that is 5 mOhm.

[Siwastaja]

If you have a reason to expect fakes (i.e., they are not from distributors but e.g. Ebay), they most likely are fakes.

Rds(on) is constant (i.e., voltage drop is linear per current) as long as die temperature stays constant. As others have said, at high currents you can expect rising Rds(on) due to heating, but as long as current is low enough so that the device does not heat up significantly, you should be pretty close to datasheet room temperature value.

[dietert1]

If you have a reason to expect fakes (i.e., they are not from distributors but e.g. Ebay), they most likely are fakes.

Rds(on) is constant (i.e., voltage drop is linear per current) as long as die temperature stays constant. As others have said, at high currents you can expect rising Rds(on) due to heating, but as long as current is low enough so that the device does not heat up significantly, you should be pretty close to datasheet room temperature value.

Yes, meanwhile i am pretty convinced that i got fakes. The transistors were bought at Mult Comercial and Mercadolivre.com.br as it can be expensive to import from digikey or mouser. One has to calculate about 150 % extra for taxes and postage. What is available locally seems to be chinaware.

[Postal2]

... Are there any rules .......?

Do search "YR1035+".
https://www.bulcomp-eng.com/datasheet/Yaorea%20YR1035_%20-%20User%20Manual.pdf

[Avelino Sampaio]

Another method that people often use is to measure the zener voltage of the internal diode. From the datasheet, we can see that we only need to apply a current of around 250µA to perform the test. I use a very simple circuit, rectifying and filtering the 120V line to obtain up to 170V and thus test several MOSFET models.

[wraper]

Another method that people often use is to measure the zener voltage of the internal diode. From the datasheet, we can see that we only need to apply a current of around 250µA to perform the test. I use a very simple circuit, rectifying and filtering the 120V line to obtain up to 170V and thus test several MOSFET models.

That circuit is dangerous. Even without it being under mains voltage, it does not have a discharge resistor. So it remains dangerous after disconnecting form the mains. 47uF is way too much as well and is another source of danger.

[Avelino Sampaio]

It was reckless of me not to inform and indicate in the schematic that I use a resistor to discharge the capacitor. Yes, the capacitor used has a very high value but it was the one I had here with a voltage of 250v.

Here is an example of a zener voltage test with three original MOSFETs.

Image 1: IRF1404 - zener voltage 40v indicated in the datasheet
Image 2: IRFB3306 - zener voltage 60v indicated in the datasheet
Image 3: IRFZ44 - zener voltage 55v indicated in the datasheet

[Avelino Sampaio]

I found a fake IRFZ44N MOSFET here. I am bringing the results of both the Rds reading and the Vds zener reading.

Image 1: IRFZ44 OK - Correct Rds reading, at 17mR.
Image 2: IRFZ44 Fake - reading of 39mR, where it should be around 17mR
Image 3: IRFZ44 Fake - Zener voltage reading of 77v, very far from the 59v of the real MOSFET.

[dietert1]

My application does not require very low Rdson nor a high Uds, and it is linear near DC (audio bandwidth). So i just bought the mosfets with the largest current rating i could get. I could use them even if they are fakes. Just wanted some kind of quality check.
The measured Rdson values of about 20 mOhm are about 4x worse than specification yet good enough. I also checked gate currents to be "zero" and the gate threshold voltages. The Uds zener voltages are next..

Regards, Dieter

[mawyatt]

The 4X higher Rdson might "hint" that the die is much smaller than expected by the part # and data sheet. This could mean a significantly less capable device wrt to power dissipation.

Best

[inse]

I would like to agree to mawyatts statement.
Without proper datasheet you won’t know the SOA curve either.
How much power do you want to dissipate, whats the application anyway?

[KerimF]

The 4X  :horse:higher Rdson might "hint" that the die is much smaller than expected by the part # and data sheet. This could mean a significantly less capable device wrt to power dissipation.

Best

Me too, I agree with 'mawyatt'.

[RoGeorge]

Except the datasheet spec is at a much larger current = 101 A (that i can't test).

Most probably they can't either.  When you see weird values like 101A, read the fine print.  Might be yet another marketing lie, like the 169A continuous current supported, when the maximum current supported by that type of packaging is only 75A.  The 169A is the calculated max possible current by a die of the same size and with ideal cooling, so bollocks, because I buy transistors, not ideal dies in ideal infinite radiators.

For example see remark 6 in the datasheet:
https://www.infineon.com/dgdl/Infineon-IRF1405-DataSheet-v01_01-EN.pdf?fileId=5546d462533600a4015355db084a18bb
or the fine print at note 1 here:
https://www.infineon.com/dgdl/Infineon-IRFS3207-DataSheet-v01_01-EN.pdf?fileId=5546d462533600a401535636708e215e

They should let the engineer write the datasheets, and the marketing should write only the bloatware flyers filled with empty emphasis words and pics.  I'm very upset about infesting engineering specs with marketing crap. 


I didn't study in full the datasheets, but I suspect the 5m\$\Omega\$ specs do not apply to the kind of test you are doing. Probably the 3-5m\$\Omega\$ is yet another one of those values that are true only when all the planets align just right.

My bet is the transistors are probably not fake, but to be sure, I would ask Infineon what are the expected measured Rds values for your particular test setup.

[Kleinstein]

The maximum current cited on the first page is usually so that 1/2 R_on * I²_max = P_tot. This gives rather unrealistic values. This starts with using a too high a P_tot and R_on at roomtemperature.

The  IRFB3207, even the real one is anyway no suitable for linear operation, e.g. in a power supply or MOSFET amplifier. The forward SOA is rather poor, as expected for a modern relatively low voltage FET that is made for switching. The IRF1405 may be a little better, but still not good - just no know to be bad. Attached is the SAO curve from the Infineon / IRF datasheet for the IRFB3207.

For linear operation one should more looks at higher voltage MOSFETs and preferrably a good size case as a TO220 is also limited in power handling. Also old types like BUZ10 have a good chance to work.  A old time favorite is the IRFP250.

[inse]

@RoGeorge:
I have seen datasheet where the maximum current or power dissipation were specified at t_junction of 25 Celsius, how to achieve this practically - “not our business“

[Postal2]

... the maximum current supported by that type of packaging is only 75A.  ...

True. Three welding points, 25A each.

[dietert1]

The application is a TEC driver. The circuit is a full bridge where the lower mosfets get switched while the upper ones run in linear mode in order to realize a current source for the TEC. Currents will be up to 3 A at a supply voltage of about 7 V.
Under these conditions even a mosfet with Rdson = 20 mOhm should survive.
Will report later, how the idea of getting a "strong enough" fake worked out. They may in fact be other, older mosfet chips similar to those BUZ... types.

Regards, Dieter

[Kleinstein]

Eben with only 7 V the SOA limit for the IRFB3207 would be just marginal for 3 A. So really a speciallized switching type. One may still not have the full 3 A and 7 V Drain-source at the same time.

[dietert1]

That IRF3207 SOA diagram is a bit difficult to understand, as they specify about 0.5 K/W as thermal resistance from junction to case elsewhere. That seems to be the same as for other TO-220 power mosfets. As far as i understand the 175 °C - 25 °C = 150 K temperature difference between junction and case assumed for the SOA diagram means a continuous load of about 300 W. Those SOA curves seem to be about avalanche capability under heavy load conditions.
Anyway i have enough mosfets to use IRF1405 for linear control and IRF3207 for switching.

Regards, Dieter