• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 46 subscribers
  • Views 883 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADC12DJ2700: How long is "extended times" measured in seconds, minutes, hours etc?

Per Magnusson
Intellectual 500 points
Part Number: ADC12DJ2700
Other Parts Discussed in Thread: , LM95233

In note 1 on page 4 of the ADC12DJ2700 datasheet, it says that:

"Powering down the high-speed data outputs (DA0± ... DA7±, DB0± ... DB7±) for extended times may reduce performance of the output
serializers, especially at high data rates"

It is hard to know if this means that one second or a year is the limit at which one risks (say 0.1% chance) a noticeable degradation. Can you clarify the times involved?

Also, it would be interesting to know what the mechanism of the degradation is. Since the part is partially powered down, I doubt it is electromigration. Is it breakdown of thin gate oxides?

  • 0
    TI__Mastermind 44680 points

    Hi Per

    The key is to minimize the percentage of time the high speed serializers are powered down while the ADC12DJ2700 is powered up. Basically this means only seconds while in this condition.

    If you are considering using power-down mode to enable your end product to have a lower power mode of operation then the ADC should instead be completely shut down by turning off the power supplies.

    If you need to use the device in a number of different JMODES with different numbers of lanes being active, you need to use the JEXTRA_A and JEXTRA_B settings as follows:

    1) For the JMODE(s) with highest number of active lanes use the device normally.

    2) For the JMODEs with a lower number of active lanes use the JEXTRA_A and JEXTRA_B settings to enable extra lanes as needed to maintain the same number as in 1) above.

    I hope this is clear.

    Best regards,

    Jim B

  • 0 in reply to Jim Brinkhurst1
    Intellectual 500 points

    Hi Jim

    Thanks for your quick response. Mentioning "seconds" helps in understanding what is meant by "extended times".

    After reading the datasheet, I do not really consider using the PD pin for power down, but my main concern is that it would be good to be able to monitor the die temperature using the diode and for accuracy this apparently needs an offset calibration which should be done while the device is in power down according to the description of the PD pin. A reasonable way of doing this would be to start the device up with PD active so that it does not start warming up and then when the FPGA and API are up and running, the board (ambient) temperature and the diode voltage can be measured before the ADC is powered up. It is however quite possible that it takes more than a few seconds for the FPGA, API and other software to get going, so this strategy seems risky.

    Is there any downside to grounding the PD pin permanently and just not enabling the power supplies until the diode reference measurement has been performed? If not, why does the description of the PD pin mention that the diode calibration should be done with PD (and power) enabled?

    I am still curios as to what the physical damage mechanism is that happens when PD high and power is applied.

    Per

  • 0 in reply to Per Magnusson
    Intellectual 500 points

    Hi Jim,

    Is there any chance you could answer the following even if you might not want to say anything about how the chip breaks if using the PD feature:

    "Is there any downside to grounding the PD pin permanently and just not enabling the power supplies until the diode reference measurement has been performed? If not, why does the description of the PD pin mention that the diode calibration should be done with PD (and power) enabled?"

    Per

  • 0 in reply to Per Magnusson
    TI__Mastermind 44680 points

    Hi Per

    Sorry for not responding to the earlier question yet. Unfortunately I can't comment on the reasons that PD cannot be used for longer periods of time, other than it degrades device performance unacceptably.

    We did want customers to be able to do calibration of the on-chip thermal diode by using the PD mechanism for brief periods of time. In my experience the recommended temperature monitoring device (LM95233 as used on the ADC12DJ2700EVM) or similar devices are quite accurate and no baseline temperature calibration is needed. In those cases there is no need to ever use the register controlled or PD pin initiated power-down feature.

    I hope this is helpful.

    Best regards,

    Jim B

  • 0 in reply to Jim Brinkhurst1
    Intellectual 500 points

    Thanks Jim for the additional explanation. Still I wonder (for the third time):

    "Is there any downside to grounding the PD pin permanently and just not enabling the power supplies until the diode reference measurement has been performed? If not, why does the description of the PD pin mention that the diode calibration should be done with PD (and power) enabled?"

    To be extra clear: What I propose is to never ever use PD (to completely avoid the mysterious degradation process), but control the enable signal to the power regulators supplying the ADC and do the (perhaps not entirely necessary) diode offset calibration while the ADC does not receive power from the regulators, but a 100 µA current is injected into the diode.

    This seems to me like it would provide the benefits of the diode offset calibration with no risk of staying in the dangerous PD state for too long. What I do not know is if there is some hidden reason my proposed procedure will not work. Maybe there is since the datasheet does not mention this method and instead suggests using the dangerous PD mode.

    Best regards

    Per

  • 0 in reply to Per Magnusson
    TI__Mastermind 44680 points
    Hi Per
    The only concern with driving the TDIODE connections prior to powering up the ADC is that the temperature monitor might force the TDIODE+ or TDIODE- pins beyond the allowed voltages listed in the Absolute Maximum Ratings table. If those voltages will not be exceeded then your plan is fine.
    Otherwise, I recommend using PD briefly (a few seconds at a time is fine, up to 5 minutes cumulative can be tolerated over the lifetime of the device) during your prototyping stages to confirm the diode measurement method you are using is accurate and has minimal offset. I expect that will be the case, and you won't need to use PD mode at all in deployed systems.
    Best regards,
    Jim B
  • 0 in reply to Jim Brinkhurst1
    Intellectual 500 points

    Hi Jim

    Thanks. Now I think I know enough to figure out a good way of doing this. I do have a negative supply available, so one option is to drive TDIODE- slightly negative to avoid the abs max ratings when the ADC is unpowered and the diode receives 100 µA.

    Best regards

    Per