1
\$\begingroup\$

I've built a simple circuit to test a combo of a MOSFET (SUP70030E) and a gate driver (TC4452): Schematics

The MOSFET should be compatible with the gate driver (if I'm not reading the datasheets wrong) and the circuit works fine - I'm switching the LED on and off in 10 second intervals.

However, the gate driver starts running very hot very soon, and I'm measuring 150 mA at the gate driver VDD all the time. That's more than 3 orders of magnitude higher that what the datasheet says, and the current stays the same with both 0 and 1 at the input.

Any idea what could be causing this?

EDIT: The MOSFET is fine (keeps cold and easily switches when connecting directly to 12V), and the driver draws 150mA even with the output disconnected.

\$\endgroup\$
11
  • \$\begingroup\$ When you say "hot", what does that mean? Did you measure it? What are you reading for a temperature? \$\endgroup\$
    – Ron Beyer
    Apr 26 '20 at 19:02
  • \$\begingroup\$ Are you certain that it's connected properly? Seeing extremely high currents typically makes me think you might have some pins swapped around. \$\endgroup\$
    – Hearth
    Apr 26 '20 at 19:02
  • \$\begingroup\$ @RonBeyer I think the problem here is the constant current draw, which is far higher than it should be regardless of what temperature the die is getting to. \$\endgroup\$
    – Hearth
    Apr 26 '20 at 19:03
  • \$\begingroup\$ @Hearth I'd agree, just wanting to look at the datasheet and compare measured to actual. It also looks like the datasheet recommends the 4.7uF cap be polarized, I'm also not sure it needs the 10k pull-down. \$\endgroup\$
    – Ron Beyer
    Apr 26 '20 at 19:09
  • \$\begingroup\$ Temperature measurement = finger on the package :) 60 degs celsius I'd say, I always turn it all off before it gets way too hot. The pull-down is probably not needed, and I have no polarized caps at hand, but I don't think that's what could be causing it, right? \$\endgroup\$
    – woko
    Apr 26 '20 at 19:25
1
\$\begingroup\$

at first: let's look at TC4451/2 datasheet: http://ww1.microchip.com/downloads/en/DeviceDoc/20001987C.pdf The TC4451/TC4452 are single-output MOSFETdrivers. These devices are high-current buffers/drivers capable of driving large MOSFETs and insulated gate bipolar transistors (IGBTs).

Then look at Functional Block Diagram, an output already are powerful mosfet. Drop your Q1 and all around, and connect led+resistor to output 6 OR 7 and + OR - accordingly. don't connect 6 and 7 together.

Second: if it just LED, ESP32 has enough 1ma current on output to drive led.

and finally: are you sure your Q1 not burnt? take it separately, connect to as on your diagram, and using var. resistor slowly increase voltage on gate from 0 to 5, at "0" it will completely shut.

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

The gate driver runs too hot.

1) is this built over a ground plane?

2) is there at least 0.1uF right at the gatedriver VDD, soldered to the GND plane, leads shorter than 3mm ?

2') if you use a socket, then inductance is a problem

3) is the Logic input source also located on the Ground Plane?

4) is the logic input smoothly swinging, with faster than 10 nanosecond edges, from 0V to +5v to 0V? are the logic inputs swinging without overshoot?

5) is the switching frequency < 1MHz

6) is the capacity on the gatedriver output less than 10,000 picoFarad? This needs to include any Miller Multiplication effect.

7) are you driving a DC load?

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

I've replaced the gate driver and it seems to work fine now with an appropriate current draw (in uA's). No idea what went wrong, but I suspect not connecting the duplicate pins initially could have something to do with it. Thanks to your answers I've also moved the caps close to the gate driver VDD so it could have helped as well.

\$\endgroup\$
0
\$\begingroup\$

The driver chip is likely pretty fast. These are intended to be used for driving power FET's in switch-mode power supplies, operating at a couple dozen kHz up to units of MHz. Whenever I hear about too much consumption in a fast circuit, self-oscillation is among the theories I need to test for. Hence my further comments:

1) you haven't shared a photo of your PCB with us.

2) do you have access to an oscilloscope?

===EDIT===

So you're on a solderless breadboard...

Speaking of oscillations, your driver chip model is non-inverting, which alone hints that there's room for positive feedback... but the "energy coupling" needed for oscillation can happen in various cunning ways. Long power suppy leads can radiate. The output FET switches apparently can "pack quite some punch" = are fast AND strong, can yield quite a bit of dI/dt.

I recall being slightly surprised by a stepper motor driver chip, that did not stay open for a long time. It had some sort of protection where it would shut off the output after some fixed period of input staying high. Like a "minimal PWM frequency requirement" from the PWM controller. But I tend to believe that this is not your problem. The datasheet mentions "local ON/OFF switch" as a particular application. This would suggest standard continuous operation (DC, active state unlimited by time).

If your goal is to test the driver chip, I wish you good luck getting to an oscilloscope. If your goal is just to blink a LED, you don't need the FET driver. I suggest that you use a small FET with V_gs_th low enough to accept drive by 5V TTL. I'd use a BS170 or 2N7000TA (note that they have different pinouts in the TO92 package). Actually your SUP70030 might be happy with 5 Volts at the gate too, but seems a bit of an overkill for just an LED :-)

\$\endgroup\$
3
  • \$\begingroup\$ 1) i.imgur.com/YR18I5Y.jpg - I've tried adding some resistors since posting this question, but apart from that the circuit is the same as before. 2) unfortunately not with the current lockdown :( \$\endgroup\$
    – woko
    Apr 26 '20 at 20:26
  • \$\begingroup\$ You should definitely put the power blocking capacitors as close to the driver chip's legs as you possibly can. And, if you need to bridge the GND and VCC pins (two of each kind together), those two traces should be pretty short as well. I guess you could improve the layout on your breadboard. Still, for fast circuitry, you should prefer soldered joints. On a somewhat tangential note (hopefully this is not your problem), FET drivers at full steam do dissipate heat. The thermals of a proper PCB would allow for that - possibly large copper areas soldered to GND and VCC. \$\endgroup\$
    – frr
    Apr 26 '20 at 21:15
  • \$\begingroup\$ Thanks for the pointers. This is my first time working with MOSFETs , so I just wanted to test it out on a breadboard, but I can see that it isn't necessarily as straightforward as hooking up a couple of ICs together. And I'm aiming to drive a 100V/200A load, so logic level MOSFETs won't cut it... \$\endgroup\$
    – woko
    Apr 27 '20 at 7:40

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.