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.

Clock Source Question

The C5505 Data Sheet states that one can either use a 32.786KHz crystal, or use a ~12MHz external clock source.

Can one not use an external 32.768KHz clock source? Wouldn't the PLL work with that, since the crystal input would be oscillating anyway, so why would the PLL know the difference?

 

My reason for considering this is that the RTC needs about 7uA minimum to keep running, and would need its own dedicated 1.05V power supply. If one needs the wakeup feature, the power consumption goes up and one needs TWO holdup supplies.

For about the same cost and effort, one can use a standard RTC chip that uses less than 1uA and also has the ability to wake up a system. So the built-in RTC was a nice idea in the C5505, but with that kind of support requirement, it's not necessarily cost-effective.

So if I use an external RTC chip, that would have it's own 32.768KHz clock output that could then be used to drive the RTC X input on the C5505. Otherwise, one needs two 32.768KHz xtals - one for the external RTC device and one for the C5505.

Any ideas and thoughts?

 

  • Oh, and I meant to state clearly that the idea would be to drive the RTC XTAL input with the external 32.768KHz clock, not the CLKIN that expects 12MHz.

     

  • Hello,

    Let me ask you some questions:

    1) Are you going to use boot-up feature?

    The bootloader code expects 12MHz to PLL intput and that is why we said that CLKIN should be 12MHz. If you do not care about the bootloading, you can use any frequency to CLKIN

    2) Are you planning to shut off entire C5505 including the built-in RTC and power it again by the external RTC when it is needed to be powered? I'd like to get a clan picture how you want to use the external RTC.

    3) The RTC module has to be powered all the time unless you shut off entire device but it does not mean that you need to provide 32.768KHz to RTC if you have another clock source to CLKIN pin. Are you going to use the 32KHz as the PLL clock source?

    Basically, in order to answer your question, I need to dig into RTC XTAL design and it will take some time. Let me dig it up for you. In the mean time, could you answer my questions?

    I'd like to let you know one thing that what you are asking is not what we have tested, so even if it is possible, we may not gurantee its stabilty because we have not tested.

    Best Regards,

    Peter Chung

     

     

  • Peter,

    To answer your questions:

    1. yes, we definitely want to use the bootup feature.

    2. If we use an external 32.768KHz clock chip, we would probably shut down the entire C5505. That's one benefit of what i'm considering - the power supply design gets simpler. If we use the internal C5505 RTC, we need special supplies just for that, and the current consumption is higher. So we would consider shutting the whole C5505 down.

    I understand what you're saying about not needing to supply the RTC with 32.768KHz if we have another clock source. But what I'm thinking is that INSTEAD of using the normal PLL with 12MHz, why can't we use the RTC PLL which expects 32.768KHz, except that instead of putting a crystal on those pins, we driver a 32.768KHz signal from another chip?

    If it's not tested and too difficult to know what will happen, we will probably not take the chance and will just use two 32.768KHz crystals in the design. A bit of a pity.

     

    Here is the trade-off. Interested in any responses:

    1. Use INTERNAL RTC:

    Benefits - No need for another RTC/calendar. Less real estate.

    Drawbacks - Need special power supply just for RTC, and if we want wakeup feature, need TWO power supply chips JUST for the RTC. Also, RTC needs 7uA, plus 17uA if we use wakeup feature. Backup battery must be quite large for extended life. Cost of two extra linear supplies - 2 ICs, about $0.75

    2. Use EXTERNAL RTC:

    Drawbacks: Adds another chip and another crystal to the design. Added cost about $1.00

    Benefits: External chip uses only 1uA max which can be supplied very easily by almost any battery. Even with a wakeup signal, external RTC needs same current. C5505 power supplies become much simpler - just shut off the whole chip.

    So internal RTC on C5505 is great to have at first look, but when we analyze a bit deeper, it needs some PSU support that's not convenient and the current consumption isn't so good.

  • Hi,

    Thank you for the information. I understand what you are trying to do. I have been asking around if an external 32KHz can drive the RTC_XI and XO. I have some information but not yet clear answer. Unfortunately, the person who I need to contact on this subject is on vation this week. Let me try to give you an answer sometime next week.

    Best Regards,

    Peter Chung

     

  • Peter,

    Forgive me if I missed the follow-on to this thread. Could you please tell me what you found out about using an external clock source instead of a crystal with the c5515 RTC?

    Thanks and regards,

    Mark Johnson

     

    Peter Chung said:

    Hi,

    Thank you for the information. I understand what you are trying to do. I have been asking around if an external 32KHz can drive the RTC_XI and XO. I have some information but not yet clear answer. Unfortunately, the person who I need to contact on this subject is on vation this week. Let me try to give you an answer sometime next week.

    Best Regards,

    Peter Chung

     

     

  • Yes, you can. Just feed 32KHz clock into RTC_XI and leave RTC_XO open. AFAIK, it is possible to use external clock source to feed all crystal-based oscillators in this way. If there existed an oscillator not allowing doing that, the 2 crystal pins would not be named XI and XO.

    What I don't know for certain is the voltage level, core level 1.xV or peripheral level 3.3V. If it is 1.xV, you may need a simple voltage divider with 2 resistors.

  • Cong,

    Thanks for the helpful reply. I agree with you that the pin naming is suggestive. The only strange thing is that, in the c5515 datasheet, it is recommended that the RTCXI and RTCXO pins are tied to CVDD_RTC and GND, respectively, if a RTC crystal isn't used. If the oscillator is the standard Pierce format, it would be odd to tie the output to GND even if the input is high. However, the datasheet does help on the other issue: CVDD_RTC levels are needed for an external clock in RTCXI. I have tried this with an EzDSP and it seems to work fine. No additional current is drawn from CVDD_RTC (above the standard 6uA) and a good-looking clock comes out of RTC_CLKOUT. I would still like to hear from TI about what should be done with RTCXO and whether there are any other issues to be aware of in using an external RTC clock source.

    Regards,

    Mark

  • Hi Mark,

    When RTC is not used at all, CVDD_RTC must be still powered (according to C55x5 RTC datasheet), so RTCXI must be tied to a known level to make the logic circuit stable and minimize power consumption. RTCXO tied to GND seems to be a further guard in that sense.

    In your modification, RTC_XO oscillates at the same frequency with RTC_XI (RTC_XO = not(RTC_XI)). Thus, it would be very strange to me if RTC_XO is recommended to be tied to somethings rather than left opened.

    Regards,

    Cong-Van