4

I've received a prototype for this PCB I described in a different thread:

First PCB design, PWM Motor Controller feedback

I've changed the design based on the feedback I got there, I changed the gate resistors to 10 Ohm, removed the capacitor after the MOS driver. After finishing the programing I've measured everything. The PWM singnal generated by the Attiny, the PWM signal generated by the MOS driver. Everything looked fine. I put a motor and it spinned based on my adjustance on AIN.

After that I hooked the motor to some load, a planetary gear 1:25 with a small hydraulic press. I managed to output the desired pressure on the hydraulic press but D4, Q1, Q2 died and the motor kept spinning. They all became 0 Ohm resistors. No smoke and no bang, eveything looked fine until I released the trigger and the motor kept spinning. Everything else seems to still work fine, after I desolder D4, Q1 and Q2, expect that the motor can't spin anymore.

My question is, why did my design fail ? I measured 10A@14V, both MOSFETs parallel should easily be able to handle that load.

From the old thread:

Schematic: schematic

PCB design: pcb design

Datasheets:

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208
Stackoverflowsuperbad
  • 173
  • 9

2 Answers2

10

They got destroyed due to inductive kickback. The TVS diode is not a correct choice, you should use a schottky diode at that position.

When the MOSFETs were switched off, the motor induced a high voltage, that was clamped with TVS, but these devices are meant to surge a short pulse, not to dump entire motor energy, so it has blown, consequently MOSFETs also blown due to HV pulse.

enter image description here

Marko Buršič
  • 23,562
  • 2
  • 20
  • 33
  • 2
    Specifically the diode limits the drain voltage to something like 29.5V above VCC. And the FET's Abs Max VDS is 30V. Problem. –  Sep 07 '20 at 13:20
  • @Brian Drummond I found this diode, I hope this will work https://datasheet.lcsc.com/szlcsc/1809011210_STMicroelectronics-STPS3030CG-TR_C172486.pdf – Stackoverflowsuperbad Sep 08 '20 at 06:26
  • @Stackoverflowsuperbad You could try with a DO-214AB diode, first. In such case you don't need to make a new PCB: https://eu.mouser.com/datasheet/2/427/sl42-1768381.pdf You could also use through hole: https://eu.mouser.com/datasheet/2/115/SBR10U45SD1-464990.pdf – Marko Buršič Sep 08 '20 at 07:53
  • I tried this: https://www.onsemi.com/pub/Collateral/FSV2050V-D.PDF and no luck. Same again, diode and MOSFETs are dead. – Stackoverflowsuperbad Sep 10 '20 at 06:07
  • @Stackoverflowsuperbad How much tim did it last? Was blown after MOSFET turn off, or turn on. Are you sure you mounted the diode with correct polarity? – Marko Buršič Sep 10 '20 at 06:48
  • @MarkoBuršič I checked the polarity many times and measured it. I tried it with two different power supplies, the first with no load, than I put as much load as I could manually (not alot) but everything worked fine. At the other power supply I tried no load again, it died after the first try at just 2-3A, less than a second. Both supplies worked with 14V. – Stackoverflowsuperbad Sep 10 '20 at 07:11
  • @Stackoverflowsuperbad So it's probably due to the PSU. Your circuit doesn't have any bus capacitor across the MOSFETs, a lab PSU also has very small because the PSU shall provide fast transient response, so without capacitance the the PSU regulation loop went mad and you got a maximum PSU voltage on your circuit, supposing you have a linear laboratory PSU. – Marko Buršič Sep 10 '20 at 07:32
7

My question is, why did my design fail ?

Q1 and Q2 have a maximum drain-source voltage rating of 30 volts. The TVS will limit the drain voltage to Vcc + 24 volts plus anything up to 10 volts more across the full range of likely motor currents. If Vcc is anything above 6 volts then you have a bad design for a motor load current of milli-amps. If the motor load current is several amps then it's basically a bad design choice.

Use a fly-back diode is my advice.

You don't need a schottky diode by the way; just a diode that can handle the peak operating current of the motor.

A brief note about the TVS diode you have chosen: -

enter image description here

  • Firstly, the data sheet - very poor. It gives little to zero information about power handling capability - as a minimum for any TVS diode I would expect to see an energy-time graph so you can properly calculate that the device can handle the inductive (or otherwise) transient your circuit throws at it. This makes me very suspicious about them.
  • The company (Born-semiconductor) is from Shenzen, Guandong and they don't appear to have much of a reputation for this type of product. I wouldn't touch them with a barge-pole.
  • They are piggy-backing on Bourns semiconductor's good reputation for TVS diodes. Bourns always produce the power derating curve below: -

enter image description here

Here's the Bourns logo so you don't get confused.

enter image description here

Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • I tried this: onsemi.com/pub/Collateral/FSV2050V-D.PDF and no luck. Same again, diode and MOSFETs are dead. – Stackoverflowsuperbad Sep 10 '20 at 06:12
  • So, it works on one power supply and not on the other according to what you reported to Marko? – Andy aka Sep 10 '20 at 07:25
  • Yes, looks like a faulty power supply. But I really really doubt that. I can't explain it. Working power supply: http://uniroi.net/a/Products/20190624/72.html. Not working power supply: https://www.peaktech.de/productdetail/kategorie/schaltnetzteile/produkt/peaktech-1535.html – Stackoverflowsuperbad Sep 10 '20 at 07:33
  • It's probably output inductance of the power supply. You can probably prove this by adding local power supply decoupling capacitors at your circuit board to act as an energy sink when you deactivate the motors thus stopping the 14 volts instantaneously rising above 30 volts (powerline inductive kick-back). Are the leads from your power source a lot longer than what you'd expect in reality? – Andy aka Sep 10 '20 at 07:42