Re: [time-nuts] TBolt: UTC PPS
Bob Camp wrote: Hi When there's a leap second, what do you want to do? Ignore it = GPS time put in an extra pulse = UTC time No, this is about which set of corrections to use. The GPS time is what the navigation solution cranks out, and UTC time is what you then correct that into for user display. Cheers, Magnus ___ 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] TBolt: UTC PPS
Magnus Danielson wrote: Bob Camp wrote: Hi When there's a leap second, what do you want to do? Ignore it = GPS time put in an extra pulse = UTC time No, this is about which set of corrections to use. The GPS time is what the navigation solution cranks out, and UTC time is what you then correct that into for user display. Cheers, Magnus When a leapsecond occurs the PPS pulse labels at the top of the minute are changed, no extra pulse is inserted. 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.
Re: [time-nuts] TBolt: UTC PPS
Tom Van Baak wrote: It also has an option to sync the PPS with GPS or UTC. I thought they were off by an integral number of seconds so I don't expect any change. Does anybody know what's going on here? The GPS broadcast message includes leap second and a0 and a1 terms which are used to forward predict UTC from GPS time, which is derived from UTC(USNO). A0 and A1 are what you'd guess: a phase offset and a frequency offset. See page 157 of the GPS ICD: http://www.navcen.uscg.gov/gps/geninfo/ICD-GPS-200C%20with%20IRNs%2012345.pdf So yes, at the ns level the two 1PPS will differ since UTC is not the same as UTC(USNO) or any other UTC(k) for that matter. Have you measured it? I would be interested to know what the delta is for your receiver today. You can also dump out the a1 and a0 corrections with binary commands on some GPS receivers. The GPS time is steered towards UTC(USNO) to be maintained within 1 us. The effective difference is in a few ns. I could dig up the webpage at USNO giving the difference if needed to. (TvB already has a mail from me on this topic) There should be no effective difference between them. Cheers, Magnus ___ 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] TBolt: UTC PPS
Bruce Griffiths wrote: Magnus Danielson wrote: Bob Camp wrote: Hi When there's a leap second, what do you want to do? Ignore it = GPS time put in an extra pulse = UTC time No, this is about which set of corrections to use. The GPS time is what the navigation solution cranks out, and UTC time is what you then correct that into for user display. Cheers, Magnus When a leapsecond occurs the PPS pulse labels at the top of the minute are changed, no extra pulse is inserted. In that case the function is completely incorrectly named. Cheers, Magnus ___ 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] What time is it anyway?
The task of the BIPM is to ensure world-wide uniformity of measurements and their traceability to the International System of Units (SI). It does this with the authority of the Convention of the Metre, a diplomatic treaty between fifty-four nations, and it operates through a series of Consultative Committees, whose members are the national metrology laboratories of the Member States of the Convention, and through its own laboratory work. The BIPM carries out measurement-related research. It takes part in, and organizes, international comparisons of national measurement standards, and it carries out calibrations for Member States. So, the BIPM is THE reference, because it result from international Meter Convention . Every country (54), who sign this convention, participate to elaborate the UTC by the mean of its own laboratories. Cordialement Alain F4GBC - Original Message - From: Gerhard Hoffmann dk...@arcor.de To: Discussion of precise time and frequency measurement time-nuts@febo.com Sent: Sunday, March 28, 2010 1:18 PM Subject: Re: [time-nuts] What time is it anyway? bullet1.gif___ 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] TBolt: UTC PPS
On Mon, 29 Mar 2010 08:24:00 +0200, Magnus Danielson wrote: Tom Van Baak wrote: It also has an option to sync the PPS with GPS or UTC. I thought they were off by an integral number of seconds so I don't expect any change. Does anybody know what's going on here? The GPS broadcast message includes leap second and a0 and a1 terms which are used to forward predict UTC from GPS time, which is derived from UTC(USNO). A0 and A1 are what you'd guess: a phase offset and a frequency offset. See page 157 of the GPS ICD: http://www.navcen.uscg.gov/gps/geninfo/ICD-GPS-200C%20with%20IRNs%2012345.pdf So yes, at the ns level the two 1PPS will differ since UTC is not the same as UTC(USNO) or any other UTC(k) for that matter. Have you measured it? I would be interested to know what the delta is for your receiver today. You can also dump out the a1 and a0 corrections with binary commands on some GPS receivers. The GPS time is steered towards UTC(USNO) to be maintained within 1 us. The effective difference is in a few ns. I could dig up the webpage at USNO giving the difference if needed to. (TvB already has a mail from me on this topic) There should be no effective difference between them. UTC(USNO)-UTC(BIPM) difference is actually swinging between +- 7ns: http://tycho.usno.navy.mil/gif/tai.gif A good summary showing the differences for the international laboratories UTC-UTC(k) for the past with last date 2010 March 12 ( and a lot more): http://tycho.usno.navy.mil/latestcircT Arnold ___ 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] TBolt: UTC PPS
The Trimble Thunderbolt is one of the GPS receivers that will report subframe 4 data. The UTC parameters are in TBolt packet 0x58 type 5. Right now, the numbers are: a0 = -0.931 ns and a1 = -5.33e-15 s/s (which is -0.460 ns/day). Thanks. I was close to typing in the same stuff. I got: -9.31323e-10 -5.32907e-15 In that packet, a0 is double precision. Does anybody know why they need all those bits? -- These are my opinions, not necessarily my employer's. I hate spam. ___ 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] What time is it anyway?
So there are 250 clocks, presumably, spread around the World and owned by the member countries of the BIPM. Their time is somehow compared centrally and an absolute time is determined from them. Each clock will then have a delta to apply to it's own time to provide the BIPM absolute time which then allows each member country to use its own clock for timing. This sounds like a logistic and bureaucratic nightmare, I will have to look further as to how this is achieved. Any pointers gratefully accepted. On 29 March 2010 01:51, Magnus Danielson mag...@rubidium.dyndns.org wrote: Arnold Tibus wrote: The answer looks to me a bit difficult reading the USNO definition : INTERNATIONAL TIME SCALES AND THE B.I.P.M. http://tycho.usno.navy.mil/bipm.html citing: ...the U.S. Naval Observatory timescale, UTC(USNO), and its real-time implementation, Master Clock #2 (MC #2), are kept within a close but unspecified tolerance of the international atomic timescale published by the Bureau International des Poids et Mesures (International Bureau of Weights and Measures [BIPM]) in Sevres, France. ...Hence, all these atomic timescales are called Coordinated Universal Time (UTC), of which USNO's version is UTC(USNO). ...The difference between UTC (computed by BIPM) and any other timing center's UTC only becomes known after computation and dissemination of UTC, which occurs about two weeks after the fact. This difference is presently limited mainly by the long-term frequency instability of UTC. UTC(USNO) has been kept within 26 nanoseconds of UTC during the past year through frequency steering of our Master Clocks to our extrapolation of UTC. So I do understand that BIPM is the world's time keeper, but there may be a difference between the UTCs of up to 26ns? First of all, that comment related only to USNOs performance, not any other lab like NIST, PTB or SP. Historically, it may be 100s of ns away if they fluke it. This is the difference in should and will supply stability. ...Since synchronization is never perfect, we provide the latest data below on the differences between UTC and the UTC of other timing centers, including USNO,... and ...All of our reference clocks are real-time approximations of UTC(USNO), and as such are denoted UTC(USNO,MC). Master Clock #2 (MC #2) is our official reference clock... So I understand this as, that the USA do refer to the time reference of USNO - and the rest of the world to BIPM directly? No. First of all, you should consider that there is two bodies in the USA, NIST and USNO. NIST is the primary body for all BIPM stuff, but USNO is a complementary body in regards to time. Then, both NIST and USNO refer back to BIPM, just as PTB in Germany refers back to BIPM and SP in Sweden refers back to BIPM. The national laboratories represent the first link in the traceability chain down from BIPM. In this regard NIST and USNO is just that, the US national laboratories. If I where to get my traceability through my national laboratory SP and I also follow the calibration and reporting rules, set forth in ISO 17025 and related, then that measurement is traceable to BIPM and let's say HP in their Santa Clara lab does the same, they relate to NIST (not USNO really) then by mutual agreements these measurements is being recognised and agreed upon. This aspect of recognised is highly related to trade and trust in trade. Since dec. 2009 the PTB in Braunschweig, Germany (with the new CSF2) and the BIPM in Sevre are the only countries running 4 of the most precise primary Cs fountain clocks, if I am informed correctly. Together they should run quite close to the time defined by BIPM I think, and according our law our official time is transmitted by the PTB. Cesium fountains is not a good measure of lab stability. Only a small fraction of all the clocks (250 is a number frequently occuring) being used. Each individual clock is being measured and weighted in. Most of them is HP (now Symmetricom) 5071A clocks with normal or high-performance tubes. The stability of these clocks is being compared and a optimum stable paper-clock called EAL is created. This is then corrected into TAI and with the decissions of IERS is corrected into UTC. TAI is being somewhat of a paper-time, but several labs realize their variant of it, referred to as TA(x) where x is replaced with lab name. Now, How do I have to interprete the readout of GPSDOs like Trimble's Thunderbolt and others PPS difference in ns to UTC? To which UTC? I suppose to the time transmitted by the US GPS SATs. The traceability of GPS time becomes: BIPM - USNO - GPS-time - Sat time - User equipment USNO provides measurements into BIPM, they then realize UTC(USNO) and steer it towards the BIPM for long-term stability. GPS time is separately maintained and is being supervised by USNO, the UTC(GPS) is then steered towards UTC(USNO). USNO provides the calibration
Re: [time-nuts] TBolt: UTC PPS
I was close to typing in the same stuff. I got: -9.31323e-10 -5.32907e-15 In that packet, a0 is double precision. Does anybody know why they need all those bits? Hal, Referring to the GPS ICD, the size of A0 in subframe 4 is a signed scaled 32 bit integer so if a receiver chooses to convert it to floating point it needs a double precision format to contain it without loss of precision. A1, on the other hand, is a signed scaled 24 bit integer and so it just fits in a single precision floating point number. So Trimble did the right thing here. What's equally interesting is to look at the binary values for A0 and A1 right now. |A0|, the time offset, ranges from 0 to 2^31 scaled by 2^-30 seconds, or 0.931 ns to 2 seconds. |A1|, the frequency offset, ranges from 0 to 2^23 scaled by 2^-50, or 7.45e-9 to 8.88e-16. Based on this we can tell the binary values for A0 and A1 today are just 1 and 6. So while the range of these two correction numbers is very wide; GPS is running so close to UTC that it's at the extreme bottom end of the resolution right now. /tvb ___ 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] What time is it anyway?
Hi It's pretty much the way any international standard unit is worked out. The various national labs come up with a best guess and then they swap around to see how the guesses all compare with each other. Ultimately they come up with a way to agree on what the average value is. The math to compute that is rarely simple. After watching the magic number for a while (and tweaking the math) they conclude that it's the best guess at what's correct. Bob -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Steve Rooke Sent: Monday, March 29, 2010 8:10 AM To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] What time is it anyway? So there are 250 clocks, presumably, spread around the World and owned by the member countries of the BIPM. Their time is somehow compared centrally and an absolute time is determined from them. Each clock will then have a delta to apply to it's own time to provide the BIPM absolute time which then allows each member country to use its own clock for timing. This sounds like a logistic and bureaucratic nightmare, I will have to look further as to how this is achieved. Any pointers gratefully accepted. On 29 March 2010 01:51, Magnus Danielson mag...@rubidium.dyndns.org wrote: Arnold Tibus wrote: The answer looks to me a bit difficult reading the USNO definition : INTERNATIONAL TIME SCALES AND THE B.I.P.M. http://tycho.usno.navy.mil/bipm.html citing: ...the U.S. Naval Observatory timescale, UTC(USNO), and its real-time implementation, Master Clock #2 (MC #2), are kept within a close but unspecified tolerance of the international atomic timescale published by the Bureau International des Poids et Mesures (International Bureau of Weights and Measures [BIPM]) in Sevres, France. ...Hence, all these atomic timescales are called Coordinated Universal Time (UTC), of which USNO's version is UTC(USNO). ...The difference between UTC (computed by BIPM) and any other timing center's UTC only becomes known after computation and dissemination of UTC, which occurs about two weeks after the fact. This difference is presently limited mainly by the long-term frequency instability of UTC. UTC(USNO) has been kept within 26 nanoseconds of UTC during the past year through frequency steering of our Master Clocks to our extrapolation of UTC. So I do understand that BIPM is the world's time keeper, but there may be a difference between the UTCs of up to 26ns? First of all, that comment related only to USNOs performance, not any other lab like NIST, PTB or SP. Historically, it may be 100s of ns away if they fluke it. This is the difference in should and will supply stability. ...Since synchronization is never perfect, we provide the latest data below on the differences between UTC and the UTC of other timing centers, including USNO,... and ...All of our reference clocks are real-time approximations of UTC(USNO), and as such are denoted UTC(USNO,MC). Master Clock #2 (MC #2) is our official reference clock... So I understand this as, that the USA do refer to the time reference of USNO - and the rest of the world to BIPM directly? No. First of all, you should consider that there is two bodies in the USA, NIST and USNO. NIST is the primary body for all BIPM stuff, but USNO is a complementary body in regards to time. Then, both NIST and USNO refer back to BIPM, just as PTB in Germany refers back to BIPM and SP in Sweden refers back to BIPM. The national laboratories represent the first link in the traceability chain down from BIPM. In this regard NIST and USNO is just that, the US national laboratories. If I where to get my traceability through my national laboratory SP and I also follow the calibration and reporting rules, set forth in ISO 17025 and related, then that measurement is traceable to BIPM and let's say HP in their Santa Clara lab does the same, they relate to NIST (not USNO really) then by mutual agreements these measurements is being recognised and agreed upon. This aspect of recognised is highly related to trade and trust in trade. Since dec. 2009 the PTB in Braunschweig, Germany (with the new CSF2) and the BIPM in Sevre are the only countries running 4 of the most precise primary Cs fountain clocks, if I am informed correctly. Together they should run quite close to the time defined by BIPM I think, and according our law our official time is transmitted by the PTB. Cesium fountains is not a good measure of lab stability. Only a small fraction of all the clocks (250 is a number frequently occuring) being used. Each individual clock is being measured and weighted in. Most of them is HP (now Symmetricom) 5071A clocks with normal or high-performance tubes. The stability of these clocks is being compared and a optimum stable paper-clock called EAL is created. This is then corrected into TAI and with the decissions of IERS is
[time-nuts] TBolt: UTC PPS
Well, mine is driven by a Tek AFG5101 arbitrary function generator. It's currently clocked by a PTB-100 rubidium time base... got the clock out of a trashcan... -- Guess I'm going to have to figure out another way to drive that stepper motor analog clock. I figured that was what it was there for :) _ Hotmail: Trusted email with powerful SPAM protection. http://clk.atdmt.com/GBL/go/210850553/direct/01/ ___ 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] TBolt: UTC PPS
Hi I suspect that generating some sort of 60 Hz sine-ish wave would open up the selection of clocks considerably. Bob -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Mark Sims Sent: Monday, March 29, 2010 12:32 PM To: time-nuts@febo.com Subject: [time-nuts] TBolt: UTC PPS Well, mine is driven by a Tek AFG5101 arbitrary function generator. It's currently clocked by a PTB-100 rubidium time base... got the clock out of a trashcan... -- Guess I'm going to have to figure out another way to drive that stepper motor analog clock. I figured that was what it was there for :) _ Hotmail: Trusted email with powerful SPAM protection. http://clk.atdmt.com/GBL/go/210850553/direct/01/ ___ 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] Making a HP 10811 better
So the cold ambient temperature was simulating the retuning of the oven temperature as described here: http://www.realhamradio.com/z3801a-turning-point.htm and the particular unit graphed in the article happened to have the oven temperature a bit high. Sounds like a sunny deck and a cooler of ice (plus your beverage of choice, of course) would be useful test equipment to check your 10811 to see if retuning the oven is necessary. Warren: On your list of things to do, checking that the crystal inflection point matches the oven temperature should probably come before adjusting the thermal gain. Ed Bob Camp wrote: Hi The best at tha cold end was a function of where the oven was relative to the turn temperature of the crystal. There's actually a bit more to it than that, but that's the simple answer. Bottom line - the temperature curve of *your* 10811 likely will be a bit different than what they show. A cooler, a small bag of ice, a thermometer, and a sunny deck in the summer will give you a temperature ramp. That and some patience will give you a pretty good idea of what your oscillator actually looks like. Other techniques may also work, most provide less of an excuse to drink beer during the experiment. Bob On Mar 28, 2010, at 4:20 PM, Ed Palmer wrote: Have you considered putting the 10811 in a deep freeze? :-) The March 1981 issue of the HP Journal includes an in-depth article on the 10811A/B. It talks about the heater, thermal gain, optimization, etc. Figure 5 on page 21 shows a graph of ambient temperature vs. frequency change. The figure shows that the lowest frequency changes occur around an ambient temperature of -20C! Maybe instead of an outer oven we need to use a Peltier cooler. http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1981-03.pdf Ed WarrenS wrote: Richard wrote approx ... This is how the Thermal gain can be increase to eliminate temperature drift over typical room variations. Thank You Great Idea. Wish I would of thought of that. But Yeah almost as good, Now I know it, can take that away. That is the type of useful and not obvious information that I was hoping for. THANKS SO MUCH That goes to the top of my list to find a procedure to do it. You are a true Nut hero. WS - Original Message - From: Richard H McCorkle mccor...@ptialaska.net To: Discussion of precise time and frequency measurement time-nuts@febo.com Sent: Sunday, March 28, 2010 11:32 AM Subject: Re: [time-nuts] Making a HP 10811 better Warren, One thing Rick Karlquist pointed out is a higher thermal gain can be realized to minimize temperature effects by optimizing the heater transistor balance. The 10811A/B Quartz Crystal Oscillator Operating Service Manual describes the balance circuit in section 8-40. The two heater transistors are not equally spaced with Q7 being closer to the crystal so rather than 50% of the current flowing thru both heater transistors the current in Q8 is set slightly higher at 57% +/- 2% in a production unit to apply equal heat at the crystal. The typical thermal gain on a production unit is on the order of 100 without optimization, but gains of 1000 or more are possible if the heater transistor balance is optimized. I was fortunate enough to obtain a set of single oven 10811-60158 units that had been optimized by an HP engineer for different temperature sensitivities and tagged as such. The unit optimized for maximum thermal gain is virtually immune to typical room temperature variations. A second unit optimized for double oven operation has much lower thermal gain and makes a great unit for testing software temperature compensation routines and outer oven designs. Richard Time to Push the reset button I hope we can all agree what one is not going to find an axes that makes the Frequency modulation that is caused by tapping on the Oscillator, or the table or the airplane, or the boat, to go away, for whatever reasons. Sorry, what I've been falsely referring to as the zero-G axes is actually the Zero Tilt angle axes. I have not had problems with My G changing short term, (BUT it would be interesting to see if I can detect the moon overhead). What I do have is some problems with the Osc tilt angle changing due to its position changing a little. It is the Zero-tilt angle axes that works over a very small change of a few degrees at most. On the other hand, the Zero tilt axes is in fact the Max G sensitivity axes. If you want the Min G sensitivity axes, so that there is no change when turning the Osc over, That is 90 deg from the zero tilt axes, which is also how you make the Osc into its best Tilt angle meter. What happens at any G value that is between +1 and -1, or is greater than +1 or -1, I have not tested for, so I'm not qualified to speculate. I am only stating that there is an axes where the Oscillator frequency is exactly the same when you turn it exactly over.
Re: [time-nuts] TBolt: UTC PPS
Or some sort of 50 Hz for the ones in other sides of the world :-) . Ignacio - Bob Camp wrote: Hi I suspect that generating some sort of 60 Hz sine-ish wave would open up the selection of clocks considerably. Bob -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Mark Sims Sent: Monday, March 29, 2010 12:32 PM To: time-nuts@febo.com Subject: [time-nuts] TBolt: UTC PPS Well, mine is driven by a Tek AFG5101 arbitrary function generator. It's currently clocked by a PTB-100 rubidium time base... got the clock out of a trashcan... -- Guess I'm going to have to figure out another way to drive that stepper motor analog clock. I figured that was what it was there for :) _ Hotmail: Trusted email with powerful SPAM protection. http://clk.atdmt.com/GBL/go/210850553/direct/01/ ___ 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. ___ 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] TBolt: UTC PPS
I suspect that generating some sort of 60 Hz sine-ish wave would open up the selection of clocks considerably. Bob Correct. See: http://leapsecond.com/pages/atomic-nixie/ /tvb ___ 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] TBolt: UTC PPS
Referring to the GPS ICD, the size of A0 in subframe 4 is a signed scaled 32 bit integer so if a receiver chooses to convert it to floating point it needs a double precision format to contain it without loss of precision. A1, on the other hand, is a signed scaled 24 bit integer and so it just fits in a single precision floating point number. So Trimble did the right thing here. Then why do they need 32 bits in the GPS packet format? A single precision floating point would work fine until the 24th bit gets turned on. That's 16,000,000 ns or 16 ms. So back in the days when they were designing the packet formats, somebody must have thought the satellite clocks might drift a long way from UTC. Maybe their control loop wouldn't be good enough to keep the clocks locked but the clocks would be stable enough for navigation. -- These are my opinions, not necessarily my employer's. I hate spam. ___ 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] What time is it anyway?
Below is a nice description of how UTC is generated including a chart showing how individual timing laboratories, BIPM and IERS are connected: http://www.npl.co.uk/science-technology/time-frequency/time/research/the-world-time-system /tvb ___ 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] TBolt: UTC PPS
When there's a leap second, what do you want to do? Ignore it = GPS time put in an extra pulse = UTC time Nice try, but that's not how it's done. They put in an extra second in the last minute of the day. 23:59:59 23:59:60 00:00:00 Humm. Isn't that backwards. Don't you want to remove a pulse? Inserting a leap-second is equivalent to stopping the clock for a tick. I suppose you could setup a microprocessor that listened to the text and mask out a pulse at the right time. That would delay the PPS pulse by the prop time through an AND gate so we could come up with time-nut level schemes to fix that. Or you could just wait for the next leap second and stick your finger in the works at the right time. -- These are my opinions, not necessarily my employer's. I hate spam. ___ 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] Making a HP 10811 better
Bert Want to share the results? More info re the experiment would be of interest as well. DaveB, NZ - Original Message - From: ewkeh...@aol.com To: time-nuts@febo.com Sent: Tuesday, March 30, 2010 6:32 AM Subject: Re: [time-nuts] Making a HP 10811 better I would not feel comfortable tearing into the unit. Controlling its environment and power to it yes. I did that with a 10811 and a 10544 years ago by drilling a 20 foot hole in my Dallas backyard lowering the units in it with support electronics in place and sealing the shaft with individual sand bags each with a numbered string, so I could pull them out one at time when I had to get to them. Did not have the equipment I have today but I did manage to borrow a 5061A for set up. Here in Miami being right on the water it is not an option. When I relocate from Miami to dryer grounds I may repeat it with a Rubidium. Bert ___ 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] TBolt: UTC PPS
Hal, The words insert a pulse is what causes the confusion in conjunction with leap seconds. There is no extra or missing pulse; no double pulse, no early or delayed pulse. The 1PPS output continues to give a pulse at a 1 Hz rate regardless if there is leap second or not. There is never any ambiguity with TAI or UTC with respect to the pulses themselves. The confusion with terminology arises from the fact that the implicit or explicit time-of-day string associated with a UTC time scale does its funny thing around a leap second. If there is a negative leap second the 23:59:59 label is missing. If there is a positive leap second the 235960 is present. You'll also see some GPS receivers or timing systems use the unofficial trick of labeling 23:59:59 or 00:00:00 twice rather than using 23:59:60. No matter how it's done, it's just the label. The pulses still come once a second. Does that help clarify the situation? /tvb ___ 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] TBolt: UTC PPS
Referring to the GPS ICD, the size of A0 in subframe 4 is a signed scaled 32 bit integer so if a receiver chooses to convert it to floating point it needs a double precision format to contain it without loss of precision. A1, on the other hand, is a signed scaled 24 bit integer and so it just fits in a single precision floating point number. So Trimble did the right thing here. Then why do they need 32 bits in the GPS packet format? A single precision floating point would work fine until the 24th bit gets turned on. That's 16,000,000 ns or 16 ms. So back in the days when they were designing the packet formats, somebody must have thought the satellite clocks might drift a long way from UTC. Maybe their control loop wouldn't be good enough to keep the clocks locked but the clocks would be stable enough for navigation. But A0, A1 is not related to SV clocks. They are for translating t_GPS to t_UTC. A_f0, A_f1 and friends are used to get t_GPS from t_SV. It seems they wasted a few bits, since the USNO goal is having t_GPS within 1us. Actual performance is as told in the some ns range. -- Björn ___ 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] TBolt: UTC PPS
or, the map is not the territory? Don Tom Van Baak Hal, The words insert a pulse is what causes the confusion in conjunction with leap seconds. There is no extra or missing pulse; no double pulse, no early or delayed pulse. The 1PPS output continues to give a pulse at a 1 Hz rate regardless if there is leap second or not. There is never any ambiguity with TAI or UTC with respect to the pulses themselves. The confusion with terminology arises from the fact that the implicit or explicit time-of-day string associated with a UTC time scale does its funny thing around a leap second. If there is a negative leap second the 23:59:59 label is missing. If there is a positive leap second the 235960 is present. You'll also see some GPS receivers or timing systems use the unofficial trick of labeling 23:59:59 or 00:00:00 twice rather than using 23:59:60. No matter how it's done, it's just the label. The pulses still come once a second. Does that help clarify the situation? /tvb ___ 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. -- Dr. Don Latham AJ7LL Six Mile Systems LLP 17850 Six Mile Road POB 134 Huson, MT, 59846 VOX 406-626-4304 www.lightningforensics.com www.sixmilesystems.com ___ 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] Making a HP 10811 better
Now if you really want some fun connect your OCXO to a one meter cable and let it swing like a pendulum clock. You should see some very nice sinusoidal FM as it goes back and forth. Devise some sort of mechanical, compressed air, or magnetic impulse system to keep it running like that all day. /tvb It works! http://www.leapsecond.com/pages/10811-g/ Bob, it would seem the frequency shifts in this swinging 10811 oscillator are all acceleration related and not due to internal temperature variations, right? I could imagine there are (longer time constant) thermal effects for a full 2g turn-over, but not with this short ~1 Hz period. In fact you can see a slight, probably temperature gradient related, frequency drift after a full 2g turn (I waited about 5 minutes after each turn): http://www.leapsecond.com/pages/ocxo-2g/TBolt-2g-6axis.gif /tvb ___ 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] Making a HP 10811 better
Hi Tom: Here's a 2g (turnover) plot for the PRS10 Rb source: http://www.prc68.com/I/PRS10.shtml#Accel Have Fun, Brooke Clarke http://www.PRC68.com Tom Van Baak wrote: Now if you really want some fun connect your OCXO to a one meter cable and let it swing like a pendulum clock. You should see some very nice sinusoidal FM as it goes back and forth. Devise some sort of mechanical, compressed air, or magnetic impulse system to keep it running like that all day. /tvb It works! http://www.leapsecond.com/pages/10811-g/ Bob, it would seem the frequency shifts in this swinging 10811 oscillator are all acceleration related and not due to internal temperature variations, right? I could imagine there are (longer time constant) thermal effects for a full 2g turn-over, but not with this short ~1 Hz period. In fact you can see a slight, probably temperature gradient related, frequency drift after a full 2g turn (I waited about 5 minutes after each turn): http://www.leapsecond.com/pages/ocxo-2g/TBolt-2g-6axis.gif /tvb ___ 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] Making a HP 10811 better
Hi Thermal shifts from tip are a very real issue when you flip around OCXO's. The acceleration is an instantaneous change. The thermal will have a time constant out beyond 30 seconds. It's pretty easy to separate the two if you have something that functions like a strip chart recorder. Stable 32 does the job very nicely... There is a second thermal issue related to just plain blowing a bunch of air over the OCXO, compared to still air. That may or may not be significant depending on just how well the 10811 is shielded in your pendulum. Bob -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of Tom Van Baak Sent: Monday, March 29, 2010 4:58 PM To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] Making a HP 10811 better Now if you really want some fun connect your OCXO to a one meter cable and let it swing like a pendulum clock. You should see some very nice sinusoidal FM as it goes back and forth. Devise some sort of mechanical, compressed air, or magnetic impulse system to keep it running like that all day. /tvb It works! http://www.leapsecond.com/pages/10811-g/ Bob, it would seem the frequency shifts in this swinging 10811 oscillator are all acceleration related and not due to internal temperature variations, right? I could imagine there are (longer time constant) thermal effects for a full 2g turn-over, but not with this short ~1 Hz period. In fact you can see a slight, probably temperature gradient related, frequency drift after a full 2g turn (I waited about 5 minutes after each turn): http://www.leapsecond.com/pages/ocxo-2g/TBolt-2g-6axis.gif /tvb ___ 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] Making a HP 10811 better
Bert wrote: would not feel comfortable tearing into the unit. I tend to agree, First rule is do no harm, which is why I start with reversible things that can be done using just external stuff. The second rule does NOT apply though, Don't fix it, if it aint broken Of course if it is already broken, or you have lots of them, or they are not yours, or you're getting paid for it, then go for it. BUT I find it amazing that you are more concerned about opening the unit than you are about sticking it down a deep hole from which it may never return if something goes wrong. I think I'd rather have mine on the bench even if it was in parts, than to lose it to the center of the earth. BUT I must Add, good idea if you want a nice constant passive green environment with a very long time constants (maybe 1/2 year TC?) ws ** From: ewkeh...@aol.com I would not feel comfortable tearing into the unit. Controlling its environment and power to it yes. I did that with a 10811 and a 10544 years ago by drilling a 20 foot hole in my Dallas backyard lowering the units in it with support electronics in place and sealing the shaft with individual sand bags each with a numbered string, so I could pull them out one at time when I had to get to them. Did not have the equipment I have today but I did manage to borrow a 5061A for set up. Here in Miami being right on the water it is not an option. When I relocate from Miami to dryer grounds I may repeat it with a Rubidium. Bert In a message dated 3/28/2010 11:52:14 A.M. Eastern Daylight Time, warrensjmail-...@yahoo.com writes: Bert asked why would I want to mess with them at all if they are already under 1E-12? Don't know about you, The reason I'd mess with them is cause I'm nuts and can make them even better. And Allan variance is not everything. Have you plotted their Freq change over time with better than 1E-12 resolution? If you are a True NUT, I'm betting you would see things that would make you sick. Things such as line freq spurs, Freq jumps, the effects of: changing temperature, changing PS, changing time, changing Load, partial injection locking to other osc friends around them, and maybe even to the changing moon. Don't forget the Basic Nut Cake pledge. If it can be made better then it is not good enough Or for the more practical answer; Cause I can make mine good enough with some help from two of its friends, named GPS and Rb, that I do not need a Cs primary standard. ws * Good morning I have a more fundamental question. Having three 10811-5071 all with Allan Variance from 1 to 100 Hz well below 1 E-12 one as low as 6 E-13 why would I want to mess with them at all? Bert Kehren ___ 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] TBolt: UTC PPS
t...@leapsecond.com said: The words insert a pulse is what causes the confusion in conjunction with leap seconds. There is no extra or missing pulse; no double pulse, no early or delayed pulse. The 1PPS output continues to give a pulse at a 1 Hz rate regardless if there is leap second or not. There is never any ambiguity with TAI or UTC with respect to the pulses themselves. Right. I was commenting about the sub-thread that was expecting an extra pulse so a mechanical clock would stay in sync over a leap second event. Since what's needed is removing a pulse (to allow for the leap second), rather than inserting a pulse, I still like the idea of poking a finger on the right place in the clock to gum up the works for one tick. -- These are my opinions, not necessarily my employer's. I hate spam. ___ 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] Making a HP 10811 better
At 05:59 PM 3/29/2010, WarrenS wrote... Bert wrote: would not feel comfortable tearing into the unit. I tend to agree, First rule is do no harm, I've got this rule that says If I don't know what the insides look like when it's working, I won't know what to look for when it breaks. :-) ___ 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] TBolt: UTC PPS
It is simple to divide the PPS to the square wave, but what is needed is a simple/surefire method of adding/removing the leap seconds. Preferably one that works automatically. Any ideas? Software. If you want to remove leap seconds, you need a way to insert an extra pulse. Fortunately, they haven't happen yet and are not likely to happen. To insert a leap second, you need to remove a pulse. That's just an AND gate. -- These are my opinions, not necessarily my employer's. I hate spam. ___ 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] TBolt: UTC PPS
Use a JK flip flop to divide by 2. During insertion of the leapsecond take J+K to zero to disable toggling of the FF. To delete leap second you could try inserting an extra clock pulse halfway between a apair of seconds ticks. If the clock cant cope with the above scheme you may need to use a DDS or equivalent to create either 59, 60 or 61 pulses for a minute depending if one needs to delete a leap second, increment the clock at the normal rate ,add a leapsecond over 1 minute. Bruce Neville Michie wrote: The most available analogue dial for PPS clocks is the common quartz clock movement. You can leave the unpowered quartz unit in circuit and just connect to the coil. The drive is a 5V square wave signal of 0.5 Hz coupled through an electrolytic capacitor in the order of 10 - 100mfd. A series resistor of a few hundred ohms may help. Each make of clock may need different values to work best, the drive is the alternating positive and negative impulse. The original unit may have had a pulse about 20ms long, but the square wave drive gives a longer tail. Over-driving is just as bad as under-driving, to work well you must optimise the capacitor and resistor. It is simple to divide the PPS to the square wave, but what is needed is a simple/surefire method of adding/removing the leap seconds. Preferably one that works automatically. Any ideas? Neville Michie On 30/03/2010, at 9:02 AM, Hal Murray wrote: t...@leapsecond.com said: The words insert a pulse is what causes the confusion in conjunction with leap seconds. There is no extra or missing pulse; no double pulse, no early or delayed pulse. The 1PPS output continues to give a pulse at a 1 Hz rate regardless if there is leap second or not. There is never any ambiguity with TAI or UTC with respect to the pulses themselves. Right. I was commenting about the sub-thread that was expecting an extra pulse so a mechanical clock would stay in sync over a leap second event. Since what's needed is removing a pulse (to allow for the leap second), rather than inserting a pulse, I still like the idea of poking a finger on the right place in the clock to gum up the works for one tick. -- These are my opinions, not necessarily my employer's. I hate spam. ___ 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. ___ 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] Making a HP 10811 better
I agree with leaving well alone, But I had a HP10811 which had no EFC control. I agonised about why this would be and bought a few varicaps, - I could not find suitable current sources and I had resistors. When I opened it I found the junction of the varicap diode, the oscillator capacitor and a resistor was dry. No solder at all! When it was soldered all worked well. Another unit was inspected and it had a lousy joint in the same location. The serial numbers of the two units were years apart. Now I am more confident with them I am thinking of trying to reset the oven temperature to suit my ambient temperature, the aged crystal and the drift of the thermistor and resistors. I may even lag the HP10811 with some insulation to reduce the internal temperature gradients and reduce the oven power. The method would be to use an external resistor which would be set to several likely values increasing the temperature past maximum frequency, then the same values decreasing the temperature. An hour or so at each temperature. Then a least squares regression to a parabola which should show the best value. cheers, Neville Michie On 30/03/2010, at 9:49 AM, Mike S wrote: At 05:59 PM 3/29/2010, WarrenS wrote... Bert wrote: would not feel comfortable tearing into the unit. I tend to agree, First rule is do no harm, I've got this rule that says If I don't know what the insides look like when it's working, I won't know what to look for when it breaks. :-) ___ 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] Making a HP 10811 better
I did retrieve it many times that is why I had all those sand bags with # strings and that time the cost of an oscillator was much higher and I tried putting the bags in and out several times before the oscillators went down there including leak test for moisture penetration. Bert In a message dated 3/29/2010 6:00:19 P.M. Eastern Daylight Time, warrensjmail-...@yahoo.com writes: Bert wrote: would not feel comfortable tearing into the unit. I tend to agree, First rule is do no harm, which is why I start with reversible things that can be done using just external stuff. The second rule does NOT apply though, Don't fix it, if it aint broken Of course if it is already broken, or you have lots of them, or they are not yours, or you're getting paid for it, then go for it. BUT I find it amazing that you are more concerned about opening the unit than you are about sticking it down a deep hole from which it may never return if something goes wrong. I think I'd rather have mine on the bench even if it was in parts, than to lose it to the center of the earth. BUT I must Add, good idea if you want a nice constant passive green environment with a very long time constants (maybe 1/2 year TC?) ws ** From: ewkeh...@aol.com I would not feel comfortable tearing into the unit. Controlling its environment and power to it yes. I did that with a 10811 and a 10544 years ago by drilling a 20 foot hole in my Dallas backyard lowering the units in it with support electronics in place and sealing the shaft with individual sand bags each with a numbered string, so I could pull them out one at time when I had to get to them. Did not have the equipment I have today but I did manage to borrow a 5061A for set up. Here in Miami being right on the water it is not an option. When I relocate from Miami to dryer grounds I may repeat it with a Rubidium. Bert In a message dated 3/28/2010 11:52:14 A.M. Eastern Daylight Time, warrensjmail-...@yahoo.com writes: Bert asked why would I want to mess with them at all if they are already under 1E-12? Don't know about you, The reason I'd mess with them is cause I'm nuts and can make them even better. And Allan variance is not everything. Have you plotted their Freq change over time with better than 1E-12 resolution? If you are a True NUT, I'm betting you would see things that would make you sick. Things such as line freq spurs, Freq jumps, the effects of: changing temperature, changing PS, changing time, changing Load, partial injection locking to other osc friends around them, and maybe even to the changing moon. Don't forget the Basic Nut Cake pledge. If it can be made better then it is not good enough Or for the more practical answer; Cause I can make mine good enough with some help from two of its friends, named GPS and Rb, that I do not need a Cs primary standard. ws * Good morning I have a more fundamental question. Having three 10811-5071 all with Allan Variance from 1 to 100 Hz well below 1 E-12 one as low as 6 E-13 why would I want to mess with them at all? Bert Kehren ___ 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.
[time-nuts] Making a HP 10811 better
Aww, what a bunch of unadventurous louts ;-) Where's the fun in that? Boldly go... but let's not talk about that core memory stack assembled under 5 tons of pressure... I wondered why those bolts were so tight... Bert wrote:would not feel comfortable tearing into the unit.I tend to agree, First rule is do no harm,I agree with leaving well alone, _ Hotmail: Powerful Free email with security by Microsoft. http://clk.atdmt.com/GBL/go/196390710/direct/01/ ___ 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] Making a HP 10811 better
or was that the capacitor storage in the IBM704? Don Mark Sims Aww, what a bunch of unadventurous louts ;-) Where's the fun in that? Boldly go... but let's not talk about that core memory stack assembled under 5 tons of pressure... I wondered why those bolts were so tight... Bert wrote:would not feel comfortable tearing into the unit.I tend to agree, First rule is do no harm,I agree with leaving well alone, _ Hotmail: Powerful Free email with security by Microsoft. http://clk.atdmt.com/GBL/go/196390710/direct/01/ ___ 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. -- Dr. Don Latham AJ7LL Six Mile Systems LLP 17850 Six Mile Road POB 134 Huson, MT, 59846 VOX 406-626-4304 www.lightningforensics.com www.sixmilesystems.com ___ 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] Making a HP 10811 better
Hi The 10811 should respond quickly enough that a few minutes (5 ~10) between steps should be plenty. If you stretch it out to far, things like aging, retrace, and ambient change can get into the data. Always return to your original resistor value to confirm that drift hasn't gotten in the way. If you decide to add more insulation, you may need to change the DC gain of the controller to compensate. Also remember that DC gain, thermal gain, and set point all interact with each other at some level. A full optimization would attack all three. If you are going to do this to an OCXO, the 10811 is a pretty good candidate. It's made with leaded parts and put together with screws. No nasty SMD's or welds to get past. Without a temperature chamber of some sort, making sure you have improved things can be difficult. Bob On Mar 29, 2010, at 7:19 PM, Neville Michie wrote: I agree with leaving well alone, But I had a HP10811 which had no EFC control. I agonised about why this would be and bought a few varicaps, - I could not find suitable current sources and I had resistors. When I opened it I found the junction of the varicap diode, the oscillator capacitor and a resistor was dry. No solder at all! When it was soldered all worked well. Another unit was inspected and it had a lousy joint in the same location. The serial numbers of the two units were years apart. Now I am more confident with them I am thinking of trying to reset the oven temperature to suit my ambient temperature, the aged crystal and the drift of the thermistor and resistors. I may even lag the HP10811 with some insulation to reduce the internal temperature gradients and reduce the oven power. The method would be to use an external resistor which would be set to several likely values increasing the temperature past maximum frequency, then the same values decreasing the temperature. An hour or so at each temperature. Then a least squares regression to a parabola which should show the best value. cheers, Neville Michie On 30/03/2010, at 9:49 AM, Mike S wrote: At 05:59 PM 3/29/2010, WarrenS wrote... Bert wrote: would not feel comfortable tearing into the unit. I tend to agree, First rule is do no harm, I've got this rule that says If I don't know what the insides look like when it's working, I won't know what to look for when it breaks. :-) ___ 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. ___ 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] What time is it anyway?
Steve Rooke wrote: So there are 250 clocks, presumably, spread around the World and owned by the member countries of the BIPM. There is a fair spread geographically, yes. See map: http://www.bipm.org/en/scientific/tai/tai.html Their time is somehow compared centrally and an absolute time is determined from them. They are compared by various methods in a network style of operation. The sites are compared against each other and the clocks of each site is compared against the clock driving the network comparison. Hence, through these steps phase differences can be cranked out. http://www.bipm.org/en/scientific/tai/clock_comparisons.html Equipment calibrations: http://www.bipm.org/jsp/en/TimeCalibrations.jsp Each clock will then have a delta to apply to it's own time to provide the BIPM absolute time which then allows each member country to use its own clock for timing. Yes, out comes the delta of each individual clock as well as the individual labs. These are reported upon regularly to be found in BIPM circular T: http://www.bipm.org/jsp/en/kcdb_data.jsp ftp://ftp2.bipm.org/pub/tai/publication/cirt.266 Useful resource: http://www.bipm.org/jsp/en/TimeFtp.jsp?TypePub=publication This sounds like a logistic and bureaucratic nightmare, I will have to look further as to how this is achieved. Any pointers gratefully accepted. I hope the above pointers get you started. They have spent some time cooking this up, yes. Toss into that, we have not really spoken too much about how they process things. I can't tell you too much about it, except that I know that their tool is called ALGOS. If someone could give me the appropriate articles I would be greatful. Cheers, Magnus ___ 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] HP10514B Mixer Terminations
Hi You can get a *much* more squared output from the mixer than the photos you show on the scope. The waveform looks a lit like a triangle wave with the tips chopped off. Normally the fastest edge happens into a capacitive load at RF that's below about 0.5 J ohms for a 50 ohm mixer. Bob On Mar 29, 2010, at 10:06 PM, Brian Kirby wrote: I have been working on a Dual-Mixer Time Difference system. In the first design type/experiment, I was using HP10514B mixers and a LT1037 preamp and a OP27 zero crossing amplifier/limiter - all a very basic setup. I obtained some fair measurements; Using 10 MHz sources, a 9. MHz offset for a 100 hertz beat, the floor of the system looked like this: 0.01 second = 1x10-10 0.1 second = 1x10-11 1 second = 1x10-12 10 second = 1x10-13 100 second = 1x10-14 1000 second = 1x10-15 10,000 second = 1x10-16 this was three days of data Running it again, with a 10 hertz beat; it looked like this; 0.1 second = 4x10-12 1 second = 4x10-13 10 second = 4x10-14 100 second = 4x10-15 1000 second = 4x10-16 I also had a lot of good suggestions From Ulrich Bangert, Bob Camp and Bruce Griffins, who I will call my mentors and thank for all the help. I went back and did some basic experiments this evening. Looking at mixer terminations. I have attached two photos - low res. The first photo named mixer_10db, is the mixer driven with +10 dbm on both ports. The o'scope is looking thru a basic RC filter of 1 kilo-ohm resistor in series with the mixer output, and on the output of the resistor is a 0.1 uF capacitor to ground. This is a mixer that is intentionally over driven to use as a phase detector. The mixer is rated +13 dbm maximum, and about everybody I have talked with (NIST and BIPM) about these mixers ran them at +10 dbm on both LO and RF ports. As these mixers are hard to find, and they are not made anymore, I would not over-drive them any further. These mixers also have some of the lowest phase noise measurements on record. The second photo named mixer_330 pF, is the same setup, except I have put a 330 pF capacitor across the mixer output. By capacitive terminating the mixer, it squares up the output of the mixer - which makes it easier to be converted to a high slew rate signal. What I found, is you want to run the minimum capacitance value for the highest beat frequency you plan to run. That way the signal stays squared up from the highest to the lowest beat frequency. I got this value by playing around by looking at the mixer filtered (RC) output at 1 hz, 10 hz, and 100 hz. When I was using 0.1 and 1 uF terminations, The 1 and 10 hertz beat was OK, but the 100 hertz beat was still a sine wave. That may be why the results above shows a difference. For a test, at 330 pF, I did try it at 1 KHz, it was back to a sine wave. So 330 pF looks good for trying to get a squared wave out of the mixer for 1, 10 and 100 hertz beats.I tried 36 pF for 1 KHz, it did not present enough capacitance to give the squared wave at 1, 10 and 100 hertz beat. We have been running email outside of Time-Nuts group as I am not sure if any of you wanted to see the project I am working on. I did not want to clutter up the forum..but if there is an interest, I can bring it back. My next plans are to start over building a new system using a much lower noise op amp, the LT1028. If the mixer terminations are OK with my mentors, I will use a LT1028 preamp set for about x15 gain and it will dump into the first set of limiter diodes. And I believe that will call for 1.6 KHz low pass filtering on the first limiter diodes. Comments ? Brian - KD4FM mixer_330pf.jpgmixer_10dbm.jpg___ 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] HP10514B Mixer Terminations
Brian You should get even better results if you replace the 330pF cap with a 1/4 wave (at the 20MHz sum frequency) coax cable open circuited stub. Thats around 2.5m of RG 58 coax for example. Connecting a series tuned circuit (at the sum frequency) across the mixer IF output should also work well. Bruce Brian Kirby wrote: I have been working on a Dual-Mixer Time Difference system. In the first design type/experiment, I was using HP10514B mixers and a LT1037 preamp and a OP27 zero crossing amplifier/limiter - all a very basic setup. I obtained some fair measurements; Using 10 MHz sources, a 9. MHz offset for a 100 hertz beat, the floor of the system looked like this: 0.01 second = 1x10-10 0.1 second = 1x10-11 1 second = 1x10-12 10 second = 1x10-13 100 second = 1x10-14 1000 second = 1x10-15 10,000 second = 1x10-16 this was three days of data Running it again, with a 10 hertz beat; it looked like this; 0.1 second = 4x10-12 1 second = 4x10-13 10 second = 4x10-14 100 second = 4x10-15 1000 second = 4x10-16 I also had a lot of good suggestions From Ulrich Bangert, Bob Camp and Bruce Griffins, who I will call my mentors and thank for all the help. I went back and did some basic experiments this evening. Looking at mixer terminations. I have attached two photos - low res. The first photo named mixer_10db, is the mixer driven with +10 dbm on both ports. The o'scope is looking thru a basic RC filter of 1 kilo-ohm resistor in series with the mixer output, and on the output of the resistor is a 0.1 uF capacitor to ground. This is a mixer that is intentionally over driven to use as a phase detector. The mixer is rated +13 dbm maximum, and about everybody I have talked with (NIST and BIPM) about these mixers ran them at +10 dbm on both LO and RF ports. As these mixers are hard to find, and they are not made anymore, I would not over-drive them any further. These mixers also have some of the lowest phase noise measurements on record. The second photo named mixer_330 pF, is the same setup, except I have put a 330 pF capacitor across the mixer output. By capacitive terminating the mixer, it squares up the output of the mixer - which makes it easier to be converted to a high slew rate signal. What I found, is you want to run the minimum capacitance value for the highest beat frequency you plan to run. That way the signal stays squared up from the highest to the lowest beat frequency. I got this value by playing around by looking at the mixer filtered (RC) output at 1 hz, 10 hz, and 100 hz. When I was using 0.1 and 1 uF terminations, The 1 and 10 hertz beat was OK, but the 100 hertz beat was still a sine wave. That may be why the results above shows a difference. For a test, at 330 pF, I did try it at 1 KHz, it was back to a sine wave. So 330 pF looks good for trying to get a squared wave out of the mixer for 1, 10 and 100 hertz beats.I tried 36 pF for 1 KHz, it did not present enough capacitance to give the squared wave at 1, 10 and 100 hertz beat. We have been running email outside of Time-Nuts group as I am not sure if any of you wanted to see the project I am working on. I did not want to clutter up the forum..but if there is an interest, I can bring it back. My next plans are to start over building a new system using a much lower noise op amp, the LT1028. If the mixer terminations are OK with my mentors, I will use a LT1028 preamp set for about x15 gain and it will dump into the first set of limiter diodes. And I believe that will call for 1.6 KHz low pass filtering on the first limiter diodes. Comments ? Brian - KD4FM ___ 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] HP10514B Mixer Terminations
Your poor results at 1kHz have more to do with the 1K +0.1uF low pass filter which has a cutoff frequency of about 1.6KHz. This will attenuate the beat frequency harmonics required for high slew rate at the beat frequency zero crossings. A filter cutoff of 16kHz (1K + 10nF) should improve the slew rate at 1KHz. Bruce Brian Kirby wrote: I have been working on a Dual-Mixer Time Difference system. In the first design type/experiment, I was using HP10514B mixers and a LT1037 preamp and a OP27 zero crossing amplifier/limiter - all a very basic setup. I obtained some fair measurements; Using 10 MHz sources, a 9. MHz offset for a 100 hertz beat, the floor of the system looked like this: 0.01 second = 1x10-10 0.1 second = 1x10-11 1 second = 1x10-12 10 second = 1x10-13 100 second = 1x10-14 1000 second = 1x10-15 10,000 second = 1x10-16 this was three days of data Running it again, with a 10 hertz beat; it looked like this; 0.1 second = 4x10-12 1 second = 4x10-13 10 second = 4x10-14 100 second = 4x10-15 1000 second = 4x10-16 I also had a lot of good suggestions From Ulrich Bangert, Bob Camp and Bruce Griffins, who I will call my mentors and thank for all the help. I went back and did some basic experiments this evening. Looking at mixer terminations. I have attached two photos - low res. The first photo named mixer_10db, is the mixer driven with +10 dbm on both ports. The o'scope is looking thru a basic RC filter of 1 kilo-ohm resistor in series with the mixer output, and on the output of the resistor is a 0.1 uF capacitor to ground. This is a mixer that is intentionally over driven to use as a phase detector. The mixer is rated +13 dbm maximum, and about everybody I have talked with (NIST and BIPM) about these mixers ran them at +10 dbm on both LO and RF ports. As these mixers are hard to find, and they are not made anymore, I would not over-drive them any further. These mixers also have some of the lowest phase noise measurements on record. The second photo named mixer_330 pF, is the same setup, except I have put a 330 pF capacitor across the mixer output. By capacitive terminating the mixer, it squares up the output of the mixer - which makes it easier to be converted to a high slew rate signal. What I found, is you want to run the minimum capacitance value for the highest beat frequency you plan to run. That way the signal stays squared up from the highest to the lowest beat frequency. I got this value by playing around by looking at the mixer filtered (RC) output at 1 hz, 10 hz, and 100 hz. When I was using 0.1 and 1 uF terminations, The 1 and 10 hertz beat was OK, but the 100 hertz beat was still a sine wave. That may be why the results above shows a difference. For a test, at 330 pF, I did try it at 1 KHz, it was back to a sine wave. So 330 pF looks good for trying to get a squared wave out of the mixer for 1, 10 and 100 hertz beats.I tried 36 pF for 1 KHz, it did not present enough capacitance to give the squared wave at 1, 10 and 100 hertz beat. We have been running email outside of Time-Nuts group as I am not sure if any of you wanted to see the project I am working on. I did not want to clutter up the forum..but if there is an interest, I can bring it back. My next plans are to start over building a new system using a much lower noise op amp, the LT1028. If the mixer terminations are OK with my mentors, I will use a LT1028 preamp set for about x15 gain and it will dump into the first set of limiter diodes. And I believe that will call for 1.6 KHz low pass filtering on the first limiter diodes. Comments ? Brian - KD4FM ___ 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] TBolt: UTC PPS
t...@leapsecond.com said: a0 = -0.931 ns and a1 = -5.33e-15 s/s (which is -0.460 ns/day). It changed at 15:53:40 Pacific time. Was: -9.31323e-10 -5.32907e-15 Now: -2.79397e-09 -5.32907e-15 -- These are my opinions, not necessarily my employer's. I hate spam. ___ 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] HP10514B Mixer Terminations
Your correct - your also keeping me up past my bedtime !- I got to be 90 miles from home tomorrow morning by 7:30 AM It looks like I got the squarest wave at 150 pF. Lesser capacitance, give a peaked sinewave, like maybe a second harmonic. Past 200 pF, it starts rounding. 150pf= XC of 53 ohms This DMTD system, will only go to 100 hertz maximum beat for my design... The scope is set to 100 mV per div, 1ms per division, I intentionally mis-triggered it to show rise and fall times. 100 hz beat signal. ANd Bruce told me how to calculate slew rate, but it has to be beated in my thick tough head. And Bruce, I'll try some coax tomorrow, if I get back home early enough - my work project may keep me late. And I plan to only use this system to 100 hertz beat, I was just playing around at 1K.I like learning. Good night.BrianKD4FM Bob Camp wrote: Hi You can get a *much* more squared output from the mixer than the photos you show on the scope. The waveform looks a lit like a triangle wave with the tips chopped off. Normally the fastest edge happens into a capacitive load at RF that's below about 0.5 J ohms for a 50 ohm mixer. Bob On Mar 29, 2010, at 10:06 PM, Brian Kirby wrote: I have been working on a Dual-Mixer Time Difference system. In the first design type/experiment, I was using HP10514B mixers and a LT1037 preamp and a OP27 zero crossing amplifier/limiter - all a very basic setup. I obtained some fair measurements; Using 10 MHz sources, a 9. MHz offset for a 100 hertz beat, the floor of the system looked like this: 0.01 second = 1x10-10 0.1 second = 1x10-11 1 second = 1x10-12 10 second = 1x10-13 100 second = 1x10-14 1000 second = 1x10-15 10,000 second = 1x10-16 this was three days of data Running it again, with a 10 hertz beat; it looked like this; 0.1 second = 4x10-12 1 second = 4x10-13 10 second = 4x10-14 100 second = 4x10-15 1000 second = 4x10-16 I also had a lot of good suggestions From Ulrich Bangert, Bob Camp and Bruce Griffins, who I will call my mentors and thank for all the help. I went back and did some basic experiments this evening. Looking at mixer terminations. I have attached two photos - low res. The first photo named mixer_10db, is the mixer driven with +10 dbm on both ports. The o'scope is looking thru a basic RC filter of 1 kilo-ohm resistor in series with the mixer output, and on the output of the resistor is a 0.1 uF capacitor to ground. This is a mixer that is intentionally over driven to use as a phase detector. The mixer is rated +13 dbm maximum, and about everybody I have talked with (NIST and BIPM) about these mixers ran them at +10 dbm on both LO and RF ports. As these mixers are hard to find, and they are not made anymore, I would not over-drive them any further. These mixers also have some of the lowest phase noise measurements on record. The second photo named mixer_330 pF, is the same setup, except I have put a 330 pF capacitor across the mixer output. By capacitive terminating the mixer, it squares up the output of the mixer - which makes it easier to be converted to a high slew rate signal. What I found, is you want to run the minimum capacitance value for the highest beat frequency you plan to run. That way the signal stays squared up from the highest to the lowest beat frequency. I got this value by playing around by looking at the mixer filtered (RC) output at 1 hz, 10 hz, and 100 hz. When I was using 0.1 and 1 uF terminations, The 1 and 10 hertz beat was OK, but the 100 hertz beat was still a sine wave. That may be why the results above shows a difference. For a test, at 330 pF, I did try it at 1 KHz, it was back to a sine wave. So 330 pF looks good for trying to get a squared wave out of the mixer for 1, 10 and 100 hertz beats.I tried 36 pF for 1 KHz, it did not present enough capacitance to give the squared wave at 1, 10 and 100 hertz beat. We have been running email outside of Time-Nuts group as I am not sure if any of you wanted to see the project I am working on. I did not want to clutter up the forum..but if there is an interest, I can bring it back. My next plans are to start over building a new system using a much lower noise op amp, the LT1028. If the mixer terminations are OK with my mentors, I will use a LT1028 preamp set for about x15 gain and it will dump into the first set of limiter diodes. And I believe that will call for 1.6 KHz low pass filtering on the first limiter diodes. Comments ? Brian - KD4FM mixer_330pf.jpgmixer_10dbm.jpg___ 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
[time-nuts] Anritsu MH-4100A xtal osc
I picked one of these up for $5 this weekend but can find almost nothing about it. Takes 117V A/C and spits out a BNC 10 MHz. The only thing on the front is a STBY / ON switch and two LEDs indicating said switches position. No surprise, nada from Anritsu. Anyone seen/had/have one ? Comments ? Did I just get a nice little box for $5 and nothing more ? -pete ___ 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.
