MCP4822 fried - possible reason (?)

[atferrari]

Testing a DAC MCP4822 (data sheet **broken link removed**) I noticed it hot to touch with 2V at the A output. Last settings transmitted through the SPI interface were to have just 0,5V at VoutA.

Additional consumption current with that IC in-circuit was as of about 50 mA. No surprise it was hot.

After discarding a short, I found that:

VoutA - pin 1 (and VoutB - pin 3 for the case) had no load resistors.

Both outputs had 4u7F tantalium caps as filters.

The software had an error that never latched VoutA when values were being progressively decreased. In other words, after going high, VoutA never came down.

The test I designed, involved cyclic alternative reset of both outputs, starting with very low values. There, VoutA was effectively (and abruptly) updated.

I corrected all that and the new chip works OK with a consumption of around 1,5 mA.

From all the above, what could be the reason for the damage? Maybe the abrupt changes with a big charged cap at the output and no resistor to drain it?

I expect to learn something from all this. Comments are welcome.

Successful design, let me tell you...

 

[Mr RB]

Doesn't the datasheet have a maximum load it can drive? It sounds like you had low impedance tantalum caps directly from the output pins to ground?

I think it would be wise to put a series resistor between the output pin and the tantalum cap, but I'm sure the datasheet will say if that is needed with a spec and/or a suggested schematic etc.

 

[atferrari]

According to datasheet

Doesn't the datasheet have a maximum load it can drive? It sounds like you had low impedance tantalum caps directly from the output pins to ground?

I think it would be wise to put a series resistor between the output pin and the tantalum cap, but I'm sure the datasheet will say if that is needed with a spec and/or a suggested schematic etc.

Yes, directly to ground.

Figure 6-1 in pag 25, "Typical Connection", shows the caps exactly like that. No resistors. That's why I did it.

Immediately above the figure, datasheet says: The tested range for stability is .001 uF to 4.7 uF. I went for the biggest ones!

Later I found that the test conditions included a 5K load resistor.

 

[Mr RB]

Thanks for the info, it appears the output driver can drive 4.7uF caps directly.

I'm not sure I understood the test you used, but you said about "cyclic alternative reset of both outputs"... If that means the output was switched from full volts to zero repeatedly like a square wave the caps look a lot like short circuits and there will be significant current every half cycle. It's rare for DACs to be used like that especially with large caps on the output and might still be worth some investigation there.

Otherwise I'm just guessing, it's unlikely your chip blew to a shorted state from any data you sent on the SPI inputs, so it's down to some overcurrent issue like maybe a shorted lead from soldering or accidental probe touch or something?