DRV8835 motor controller burns

[rasul]

Hello to everybody! 
I got a really big issue with DRV8835 motor controller and can't handle it over month  . Please help me!
I did some kind of game based on Arduino and Pololu zumo bot shield.
You can watch it here:

As you can see these toy cars are used in a really agressive manner. There are a lot of motors stall and reverse conditions.
After few days of work DRV8835 on these cars becomes not working. Soldering another DRV8835 instead broken one fixes the problem for another few days.
Here the type of motor I use https://www.pololu.com/product/2215. 120 mA with no load and 1.6 A at stall.
And DRV8835 which can supply up to 1.5-A of output current. With internal shutdown functions provided for over current protection, short circuit protection, under voltage lockout and overtemperature.
I did some measurments using oscilloscope and found that motors current consumption using my battaries are not higher than 600 mA even in stall conditions.
Howevever I got this graph on 0.68 Ohm resistor connected in series with Vbat terminal. I used two motors simultaneously reversing directions at full speed every few seconds with no delay time to simulate worst using conditions. To calculate one motor current your should substract 200 mA (Arduino and other stuff consumption) and divide by 2.

You can see here that here some spikes with current about 3 A on each motor.
Voltage measurements showed that there is no voltage spikes on any supply terminal. I have a lot of noise of course on Vbat but Vrms even lower than with no load.   
How these can happen with DRV8835 with its internal shutdown functions for over current protection?
Do I miss something fundamental?
Are there any way to save my motor controller from burning for my next design using this motors?
I feel myself so stupid right now...

[michaeliv]

The current protection is not instantaneous, it takes some time to kick in. Seems to be 1 micro second, from the datasheet. So, you can have high spikes that will show up and may damage your controller if high enough. the datasheet mentions that overcurrent protection works only up to 3.5A, so if you exceed this theoretically you will damage your controller.
Your motor is also probably back-feeding into the controller which may damage it.
If power consumption is not an issue, you can add a current limiting resistor in series with the motors to limit max current, and / or a schottky diode to limit back-feeding into the controller.

[rasul]

Thank you very much michaeliv for response!
I guess I will use more powerful motor driver.
Can you advise me driver which can handle this motor without a problem?

[michaeliv]

DigiKey has nice filtering for controllers, you can browse and look at the datasheet for something with at least 7A current protection that also matches your motor.
http://www.digikey.com/product-search/en/integrated-circuits-ics/pmic-motor-drivers-controllers/2556626?FV=fff40027%2Cfff802d2%2C1b80362&mnonly=0&newproducts=0&ColumnSort=-129&page=1&stock=0&pbfree=0&rohs=0&k=&quantity=10&ptm=0&fid=0&pageSize=500

[matkar]

Are you sure the reason for failure lies in overcurrent? Do the drivers overheat?
I'd look into overvoltage first. How is your driver protected against induced voltage spikes? Some TVS diodes on motors might be a good start.

[rasul]

Dear matkar and michaeliv, thank you very much for your responses!
Sorry for beeing late.
matkar I did put 9v zener diode on Vmotor rail and 47u capacitor. It didn't help. Did I something wrong? Also I observed that during full speed reversing there were not any voltage spikes - conversely the voltage were getting lower (they are 4AA battery supplied).
I have not any experience in motor controllig applications. Maybe I don't understand something important.
michaeliv why 7 amps? Is it like a good rule of thumb? Is it an overkil for 1.6A stall current motor?

[matkar]

I meant a diode on the motor terminals. Like here: http://www.rmcybernetics.com/images/tutorials/tvs-diode.jpg
I imagine you spin the motors both ways so a bidirectional TVS is required. Choose something from: http://si.farnell.com/webapp/wcs/stores/servlet/Search?catalogId=15001&langId=386&storeId=10175&categoryId=700000006296&showResults=true&aa=true&pf=110452410
DO214 size should be OK. Look for a diode with standoff voltage just above your supply voltage and clamping voltage below absolute maximum driver output voltage.
It is a good idea to place a small capacitor across motor terminals as well to take the edges off.

[rasul]

matkar thank you again!
Your point seems really interesting. I will try to use TVS diodes to check if it will save my controller.
Looks like I should check for voltage spikes not a Vmotor pin but motor teminals.
I going to choose and order some TVS diodes. 

[matkar]

While you're waiting for TVS diodes you could try with two zenners mounted back to back in series across motor terminals. They are usually slower than TVS but they might be fast enough in your application. Place a few nF ceramic capacitor in parallel to the zenner pair as well.