THS3491: Part oscillates when warm

Hi Steve,

This circuit is very sensitive to the way it is configured and can easily result in an oscillation. I was wondering if you could share the circuit and layout you implemented. Are you using the same circuit/layout as the reference we have available? I was also wondering if you can share the load being used with this +/-10V output. Is one of the devices going into thermal shutting down by any chance?

Best Regards,

Ignacio

Hi Ignatio, I can certainly take a look if you can send me a link to upload. BUT trace lengths are kept very short and the issue only occurs when the ths3491's get warm. Being a bootstrap and all parts are very close it is difficult to tell which part is overheating. the supplies are =/-31V and the waveform was +/-20v 2MHz output square wave and the supply currents were around 50mA each so an average of 3.1W of power so at 11C/W heat sink I'd figure a temperature of around 55deg/C board temperature. Also don't think it was shutting down problem looked more like random oscillation asynchronous to the output frequency. Also this was still air with a single Digikey HS406-ND heatsink as used on your evaluation board.  BUT if it were thermal shut down I'd expect to see it either run or shut down but not continue to give square wave with almost full amplitude oscillation superimposed. Or are you saying the thermal shutdown is momentary without much hysterisis ?

Best,

Steve

Hi Steve,

There is a schematic for the circuit we implemented in our bootstrap design. I added the link to the page below. Could you share the schematic of your circuit as well as the signal you see when the device is in this oscillating state? I agree it does seem to be a stability issue not so much the device shutting down as you see your original signal with an oscillation superimposed. 

TIDA-060033 reference design | TI.com

Best Regards,

Ignacio

Hi Ignatio, my circuit is virtually identical to the TI reference design except for resistor values. We use all 3K Ohm resistors for the bootstrapping to minimize capacitive lag. Our gain = x4, Rf =1536 Ohms which should lower the loop bandwidth so should be more stable not less. Also the reference design burns a lot of extra power in Rf. For 25V out Rf has 22.5V drop so power in Rf=1Watt so prefer lower bandwidth lower power of our version unless high Rf is the cause of the problem ? If you still need a copy of the layout send me an email or upload link. Best Steve

HI Steve,

That is correct the increased Rf should result in a more stable circuit. I sent a friend request so you could send me your layout privately.

Best Regards,

Ignacio