Hi, we're seeing random failures of ADS1274 converters in our application. There does not appear to be any problem with supply voltages going out of spec, and our drivers are very similar to the eval board except we have Schottky diodes to prevent signals from going much below AGND. The ADC talks to an FPGA. We don't see any correlation with temperature- we've run it up to Ta > 60°C in our environmental chamber several times. We've set scopes to capture transients on the supply lines and don't see anything odd.
In this set of boards, the therm pad is floating and we don't deliberately sequence the supplies (they are derived similarly to the eval board).
The symptom is that one channel fails (for example, channel 2), producing a "railed" output equivalent, though the digital output and analog inputs look fine.
Any ideas?
Hi Spehro!
Welcome to our forum! Can you possible post a schematic snippet of your analog front end and give us a little more detail about how you are using the ADS1274? If you can provide details on the operating mode, TDM vs Discrete output, speed/oversampling details; that would be helpful. When you say the part 'fails', is this just an erroneous conversion result or is the ADS1274 physically damaged (i.e. needs to be removed/replaced)?
Regards,
Tom
Hi, Tom:-
Mode: Mode 01, high resolutionOutput: Frame sync
We have a mix of ADS1274 and some ADS1278 in the same circuit. These are the most recent occurrences:
Failure occurrence 1: No data for any channels.Failure occurrence 2: Serial data for one channel (it may have been ADC 2, but that is not certain) was unchanging regardless of input. Data for other channels appeared to be correct.Failure occurrence 3: Serial data for ADC2 corresponds to full-scale positive value, regardless of any input voltages on the associated pins. The data for other channels is still present, and appears correctly correspond with the actual values.
I've attached schematics. Not shown are schottky diodes from each analog input to ground to clip negative signals.
Hi Spehro,
Were you ever able to solve this?
Generally, the powerpad is directly connected to the die substrate, and that a connected should be made to the lowest potential, in this case GND. This is not always the case in that the die may not be directly attached to the power pad but rather connected with some sort of thermally conductive, not electrically conductive, material. Regardless, In the case of the ADS1278 layout, we recommend having the powerpad connected to the GND potential of the board. On our EVM board we use 9 Vias to connect the ground to the power pad. I tested a part to verify that, with this package design, the power pad is connected to the substrate with a non electrically conductive material. Though, I would recommend tying it to GND, it should not cause one channel to fail at powerup.
Tony Calabria
Hi, Tony:-
Apparently not- it seems to have happened again last week (all channels dead). We'll replace the part and see if that fixes it.
We had some suspicion that it could be a transient on the power supply voltages and added clamps, but that does not seem to have made any difference (we never observed anything on the PS rails, even over extended testing). Strange that it does not seem to happen at power-up. Any other suggestions?
Best regards,
Hi, Tony:- No, not solved. It's happened twice more. My working hypothesis right now is that it's power-supply related (the supplies are galvanically isolated) but the fact it usually only kills one channel of the ADC at a time has us scratching our heads.
Hi again Spehro,
Sorry to hear that you are still having issues. One thing to check perhaps is that the analog input voltage is not exceeding the power rail. If you by chance have an analog signal applied before the rails are fully established, you could blow out the analog input. Please let us know when/if you do find the problem.