• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 97 subscribers
  • Views 1749 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.

Importance of common OSC and RTC GND on AM3352 PCB artwork

LeonardEllis
Guru 71526 points

During a schematic Design Review, the following comment was made:   Refer to AM335x Errata – Advisory 1.0.30. OSC_GND and RTC_GND should not be isolated grounds and you should connect the VSS_OSC and VSS_RTC terminals and respective crystal circuit component grounds directly to the common PCB digital ground plain.    .

If the GNDs are not connected together, and the OSC components are not tied directly to the GND plane, could this cause RTC RESET when switched to RTC-only mode?  What we are seeing with this design is an occasional RTC being cleared (RESET) when changing either TO or FROM RTC-only mode.

~Leonard

 .

  • 0
    TI__Guru**** 393215 points
    Hi Leonard,

    Advisory 1.0.30 is mainly related to erratic clock behavior. Have you monitored the RTC reset line for possible spikes when the issue happens? I would suggest that you also tie the oscillator GNDs to common GND with zero ohm resistors to see if this will eliminate the issue.
  • 0 in reply to Biser Gatchev-XID
    TI__Guru 63125 points

    Leonard,

    Several customers experienced random unexpected jumps in time, forward and backwards, when using the AM335x TIMERs that were being clocked directly from the oscillator outputs.  Most customers were seeing an issue when using the 32KHz oscillator as the clock source.

    I’m not aware of anyone having issues with the RTC, but it may be possible.

    We determined the original ground connection recommendation allowed noise to easily couple into the crystal circuit which could generate clock glitches as the slow changing, noisy input signal crossed through the switching threshold of the oscillator input buffer.  These clock glitches can produce very unpredictable behavior in digital circuits since the short duration glitch can over-clock the synchronous circuits.  This occurs when the glitches occurs at random times relative the expected periodic clock edge.  Clock glitches could cause the symptom you describe, but they normally create multiple symptoms with unpredictable results.

    The best way to determine if this is the cause, select a board that reliably reproduces the problem and temporarily replace the crystal circuit with a 1.8v LVCMOS oscillator that produces a reference clock with fast rise/fall edges.

    It is possible clock glitches is causing the problem you describe, but sounds like something else.  They may need to look at the RTC reset source to make sure noise is not coupling to it and resetting the RTC.

    Regards,
    Paul

  • 0 in reply to peaves
    TI__Guru 71526 points

    Is there a way to measure/'scope capture either of these events?

  • 0 in reply to LeonardEllis
    TI__Guru 63125 points

    You cannot measure oscillator glitches since they only exist inside the AM335x device.  They will not be able to see the glitches, even when routing the oscillator output to a CLKOUT pin because the output buffer connected to the CLKOUT pin is much slower than the internal clock signals and filters the glitches. 

    The best way to determine if the clock is getting glitches, is to temporary provide an alternate LVCMOS reference clock with very fast rise/fall times.  If the problem goes away, you know noise was coupling into the crystal circuit.

    It may be possible to capture a unexpected RTC reset on a very high bandwidth scope if they use a high bandwidth low capacitance FET scope probe that is connected directly to the RTC reset pin and nearest ground without connecting any wire leads between the probe and test points.

    Regards,
    Paul

  • 0 in reply to peaves
    TI__Guru 71526 points

    Thank you for this very useful and concise information.  Very helpful and informative!