Re: [time-nuts] GPS: ADEV or MDEV?
In a message dated 4/23/2007 20:51:28 Pacific Daylight Time, bruce [EMAIL PROTECTED] writes: Of course one can always resort to annealing out the retrace by controlled thermal cycling of the reference as Fluke (optionally) do in their 7000 series references whenever continuous power is lost. These 10V references use selected LTZ1000's to achieve a drift of around 1ppm/year. Bruce Nice performance on the Fluke. Agree on the thermal sensitivities of the DAC, reference etc. Long term Zener aging not so important for a GPSDO of course. BTW: I know of a test lab that officially tested a Fluke GPSDO for someone I know only to find it was not better than 1E-09. They maintained that the unit was not broken, just bad... bye, Said ** See what's free at http://www.aol.com. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
* The Vref output of most OCXO's is from a Zener diode inside the can. These typically have aging, thermal sensitivity and very poor voltage accuracy, and there are much better monolithic high-precision, low-tempco voltage reference available on the market now (Digikey etc). Depending on the internal design of the OCXO, it may help the noise and stability to add an external 100uF or larger tantalum cap to the Vref pin, since it's voltage is likely used as the power supply for the Oscillator etc. If you want the lowest tempco and drift available try using an LTZ1000 (used in HP 8 1/2 digit DVMs). Question - does Vref drift matter for a GPSDO? It seems it is either too low to be of concern or the effect of Vref drift is in fact indistinguishable from OCXO frequency drift and is thus transparently compensated by any 3rd order loop? I don't know for sure, but is 1 ppm/day drift in Vref typical? If so given an EFC gain of 1 Hz / 10 MHz / 10 V = 1e-6 / V the net drift per day due to Vref is 1e-12 which is 50x below the quartz drift spec of say a hp 10811A (5e-10/day). Is the zener drift plot in Analog Devices AN-713 typical? /tvb ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Tom Attached images are drift plots for: 1) VRE430 reference utilising a buried zener. 2) LM199 3) LTZ1000 Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Question - does Vref drift matter for a GPSDO? It seems it is either too low to be of concern or the effect of Vref drift is in fact indistinguishable from OCXO frequency drift and is thus transparently compensated by any 3rd order loop? I don't know for sure, but is 1 ppm/day drift in Vref typical? If so given an EFC gain of 1 Hz / 10 MHz / 10 V = 1e-6 / V the net drift per day due to Vref is 1e-12 which is 50x below the quartz drift spec of say a hp 10811A (5e-10/day). Is the zener drift plot in Analog Devices AN-713 typical? /tvb Tom Tempco (if the reference isnt in a temperature stabilised environment) and noise can be much more important if the drift is low enough. That drift plot in AN713 is the for a bandgap reference. However the attached plot from Woodward J Eicke's in April 1964 ISA transactions paper shows similar drifts for discrete zener reference diodes. Buried zeners can be much better. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Tom Attached images are drift plots for: 1) VRE430 reference utilising a buried zener. 2) LM199 3) LTZ1000 Bruce OK, thanks very much. What I read from the slopes in those long-term plots are Vref drift rates on the order of: -6 ppm / 1000 hours (= 2e-12/d) -2 ppm / 1000 hours (= 6e-13/d) -2 ppm / 3000 hours (= 2e-13/d) +6 ppm / 5 months (= 5e-13/d) +10 ppm / 12 months (= 3e-13/d) -1 ppm / 10 days (= 1e-12/d) -1.5 ppm / 30 days (= 6e-13/d) These all fall into the range of 2 to 20 e-13 / day drift. It confirms my hunch then that Vref drift is a non-issue for a GPSDO. Tom Tempco (if the reference isnt in a temperature stabilised environment) and noise can be much more important if the drift is low enough. Can someone do a similar numerical analysis of these two factors: tempco and noise? It's not that I don't believe you; it's just that I'd rather see numbers and plots than words. That drift plot in AN713 is the for a bandgap reference. However the attached plot from Woodward J Eicke's in April 1964 ISA transactions paper shows similar drifts for discrete zener reference diodes. Buried zeners can be much better. Bruce Wow, a very long-term plot. I see Vref drift rates like: +75 ppm / 500 days (= 2e-12 / d) -25 ppm / 400 days (= 7e-13/d) -60 ppm / 100 days (= 7e-12/d) -100 ppm / 700 days (= 2e-12/d) So that one is not too unlike the first three plots. I think we can put the GPSDO Vref drift issue to rest. /tvb ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Tom Van Baak wrote: Can someone do a similar numerical analysis of these two factors: tempco and noise? It's not that I don't believe you; it's just that I'd rather see numbers and plots than words. Tom When the reference is temperature cycled hysteresis may be significant. Hysteresis data is harder to find but its much worse for bandgap references than buried zener references. Attached plots are for low frequency noise of various IC zener references. If one ensures that the zener noise is adequately low pass filtered, then in most cases its effect is probably insignificant. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Tom Attached is some hysteresis and other data for various references. NB take the claimed XFTE reference ADR293 drift figure with a large dose of salt, its 2 orders of magnitude lower than ADI claim. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
In a message dated 4/23/2007 06:05:12 Pacific Daylight Time, [EMAIL PROTECTED] writes: -6 ppm / 1000 hours (= 2e-12/d) -2 ppm / 1000 hours (= 6e-13/d) -2 ppm / 3000 hours (= 2e-13/d) +6 ppm / 5 months (= 5e-13/d) +10 ppm / 12 months (= 3e-13/d) -1 ppm / 10 days (= 1e-12/d) -1.5 ppm / 30 days (= 6e-13/d) These all fall into the range of 2 to 20 e-13 / day drift. It confirms my hunch then that Vref drift is a non-issue for a GPSDO. Hi Tom, I agree. What makes life even better is that these are probably also worst-case numbers from the datasheets. Typical numbers can be better :) Also, drift typically does slow over long time frames. What is an issue is the tempcos of the reference, the PCB (thermocouple effects on the leads etc), the DAC, and the low pass filter. Active DAC low pass filters add temperature dependent offsets, as well as their own noise, drift, and PSRR's etc. BTW: the retrace hysterisis of typical very high-quality OCXO's is speced at 1E-08 (24 hours off, 2 hours on) that's probably a lot worse than the voltage reference zener retrace will be. A typical 20ppm Hysterisis on a 5V Zener with the OCXO having a +-22Hz (0-5V) adjustment range gives 4.4E-011 retrace. Several orders of magnitude better than these particular OCXO's inherent retrace by itself. bye, Said ** See what's free at http://www.aol.com. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Can someone do a similar numerical analysis of these two factors: tempco and noise? It's not that I don't believe you; it's just that I'd rather see numbers and plots than words. Tom attached plot of zener noise is also from Eickes 1964 paper. Bandpass is a little unclear as its determined by the response of a galvanometer Bruce Good. Thanks for the cool plots. OK, my eyeball sees: 5 uV peak-peak, maybe 2 uV rms. 10 uV peak-peak, maybe 4 uV rms. 1 uV peak-peak, about 0.2 uV rms 4 uV peak-peak, guess 1 uV rms 3.5 uV peak-peak, 1 uV rms 10 uV peak-peak, 1 uV rms 3 uVpp per spec This suggests that one could get Vref noise down to about 1 uV (or better if you measure rms). Does anyone on the list have actual experience with this issue? Anyway, assume a full 1 uV of noise and assume a 10811A with an EFC sensitivity of 1e-6/volt. Then Vref noise translates to 1e-12 in frequency -- which is worth considering since it's in the same ballpark as good OCXO noise. Maybe not the dominant factor, but a potential contributing factor. For example, below is a 10-minute frequency plot of a typical hp 10811A which has a low ADEV of about 2e-12 (tau 1 to 10 seconds). The scale is 1e-11 / division. http://www.leapsecond.com/museum/hp10811/log14669v.gif So you can imagine that Vref noise on the order of 1e-12 might just start being noticeable in the plot, meaning that if you want short-term performance down in the low 12's and if you have an OCXO that even gets that low, then you want to keep your Vref (and DAC, filters, etc.) clean to microvolt levels. It's just a back of the envelop calculation so I'm not really sure. Someone correct me if you see something wrong. Related - does anyone have equipment in your home/lab that can directly measure, at uV or sub-uV levels, noise on the EFC line? I wonder how clean a Z3801A EFC is, for example. Also an idea - is there an easy way to artificially increase Vref noise, by say 10x, so one could see if that change made any measurable change in OCXO output ADEV or phase noise? /tvb ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
In a message dated 4/23/2007 11:03:07 Pacific Daylight Time, [EMAIL PROTECTED] writes: So you can imagine that Vref noise on the order of 1e-12 might just start being noticeable in the plot, meaning that if you want short-term performance down in the low 12's and if you have an OCXO that even gets that low, then you want to keep your Vref (and DAC, filters, etc.) clean to microvolt levels. It's just a back of the envelop calculation so I'm not really sure. Someone correct me if you see something wrong. Hi Tom, the 10811 I had used in the past had about +-2Hz deviation for a 0 - 5V EFC... So 8E-08 sensitivity per volt. That very low sensitivity (10x lower than MTI parts) also gives a 10x higher EFC noise immunity. bye, Said ** See what's free at http://www.aol.com. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Hi Tom: Appendix H of Linear App Note 86 A Standards Lab Grade 20-Bit DAC with 0.1ppm/C Drift on pdf page 43 describes the 1,000 X amplifier they used and the need for an out of production analog scope with to see sub 1 uV noise as well as the need for a totally shielded measurement system. In table H4 (pdf pag 45) they list the SR560 as a suitable amplifier that includes the needed filtering. http://www.thinksrs.com/products/SR560.htm I don't have this stuff but was just reading the app note after the reference here to the LTZ1000 DC reference. It uses their 24 bit A/D converter and a PIC with a couple of D/A converters to get the 20 bits. It could be modified to get more bits with less precision. One of the things I've looked at is how do you generate the EFC control voltage for a GPSDO. Stock D/A converters when scaled for the full voltage range of the oscillator will be way too coarse. If you use a manual pot to allow them to run where the LSB is a little smaller than the desired frequency step size then you need to set the pot maybe every 6 months or so to follow the aging of the crystal. The FTS4060 combines two 16 bit D/A converters and only needs manual tweaks about once a year. Have Fun, Brooke Clarke http://www.PRC68.com http://www.pacificsites.com/~brooke/PRC68COM.shtml http://www.precisionclock.com Tom Van Baak wrote: Can someone do a similar numerical analysis of these two factors: tempco and noise? It's not that I don't believe you; it's just that I'd rather see numbers and plots than words. Tom attached plot of zener noise is also from Eickes 1964 paper. Bandpass is a little unclear as its determined by the response of a galvanometer Bruce Good. Thanks for the cool plots. OK, my eyeball sees: 5 uV peak-peak, maybe 2 uV rms. 10 uV peak-peak, maybe 4 uV rms. 1 uV peak-peak, about 0.2 uV rms 4 uV peak-peak, guess 1 uV rms 3.5 uV peak-peak, 1 uV rms 10 uV peak-peak, 1 uV rms 3 uVpp per spec This suggests that one could get Vref noise down to about 1 uV (or better if you measure rms). Does anyone on the list have actual experience with this issue? Anyway, assume a full 1 uV of noise and assume a 10811A with an EFC sensitivity of 1e-6/volt. Then Vref noise translates to 1e-12 in frequency -- which is worth considering since it's in the same ballpark as good OCXO noise. Maybe not the dominant factor, but a potential contributing factor. For example, below is a 10-minute frequency plot of a typical hp 10811A which has a low ADEV of about 2e-12 (tau 1 to 10 seconds). The scale is 1e-11 / division. http://www.leapsecond.com/museum/hp10811/log14669v.gif So you can imagine that Vref noise on the order of 1e-12 might just start being noticeable in the plot, meaning that if you want short-term performance down in the low 12's and if you have an OCXO that even gets that low, then you want to keep your Vref (and DAC, filters, etc.) clean to microvolt levels. It's just a back of the envelop calculation so I'm not really sure. Someone correct me if you see something wrong. Related - does anyone have equipment in your home/lab that can directly measure, at uV or sub-uV levels, noise on the EFC line? I wonder how clean a Z3801A EFC is, for example. Also an idea - is there an easy way to artificially increase Vref noise, by say 10x, so one could see if that change made any measurable change in OCXO output ADEV or phase noise? /tvb ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Tom Van Baak wrote: Also an idea - is there an easy way to artificially increase Vref noise, by say 10x, so one could see if that change made any measurable change in OCXO output ADEV or phase noise? /tvb Tom 1) Increase the cutoff frequency of any low pass filter used. 2) With a discrete zener reference is to use a very low zener current. 3) Use a reverse biased base emitter junction of a small signal transistor as a zener at low current (don't reuse the transistor for small signal applications afterwards). 4) Inject noise from say a pseudorandom noise generator via a resistive network in the EC line. If you havent got a pseudorandom noise generator (or one of the pseudorandom noise generator chips that chips National Semiconductor used to sell) you could try the method suggested in: http://www.edn.com/article/CA200385.html Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV? (Tom Van Baak)
Hi Colin: That's one sweet meter. Years ago at work I used them and we had both flavors. One used the old definition of the volt and the other used the new volt. Although our cal lab was not good enough to calibrate them we could feed them signals and measure their readings. They differed by the difference in the definition. Quite amazing. They also can be used as a kind of sampling scope. As I remember it the data is arranged in a strange way. There might be a driver available from Agilent to do the scope thing. Have Fun, Brooke Clarke http://www.PRC68.com http://www.pacificsites.com/~brooke/PRC68COM.shtml http://www.precisionclock.com Colin Bradley wrote: My 3458A can make measurements down to the 10nV level. At 7 1/2 digits it can do 60 readings a second. At 8 1/2 digits it slows down to 6 readings a second. The real question is the frequency of the noise. I have yet to interface it to HP-IB bus. I have several Z3801's I could try it on. Colin Good. Thanks for the cool plots. OK, my eyeball sees: 5 uV peak-peak, maybe 2 uV rms. 10 uV peak-peak, maybe 4 uV rms. 1 uV peak-peak, about 0.2 uV rms 4 uV peak-peak, guess 1 uV rms 3.5 uV peak-peak, 1 uV rms 10 uV peak-peak, 1 uV rms 3 uVpp per spec This suggests that one could get Vref noise down to about 1 uV (or better if you measure rms). Does anyone on the list have actual experience with this issue? Anyway, assume a full 1 uV of noise and assume a 10811A with an EFC sensitivity of 1e-6/volt. Then Vref noise translates to 1e-12 in frequency -- which is worth considering since it's in the same ballpark as good OCXO noise. Maybe not the dominant factor, but a potential contributing factor. For example, below is a 10-minute frequency plot of a typical hp 10811A which has a low ADEV of about 2e-12 (tau 1 to 10 seconds). The scale is 1e-11 / division. http://www.leapsecond.com/museum/hp10811/log14669v.gif So you can imagine that Vref noise on the order of 1e-12 might just start being noticeable in the plot, meaning that if you want short-term performance down in the low 12's and if you have an OCXO that even gets that low, then you want to keep your Vref (and DAC, filters, etc.) clean to microvolt levels. It's just a back of the envelop calculation so I'm not really sure. Someone correct me if you see something wrong. Related - does anyone have equipment in your home/lab that can directly measure, at uV or sub-uV levels, noise on the EFC line? I wonder how clean a Z3801A EFC is, for example. Also an idea - is there an easy way to artificially increase Vref noise, by say 10x, so one could see if that change made any measurable change in OCXO output ADEV or phase noise? /tvb __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Tom Van Baak wrote: Also an idea - is there an easy way to artificially increase Vref noise, by say 10x, so one could see if that change made any measurable change in OCXO output ADEV or phase noise? /tvb Tom Another possibility is to use a low noise instrumentation amplifier with a gain of 10. Connect the noninverting input to Vref, connect the inverting input to a low pass filtered (gain = +1) version of Vref and connect the reference pin of the instrumentation amplifier to Vref. The output of the instrumentation amplifier will be equal to Vref with about 11x the noise down to a frequency determined by the low pass filter cutoff frequency. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
attachment did not make it, there it is: http://www.ko4bb.com/ham_radio/Manuals/Tektronix - 7L5/Tek-7L5-PhaseNoise.png Didier Didier Juges wrote: Tom Van Baak wrote: Related - does anyone have equipment in your home/lab that can directly measure, at uV or sub-uV levels, noise on the EFC line? I wonder how clean a Z3801A EFC is, for example. Here is the spurious/noise spec for the Tek 7L5 spectrum analyzer plug-in (I have a working sample with L3 plug-in, not been calibrated for a while :-) Didier KO4BB ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
In a message dated 4/23/2007 18:54:54 Pacific Daylight Time, [EMAIL PROTECTED] writes: If an unheated reference exposed to wide ambient temperature swings is used then hysteresis may be significant when the OCXO is continuously powered. The significance and magnitude of the effect depends on the range of ambient temperature variations experienced. Bruce Hi Bruce, in an 'uncontrolled' environment like that one would not get 2-20ppm retrace on the Zener (as speced in the table sent to this group earlier) I would think. bye, Said ** See what's free at http://www.aol.com. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Dr Bruce Griffiths wrote: Tom Van Baak wrote: Bruce, I'm getting the idea you don't like _any_ of the hobbyists GPSDO's that have come out in the last ten years... So I'm curious what then would qualify as a well designed GPSDO in your opinion? (and please don't bring up the Quartzlock thing; it's a hundred times more expensive than the GPSDO's we're talking about here). /tvb Additional desirable feature. 12) provision for measuring OCXO case temperature and power supply current to allow compensation for variable internal wiring voltage drops for those OXCOs that do not allow Kelvin sensing of the EFC varactor voltage. Some OCXO's (FTS1200 etc) do not have independent sensing of the EFC varactor differential voltage, so the temperature dependent voltage in the internal ground wiring drop due to the variable oven current will modulate the OCXO output frequency when the DAC ouput voltage is fixed. If this effect is modelled sufficiently accurately then the DAC output can be adjusted to compensate. For example with 0.1 ohms resistance in the ground wire shared by the EFC varactor and the heater a variation of say 1mA in the heater current will alter the effective EFC varactor voltage by 100uV, which is far from negligible. In any case a heater current stable to 1mA or better is somewhat unlikey unless the ambient temperature is also controlled to better than 1 degree. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
12) provision for measuring OCXO case temperature and power supply current to allow compensation for variable internal wiring voltage drops for those OXCOs that do not allow Kelvin sensing of the EFC varactor voltage. Some OCXO's (FTS1200 etc) do not have independent sensing of the EFC varactor differential voltage, so the temperature dependent voltage in the internal ground wiring drop due to the variable oven current will modulate the OCXO output frequency when the DAC ouput voltage is fixed. If this effect is modelled sufficiently accurately then the DAC output can be adjusted to compensate. Does anybody have a list of OCXOs with a separate ground pin for the oven? Some of the OCXO data sheets I've looked at have a Vref output. Seems like a good idea. They already have good temperature control and that's most of the trouble with getting a stable reverence voltage. Do any OCXOs also include the DAC internally? Isn't that the obvious next step? -- These are my opinions, not necessarily my employer's. I hate spam. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
In a message dated 4/22/2007 01:11:30 Pacific Daylight Time, [EMAIL PROTECTED] writes: Does anybody have a list of OCXOs with a separate ground pin for the oven? Some of the OCXO data sheets I've looked at have a Vref output. Seems like a good idea. They already have good temperature control and that's most of the trouble with getting a stable reverence voltage. Do any OCXOs also include the DAC internally? Isn't that the obvious next step? Hi guys, Some comments: * The Vref output of most OCXO's is from a Zener diode inside the can. These typically have aging, thermal sensitivity and very poor voltage accuracy, and there are much better monolithic high-precision, low-tempco voltage reference available on the market now (Digikey etc). Depending on the internal design of the OCXO, it may help the noise and stability to add an external 100uF or larger tantalum cap to the Vref pin, since it's voltage is likely used as the power supply for the Oscillator etc. * Dithering the DAC to get better resolution is not a good idea, since it will create spurs. If not through the low-pass filter into the EFC pin, then through the power supply or ground rails into the PCB, or through all that digital noise being generated by the constant updates of the DAC. It's better to cascade two dacs through a matching network (coarse/fine DAC) and have as few digital traces switch as possible. These DAC's as well as the matching resistors, and especially the DAC voltage reference should be as low tempco as possible. For the real time nut, the DAC's and Voltage reference will be selected to have canceling Temperature Tempco's :) One advantage of the cascaded DAC is that we can use 14, 16, or even 20 bit DACS and their DNL/INL only needs to have 12+ bits of accuracy to make the compound DAC linear enough for EFC control. For the dithered single DAC approach the DNL/INL is totally critical (better than 1 LSB needed) to make the dithering work well. * Putting Digital circuits (read DAC) into the OCXO can is not a good idea, we really don't want high edge rate (10ns rise/fall time) inside the analog OCXO can. It's gonna put noise on the sine wave. Plus it requires a lot of additional pins (well, at least one). * We investigated the effect of varying OCXO current on the single-ground-pin Euro can (BTW: whoever designed that can without Kelvin Sensing was just not thinking right at the time). One way around the problem is simply to solder a thick ground wire to the can's case next to the VCC pin through which the oven current will pass, and use the main ground pin (placed far away from the VCC pin, another stupid idea) as the EFC Kelvin return. But keep the soldering time very short since the oven may overheat otherwise. As Bruce mentioned, measuring the oven current and compensating for it electronically is another way to fix the problem. This is exactly what we do on the Fury GPSDO, with a highly accurate 24 bit Sigma Delta instrumentation ADC. unfortunately that part and the sensing element have their own very small tempco's of course. With careful layout, ensuring that the oven current does not pass in the direction of the DAC/voltage reference etc it is possible to mitigate the Kelvin sensing problem of Euro-can OCXO's to negligible levels. * Bruce forgot to mention the use of smart algorithms like Kallmann filtering to measure, and compensate for aging and temperature effects, this is especially important for hold-over performance and also to minimize errors due to diurnal temp changes etc. The algorithm should be as proactive as possible, so that only small errors need to be corrected by GPS control (such as the rate of change of the aging rate). * maybe a bit above home-brew budgets, but being the lowest-cost new-with-warranty GPSDO on the market as far as we know, the Jackson-Labs Fury GPSDO almost fulfill's or exceeds all of Bruce's requirements :) I think our entry level pricing is lower than most Z3801A's on Ebay these days * bye, Said ** See what's free at http://www.aol.com. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
I made a hardware sawtooth remover a couple of years ago as an experiment, but where I used it I didn't seen much difference. I just checked again recently, but the sawtooth errors seem very small compared to overall instability, even for just an M12 to Rb at a fairly stable temperature. The larger errors of a few ns (over 1000s) which are occasionally visible when the M12 is temperature controlled seem regularly to be exceeded by other instabilities in the pps output. There are plenty of times when sawtooth removal would be of use/interest, but in a typical GPSDO which has a heap of other errors, I do wonder what improvements in performance would actually be seen in the output from the oscillator - which is all that a lot of people are really interrested in. Angus. I removed the sawtooth error also as an experiment and still have yet to do a closed loop test with my GPSDO. I have a MTI-260 OCXO which auto adjusts the oven set point every few days by using a transient temperature impulse (I believe). I am waiting for convergence of the oven temperature before I change my setup and do a closed loop test - frustration increasing - still waiting. What got me started down the path of sawtooth elimination was a correlation between the stability of the least significant digit of my frequency counter an HP 53132A (driven by the Z3816A) and the logged pps to gps error. If you think about a control system the better the driving signal the smaller the servo loop error which results in a more accurate system. I can see a few nanosecond jump in the sawtooth when a SV is added or removed from the pps solution mostly because my antenna is rarely at the surveyed location. The survey math was intended for SA removal and is probably inappropriate now. The statistical MODE position from a night survey and a fudge factor to lower the antenna height would be my bet. In truth the calculated antenna height is dynamic from day to night for my gps receivers. Peter ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Dr Bruce Griffiths wrote: Said [EMAIL PROTECTED] wrote: Hi guys, Some comments: * The Vref output of most OCXO's is from a Zener diode inside the can. These typically have aging, thermal sensitivity and very poor voltage accuracy, and there are much better monolithic high-precision, low-tempco voltage reference available on the market now (Digikey etc). Depending on the internal design of the OCXO, it may help the noise and stability to add an external 100uF or larger tantalum cap to the Vref pin, since it's voltage is likely used as the power supply for the Oscillator etc. If you want the lowest tempco and drift available try using an LTZ1000 (used in HP 8 1/2 digit DVMs). In those cases where a thermally stabilized reference is undesirable (power consumption, power dissipation, thermal stabilization time at power up), see Thaler www.thaler.com They have the best (as far as I know) thermally compensated references and the best initial accuracy. Didier KO4BB ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
In a message dated 4/22/2007 06:30:46 Pacific Daylight Time, [EMAIL PROTECTED] writes: If you want the lowest tempco and drift available try using an LTZ1000 (used in HP 8 1/2 digit DVMs). In those cases where a thermally stabilized reference is undesirable (power consumption, power dissipation, thermal stabilization time at power up), see Thaler www.thaler.com They have the best (as far as I know) thermally compensated references and the best initial accuracy. Didier KO4BB Hi Didier, Bruce, Thanks for the pointers! the Thaler ref is pretty nice, low noise too. The LT part is a bit complex to properly put into an application, and both are probably not that cheap. I like Intersils' new X60008 reference line, cheap, pretty good stability, very good accuracy, and completely self-contained. bye, Said ** See what's free at http://www.aol.com. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
The VP is still distinctly worse until tau 1000sec. The plots are not conclusive evidence that correcting for the sawtooth error isn't advisable. What about hanging bridges and similar artifacts? Bruce It would be interresting to see exactly how much the sawtooth on an M12 affects actual GPSDO setups. I made a hardware sawtooth remover a couple of years ago as an experiment, but where I used it I didn't seen much difference. I just checked again recently, but the sawtooth errors seem very small compared to overall instability, even for just an M12 to Rb at a fairly stable temperature. The larger errors of a few ns (over 1000s) which are occasionally visible when the M12 is temperature controlled seem regularly to be exceeded by other instabilities in the pps output. There are plenty of times when sawtooth removal would be of use/interest, but in a typical GPSDO which has a heap of other errors, I do wonder what improvements in performance would actually be seen in the output from the oscillator - which is all that a lot of people are really interrested in. Angus. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
There are plenty of times when sawtooth removal would be of use/ interest, but in a typical GPSDO which has a heap of other errors, I do wonder what improvements in performance would actually be seen in the output from the oscillator - which is all that a lot of people are really interrested in. What's the time constant on the PLL filter? How many of the other errors will it mask? I'd expect it to mask most of the sawtooth errors - all but the hanging bridges that are wider than the PLL time constant. That's what makes them so interesting. Has anybody gone hunting for hanging bridges to see what the statistics on their width look like? -- These are my opinions, not necessarily my employer's. I hate spam. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Hal Murray wrote: There are plenty of times when sawtooth removal would be of use/ interest, but in a typical GPSDO which has a heap of other errors, I do wonder what improvements in performance would actually be seen in the output from the oscillator - which is all that a lot of people are really interrested in. What's the time constant on the PLL filter? How many of the other errors will it mask? I'd expect it to mask most of the sawtooth errors - all but the hanging bridges that are wider than the PLL time constant. That's what makes them so interesting. Has anybody gone hunting for hanging bridges to see what the statistics on their width look like? Hal One has to be careful not to misuse such statistics to mask the shortcomings of a poorly designed GPSDO. Hanging bridges, their width and their occurrence frequency are not likely to be random, they will tend to be correlated in a complex manner with temperature. Only the short term instabilities of the GPS receiver oscillator and its ageing corrupting the correlation. Consequently such statistics will be specific to a particular thermal environment and receiver. It would be misleading to apply them to other receivers and thermal environments. Since the fix is a relatively trivial piece of software when the sawtooth correction data is available, why not do it? Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
One has to be careful not to misuse such statistics to mask the shortcomings of a poorly designed GPSDO. Bruce, I'm getting the idea you don't like _any_ of the hobbyists GPSDO's that have come out in the last ten years... So I'm curious what then would qualify as a well designed GPSDO in your opinion? (and please don't bring up the Quartzlock thing; it's a hundred times more expensive than the GPSDO's we're talking about here). /tvb ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Tom Van Baak wrote: Bruce, I'm getting the idea you don't like _any_ of the hobbyists GPSDO's that have come out in the last ten years... So I'm curious what then would qualify as a well designed GPSDO in your opinion? (and please don't bring up the Quartzlock thing; it's a hundred times more expensive than the GPSDO's we're talking about here). /tvb Tom If one insists on using a GPS timing receiver rather than using carrier phase disciplining techniques, then anything that: 1) Has a one shot phase of sufficient resolution (and accuracy) (1ns or preferably a little better for an M12M ) so that performance is limited by that of the GPS receiver rather than the phase detector. 2) Corrects for sawtooth error (when correction data is made available by the GPS receiver) in software (hardware correction also acceptable but probably unnecessarily expensive). 3) Uses statistical filtering to eliminate phase error outliers. 4) Has adjustable PLL parameters so that the disciplining algorithm can be tuned to obtain the best performance from a particular receiver and the OCXO or other source being disciplined. A third order loop or equivalent is perhaps desirable to correct for linear frequency drift. 5) Uses a sufficiently high resolution monotonic low noise DAC (or equivalent DDS system) that the OCXO short term stability limits the performance. 6) Doesn't rely on the relative phase of an independent oscillator being random with respect to the PPS signal or the OCXO signal. This is rarely satisfactory unless one makes heroic efforts to isolate the oscillator from the PPS and the OCXO. Injection locking of the independent oscillator is almost inevitable unless there is sufficient isolation. 7) Allows the GPS receiver data, PLL loop parameters and phase error statistics etc to be monitored/logged. 8) Indicates when the GPS receiver data is too noisy/unavailable for disciplining so that the OCXO is operating in holdover mode. 9) Uses synchronisers where needed to effectively eliminate metastability as a significant concern. None of these requirements is particularly difficult or expensive to implement. The most difficult being the high resolution phase detector, particularly if one has stringent holdover requirements. Constructing a phase detector with a range of say 1us and a sub nanosecond resolution is relatively easy and inexpensive , extending the range to greater than 1us is somewhat more difficult/expensive unless one uses a 1GHz clock. The high resolution DAC is also a bit of a challenge in that high resolution audio DACs have relatively poor tempcos which means that their temperature needs to constant to within a fraction of a degree for a a time interval of around the loop response time to avoid degrading the short term stability. It may be better to use a lower tempco higher resolution monotonic (e.g. string DAC) DAC and just dither its output. If a suitable low pass analog filter is used between the DAC and the OCXO frequency control input then the residual dither amplitude at the filter output can be extremely small. Optional but desirable characteristics (particularly when disciplining well aged OCXOs): 10) Allow disciplining of an OCXO whose frequency can no longer be adjusted to nominal. 11) Include an offset DDS system so that a output at the nominal frequency is available. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
From: Christopher Hoover [EMAIL PROTECTED] Subject: Re: [time-nuts] GPS: ADEV or MDEV? Date: Sat, 21 Apr 2007 20:06:59 -0700 Message-ID: [EMAIL PROTECTED] Bruce wrote: 6) Doesn't rely on the relative phase of an independent oscillator being random with respect to the PPS signal or the OCXO signal. This is rarely satisfactory unless one makes heroic efforts to isolate the oscillator from the PPS and the OCXO. Injection locking of the independent oscillator is almost inevitable unless there is sufficient isolation. I am surely exposing my ignorance here, I don't completely understand this. The independent oscillator is being used to measure the TI of the PPS, is that correct? If so, let's presume this oscillator is clocking a free running counter, the value of which is being captured on the edge of the PPS and sent to a processor (for sawtooth correction, filtering, etc.). How does injection locking occur? The PPS signal triggers the trigger and pulls the supplies and this inject into the oscillator. If you think according to similar lines there is more in the signal integrity line. Cheers, Magnus ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Surely its better to detect the pseudorange instability in software (if one has access) rather than attenuating the signal from all SVs. Another option is to use multiple GPS antennae operating as a phased array. Bruce Of course but I think that approach is part of sub nanosecond time transfer accuracy. Attenuation is simple and the results are descent. The signal is less than 6dB above the receiver noise floor to acquire SV. Unfortunately the SV do not have matched output power so they are acquired and weighted unequally. Also slow ionospheric disturbances are not rejected from the pps solution. Not perfect but still good. I made a half hour WinOncore12 .bin file april13_-10dB.rar that is extremely boring for the first half and then somewhat revealing as SV are dropped, recaptured and then finally lost. WinOncore is required to play it. Also thank you Didier for the use of your ftp http://www.ko4bb.com/~manuals/ Peter ___ time-nuts mailing list [EMAIL PROTECTED] https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
From: Peter Schmelcher [EMAIL PROTECTED] Subject: Re: [time-nuts] GPS: ADEV or MDEV? Date: Sat, 14 Apr 2007 02:40:57 -0700 Message-ID: [EMAIL PROTECTED] Surely its better to detect the pseudorange instability in software (if one has access) rather than attenuating the signal from all SVs. Another option is to use multiple GPS antennae operating as a phased array. Bruce Of course but I think that approach is part of sub nanosecond time transfer accuracy. Attenuation is simple and the results are descent. The signal is less than 6dB above the receiver noise floor to acquire SV. Unfortunately the SV do not have matched output power so they are acquired and weighted unequally. The difference in path-length and thus space attenuation is compensated by the antenna pattern such that it emits stronger to the rimb of its earth view than central. That evens out power-shifts due to space attenuation for the viewer. Otherwise we would have yeat another problem to deal with. Cheers, Magnus ___ time-nuts mailing list [EMAIL PROTECTED] https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
On Sat, April 14, 2007 13:32, Magnus Danielson said: From: Peter Schmelcher [EMAIL PROTECTED] Subject: Re: [time-nuts] GPS: ADEV or MDEV? Date: Sat, 14 Apr 2007 02:40:57 -0700 Message-ID: [EMAIL PROTECTED] Surely its better to detect the pseudorange instability in software (if one has access) rather than attenuating the signal from all SVs. Another option is to use multiple GPS antennae operating as a phased array. Bruce Of course but I think that approach is part of sub nanosecond time transfer accuracy. Attenuation is simple and the results are descent. The signal is less than 6dB above the receiver noise floor to acquire SV. Unfortunately the SV do not have matched output power so they are acquired and weighted unequally. The difference in path-length and thus space attenuation is compensated by the antenna pattern such that it emits stronger to the rimb of its earth view than central. That evens out power-shifts due to space attenuation for the viewer. Otherwise we would have yeat another problem to deal with. Do the HP GPSDOs not have a tuneable maskangle? Or is the receiver in 1SV timing mode? and do not move to another before it comes to close to the horizon? Maskangles are used to chose an angle vs the horizon that you want all your used SVs to be above. Was I missing something here? -- Björn ___ time-nuts mailing list [EMAIL PROTECTED] https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Surely it would be much simpler (in principle at least) to just to add sufficient Gaussian phase noise to the GPS receiver clock so that potential coherence problems are virtually eliminated. The noise only really needs to be added to the PPS positioning algorithm/hardware. My two cents. Masking the sawtooth is bad news. If you log the pps output from a Z3816A you get the incorrect impression that the OCXO and/or the atmosphere is unstable for time spans of tens of minutes. My Z3816A pps logs look typical sd=10ns a few 50ns spikes the EFC ad=2 sd=5 etc. But use the OXCO to synthesize a low noise LO for a second gps and the pps becomes very stable a couple of ns. In the Z3816A the gps LO crystals 0.1 second tau and the sawtooth interact and do bad things. It is best to eliminate the problem and then to move on to new things. I am attenuating my HP hockey puck antenna approximately 10dB to remove SV with range instability from the gps pps solution. The current setup acquires SV at about 30 degree and can then track them down to about 20 degree. The idea is that transit time instability modulates the SV signal requiring a greater tracking filter bandwidth which reduces the SNR hopefully below tracking. It is not perfect - testing continues. Peter ___ time-nuts mailing list [EMAIL PROTECTED] https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Dr Bruce Griffiths wrote: Peter I only intended to indicate that not all GPS timing receivers need exhibit hanging bridges. The Trimble Resolution T for example may (or may not) exhibit hanging bridges. If anyone has any evidence either way it would be interesting to look at it. If one also logs the PPS correction then you haven't lost any information due to the randomisation of the PPS error. Surely its better to detect the pseudorange instability in software (if one has access) rather than attenuating the signal from all SVs. Another option is to use multiple GPS antennae operating as a phased array. Bruce I believe the Thunderbolt should not exhibit hanging bridges either, since it's OCXO is used to clock the CPU. I have yet to demonstrate that. It would be helpful to have another GPSDO design to compare it against, and I have not come around to finishing my own home-brew design yet. The interesting part is that the Thunderbolt is actually a simpler system. Hanging bridges are due to the beat (or lack thereof leading to a constant phase offset) between the OCXO and the CPU clock if I understood well. If these two are one and the same, the problem goes away. Didier ___ time-nuts mailing list [EMAIL PROTECTED] https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Hal Murray wrote: Surely it would be much simpler (in principle at least) to just to add sufficient Gaussian phase noise to the GPS receiver clock so that potential coherence problems are virtually eliminated. It doesn't look reasonable to me, but I'm not good at this sort of math. The clock is, say, 40 MHz. That's 25 ns, so the output jitter will be 25 ns p-p. If you are at the top of a hanging bridge, you have to add enough noise to hide a 12.5 ns offset. Will a typical GPS receiver still work with that much trash on its clock signal? What sort of noise spectrum would you use? Hal The noise only really needs to be added to the PPS positioning algorithm/hardware. In principle (if you can get at the firmware) it would be possible to implement noise shaping using a sigma delta modulator (possibly with added noise dithering to break up idle patterns - like hanging bridges) so that the short term PPS positioning noise increases but the average position is more stable. This may be useful when a long time constant PLL is used to discipline an oscillator with the PPS output of the receiver. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
From: Dr Bruce Griffiths [EMAIL PROTECTED] Subject: Re: [time-nuts] GPS: ADEV or MDEV? Date: Tue, 10 Apr 2007 17:49:03 +1200 Message-ID: [EMAIL PROTECTED] Poul-Henning Kamp wrote: In message [EMAIL PROTECTED], Brooks Shera writes: The impact of time averaging to suppress white phase noise is illustrated by a new plot TVB has created and placed on his website http://www.leapsecond.com/pages/3gps/gps-adev.gif. These revealing plots show that just a few hundred-second time average has largely removed the performance difference between the aged VP and the newer M12+ and has almost entirely removed the benefit of sawtooth correction (CNS-CNS II). These important results are not evident in the usual ADEV plot. This is simply not true. No averaging period can ever guarantee you will not get a constant offset from the sawtooth. You may get rid of the noise, but not the bias. Poul-Henning Not to mention the case of the GPS receiver with a pathologically stable local oscillator frequency that generates lots of hanging bridges and little sawtooth. Since software sawtooth correction is virtually free why not do it, when such data is made available by the GPS receiver? Hoping that the local oscillator frequency is sufficiently unstable isn't good engineering, one should design to accommodate the worst case possibilities. Ovenize it to control the sawtooth frequency? Temperature controlled oscillator! :-) Should be fairly simple to acheive, really just a FLL. Cheers, Magnus ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
[context is avoiding hanging bridges] Ovenize it to control the sawtooth frequency? Temperature controlled oscillator! :-) Should be fairly simple to acheive, really just a FLL. I was wondering about that a while ago. Is the basic idea feasible? Assume you had a good crystal with an A/D driving a voltage control input and wired that up to your GPS unit. You would need to lock that crystal to the PPS output from the GPS, but you don't want an exact lock, you want to be off by a fraction of a HZ, say 1/10th or 1/20th. One problem is that as soon as you lock it up, the normal results are no longer random. You could easily get locked on so that your 10 or 20 samples per sawtooth are biased by up to 1/N of the sawtooth. As long as N is big enough that's probably OK. It might be better to lock on the the steep edge of the sawtooth, getting samples from just before and just after the big step. -- These are my opinions, not necessarily my employer's. I hate spam. ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Hal Murray wrote: [context is avoiding hanging bridges] Ovenize it to control the sawtooth frequency? Temperature controlled oscillator! :-) Should be fairly simple to acheive, really just a FLL. I was wondering about that a while ago. Is the basic idea feasible? Assume you had a good crystal with an A/D driving a voltage control input and wired that up to your GPS unit. You would need to lock that crystal to the PPS output from the GPS, but you don't want an exact lock, you want to be off by a fraction of a HZ, say 1/10th or 1/20th. One problem is that as soon as you lock it up, the normal results are no longer random. You could easily get locked on so that your 10 or 20 samples per sawtooth are biased by up to 1/N of the sawtooth. As long as N is big enough that's probably OK. It might be better to lock on the the steep edge of the sawtooth, getting samples from just before and just after the big step. Hal Surely it would be much simpler (in principle at least) to just to add sufficient Gaussian phase noise to the GPS receiver clock so that potential coherence problems are virtually eliminated. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Brooks Shera wrote: In view of recent interest in the Allan Deviation of GPS-based 1 pps time, it should be mentioned that the calculation of ADEV is based on a statistical model which is not completely appropriate for noise sources present in GPS signals and their decoding hardware/software. For example, white phase noise, which results from quantization effects in GPS receivers (sawtooth) and in counter-based TICs. The result is that ADEV plots can be misleading if used to compare the performance of clocks or signals with dissimilar noise properties. Allan pointed this out in his 1981 paper in 'Proc. 35th Ann. Freq. Control Symposium', and proposed a modified ADEV (MDEV) as a solution. He wrote: A direct application for using the modified allan variance recently arose in the analysis of atomic clock data as received from a GPS satellite. ...Using Mod AV we can tell that the fundamental limiting noise process involved in the system is white noise PM with the exciting result that averaging for four minutes can allow one to ascertain time differences to better that one nsec. The impact of time averaging to suppress white phase noise is illustrated by a new plot TVB has created and placed on his website http://www.leapsecond.com/pages/3gps/gps-adev.gif. These revealing plots show that just a few hundred-second time average has largely removed the performance difference between the aged VP and the newer M12+ and has almost entirely removed the benefit of sawtooth correction (CNS-CNS II). These important results are not evident in the usual ADEV plot. Such a time averaging effect is not entirely unexpected. In the long run, the averaged time jitter in the transmission path between the satellite and the ground must approach zero. What IS surprising is that the noise averages to zero rather quickly and for purposes that require time filtering, e.g. a GPSDO, it can safely be ignored. Regards, and have fun, Brooks ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts Brooks The plot you are referring to is surely: http://www.leapsecond.com/pages/3gps/gps-adev-mdev.gif The VP is still distinctly worse until tau 1000sec. The plots are not conclusive evidence that correcting for the sawtooth error isn't advisable. What about hanging bridges and similar artifacts? Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
Re: [time-nuts] GPS: ADEV or MDEV?
Poul-Henning Kamp wrote: In message [EMAIL PROTECTED], Brooks Shera writes: The impact of time averaging to suppress white phase noise is illustrated by a new plot TVB has created and placed on his website http://www.leapsecond.com/pages/3gps/gps-adev.gif. These revealing plots show that just a few hundred-second time average has largely removed the performance difference between the aged VP and the newer M12+ and has almost entirely removed the benefit of sawtooth correction (CNS-CNS II). These important results are not evident in the usual ADEV plot. This is simply not true. No averaging period can ever guarantee you will not get a constant offset from the sawtooth. You may get rid of the noise, but not the bias. Poul-Henning Not to mention the case of the GPS receiver with a pathologically stable local oscillator frequency that generates lots of hanging bridges and little sawtooth. Since software sawtooth correction is virtually free why not do it, when such data is made available by the GPS receiver? Hoping that the local oscillator frequency is sufficiently unstable isn't good engineering, one should design to accommodate the worst case possibilities. Bruce ___ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts