[Moved] Analog Power Amplifier - OPA549 Shuts Down Unexpectedly!

[tms8c8]

Since this is my first post, a little background: I am a physicist but do a fair amount of microcontroller/embedded system development for both my small (hobby) business and my full time job. I also find myself doing a variety of other EE type tasks.

I recently built a special process control board around an 8-bit AVR microcontroller (ATMega328). Essentially, it is a digitally controlled analog power supply. I write values to a 12 bit DAC based on sensor input and this value is amplified by a TI OPA549 to drive a resistive load.

The load is about 25 ohms. My controller is designed to switch the amplifier power bus from 12V to 24V so the effective linear range is about 0 - 20 VDC. When the DAC value corresponds to an output of great than 10 VDC, the controller switches the amp power bus from 12V to 24V. This is done because the device is expected to run the majority of the time at an output of about 10V. With the 12V supply, this is less power dissipated than when running with the 24V supply.

Now the problem:

When I am running for a period of time, the OPA549 goes into thermal shutdown. It is mounted to a chunk of scrap aluminum via acrylic thermal pad in the prototyping stage, but neither the block nor the tab gets particularly hot. By my figuring, it only needs to dissipate about 0.8W when running at 10V (12V bus on > 2V drop across amp, 0.4 amps to load).

Despite the above facts, the status pin goes low and it won't reset unless I toggle the power. Any idea what is going on??

BTW, the status pin is connected to a dedicated ADC channel through a 470 ohm resistor. Attached is a schematic of the offending parts. Shown are actually two OPA549 and an extra 48V switch, but the prototype version only has 12V or 24V capability. Also, this was a very rushed project so the schematic is messy.

View attachment Micro_and_coms.pdf
View attachment PowerAmp.pdf

 

[tms8c8]

After spending some time on this, I believe the problem may have to do with switching the power rail. I suspect that the power rail dives to 0 for a short period of time during a switch (or during unusual power-up conditions where the 5V bus comes up before the power rail). When this happens, the amp seems to go into thermal shutdown and not recover, even though it isn't hot. I cannot verify this with a scope at the moment because the hardware is inaccessible.

 

[crutschow]

You could avoid the any momentary dropout during voltage switching if you connect the +12V bus to the amplifier through a Schottky diode (for minimum voltage drop). Then when you switch in the +24V bus the voltage will simply rise from 12V to 24V with no transient, with the diode isolating the +12V bus.

 

[tms8c8]

You could avoid the any momentary dropout during voltage switching if you connect the +12V bus to the amplifier through a Schottky diode (for minimum voltage drop). Then when you switch in the +24V bus the voltage will simply rise from 12V to 24V with no transient, with the diode isolating the +12V bus.

Diodes are planned for the next round of PCBs. Good thought on the Schottky - novice that I am, I was thinking as a physicist (ideal diode) and forgot about Vforward.