[time-nuts] Oscillator Buddy
What may make a nice little nut-project is a simple PIC processor type Oscillator Buddy circuit that would reduce the effects of environmental and time caused oscillator frequency drift errors using the oscillator's EFC analog input. What makes a nice Time-Nut Oscillator candidate for this project is one that is repeatable AND predicable. I am continually ageing and testing several good oscillators including single and dual oven HP10811s and LPRO Rb to find the most predicable repeatable ones. On these type of selected Oscillators, after minimizing the frequency uncertainty errors due to basic things like power supply effects and RF load changes and keeping them in a reasonable stable environment, the main errors I see are: 1) ADEV noise at tau = 1sec for OXCOs and at 100 sec for RBs (typical range is from 0.2e-12 to 2e-12) 2) Temperature coefficient (the typical range is from +- 1e-13 to 1e-10 per degC) 3) Ageing rate (typical range is + - 1e-13 to 1e-10 per day) 4) Barometric pressure (Typical effect TBD) I have found it is possible to reduce the environment and time caused freq drift errors by 10 to 1 or more on some of the better selected oscillators using open loop correction techniques. Disciplining the Osc with GPS in a closed loop (aka Tbolt GPSDO) is the typical and most effective way to reduce the effects of #2, 3, 4 above, but that does have some limitations and the better the disciplined osc is open loop then the better the closed loop results. Any thoughts and comments on an Oscillator Buddy project. ws ___ 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.
Re: [time-nuts] Oscillator Buddy
Hi Warren: I looked into using a PIC to make a GPSO and there are two problems: (1) is the number of bits you can get in the D/A converter. The best solution I've seen is used in the Quantic Q5200 GPS receiver which has what amounts to an internal GPSDO with 48 bits in the oscillator drive. It's in their US Patent 5440313 (link on Q5200 web page). http://www.prc68.com/I/Q5200.shtml (2) Noise on the control voltage. When I was at HP/Agilent a neighboring engineer was THE man on their 4352 VCO/PLL tester. The D/A converter in it and in the combo boxes (4395 4396 among others) not only had a lot of bits it was also extremely quiet. Getting both of these is far from simple. http://www.prc68.com/I/4395A.shtml Have Fun, Brooke Clarke http://www.PRC68.com http://www.End2PartyGovernment.com/ WarrenS wrote: What may make a nice little nut-project is a simple PIC processor type Oscillator Buddy circuit that would reduce the effects of environmental and time caused oscillator frequency drift errors using the oscillator's EFC analog input. What makes a nice Time-Nut Oscillator candidate for this project is one that is repeatable AND predicable. I am continually ageing and testing several good oscillators including single and dual oven HP10811s and LPRO Rb to find the most predicable repeatable ones. On these type of selected Oscillators, after minimizing the frequency uncertainty errors due to basic things like power supply effects and RF load changes and keeping them in a reasonable stable environment, the main errors I see are: 1) ADEV noise at tau = 1sec for OXCOs and at 100 sec for RBs (typical range is from 0.2e-12 to 2e-12) 2) Temperature coefficient (the typical range is from +- 1e-13 to 1e-10 per degC) 3) Ageing rate (typical range is + - 1e-13 to 1e-10 per day) 4) Barometric pressure (Typical effect TBD) I have found it is possible to reduce the environment and time caused freq drift errors by 10 to 1 or more on some of the better selected oscillators using open loop correction techniques. Disciplining the Osc with GPS in a closed loop (aka Tbolt GPSDO) is the typical and most effective way to reduce the effects of #2, 3, 4 above, but that does have some limitations and the better the disciplined osc is open loop then the better the closed loop results. Any thoughts and comments on an Oscillator Buddy project. ws ___ 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. ___ 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.
Re: [time-nuts] Oscillator Buddy
Brooke One simple way I'veused to eliminate the two problems you brought up is by Greatly restricting the tuning effect of the Dac voltage. Only big downside is you then may have to manually reset the nominal Osc freq every once in a while if the Dac gets near a limit. With a little thought an 8 bit Dac is more than enough to make a great GPSDO out of a good Osc, as long as the Osc also has a course freq tuning method like the HP 10811 does. But what I was suggesting is not a GPSDO, but an open loop freq compensator. Simplified example: Make the analog output of the PIC equal to (K1 times measured temperature) + (K2 times lapsed time) With that, K1 can be used to cancel the Osc tempCoef and K2 the daily ageing rate. Having fun, always thanks ws ** Hi Warren: I looked into using a PIC to make a GPSO and there are two problems: (1) is the number of bits you can get in the D/A converter. The best solution I've seen is used in the Quantic Q5200 GPS receiver which has what amounts to an internal GPSDO with 48 bits in the oscillator drive. It's in their US Patent 5440313 (link on Q5200 web page). http://www.prc68.com/I/Q5200.shtml (2) Noise on the control voltage. When I was at HP/Agilent a neighboring engineer was THE man on their 4352 VCO/PLL tester. The D/A converter in it and in the combo boxes (4395 4396 among others) not only had a lot of bits it was also extremely quiet. Getting both of these is far from simple. http://www.prc68.com/I/4395A.shtml Have Fun, Brooke Clarke http://www.PRC68.com http://www.End2PartyGovernment.com/ WarrenS wrote: What may make a nice little nut-project is a simple PIC processor type Oscillator Buddy circuit that would reduce the effects of environmental and time caused oscillator frequency drift errors using the oscillator's EFC analog input. What makes a nice Time-Nut Oscillator candidate for this project is one that is repeatable AND predicable. I am continually ageing and testing several good oscillators including single and dual oven HP10811s and LPRO Rb to find the most predicable repeatable ones. On these type of selected Oscillators, after minimizing the frequency uncertainty errors due to basic things like power supply effects and RF load changes and keeping them in a reasonable stable environment, the main errors I see are: 1) ADEV noise at tau = 1sec for OXCOs and at 100 sec for RBs (typical range is from 0.2e-12 to 2e-12) 2) Temperature coefficient (the typical range is from +- 1e-13 to 1e-10 per degC) 3) Ageing rate (typical range is + - 1e-13 to 1e-10 per day) 4) Barometric pressure (Typical effect TBD) I have found it is possible to reduce the environment and time caused freq drift errors by 10 to 1 or more on some of the better selected oscillators using open loop correction techniques. Disciplining the Osc with GPS in a closed loop (aka Tbolt GPSDO) is the typical and most effective way to reduce the effects of #2, 3, 4 above, but that does have some limitations and the better the disciplined osc is open loop then the better the closed loop results. Any thoughts and comments on an Oscillator Buddy project. ws ___ ___ 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.
[time-nuts] Oscillator Buddy
Warren, I agree with other responders that a closed loop approach to OCXO minding is feasible using a simple micro and D-A. I designed a system which has as its main focus low ADev and reasonably good frequency accuracy, without necessarily requiring the tight timing of a commercial GPSDO or the use of high resolution D-A. You might consider my system 'loose' control, as it relies on the OCXO being pretty good in the first place. It just corrects for long term effects such as ageing. What I came up with may be helpful. Essentially it phase locks the OCXO to GPS 1pps, but does so with extremely long integration times (hours), and with modest gain, so provided the OCXO is not subject to sudden temperature extremes or has an excessive ageing rate, will provide good control. As others have suggested, if you limit the control range (in my case I have a manual coarse setting) you can achieve very good closed-loop control with limited D-A resolution. I used 15 bit PWM, using a 12-bit timer and 3 bits of dither, providing a range of under 1ppm and a step size of about 3e-11. The system has no predictive capability, but does achieve (with a good OCXO) quite good holdover performance. By studying the control voltage over the last year, I've determined by extrapolation that manual intervention (coarse adjustment) will be unlikely within 10 years. I used a small AVR processor, did all the counting and maths in the processor, and operated the analog circuitry from the internal reference in the OCXO, resulting in very good thermal performance. The design is described in some detail at http://www.qsl.net/zl1bpu/MICRO/SIMPLE/SimpleGPS.htm. Regards, Murray Greenman ___ 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.