Sharing a GPS PPS signal consistently...

Hi guys,

I have what may be a tricky question regarding multiplexing / sharing a PPS signal among many boards at once. Here is some background:

I am currently rolling a board that incorporates a CFS206 RTC crystal whose frequency changes as a function of temperature. Some preliminary testing seems to suggest that I can achieve a very high correlation between the offset / error in real life vs. a open loop correction formula. The board uses a HTU21 sensor to determine the temperature and a slowly-rising board temperature and a GPS’ PPS signal, followed by an on-board least-squares formula to determine the factors in the quadratic equation. The factors are then written to EEPROM.

The correlation is pretty good from about freezing to 40*C - R^2 is slightly above 0.999 and hence the PPM error in this very limited sample is very low - about 0.2PPM on any particular sample and 0.03PPM cumulative over the course of 2 hours of sampling. However, promising these preliminary results, it’s not practical to hook up an individual GPS receiver to every board just for the PPS signal. So, what is the best approach to sharing a PPS signal among many boards? Ideally, I’d have a whole bunch of the boards inside an enclosure that can be temperature controlled while the GPS receiver is outside the box, pulsing away.

The current receiver module is a UP501 from Fasttrax, which is powerful enough to drive a LED high every time it pulses. But I’m open to alternatives, ideally ones that operate at 3.3V to be compatible with the CPU. Ideally, nice clean interrupts can then measure the MCU clock and the RTC clock errors as a function of changing temperatures inside the box.

I’m not sure why sharing 1PPS signal is an issue, isn't it just a problem of using a buffer with larger driving capability for the higher fan-out?

Length of the trace from GPS 1PPS to your measurement circuit on different boards should not be an issue, need not be balanced with equal length. Two 1PPS positive edges establish an accurate reference of 1 second duration. You count how many 32768Hz clocks in such 1 second interval to determine RTC crystal ppm drift. As 1 second interval is not enough to count, you count over a longer time but needs to be in temperature controlled chamber.

So is sharing 1PPS signal across boards still an issue? Or am I missing something?