About Mosfet Matching / Paralelling

[Botvink]

It has been discussed in the forum before, but I understood its importance better with a video I came across today.

In my own driver (My Brushless Design for Hall Effect Type Motors, Looking for Suggestions) , although rarely happened, one of mosfets would break down and short circuit, So, I learned that it is also necessary to match mosfets.

 

[mxlemming]

I advise you to use modern MOSFETs from good brands. This matching business is not sensible, not indicative of state of the art and not suitable for serious production.

If you're facing issues with one fet of 3 being hotter than the others, just ask yourself if you really want this hassle in your life or if it's best to just swap to another mos.

Video above is 1990s package MOSFETs. If you use 1990s tech then expect 1990s problems.

 

[Logic11]

GaN reference design for pedal-assisted e-bikes..??

Why no talk on Gallium Nitride GaN transistors/power electronics??

Gallium nitride electronics could drastically cut energy usage in data centers, consumer devices. "...GaN transistors have at least one-tenth the resistance of such silicon-based transistors, according to the company. This allows for much higher energy-efficiency, and orders-of-magnitude faster...

endless-sphere.com

 

[mxlemming]

GaN reference design for pedal-assisted e-bikes..??

Why no talk on Gallium Nitride GaN transistors/power electronics??

Gallium nitride electronics could drastically cut energy usage in data centers, consumer devices. "...GaN transistors have at least one-tenth the resistance of such silicon-based transistors, according to the company. This allows for much higher energy-efficiency, and orders-of-magnitude faster...

endless-sphere.com

Interesting idea but if you consider where GaN is in the development cycle, the problems are still being solved. I think I built the only documented GaN stage on endless sphere. It worked fine but I couldn't see any compelling reason to use the GaN over silicon MOS. The switching waveforms were a bit zingy. The GaN FETs were very expensive compared to equivalent silicon. The performance improvement due to dead time being very very low was noticeable but not especially useful... The claim of 1/10 the resistance of equivalent silicon is laughable. Not even if you count by chip size and disregard cost and all the supporting componentry which is basically identical...

 

[Botvink]

Hmm, but it is still discussed at NXP docs, Not sure about it is old tech. Especially launched some expensive mosfets for them.. PSMN2R3-100SSJ (N-channel 100 V, 2.3 mOhm ASFET with enhanced dynamic current sharing in LFPAK88)

I agree on that type mosfet TO247 is old, but still many users making drivers with them.

Though, Apart from theory, a demo and tests should be done in a real circuit...

 

[mxlemming]

You won't catch me using NXP, Toshiba, vishay, AOsemi and others any more. This link you just made... Better current sharing in ASFET... Seems like something infineon did for years.

I had an 18 fet power stage on my desk the other day, I'd used AOTL66518 since they're the best rated 150V toll FET. 1/3 on one phase was significantly hotter, and other phases visible out of balance with the IR camera. It eventually blew way sooner than expected. Nothing wrong with the layout, gate being driven fine... All had beads and resistors... Very symmetric layout... Swap to infineon and problem solved.

Different batch with these FETs just self cooked while just switching with tiny current.

NXP have a history of laughable specs, way beyond the nonsense other manufacturers claim. They also make packages different and incompatible with the standard ones... Why? I'm guessing so you're locked in once your design is done and it's hard to compare to other brands.