[EMAIL PROTECTED] wrote: > Hi Bruce, > > I wonder how the analog switches' tempco would affect the performance. They > usually do have 10 - 100 Ohms series resistance if I remember correctly. > > The Temex OCXO is achieving STS of parts in E-013, it would be unfortunate > to lose that performance due to external circuitry. > > At these extremely small values, I would guess that the finite > off-resistance of the switch, and the leakage currents through the switches' > transistors > would also degrade the performance. Maybe a mechanical relay would work > better? But that would cause a mechanical vibration that could jerk the > crystal. > > Also, I guess one would have to add a small RS-232 micro to decode the > Validity bit from the GPS and do the actual switchover. > > Here is an interesting factoid: Magnus mentioned that many GPS receivers do > strange things when they go into holdover. I can only concur to that. The > M12+ > and M12M are really excellent in this, but other receivers are not and may > give you a couple of 1PPS pulses that are way off, say 200ns or much more > before they shut up. Not so nice. > > bye, > Said > > Said
If your worried about analog switch tempco then the classic solution of using an noninverting buffer after the selection switch works well. The tempco of the analog switch in series with the PI filter elements is relatively insignificant in that it only affects the PLL natural frequency and damping and not the filter dc output. Leakage current shouldn't be a problem in the selector switches, unless the switches get really warm. A custom discrete switch could be used to eliminate leakage problems by ensuring that the voltage across the last off switch in the switch chain is very small. One could always use the output of the tracking ADC to drive the EFC input, disabling tracking during holdover and reenabling when the GPS reference is again stable. If one monitors the message containing the estimated variance of the PPS output one may be able to circumvent such effects. Bruce _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.