2
\$\begingroup\$

Overview: I have an electric quad that I have heavily modified. The motor will need to be replaced every now and then because I am driving a 48 volt motor with a 72 volt supply that can give extremely high currents (6 12V motorcycle batteries) . This is no problem as I dont mind getting a new motor periodically. I am looking to exploit the limits of this small vehicle!

(components below linked to Mouser datasheets)

Gate driving Problem: I am using a microcontroller to drive a Gate Driver to drive a MOSFET that powers a DC motor (with Flyback Diode of course). There are two power supplies; one to power the motor (72V) and the other one powers the control circuitry (including Gate Driver) (12V). MOSFET Gate is pulled to ground with a 5W 220ohm resistor (probably too low/unnecessary resistance). Everything shares a common "star" ground.

Every time I try to test this circuit I notice 3 things happen: (1) The circuit seems fine initially at very low throttle before the motor even start to turn (I can hear the 500HZ PWM signal hum). (2) When throttle is increased slightly more to rotate the motor, the Gate Driver]1 BLOWS UP! (3) The MOSFETs are then ruined as Gate and Drain are shorted. As well as Source and Drain.

I have a large toggle switch to shut off the circuit in an emergency. The switch connects the Source on the MOSFET to Ground (turning it off breaks the motor circuit mechanically and reliably in an emergency event).

Instinctive Solution?: Add a Diode between Source and Drain on the MOSFET? Increase PWM frequency? Reduce the Gate-Source resistor to 10K. Add a 220ohm resistor between the Gate Driver output and MOSFET? All the above?

MOSFETs

CONTROLLER

\$\endgroup\$
  • 2
    \$\begingroup\$ We need a schematic. There's a button on the editor toolbar. \$\endgroup\$ – Transistor Mar 10 '18 at 23:56
  • 2
    \$\begingroup\$ Better post a full schematic. What type of motor are you using? Series motor? Series motors have huge start currents, you need to manage your current and not simply do a time based PWM. \$\endgroup\$ – Daniel P Mar 11 '18 at 1:54
3
\$\begingroup\$

Each FET has about 4" of wire in the Source, or 100 nanoHenries.

Should you be switching a current of a mere 10 amps in each FET, in 10 nanoSeconds, the inductive kick will be

Vinductor = L * dI/dT

Vinductor = 100nH * 10 amps / 10 nanoSeconds

Vinductor = 100 volts in the Source wiring.

Thus your FET gates experience 100 volt spikes.

As the gates go short, the spikes are connected to the PowerDriver IC, and destroy that IC.

USE a sheet of copper foil under your high speed (power driver) and its Surface Mount bypass caps, and the same sheet of copper foil under your several power MOSFETS. On back side, you should install some copper bus-bars to handle the 100 amps you expect to switch.

Each millimeter of FET lead (source, drain, or gate) or of wiring or thin PCB trace, is approximately 1 nanoHenry inductance; the formulas also depend on cross-section of the FET lead, or the bond wires inside the plastic, or the copper wiring (thin) I see in your photos; very wide foil has less inductance, with a natural-log dependency; GND plane over VDD plane will reduce the plane's inductive contribution by 10:1 (from memory, this is my rule of thumb for planes), but the other "wires or leads" still add ~1nanoHenry/1milliMeter.

============================

By the way, you at present have no means of encouraging the FETS to share those high currents. Try 0.01 ohms in the source, which is 20 squares of default thickness copper foil for the default foil weight of 1 ounce/foot^2.

You are at the mercy of how matched the FETS are, if they are at the same temperature and gate-drive voltage during the turnoff and turnoff voltage slewing.

At 10 amps per FET, that I*R drop produces 0.1 volts across the 20 squares, and 0.1 volts is plenty of signal to change the output of an analog comparator. [I had error in prior sentence; I'd written "produces 0.2 volts across."] Allocate one analog comparator per FET; combine the outputs with a 4-input NAND or 8-input NAND, that NAND connected to "SET" pin of a latch, and the latch output controls the "Enable" pin of your Gate Driver.

If you attempt to monitor the current in individual FETS, the intense and rapidly changing electron movement causes intensely-fast changes in the electric fields (some of which gets labeled "magnetic field") and simply measuring that 0.1 volts across 0.01 ohms may be impossible.

Suppose you make the Source-resistor 0.1 ohms. Then at 10 amps and 1 volt across the resistor, its power is one watt. Now you have a heat removal issue. Lateral (sideways) movement of heat thru FR-4 epoxy-fiberglass is very poor, so you need a heat-removal plane under the source resistor.

The gate-driver IC cannot have long leads; inductive spikes/kicks will kill it.

Draw and post a schematic, with all high current and fast-changing current paths indicated; you need to think about managing the inductive spikes/kicks; both ends of any diodes need low inductance.

You have a combined mechanical / inductance / high-current / fast-edges / heat-removal / bypass-capacitor-placement challenge. Draw lots of sketches as you think about this. Memorize the speeds of the FETs and the gate-driver ICs; examine the circuits provided by the manufacturers for R+C time-constants; are the R+C components setting the edge speeds? to slow down the edges and thus reduce the inductive risks?

\$\endgroup\$
  • \$\begingroup\$ Thank you for your excellent elaboration on the matter! Emphasis on reducing the inductance has been firmly noted. To clarify, you also suggest driving each mosfet with its own driver IC? Also will adding a diode between the source and drain of each mosfet help keep the spikes at bay, if so could the aforementioned flyback diode be used? In addition, I believe the aluminium heatsink/bus bar should suffice for handling the high currents as it is connected to the back of the mosfets (drain) and goes right to the motor. \$\endgroup\$ – Lance Sobhani Mar 11 '18 at 2:39
  • \$\begingroup\$ Aluminum will oxidize immediately upon exposure to air. Thus a clean and conductive metal-metal contact will not happen. Use some thick or wide copper. By the way, draw a schematic for stackexchange people to review. Note the dimple around the mounting hole (unused hole) in your first photo; that warping of the aluminum prevents a good contact for both heat transfer and for electrical low-ohm path; avoid the dimples. Separate driver per MOSFET? depends on how fast you want to switch, and "fast" causes more risks with inductive spikes/kicks/bounce/flyback (notice these are identical risks). \$\endgroup\$ – analogsystemsrf Mar 11 '18 at 12:16

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.