0
\$\begingroup\$

I'm considering a design that uses an MCP7940M RTC and a PIC16F15223. The clock section for the RTC shows

RTC clock driver

I'm considering using a CFS-20632768EZBB 32.768 kHz tuning fork style crystal with CL = 6 pF.

I'm deciding whether to either drive the PIC from the RTC clock, or have it run its own low-frequency, low-power internal clock. LFINTOSC is much less accurate but that wouldn't actually matter.

If I understand this correctly:

load equation

and I set Cx1 = Cx2 = Cx, then Cx would be at most 12 pF, less accounting for board stray capacitance. If I use this value for the circuit shown in 4-2, and also connect X2 to the PIC's CLKIN, would there be any adverse effects? In the PIC's spec I'm unable to find a clear definition for the input capacitance of CLKIN, but maybe it's 5-50 pF if it's the same as parameter D380 \$C_{IO}\$. If this is true and 50 pF is added to the load capacitance equation, it probably puts the RTC clock driver way out of spec, right? (The RTC asks for CL <= 9 pF.) Does this imply that the only practical clock option for the PIC is its LFINTOSC?

\$\endgroup\$
5
  • \$\begingroup\$ Oh, man, I'm probably fighting a losing battle here, because I'm seeing a lot of this lately. "The oscillator is a CFS-20632768EZBB". No. No no no no no no no. The crystal is a CFS-20632768EZBB. An oscillator is an active circuit that you give power and possible control signals, and which generates a periodic signal (i.e. sine or square wave). A crystal is just a resonator unless you make it part of an oscillator. Please please please unconfuse yourself, and work to unconfuse the world on this point. \$\endgroup\$
    – TimWescott
    Aug 26, 2021 at 1:01
  • \$\begingroup\$ @TimWescott Fine? I guess I wasn't confused on the matter and already understand the crystal to not be an oscillator on its own; there's no need to panic. I've edited the question. \$\endgroup\$
    – Reinderien
    Aug 26, 2021 at 1:05
  • \$\begingroup\$ Thank you for the edit. I'm leaving the above comment stand because, well, there are four lights. \$\endgroup\$
    – TimWescott
    Aug 26, 2021 at 1:06
  • \$\begingroup\$ This is the third time I've run across it this week. It's either a blip and I'm overreacting, or it's a trend and I'm too late. But I'm gonna try. \$\endgroup\$
    – TimWescott
    Aug 26, 2021 at 1:07
  • 1
    \$\begingroup\$ @TimWescott lol ... your first comment made me think of this vimeo.com/338492037 \$\endgroup\$
    – jsotola
    Aug 26, 2021 at 1:07

1 Answer 1

2
\$\begingroup\$

In general, the rule of thumb for oscillator circuits is to not mess with them unless you know what you're doing.

In the case of a chip manufacturer's proprietary circuits, there's often not enough information there to know whether the circuit will be successful or not.

If you wanted to do something like run two processors off of a "normal" crystal input (not a 32kHz low-power crystal), then I'd recommend that you just use a crystal oscillator, or build one up with a 74HCU04, and feed both processors from that.

In this case you presumably want low power. If you have the time, you could investigate doing exactly the above: buying or building your own 32kHz oscillator, and sending the result to both chips. If I were doing this, my very first concern would be that the power consumption would -- comparatively speaking -- go through the roof. By definition, those crystals are used where very low power consumption is paramount, so the oscillator circuits are designed to operate at very low power levels themselves. I suspect that if you rolled your own, it would be very difficult to achieve the sort of microamp current draws that the processor and RTC are achieving here.

\$\endgroup\$
1
  • \$\begingroup\$ That's good perspective. In this case power isn't all that important, and the only reason I went with the 32 kHz is that that's the only frequency that the RTC accepts. The separated driver is an interesting idea. \$\endgroup\$
    – Reinderien
    Aug 26, 2021 at 1:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.