ADS1218: ADS1218

Hi Haresh,

Welcome to the E2E forum!  First of all, which page in FLASH are you using?  Is it PAGE 0?  If so, the contents of the FLASH PAGE 0 is read at power-up (or for a RESET) and placed in RAM.  Are you sure that the FLASH is corrupted and there was not an issue at power-up that may have corrupted the RAM when reading the FLASH?  In other words, are the registers settings in RAM corrupted or is it truly the FLASH?

Generally the FLASH is very robust if programmed correctly.  The fosc must be within the range of 2.3MHz and 4.13MHz.  Also, the SPEED bit must be set correctly for the fosc being used when writing the FLASH.  SPEED = 0 for fosc between 2.3MHz and 3.12MHz, and SPEED = 1 for fosc between 3.13MHz and 4.13MHz. 

It may be possible for the FLASH to get corrupted if the supply ramp is not monotonic (steadily increasing with no negative dips).  If the inputs to the ADS1218 or the supplies have ringing of overshoots or undershoots there could also be an issue.  This type of behavior can exist if ground traces are used as opposed to a good ground plane.  Another possible issue could be inadequate supply bypassing (or decoupling).

Best regards,

Bob B

Thanks for the support. To your query " Are you sure that the FLASH is corrupted " The issue is confirmed to be a flash corruption of page 0 even after ensuring correctness of procedure and parameters for flash write. We were using only page 0 in our application.
Further analysis, brings to the conclusion that the FLASH memory has failed ( not just corrupted). When we now try to write to page 0, its getting written to page 2. similarly during write to page 2 , page 0 is getting written though it is not intended. This observation holds good for page 1 & 3, 4&6 etc. What could be the probable reason for this flash failure.

Hi Haresh,

A write to the wrong page of flash sounds more like a communications issue as opposed to an internal device decode issue. Do you have scope or logic analyzer shots of your communication? It would be very helpful for me to review the communication for any possible glitches during the flash reads and writes that could cause an issue. Showing me your schematic would also be helpful.

If the ADS1218 was working initially, it would be useful to look into what the operating conditions were before (when working) and what may have changed when the issue first presented itself. You initially mentioned a power off/on and then the flash values were corrupted. Is it possible that one of the Absolute Maximum Specifications were violated? For example, a powered sensor connected to the ADS1218 analog input(s) while the ADS1218 was powered down? Another consideration is if any changes were made to the software between working and non-working conditions.

Best regards,
Bob B

There are 16 ADS1218 devices in our application board with SPI tied together to a master microcontroller. The issue observed is not due to communication protocol error as only one ADC has shown the abnormal behavior. we have a production line for this package and this is the first of its kind observation.

We produce below our FAILURE ANALYSIS note

The board was powered and the configuration values stored in ADC flash memory
(First 16 bytes of page 0) of the channel were verified and found incorrect. Initial programmed values for the channel was 48, 01, 07, 00, 00, 00, 00, 00, 00, 71, 3B, A7, FF, 19, C5, 92. However after one of the power on, ADC output was incorrect. Subsequently the First 16 bytes of page 0 read were AA, 55, AA, 55, AA, 55, AA, 55, AA, 54, AA, 55, AA, 55, AA, 55. It may be noted that default values written in other pages ( except page 0) is a pattern of AA,55 alternatively during initial verification of the device( as a part of our health check procedure). All other ADCs ( remaining 15) were verified and found to be the expected programmed values. After the above incidence, the correct contents were written to flash and channel functionality verified ok with power OFF and ON. The observation occured only once.

Based on our failure analysis board recommendation, the package was powered and put ON and OFF 20 times and flash page 0 corruption did not occur. The waveforms on the 5V line were captured. WREN lines were captured. No anomaly observed. With the above tests, corruption of flash could not be recreated.

Subsequently, it is also observed that when the entire flash memory was read, the contents of FLASH page 0 and flash page 2 are same, although page 2 is not intentionally written. Write and reads to other pages of flash were done. It is observed that when page 0 is written, the same contents are being written to page 0 and page 2, although page 2 is not indented to be written. Also when page 2 was written, page 0 also is getting written with same values. Similar observation existed for other pages. (eg: pages 1&3, pages 4& 6 are getting written in pair although only one page is intended to be written).

The failed device was removed from board and new device was wired and channel functioning (including flash) ok without any change in software or circuit.

How can we explain the observation( abnormal flash behavior). Can we categorize this as a random device failure?

Hi Haresh,

I'm very sorry you are facing this issue.  I have not known of any flash issues with the ADS1218 accept when not following the correct write timing (as I stated in my first response).  I also stated making sure that the power supplies are clean.  We have seen on some different flash devices (not the ADS1218) that this can be an issue.

However, the issue with different pages having contents not intended appears to be some kind of addressing decode issue.  As this is not an issue that I have ever come across before I would say this is a random device issue.  Can you give me the information (or a picture) of the top of the chip so I can do some tracking history?  Also, how was the ADS1218 purchased (directly from TI, distributer, etc.)?

Thanks,

Bob B