Part Number: ADS1258
Good afternoon,
We noticed a very small percentage of boards were having startup issues and thought that it was related to the crystal. The first symptom we saw was that after setting the START, DRDY never went low. While investigating, we saw that on boots that this behavior was observed, CLKIO never output and we diagnosed the crystal startup to be the issue. We also noticed that this effect (on the boards that showed this issue) was aggravated by the cold (we started testing between 0C and -20C to validate potential solutions).
As the investigation continued, we noticed that particular ICs were extremely sensitive. Using the negative resistance model for crystal startup safety factor (outlined in section 4.2 of slaa322d.pdf) we determined that our "good" boards had a safety factor greater than 8 (even at -20C) while our "bad" boards had a safety factor of less than 2. After a multitude of tests, we eventually reworked two boards (one good and one bad) to swap the ICs and found that the problem followed the IC (which allowed us to conclude that it wasn't parasitics of the board or tolerances of the surrounding components).
In the end, we decided to pursue an external oscillator as we were worried that the difference between a "good" and "bad" IC might have to do with the oscillator circuit.
This is when we discovered something quite interesting.
The impedance of the CLKIO pin (on "bad" ICs) seems to be much lower at cold temperatures - enough to prevent boot up. I took a cold board and measured the CLKIO waveform while it warmed up. In the following waveform, CH1 (yellow) is AVDD; CH2 (blue) CLKIO input; CH3 (purple) EXTOSC output; CH4 (green) DRDY. Note that there is a 50Ohm series resistor between the external oscillator output and the CLKIO input. It seems to show that as the board warms up, the CLKIO impedance seems to change, allowing CLKIO to swing between the DVDD/DGND rails and finally the conversions start (DRDY pulled low).
This is only a small amount of ICs that have this behavior (something in the order of 1/30). That being said, we are seeing this directly out of production (controlled handling), so we do not expect it to be damage to the IC. Because so many people have reported crystal start up issues that seems to disappear then reappear, I am worried that this might be the underlying cause.
Have you had any one else that has seen issues with the CLKIO input? Do you have any ideas to continue the investigation?
We are looking for a solution to guarantee we will not have production rejects/reworks or even in-field failures due to this issue.
Best regards,
