4
\$\begingroup\$

I have come across this question a lot in the design of a special 3 phase motor controller I plan to build this summer and I think I need to ask it directly to solve it.

I need to be able to modulate 3 phase ac with mosfets with a 12 to 24 volt dc supply voltage. I have this figured out for the most part I want to use only n type mosfets for phase symmetry. Driving them with a high power source relative to the main supply they are switching. I will use optical coupling to the MCU.

However I need to drive between 8 and 200 amps, the driver boards can be made separately in different capability ranges. But I don't think I can get around wiring mosfets in parallel this device also has to conserve space and weight.

What is the best way to create the driver circuit preferably out of surface mount parts?

\$\endgroup\$
1
\$\begingroup\$

You might investigate 'insulated gate bipolar transistor' (IGBT) technology. The high power 3-phase motor drives I've encountered (several 100's of volts x upwards of 100A) have been built around IGBT's, and were quite compact.

\$\endgroup\$
  • \$\begingroup\$ IGBT is fantastic. \$\endgroup\$ – minimer Jun 9 '11 at 1:34
  • \$\begingroup\$ -1: IGBTs are standard for motor drives, but in the high voltage range. The OP has stated the supply voltage is 12-24V. For that range, MOSFETs are far more appropriate. If you're going to recommend IGBTs, be more clear that they really only provide a benefit at high voltages. \$\endgroup\$ – Jason S Jun 10 '11 at 12:48
  • \$\begingroup\$ @Jason S - this was not so much a recommendation, as merely a point of information. Admittedly IGBTs are high voltage devices (please note "100's of volts in the above text) and the OP stated only 24V; but the OP also expressed a desire to conserve space, which IGBTs do admirably, and I felt it would add value to our content, to leave mention of IGBT tech here as an aid to those readers that might pass by here in future, who may not be voltage domain constrained. Thank you for voting. \$\endgroup\$ – JustJeff Jun 10 '11 at 22:18
  • \$\begingroup\$ I'll remove the -1 if you elaborate on your answer. I don't generally downvote, and it's not personal, but it doesn't seem like the appropriate answer for the OP's question in this case (although he/she seems to think otherwise by accepting it). IGBT's conserve space only when their power capability is an improvement over other options; we had a power electronics engineer do an analysis at our company a few years ago at the 3-5kW level, and the equivalency breakpoint for DC bus voltage between IGBT and MOSFET was ~200-300V; >300V IGBTs clearly do better, and <200V MOSFETs clearly do better. \$\endgroup\$ – Jason S Jun 10 '11 at 22:33
4
\$\begingroup\$

You can still get away with using single mosfets if you use something like the IRF7739 which is rated for 40V and 375A. The high currents will require that you use some sort of heat-sinking, but if you run at relatively low currents you can probably get away without using any. When manufacturing the PCB, use a 2oz copper pour with wide polygons instead of regular traces to get your current where it needs to go.

High current mosfets have the side effect of having a very large gate charge which requires that you use a gate driver IC. You could either use three half-bridge drivers like the LM5107, or you could use a three-phase bridge driver such as the FAN7388 or the IRS2336. You can then isolate the driver section using something like the ISO7220 for each pair of PWM signals.

\$\endgroup\$
  • \$\begingroup\$ I don't know if we can get away with that much copper. This thing ideally would weigh an ounce though admittedly i think weight exceptions could be made for the 200 amp version. What draw backs are there to lots of tiny mosfets in parallel especially ones designed to be implemented that way. It reduces gate resistance but in creases gate charge rite? \$\endgroup\$ – minimer Jun 7 '11 at 11:58
  • \$\begingroup\$ Thank you for the automotive grade mosfet i found its new big brothers. AUIRF7739L2 and the AUIRF7665S2. I'm sure if i don't use it this time ill use it one day. \$\endgroup\$ – minimer Jun 7 '11 at 12:05
  • \$\begingroup\$ @minimer: Yes, you can get away with that much copper. "2 oz copper" is the common term for 71 μm thick copper, aka "2 ounces per square foot". It doesn't actually weigh a literal 2 ounces unless your board is a full square foot in area. \$\endgroup\$ – davidcary Jun 20 '11 at 4:16
1
\$\begingroup\$

RocketPenguin has given you a good answer, but I want to point out a additional issue. It may sound nice that a FET is rated for 375A or whatever, but check all the specs very carefully. Often the maximum current spec of a FET is unattainable due to other constraints. Look at the Rdson and multiply that by the 375A figure. That is the power the FET will dissipate at that current. I haven't checked that particular part, but often that is considerably higher than the maximum safe power dissipation, even with a good heat sink. The maximum current rating is therefore only for short pulses, but otherwise not a very useful number.

\$\endgroup\$
  • \$\begingroup\$ Well besides ridiculous gate charge what are serious disadvantages to multiple smaller mosfets in parallel we have purchased the best motor controllers available and they are all manufactured with multiple mosfets in parallel and im open to the idea. But i dont know what to do about the gate charge. \$\endgroup\$ – minimer Jun 7 '11 at 11:55
  • \$\begingroup\$ Your rite by my math i can only push about 100 amps in my application but this is getting really close maybe two of these in parallel but in not sure i like the delay time rise time and shut of time. Its more than 100 ns all together and this think has huge rds on resistance. \$\endgroup\$ – minimer Jun 7 '11 at 12:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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