• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Subscribers 62 subscribers
  • Views 731 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.

UCD9090: Dependencies off not working

Clément Letonnelier
Expert 2826 points
Part Number: UCD9090


Hi,

We are using the UCD9090A to sequence our secondary power supplies.

We have the following power supplies:

  • Power rail 1 is our main input power / Has no enable
  • Power rail 2 is our 5V secondary voltage / Has an enable
  • Power rail 3 is our 3V3 secondary voltage / Has an enable

We want to achieve the following:

  • Power rails 2 and 3 shall start when Power rail 1 is in Power Good state
  • Power rails 2 and 3 shall stop when Power rail 1 is in not in Power Good state

To do that, I have done the following configuration:

  • Set the adequate thresholds for Power Good ON and OFF thresholds on the Power rail 1
  • Set Power rail 1 in ON dependencies in Power rails 2 and 3 configuration
  • Set Power rails 2 and 3 as OFF dependencies in Power rail 1

The startup phase works properly, when Power rail 1 raises above the Power Good ON threshold, both Power rails 2 and 3 enable goes high and the rails start.

However, when Power rail 1 goes goes below the Power Good OFF threshold, the Power rails 2 and 3 enable stays high and the rails doesn't shut down.

I thought I had made a mistake in configuring the UCD9090A so I removed the OFF dependencies from Power rail 1 and set Power rail 1 as OFF dependencies in Power rails 2 and 3 configuration.

Same result. I was able to somehow achieve that behavior by using under voltage faults but that's not the intended configuration...

Am I doing something wrong or is there a bug there ?

Also, for ON dependencies, I was trying to have a fourth power rail in order to have power rails 2 and 3 starting only when both the Power rail 1 and the Power rail 4 are in Power Good state so I put both those rails in ON dependencies of Power rails 2 and 3.

However, those rails seem to start when one of the two dependencies is met while I was expecting it to wait for both.

That's another strange behavior. What did I do wrong ?

Looking forward to some explanations,

Thanks

Clément

  • 0
    TI__Guru 67130 points
    Sequencing on/off dependeincecs will not initialize the power up or power down. they are the conditions to be met when the the power up or down request is issued.
    That is why when Power Rail 1 goes below PG_OFF, rail 2 and 3 are not off when rail1 is set to the turn off dependencies.
    UV fault is the way to turn off rail2 and 3. why is this approach not prefered?

    As for your 2nd question, i would like to get your project file? rail 2 and 3 shall wait until both rail 1 and 4 are above power good threshold.
    Regards
    Yihe
  • 0 in reply to yihe
    Expert 2826 points
    Not sure to understand your answer, we tried both cases of setting off dependencies and the result is the same.
    If it is not supposed to sequence off the power supplies then why are they off dependencies ?

    I forgot to state that we are in auto enable mode.

    UV fault approach was not foreseen as have other conditions where we want to sequence off our power supplies where we only want to have one resequence for safety purpose.
    If we were to implement UV fault approach as our monitoring power supply is powered by a "saved" power, we would not restart once the unsave power goes back.

    Regarding your question related to the project file, I'll have to get back an old version as we changed the way we did it.
    I'll get that to you tomorrow.

    Regards,
    Clément
  • 0 in reply to Clément Letonnelier
    Expert 2826 points

    Hello,

    Has someone been able to review my answer to see what's wrong with our usage or understanding ?

    Thanks,

    Clément

  • 0 in reply to Clément Letonnelier
    TI__Guru 67130 points

    As said in the original post, the off dependenceis are conditions and they will not initialize the power off.  That is how the features is defined and designed.

    UV fault is the way to go.

    let me ask you a question. after rail1 is below  power good off, rail #2 and #3 will be off sequentialy. After these, how would you bring the both rail 1-3 back? Do you want to they stay at off?

    Could you please describe this with more details?

    "If we were to implement UV fault approach as our monitoring power supply is powered by a "saved" power, we would not restart once the unsave power goes back."

    Regards

    Yihe

  • 0 in reply to yihe
    Expert 2826 points
    Just for my understanding, can you please clarify then what OFF dependencies are meant for ? I can't seem to get it from your first post.

    Once rail #1 is below power good off, we want rail #2 and #3 to come back when rail #1 goes above power good on.

    Wrt the following:
    "If we were to implement UV fault approach as our monitoring power supply is powered by a "saved" power, we would not restart once the unsave power goes back."

    We have the below requirements (#2 rail here but similar requirements for #3, obviously I have changed the text to cope with our #1/#2/#3 example above):
    1/ UCD shall disable the power supply rail #2, ie. de-assert enable output, when either the rail #1 goes below power good off or an over-voltage fault is detected on it for more than 500µs.
    2/ When one of the controlled power supply rail is disabled due to an over-voltage fault detection, UCD shall automatically disable all other controlled power supply rails.
    3/ When one of the controlled power supply rail is disabled due to an over-voltage fault detection, UCD shall automatically try to re-enable the faulty lane and the other controlled power supply rails after 1ms.
    4/ If a re-enabling attempt fails, UCD shall not re-enable any controlled power supply rails until UCD is reset.

    So basically, what I was trying to state in the sentence you asked explanations for is that in case of over-voltage fault we only want one resequencing attempt and then we want all rails to be shut down for good (until next global power-up sequence) if the resequence attempt fails (ie. fault still present).
    Therefore, if we use the under voltage fault for power-off sequencing (which we have to as per your posts), I was concerned by the fact that we would be blocked due to that other configuration (ie. 1 failed attempt on OV => gone for good).

    Hope I clarified a bit my initial sentence.

    Clément
  • 0 in reply to Clément Letonnelier
    TI__Guru 67130 points
    I would recommend section 4.7 and 4.8 www.ti.com/.../slua815a.pdf to understand features of sequencing dependencies.
    what your described is a typical usage of the fault resequencing.
    what you concern is that the only one resequencing try will be exhaused by the under_voltge fault which block your other re-sequencing. but if the under_Volage can not recovered, do you still want to do the re-sequencing for other rails's OVER VOLTAGE fault? your re-sequencing for other rails would not solve the under voltage case anyway. your system is still not healthy.
    Regards
    Yihe
  • 0 in reply to yihe
    Expert 2826 points

    Good morning,

    Thank you for your answer, reading 4.7 and 4.8 along with some other paragraph of the document clarified  your answer. We should indeed be using the under voltage feature or change in our  V2 design to be CONTROL based in order to finely manage our sequence ON/OFF.

    Under voltage is on our primary power input so if we loose it, it means we are going to power off state. So it can be recovered, through a power on.

    Anyways, I think you gave me clear information so I'll now review them and apply a solution accordingly. I'll come back to you in another post if I happen to have further questions.

    Thank you for your help,
    Clément