ADAU1466 overheat ?

Hi !

I would be interested in what happens to an ADAU1466 if it overheats, to compare that to the strange behaviour that I am currently experiencing.

In the current revision of my DSP board, I am still using the same supply scheme as on the eval board; meaning an ADP3338 regulator for 3.3V logic and a STD2805 transistor to regulate the core supply (DVDD). 

I'm guessing that is the main source of heat because it's fairly high current. It's hard to tell how much comes from the DSP itself, because the board is thermally optimised and heats up very evenly  That is of little use however, since it evenly becomes untouchably hot.

I'm starting to regret not having switched to switch mode regulators right away.

Anyways; I have noticed that the DSP starts acting up one it gets warm. Im not 100% sure however, if heat is the cause of the issue.

What happens is, that all inputs are showing pulses of full scale signals even if none are present. This happens simultaneously on all inputs. 

Outputs remain functional, but highly distorted (if an internal sine generator is used for example )

Communication via I2C remains functional.

These problems disappear if the board is turn off, cooled down and restarted. Restarting it immediately (while still hot) doesn't solve the problem.

I'd like to know if this behaviour can be attributed to overheating, or if one of you has any other experience to share in regards to this issue.

Thanks !


  • 0
    •  Analog Employees 
    on Jan 13, 2020 3:11 PM over 1 year ago


    In my experience, the '66 just mutes when it overheats, though the behavior is technically undefined. This is because the logic slows down and stops meeting timing requirements.Can you measure the package surface temperature?

    I'm not exactly sure what you mean by the inputs becoming full scale pulses. Are you still able to read this back through the slave port? If so, then most of the IC is still operational, and I would look for other explanations.

    At what clock speed are you running the inputs? For example, it wouldn't be too surprising for a TDM16 signal to stop making time where a stereo signal might, conceivably, still work

    My instinct is that some other part of the circuit is having trouble, though. Most likely, the core logic running at 300 MHz would stop making timing before the serial ports,.