Re: [time-nuts] GPSDO using 100Hz
Bruce Griffiths answered: Its difficult to make much useful comment as you provide few measured results. With an M12+T or equivalent the ADEV of the PPS output (without sawtooth correction) goes below 1E-10 at Tau 200 sec or so. Thus with an optimized GPSDO it wont take an hour or so to achieve 1E-10 stability. However a single shot phase error measurement system resolution of around 1ns or so is usually required. Take a look at the GPSDO ADEV plots at: http://www.leapsecond.com/pages/gpsdo/ Where the ADEV for various GPSDO remains below 1E-11 over the Tau range of [0.1s, 100,000s] The achievable performance also depends a great deal on the quality of the OCXO used in the GPSDO. If the digital phase error measurement techniques you have been comparing your system with have inadequate resolution it will tale longer for the measured ADEV to fall below 1E-10. If the OCXO used has a relatively high ADEV at low values of tau then it may well required averaging over very long time intervals to achieve an ADEV below 1E-10. More detail is required before an analysis of the performance of your system is possible. Bruce * Bruce, Thanks for the feedback, Good information to know, but you seemed to missed my point and question. Yes I am comparing to SIMPLE, and home built type, non optimized, trackers like Brooks Shera's 10 Mhz GPS Frequency Standard. I understand, and even wrote, NIST reports about a 10ns uncertainty with a 10 minute average, which would give a 1e-8 / 600, or 1.5 parts in 1E-11 possible in 600 seconds, 1e-10 in 100 seconds. Its safe to assume NIST is not using Shera's unit, which I believe adds an additional 24 ns or is it 41.7 ns uncertainty to each 1 second reading. Besides asking if anyone is using the 100Hz output, I would like to ask why don't the generally available GPSDO use the 100Hz, which can give about 1 ns of certainly with a simple PLL and analog RC filter, instead of the using the 1 sec which has more like 100 ns of uncorrected uncertainty in it and must use a processor? Also I should comment that on LeapSecond.com http://www.leapsecond.com/pages/gpsdo/ you stated where the ADEV for various GPSDO remains below 1E-11 over the Tau range of [0.1s, 100,000s]. This Tau has NOTHING to do with the tracking time constant that the GPSDO is set to, which is usually recommended to be set for a TC of several hours for optimal results. BTW most of what they have plotted is the results of NOT setting the GPSDO tracking TC slow enough. This is why the Allan Deviation increases in the tracking mode at mid averaging times. The 1 second GPS tracking signal is adding noise, which pretty much makes my point that 1 PPS signal is not so good to use if you want good fast results. Warren ___ 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] GPSDO using 100Hz
Ulrich Bangert Thanks for the great Information. UR) James Miller, G3RUH, also uses the 100 Hz signal Thanks, that is what I was asking for, other people that were using the 100Hz. Now if I could just figure out how to post a response under an existing thread, and not have it start a new one I'd be happy. Concerning your other comments: UR) it's neither specifically new nor ... I did not mean to imply that the 100Hz was new or its use was discovered by me, Mostly I'm wondering why it is so little used, and I'm looking for other people using it. Something that I can do with the 100Hz that you probable have not seen before is the ability to make a GPSDO similar to Brooks Shera's processors based unit using just 3 standard off the self 74HC IC's and a pretty crappy Osc. UR) The key error in your idea is the assumption that every 100 positive slopes of your 100 Hz is identical with the start of a new second, which is wrong when you look at it at a second to second base. Correct that my assumption was that every 100 pulses gives a one second pulse at the same location.You are incorrect that when this is not true, it is a key error in my idea. I did see that the 100Hz pulse that comes out at about the time as the 1Hz sawtooth correction is updated seems to have the same amount of phase error as reported for the 1 Hz, best I could tell. I do know it is sometimes true, I'll check that out better and verify that it is not always true on my unit. Thanks for the information. BUT, NO matter, because the max jitter error of around 100ns is the same in both of the 100Hz and the 1 Hz signal, and the 100Hz error is nonaccumative. That is when I average 100 points each with an uncertainty of 100ns I get at least the Square root of 100 better resolution If they where truly random, which they are not, they have the same basic ramp type phase response as the 1 Hz sawtooth, just much faster so averaging works much better than if they where random. What I have found is that the net effect is for the most part that averaging 100 100Hz signals will give about the same error result as averaging 100 One second pulses. If you want to proof it to yourself use a digital scope with an average function and look at the results. (there are some rare exceptions, I skip over for now) For another example, consider what the results would be if you just used every 100th 1 second pulse. You'd end up with 100ns uncertainty in 100 seconds rather the under 10 ns you get by averaging 100 1 second pulses (most of the time). UR) What is true is that the MEAN frequency of the 100 Hz is locked to the GPS and that is why your fast PLL works as well as James Miller's. What I tried to state is that the 100Hz is fast enough to make a simple analog PPL, something that is not very easy to do with the 1 Hz signals. Thanks for the name, I'll check more on how he is doing his simple GPS tracker. It looks like he is using the 10KHz output of the Jupiter GPS. Unfortunately my oncore does not have a 10KHz output. Thanks, Warren *** James Miller, G3RUH, also uses the 100 Hz signal. So it's neither specifically new nor the philosopher's stone that you may perhaps believe to have found. The PPS and the 100 Hz signal come from the SAME oscillator. The jitter in the PPS comes from the fact that the receiver logic decides WHICH slope of the oscillator signal comes next close to the TRUE point of time where the PPS should be generated. The key error in your idea is the assumption that every 100 positive slopes of your 100 Hz is identical with the start of a new second, which is wrong when you look at it at a second to second base. What is true is that the MEAN frequency of the 100 Hz is locked to the GPS and that is why your fast pll works as well as James Miller's. Best regards Ulrich Bangert ___ 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] GPSDO using 100Hz
This is my first listing so don't know if I'm doing correctly. Question is: Has anyone done any work using the 100Hz GPS output, instead of the 1 Hz output? The reason I ask is because I am in the process of cleaning up my SIMPLE GPS Freq Phase lock tracker breadboard that does about the same as most, but in a different way. By using just a few basic standard 74HCxx IC logic gates and Flip-Flops with RC's I phase lock a low accuracy VCXO to give me a high accuracy 10MHz reference. This is something I built for my own use, to check the accuracy of my 10 MHz freq standard and to check the accuracy of the 60KHz WWVB signal. I found out that the most important trick to keep the GPS tracker simple (GPSDO), fast and low power, and do it without the usual microprocessor stuff, was to use the 100Hz output instead of the 1HZ output from the Motorola Oncore receiver board. Basically this allows simple logic circuits to give 10 to 100 times better results, such as phase noise of around 1 ns instead of 100ns from second to second, cold turn on and full accuracy lock of the tracking osc in well under a minute (with GPS left on). It also has the ability to track the GPS phase errors at about a 10 Hz rate which gave me some new interesting insight, such as the 1 sec sawtooth error that the Oncore reports on, that would usually requires the use of a computer to process, is no longer very relevant. I found out the 1 second reported phase error is just an alising artifact of a higher freq phase dither that can be completely filtered out with a 1 second Time constant RC. (i.e RC replaces microprocessor). It generally takes about an hour Tracking time constant to get good 1e-10 Freq stability using the 1 Second GPS signal, and more like 3 to 24 hours as not degrade a really good Oscillator. I am getting such good results with the simple high speed RC PLL, I am now doing some test to see if there are new ways to reduce the Tracking time. It would seem it should be possible to get 1e-10 in more like a 100 seconds instead of an hour. This would then reduce the required stability of the tracking Oscillator by an order of magnitude or better. The simple PLL tracks the 100Hz within about 1 ns and can lock up in 10 seconds, and the GPS signals themselves don't seem to vary by more than about 10 ns with 10 minute and 1Hr average times. (from http://tf.nist.gov/service/gpstrace.htm) I have not found any GPS signal accuracy data for shorter average times, so that is what I'm presently plotting. One of the other many benefits of fast GPS tracking may be the possibility of a more mobile Disciplined Oscillator. If anyone has done any work using the 100Hz Oncore GS signal to do frequency tracking, I would be very interested in further discussions. Warren ___ 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.