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.

OMAP-L138: CVDD vs RVDD without DVFS

Part Number: OMAP-L138
Other Parts Discussed in Thread: TPS65070

According to this guide... processors.wiki.ti.com/.../OMAP-L138_Hardware_Design_Guide

For supporting different voltage operating points, the following rules must be followed:

  • CVDD should not exceed RVDD
  • RVDD should not drop below 1.2V regardless of CVDD voltage or operating point

1. Can CVDD exceed RVDD if DVFS is not implemented (static supplies) ie. CVDD=1.3V; RVDD=1.2V ?

2. Can RVDD be operated anywhere above its Vmin=1.14V as per datasheet, or is 1.2V the actual Vmin as per above?

We have an old PCB design that is "pre-456MHz" but populated with 456MHz parts, and ties RVDD to the other 1.2V static supplies...

  • Hi David

    1) No CVDD should not exceed RVDD in any any condition.
    2) RVDD Vnom is 1.2V , Vmin is 1.14 V as listed in the datasheet.

    If you are using a 456 MHz device *and* planning to run the device at 456 MHz , RAM needs to be at 1.3V Nom as listed in the datasheet.

    Even thought the below post is a bit confusing , you might find it helpful
    e2e.ti.com/.../1922923

    Let us know if you have any follow up questions.

    Regards
    Mukul
  • Thanks, Mukul. The post you quoted above is actually what prompted this thread...

    To clarify, can RVDD be operated anywhere above its 1.3V Vmin operating point (1.25V as per datasheet) when CVDD is at say 1.300V, or is the valid 1.3V operating range for RVDD secondary to RVDD > CVDD as a hard rule.

    In other words, is RVDD = 1.290V and CVDD = 1.295V an illegal operating point?

    We're using the TPS65070 to supply rails, and are concerned about factoring in the voltage setpoint % tolerances...
  • Hi David
    I know the tolerances is tight, in general the expecation is that RVDD should be the same voltage as CVDD, unless CVDD is lower than 1.2V nominal (RVDD has to be at least 1.2V nominal)

    >>In other words, is RVDD = 1.290V and CVDD = 1.295V an illegal operating point?
    0.05V , I am think it probably will work ok, but will still be out of data sheet spec and we do not verify functionality - so cannot give you an official Ok for this.

    Hope this makes sense?

    Regards
    Mukul
  • Hi Mukul,

    The example I brought up in my last post is well within the datasheet spec for the 1.3V operating point for both RVDD and CVDD, but depends on how the supplemental RVDD > CVDD rule (not in the datasheet) is interpreted.

    The setpoints in this case are RVDD = 1.3V, CVDD = 1.3V. This follows the spirit of the RVDD > CVDD rule if it only applies to the operating point definition (full Vmin-Vmax range), but fails if it also applies to the measured voltages.

    In light of the supplemental RVDD > CVDD rule, suggesting a customer always operate RVDD at 1.3V is a nebulous statement if CVDD can be greater than RVDD by a few millivolts (due to supply tolerances) and be within the datasheet spec.

    I'd suggest you pass this up the chain, as this is pretty important stuff which should yield a concise answer. We have several canned L138/TPS65070 designs riding on the interpretation of the RVDD > CVDD rule.

    Thanks

  • Hi David

    Can you further share your power tree with TPS65070 ?

    >>I'd suggest you pass this up the chain, as this is pretty important stuff which should yield a concise answer. We have several canned L138/TPS65070 designs riding on the interpretation of the RVDD > CVDD rule.

    I manage this device family's escalations specifically from a architecture standpoint , so there is no one to pass this up to.

    The reason the guideline was put in the wiki is that based on the isolation logic design and RAM supply design, in the extreme case if RAM supplies were 1.2V and CVDD was 0V , this could potentially could cause excessive current on the array supply. It likely wouldn’t be destructive, but our general position is to avoid un-controlled or un-expected current draw since this can cause system level problems/failure.

    What you have is slightly different and that is why I said it is likely ok. Potentially more so if your application is on 24/7 and does not go through multiple power cycles a day etc.
  • Hi Mukul,

    We're using the standard recommended power topology for the TPS65070/OMAP L138 as per page 75 of SLVS950G. I'd be happy to forward you a set of schematics if you can specify where...

    As an example of our quandary, if we program VDCDC3 (CVDD) to 1.2V, and VLDO2 (RVDD) to 1.2V (a legal 375MHz operating point), this meets the datasheet Vmin & Vmax requirements for each rail.

    However, it is easily possible that VDCDC3 exceed VLDO2 by 10-20 millivolts on any particular board, and this violates the supplemental "CVDD should not exceed RVDD" rule depending on how it is interpreted.

    The example above is a pretty standard architecture and operating point for the OMAP L138.

    Does the rule apply only to whether a nominal operating point is legal (as per Clay Turner's post you referenced above), or does it apply to measured CVDD and RVDD voltages (which further narrows Vmin-Vmax beyond the ranges in datasheet)?

    Sorry for beating this to death, but I can't think of a more concise way of expressing the issue (and the dilemma) here.

    Thanks.

  • Hi David

    >>Does the rule apply only to whether a nominal operating point is legal (as per Clay Turner's post you referenced above), or does it apply to measured CVDD and RVDD voltages (which further narrows Vmin-Vmax beyond the ranges in datasheet)?

    the rule applies to what nominal operating points.  

    >>Sorry for beating this to death, but I can't think of a more concise way of expressing the issue (and the dilemma) here.

    No issue, I appreciate the intent on due diligence for clarifications. 

    Regards

    Mukul 

  • I appreciate the response, Mukul.

    So...I need clarification on the following point you answered earlier...

    >>In other words, is RVDD = 1.290V and CVDD = 1.295V an illegal operating point?

    0.05V , I am think it probably will work ok, but will still be out of data sheet spec and we do not verify functionality - so cannot give you an official Ok for this.

    This would appear to be a perfectly valid operating point well within the datasheet spec for CVDD = 1.3V, RVDD = 1.3V operating point, so I'm confused by your answer.

    For this same operating point, RVDD = 1.260V and CVDD = 1.340V is also within the datasheet spec (a difference of 0.08V) - even though CVDD > RVDD...

  • Your subsequent response

    The setpoints in this case are RVDD = 1.3V, CVDD = 1.3V. This follows the spirit of the RVDD > CVDD rule if it only applies to the operating point definition (full Vmin-Vmax range), but fails if it also applies to the measured voltages.

    Clarified for me what you are doing. So my previous response does not hold . I am just trying to ensure design is not such that CVDD nominal is 1.3V and RVDD nominal is 1.2V . I also explained the intent of the line in the wiki to ensure we comprehend the worst case situation of 1.2V on CVDD and 0 on RVDD.

    Regards

    Mukul 

  • Thanks, Mukul - that's what I suspected.

    One final query related to this...

    During power sequencing, there is a brief condition where CVDD = 1.3V while RVDD = 0V (CVDD turns on before RVDD). I assume this falls outside the scope of the CVDD < RVDD rule?

  • The power up sequence in the datasheet has the following in the end

    RESET must be maintained active until all power supplies have reached their nominal values

    So the condition you are describing, if with RESET/PGOOD  maintained till ramp is ok

    Regards

    Mukul

  • Thanks for your help!