Forums

Rail 12 does not have Voltage Monitor (Or Current Monitor or Temperature Monitor) + VOUT setting is all zero + Fault Response is set to VOUT_UV = Shutdown Immediately with Restart up to 1 time =>

This might be the issue : When fault occurs on Rail 11 for 5 – 10 ms, the glitch filter is 2 ms on Rail 11 => After 2 ms glitch time, rail 11 considered having fault => shutdown all slaved rails (1-7, and 12), Rail 12’s TOFF Delay is 8 ms but TOFF MAX WARN LIMIT is set to 0 => it is not clear when Rail 12 is leaving the power good state (8 ms or 0 ms); Beside, The Fault Response is set to VOUT_UV with shutdown immediately and Restart 1 time and Vout is set to all zero => rail 12 is ON or OFF,  VOUT_UV fault or not is un-predictable. The fault duration of 5-10 ms here might be the issue (2 ms glitch time + 8 ms delay time for rail 12).

Suggestion: Remove Rail 12 out of Rail 11’s FSS; Set Rail 12’s VOUT_UV fault response to ignore instead of Shutdown with retry.

Please see section 6.9 of the UCD90xxx Command Reference, especially this ppg noted below.

For devices other than the UCD90120 and UCD90124: When there is not a voltage monitor assigned to a rail, this time is used to determine when a rail has reached the power-good state after it is enabled. In this case, setting TON_MAX_FAULT_LIMIT to 0 will result in no delay between enabling the rail and entering the power-good state.

Rail 12 was used as an enable for the UCD9222.

I've removed this rail, and used an logic controlled GPO for this function.

But the problem still exist. I've attached the new configuration file.

I will try to emulate your system but it may take a few days to do so and duplicate the issue. In the meantime, if you are able to get any more detail with regard to the behavior please pass it on. For example, how long does the rail 11 pulse need to be before normal operation is seen? This may provide a clue as to how the firmware is operating.

3 scope images below are included:
top: 10.5 ms, power rails don't restart (Sequencer fails)
Middle: 10.5 ms, power rails restart fine
Bottom: 2-puls, power rails restart fine

OK, thanks.

Hi Eric,

I am working with UCD90120A device. I am wondering whether this is something, I need to consider on the UCD90120A device operation?

Hi Suthan,

Yes it is possible that this exists with UCD90120A too. I am still investigating to see whether any firmware rev difference exists between the two which could cause the observed behavior.

Hi Eric,

Can you tell me the status of this problem?

Is this problem solved now?

Thanks in advance,

Jacob

This problem can be re-visited if the firmware is updated.

Hi Eric,

I've tested now with a new, native UCD90124A device, with firmware v2.4.0.0, but the problem still exist.

Have you new firmware for me without this bug?

Thanks, Jacob

Hello Jacob,

Can you contact your local TI sales office and submit this E2E HTML thread so they can work with the appropriate internal personnel?

Hi Jacob,

Can you please provide me your original project file again. I want to look into the sequencing once.

Hi Spandana,

Here's the project file again:

3583.UCD90124A @ Address 117 Project - kopie.xml

Hi Jacob,

I'm thinking maybe the re-sequence option is causing this problem. You have configured to unlimited tries on re-sequencing and the time between re-sequencing is about 100ms. When the fault on Rail 11 is getting off, the raisl try to re-sequence in 1000ms but Rail1 doesn't get turned on because the ton delay is set to 3000ms on this rail. Also some other rails are dependant on Rail1 for their turn-on. So even these don't get turned on until 3000ms. But in this Interval since the Rail11 has recovered from the UV fault it attempts to re-sequence but unable to do so.

I'll try on a board in my lab by editing the re-sequence and see if it works.

Why this isn't a problem on UCD90124 is beacuse there is no Unlimited re-sequences available on this. I'll update you more soon.

Thanks

Hi Jacob,

Did you get a chance to work on this and edit the re-sequencing? or do you have any inputs for me?

Thanks, Spandana