0
\$\begingroup\$

I'm considering making a replacement solution for an automotive fan output stage. The original part gets quite warm and tends to be unreliable - it's cooled with massive fins and sits directly in the air stream of the fan it controls to keep it cold.

I found modern TO-220 MOSFETs with suitable voltage ratings with Rds(on) values of less than 2 mOhm, which should result in much less heat and much simpler cooling than the original part - it'll have to switch up to 40 A at 14 V.

My question: How good have MOSFET developments been in the last 2-3 decades with regard to on-state resistance? Were sufficiently low-resistance MOSFETs just not available back then such that an extensive cooling solution was required?

\$\endgroup\$
6
  • \$\begingroup\$ There has been technology breakthrough, the so-called "hexfet" brand has been proven a market success and competitors have created products on similar technology to try to take some market share. But I think the general answer is yes , it is both cheaper and more available today low rds on power mosfets with crazy current capacities 60A+ in just a d2pak \$\endgroup\$
    – crasic
    Dec 2, 2020 at 8:13
  • 1
    \$\begingroup\$ Certainly significant, why don't you just compare the characteristics of your part with modern devices? \$\endgroup\$
    – Damien
    Dec 2, 2020 at 8:13
  • \$\begingroup\$ The topology of a modern power mosfet is highly parallelized gate structure, one analogy is it is many mosfets in parallel resulting in low effective rds , however gate capacitance suffers and so intermediate drivers have had to have been beefed up in some cases e.g. switching applications. \$\endgroup\$
    – crasic
    Dec 2, 2020 at 8:17
  • \$\begingroup\$ Electrical Power I^2*RdsOn * Rth (thermal resistance sum of interface and heatsink) = Temp rise ['C] Pd= 40^2*2m=3.2W but the SMD FETs have improved choices (tens of thousands ) over TO-220 for same currents.. Generally chose current rating much higher than actual operation and be sure to add flyback diode. Also rapid start of motor may be 400A \$\endgroup\$
    – Tony Stewart EE75
    Dec 2, 2020 at 8:27
  • 1
    \$\begingroup\$ I'd just add my two cents, I had a similar issue on a Citroen fan control heating up way too much. It turned out that the MOSFET was used as a linear voltage regulator and was designed to only work within the fan stream, just like yours. No need to say that with a linear regulator RDS(on) plays no role \$\endgroup\$
    – carloc
    Dec 2, 2020 at 9:27

1 Answer 1

Reset to default
3
\$\begingroup\$

The automotive industry is not particularly known for seeking out best-available solutions, rather they are very much into cost minimization. If the solution saves a bit of money and does not noticeably affect the gas mileage EPA numbers or warranty claims they will likely go that way.

While there have been improvements in MOSFETs, there is a basic equation that for a given technology and voltage rating the Rds(on) is inversely proportional to the active area of the silicon consumed, which obviously affects the cost and maybe even the packaging.

I successfully replaced a bad power MOSFET in a Nissan crossover fan controller module with a significantly beefier one, but I did confirm that the gate charge was not too much higher than the original- the drive circuit was not available to analyze. YMMV. If I was making something to sell I would have done a lot more analysis and measurements.

\$\endgroup\$
5
  • \$\begingroup\$ While the underlying economics of Rds(on) is straight forward, increasing silicon area requires addressing point defects, in this particular case it required significant development in the design to accommodate a more cost-effective process. \$\endgroup\$
    – crasic
    Dec 2, 2020 at 9:41
  • \$\begingroup\$ @crasic Are there actually significant yield penalties in commercial size MOSFET chips to go, say, 2 or 4x the area? Of course OP is asking about processes decades ago, which probably were not as good. \$\endgroup\$
    – Spehro Pefhany
    Dec 2, 2020 at 9:59
  • 1
    \$\begingroup\$ +1 for the incredibly appropriate use of "YMMV". \$\endgroup\$
    – Lou
    Dec 2, 2020 at 10:17
  • \$\begingroup\$ Not today I would say. My understanding is that the process improvement is also coupled with scaleable design. For example, as far I as I understand hexfet power mosfets were introduced ca 2000. The architecture is fault tolerant in that that point defects in gate oxide don't cause serious problems. However since I can't confirm this "industry hearsay" without further research it is not answer worthy. \$\endgroup\$
    – crasic
    Dec 2, 2020 at 10:19
  • 1
    \$\begingroup\$ Engineers,. when designing, have in mind a lot of parameters, not only cost. Replacing unreliable parts may cost more. Changing some elements in balanced system may destroy the balance. If you replace some part but it is made by not reliable manufacturer? Automotive environment is very tough and all parts are tested before come to mass production. \$\endgroup\$
    – user263983
    Dec 2, 2020 at 12:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.