2
\$\begingroup\$

i made this BLDC motor controller, here is schematic enter image description here

But TC426 Mosfet drivers keeps blowing out when i connect a load, after drivers fail, mosfets are still working, only drivers blows out, I checked all voltages, they are fine. I am using 33kHz pwm modulation for each phase and mosfets are IRF44N. I can't find a problem, but i was thinking that i need to put resistor between mosfet gate and driver? Could that be a problem? enter image description here

Thanks for everybody who answered. While you were answering, i tried to do some more testing and i have started to see some pattern, last three times first mosfet driver failed, in schematic it is left driver. I will double check all connections, but I think i will redraw PCB taking all your advice because i am tired going to electronics shop every day and spending money there for drivers! Thanks everybody

UPDATE I added capacitors to each driver and change mosfets to more powerfull ones, now everything works, thanks to everybody!

\$\endgroup\$
6
  • 2
    \$\begingroup\$ A gate resistor is always a good idea. \$\endgroup\$
    – Hearth
    Apr 27, 2017 at 15:19
  • \$\begingroup\$ Can you share your layout? \$\endgroup\$
    – Phil Frost
    Apr 27, 2017 at 15:26
  • \$\begingroup\$ @PhilFrost I added layout!And i have added last two traces. It is single side board \$\endgroup\$
    – Eižens
    Apr 27, 2017 at 15:31
  • \$\begingroup\$ Items 3.1 and 3.2 in the spec may be tripping you up too. \$\endgroup\$
    – Trevor_G
    Apr 27, 2017 at 15:50
  • \$\begingroup\$ Is the gate drive from the TC426 high enough to control the high-side N mosfets (Q1, Q2, Q3)? This is outside my normal realm of design, but I would think that you need the TC426 output to be at least the same Voltage as Vcc, and even higher if using N Mosfets. Perhaps you could use P Mosfets, and keep the outputs of the TC426's at Vcc? Or maybe add voltage doublers on the outputs of the 426's. \$\endgroup\$ Apr 27, 2017 at 16:01

4 Answers 4

0
\$\begingroup\$

I'd be looking at the signal rise time at the inputs of the TC426.

The TC426 has a mosfet input, meaning it is capacitive. You have not indicated the MCU voltage, or how long that cable is, but if the rise time is slow at the TC416 end, it will leave the TC426 output shorted for a considerable duration.

Running the signal ground back through the power ground is also a bad idea. If there is a significant rise on that ground it may drop your digital input into the grey area, again creating a short inside the TC426. You should consider decoupling those control signals, perhaps with opto-couplers to keep the signal ground completely separate from the power ground.

Also, more capacitor storage is needed close to the TC426s as indicated in the spec sheet.

\$\endgroup\$
1
  • \$\begingroup\$ @Eižens, following on with Trevor's answer here, the datasheet does state "The input levels should not be maintained between the logic ‘0’ and logic ‘1’ levels. Average power dissipation will be reduced by minimising input rise times.". Can you put a buffer logic gate per TC426 input on this board to drive those inputs with very fast rise/fall times. A buffer with a Schmitt trigger input cuts its own susceptibility to slow signals coming up the cables. Look at the SN74LVC2G17 Dual Schmitt-Trigger Buffer or similar. Put 4K7 pull-downs on the buffer inputs. \$\endgroup\$
    – TonyM
    Apr 27, 2017 at 17:50
1
\$\begingroup\$

The TC426 FET driver ICs do not need a series gate resistor as they are designed for the capacitive loading of direct connection to a FET gate.

However, the TC426 data sheet also makes it very clear that unused driver inputs must be connected to VDD or GND and must not be allowed to float.

It looks like you are driving the TC426 inputs from another circuit board. If this lets the inputs float, it could well be why they are destroying themselves.

If you are driving these inputs from a microcontroller (MCU), for example, there will be a delay on start-up while the MCU goes through reset then configures the I/O pins as outputs. Before that, the I/O pins will be configured as inputs and be high impedance, leaving them floating.

I would add pull-down resistors to all inputs on this board to remove this problem. The TC426 input leakage current is +/- 1 uA and lets assume a generous 50 uA leakage from whatever driver you have on this. A 4K7 pull-down would put about 0.235 V on a TC426 input, well inside its 0..0.8 V range for logic low.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ Yup. I'd also be concerned the signal ground is back through the power ground and insufficient decoupling. \$\endgroup\$
    – Trevor_G
    Apr 27, 2017 at 15:57
  • \$\begingroup\$ After first fail, i added external 10k pull down resistors (i forgot to add resistors to altium schematic) but still they got blown out. \$\endgroup\$
    – Eižens
    Apr 27, 2017 at 16:02
0
\$\begingroup\$

With your layout, loop area for the currents that will be switching the MOSFETs on and off are large. This means they will have high inductance, which will limit your switching time. Combined with the capacitance of the gate it can also cause ringing, and that ringing can be high enough to damage the MOSFETs or the driver. This may or may not be your failure mode, but it may merit attention anyhow.

Minimize that inductance:

  1. Move the driver closer to the MOSFET, and making the trace from the driver to the gate as short as absolutely possible.
  2. Next, make the trace from the driver ground to the MOSFET source also as short as possible, and run directly beside the gate trace.
  3. Lastly, put a low inductance capacitor (chip capacitors are good) as close as possible between Vdd and ground for each driver.

This means the currents which switch the MOSFET have a small loop area as they flow through the decoupling capacitor, through the MOSFET between gate and source, and through the driver IC. Since inductance is proportional to loop area this minimizes the inductance, which reduces switching time, reduces ringing, and increases efficiency.

Section 3 of the datasheet has a little bit on this, though it's very brief. You might go through this IR paper on MOSFET basics to get some more detail on the issues mentioned there.

\$\endgroup\$
1
  • \$\begingroup\$ You're right, I missed that. I'll delete that part. \$\endgroup\$
    – Phil Frost
    Apr 27, 2017 at 16:27
0
\$\begingroup\$

I've had MOSFET drivers destroy themselves, by oscillating.

Why would a circuit oscillate? Huge power gain. If the GND (shared by Vin and Vout, right?) bounces by 1.5 volts, you certainly are in the forbidden region. I'd target 0.5v bounce or less.

A mere 10nanoHenries and 1amp/nanosecond produces 10 volts Ground bounce.

10 nanoHenries is about 1cm (0.4inches of wire).

I'd provide GND to those drivers like this:

schematic

simulate this circuit – Schematic created using CircuitLab

On circuits like this (for most circuits, in truth), lay out the GND first. And protect that GND plan.

Learn how to provide "local batteries" for each IC, particularly when the IC must briefly provide amps of current for a few nanoseconds. A sizable cap by each IC, with a bead or 1_Ohm resistor to the global power, protects each IC from interference with other ICs.

\$\endgroup\$

Your Answer

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

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