Dear TI,
On one of our old product, we are now using the DAC8812 (we were using AD5545 for many years). However, sometimes we have communication issues with the TI's DAC. We notice the problem by measuring the output level of more than 10mV of offset error (around 6-7 LSB).
If we use the IC in a reduced system, everything is working perfectly. When we are using a normal configuration on the same system, we have a problem on about a third of the IC, and always in the B Channel. The problem does not always occurs at the same IC's location from one board to the other. We tried to swap a 'good' and a 'bad' IC and the problem followed the 'bad' IC. We tested with a few systems and it is always the same behavior.
For debugging purpose, we reduced the SPI speed from 50 MHz to 25 MHz to ensure a respect of the hold and setup time. We measured the signals and the levels are above 3V for an '1' and below 0.3V for an '0'. The timing is good (>17ns of hold and setup) and the data are correct. We didn't notice differences when we were using the reduced system or the normal system.
The power is a clean 5V and we only notice a drop of 20mV when we were using the normal system. Noise level is good. The reference voltage is exactly the same (<0.5mV) between the systems' options.
Finally, when we momentarily shorted the DAC 5V power supply when SPI was working, we found that the IC is now working perfectly in all systems. It seems that breaking something in the IC solves the problem.
Does a weakness exists in the DAC design?
Is it only a bad batch of IC that I got?
Is it safe to ship IC that have been momentary shorted and is now working has expected, to our customers if we don't care about ESD?
Thank you
Francois
