Parallel power mosfets

[Wolfgang]

Hi,

I've seen it but the DC SOAR is almost too good to be true. Its an old device (not recommended for new design), so maybe it works.

[ogden]

Maynuo and similar loads uses IRFP250N

Discussion:

https://www.eevblog.com/forum/blog/eevblog-1023-rigol-dl3021-electronic-load-teardown/

Comment section:

[Wolfgang]

... the RIGOL loads do the same. I dont know if they screen their parts for coming from the old days with the larger die sizes,
or they just use some many in parallel so they are (hopefully) far enough away from the spirito borders (even if they have no specs) ...

[exe]

Maynuo and similar loads uses IRFP250N

I've seen IRFP150N (similar device) in some linear applications from "reputable" vendors. It seems to be popular, I wonder why. Do they know something not mentioned in the datasheet? Or just a cargo cult? Rigol DP832 (power supply) uses IRFP260N, which looks to me a device from the same series.

[exe]

I think I have a clue. Infineon's site says it's made using planar technology.

[Wolfgang]

Hi, the IRF510 is an ancient part with a good compromise between switching speed and SOAR robustness.
When you look at Fairchild, e.g., they had a DC SOAR specified for their IRF510A, so linear operation for those is not illegal.
Just dont load it too hard (because of its high RthjCase), and it will survive switching and linear operation.

[OM222O]

Guys I still haven't gotten answer to my main question, how do you parallel mosfets with load balancing?
Do you suggest using a quad op amp setup with a gain of x0.25 to drive each mosfet individually?

I appreciate the recommendations for the mosfet, but let's focus on that question.

[duak]

Here's what I've found with loads which seems to be corroborated above.  It's also used in one hp load I found the schematic for.  The better (least bad) approach is to use an op-amp with a source sense resistor for each FET.  The gain, ie. transconductance or Iout/Vin should be 1/4 of the total.  There should be another current sense for the overall current.

As noted before, a higher voltage earlier technology FET seems to survive better here, I suspect because the channel resistance can be higher than a lower voltage FET and so there is some inherent current sharing between the cells in the FET.

Cheers,