@@scowell If I remember correctly it also went from "oh look, it's a NASA cartridge" to "we restored an entire AGC to drive our homemade Apollo mission simulator" :) I don't have even words to describe how much I appreciate this channel other than #yolo
I once worked at a naval air station. They had "labs " for civilian engineers. They did something to shift the frequency from 60hz to up and down by about 15% . It killed most of the equipment that I was responsible for and caused havoc. It took days before I figured out what was going on. No one checks the frequency of the mains going into various equipment rooms, no one. I eventually did. Wow, what a difference a few Hertz makes. Air to ground, ground to ground, navigation beakins and weather station. All took a dump over less than 15/30% total frequency deviation. You have neat toys, I can't do it anymore but, I can appreciate. Thx.
Отличная работа, никогда не перестаете удивлять! Хорошо что хоть кто-то помнит про миссию Союз-Апполон, такое ощущение что сейчас все игнорируют это огромное достижение космонавтики.
Thanks! Translation by courtesy of Google translate: "Great job, never cease to amaze! It is good that at least someone remembers the Soyuz-Apollo mission, it seems that now everyone is ignoring this huge achievement of astronautics."
My mom did translation of a lot of the in-flight docs from Russian to English. Always found it fascinating that she could she could translate the docs but pronouncing Russian was always really tough for her. My dad worked for Boeing out at Johnson Space Center during that period. He told us that the US had to build the docking module since the Soviet Soyuz spacecraft couldn't handle the necessary precision docking required. Always wondered the details on that. But, I was a young kid at the time.
@@68Polara500 The Apollo service module was grossly over powered - the original specification had it launching from the lunar surface. That gave a lot of reserve power that was never used, to change orbit. Soyuz needed more precise launch parameters to dock in orbit due to limited delta v.
@@68Polara500 The story with the docking module is actually pretty strange and amusing. It had to be designed as a female to female adapter, because traditionally a docking system would be male > female, and neither country wanted to be female. (I suspect both were designed to be "male" from day 1 regardless). Supposedly the soviet engineers on the project started calling it "Hermaphrodite", and this term had to be explained to a high ranking military official that happened to visit. I would imagine soyz could handle the precision given that they achieved totally automatic docking quite a bit earlier, back in 67.
Полёт Союз-Аполлон был в 1975 году, а в этих часах на радиодеталях 1983 год. В 1973 году в СССР уже производились электронные наручные часы. Спрашивается, зачем отправлять в космос (где каждый грамм на вес золота) часы с массой утюга и энергопотреблением утюга?
When I was a kid I very much wanted an HP-85, I had an HP-47c and spent so much time with it you might call it indecent. I wrote such crazy program’s for it, I was always maxing out the memory. Mostly celestial navigation and complex matrix arithmetic. Those were the days. Charlie
In 1983 as a newly minted engineer, I designed a piece of test equipment at GE Aerospace that had an embedded 8085uP. It was managed by an HP 85, via the GPIB interface. I had a little difficulty handling the noise on the GPIB handshaking lines - but some seasoned engineers helped me solve it. Great memories, seeing that HP 85 in use! Great video and analysis, as usual.
Assuming you are serious, there is indeed a correction for gravity on Quartz oscillators. But it's minuscule. For my HP oven Quartz oscillator for example, it's on the order of 25 mHz at 10 Mhz. You can measure it (and I did!) by simply flipping the instrument over. The normal Quartz oscillators are not stable enough to allow a meaningful measurement. But assuming you actually could, I imagine on the order of 2.5 mHz at 1 MHz. Not going to get your 17 seconds back I am afraid... I willing to test in orbit if someone pays for my ticket :-)
I got to see the atomic clock at gold Stone in California. Back in 1983-84. My mother was dating a guy that worked there, who also would send and receive messages to the voyagers. I was told..."At the time" the atomic clock there was the second most accurate clock in the world. 1 millionth of a second off a year or 1 second every million years! I think that's how I remember it? I was only 13 or 14 yrs old. I am still friends with this guy today. Watching you guys work on this, brings back old memories!☺️👍
This is honestly some of the best content I have seen all year on TH-cam. Your clock got me subscribed and I have enjoyed the entire journey so far. Looking forward to the next video.
Absolutely excellent, it was a pleasure to watch the whole series of Soyuz clock videos. It's great you made the clock fully operational as opposed to it being an interesting item sitting in a glass display box.
I know, I know, I was born in Belgium. A touch of "francophonie" I should have said. And of course that means the Belgians made it first to the Moon! RIP Hergé.
@@CuriousMarc belgian here too, I second that. Though I meet a lot of people not knowing where Belgium is or that it is a country. Out of curiosity, where were you born in Belgium?
With an egg strapped to the top of the clock on a "timer" and being cooked by activating a Spark gap microwave generator with the "German" sign in the background!
Your HP8116 amplitude modulation is not phase locked to the gating signal. The 800hz being gated by the cesium (via the 3325) starts at a random point on the sine wave, this has got to be a big source of the noise into the soyuz clock, because its integrator is going to have very different time delay based on the starting point of the 250ms wide sine pulse waveform. If you used an arbitrary waveform source, like an HP33250 (if I am remembering right) you could have the cesium 2hz waveform act as a start and stop command for a sine burst that always starts at the same point on the sine wave. Or could you use a gated 800 square wave input into the soyuz clock? and generate it from a higher frequency clock that is locked to the cesium as well, perhaps using two 3325s, one at 2hz and one at 800hz both driven by the 5MHZ cesium reference?. I do love the HP85, I used one of those back in the day until the belts and rollers disintegrated in the tape drive. Maybe I'll dig into some old equipment now that I am stuck at home. Thanks for your enjoyable videos.
Yes, you can clearly see that's one of the biggest contributor to the clock recovery signal jitter. Which is all fine of course. Astronauts are not sensitive to their seconds signal phase noise...
@@CuriousMarc Well, if you had a pair of 3325A's you could get around the problem. Clock both off the 5 MHz and use one to gate the other. Of course a TTL or HCMOS divider chain off the 5MHz to generate both the 800 Hz and 1 Hz would pretty much eliminate the phase noise completely. I can't recall if HP made some programmable counters in that era that could be used as programmable frequency dividers, but I think they or Tektronix did.
They did one just for that very application, to divide the Cesium and generate very low phase noise clock timing. That’s the unit you see on top of the flying clock rack. But we don’t have it...
I really glad that clock was end up Your desk instead some random collector - as You was able to create so much interesting content based on that! have watched all series an totally enjoyed!!!
Marc, it would be cool if you could take a micrometer to that clock so that someone can make an accurate computer model that can be 3D printed or CNCd. Or just to further document this piece of engineering history. Love the videos!
I think you should verify the HP calculations by punching the raw data onto paper tape, convert that to cards, write a IBM 1401 Fortran program to re-do the calculations and then run that on the IBM 1401 computer.
I am not sure about that the external clock source is designed to be feed from some reference on Soyuz board. I am thinking about possibility that theese 2Hz pulses came from some radio receiver, which could lock the clock to some radio clock station on the earth. Using A1 modulated signal to interconnect it with some clock reference inside soyuz simply makes no sense and looks too complicated to me. And on top of that. If I am right and this was connected to some radio equiptment instead of some local reference, it opens the posibility, that there was more commands than "increase ½second", e.g. "first second in minute" or something similar. I do not know how much this was reverse-engineered, so I do not know if this is possible, or not.
Fun measurement setup, but while it would have made made no difference on the long term graphs, you could probably get substantially better single-period stability if something other than the 8116A was used to generate the tone burst. Since it's using an integrator to sense the start of the burst, the phase the burst starts at is critical to eliminate leading edge timing jitter, and the 8116A can't be locked to an external reference! You could see the phase of the beginning of the burst drift on your scope captures, and if you used an instrument that either started the burst on the trigger (rather than just gating a free running oscillator) or which was reference locked to the same atomic source, you could make sure that every burst started with the same phase, so that the integrator's rise would be much more precisely controlled (I am assuming that the soyuz clock isn't measuring the time between the beginning and end of the integrator and is instead timing based off of the rising or falling edge of the integrated burst signal). Would be interesting to see an 11 or 12 digit/s counter thrown at it too, though not period appropriate, to really flesh out the noise floor measurement for the synthesizer, but I suppose I am a bit of an amateur time nut.
So did the Soyuz program have equivalent frequency generator soviet equipment onboard? or did they just use the internal clock? is there a way to send a radio signal to sync to Moscow time? How did the Soviets provide an appropriate level of precision to allow the clock to function in a useful manner? If I understood correctly, some of the inaccuracy derives from the 60hz grid frequency being slightly inaccurate over short periods of time, is that correct?
I've been watching this series and I am curious. Mind you, I'm not an electrical engineer, but I have an interest in accurate time bases for clocks. How would what you have done here, in the long term, compare to the accuracy of using the AC power grid frequency as a time base (I'm in the US where it's 60 Hz.) I think it's interesting as I have AC motor powered clocks that, as long as the power is there, that I nearly never have to set. Very enjoyable series, by the way, and thank you for presenting it!
That is true, because the power grid engineers actually have a clock similar to your in their control room besides a very accurate clock (probably synced to standard time via Internet or GPS). When they control the line frequency (by ordering more or less power into the grid from the power stations) they have two targets: 1. In the short run (seconds), keep the frequency as close to 60 Hz as possible 2. In the long run (hours), make sure that the power grid clock is not fast or slow as compared to the true clock The second target is there to cater for the dwindling amount of line synced clocks like yours still in existence.
I assume the whole problem is that the 800Hz and Hz signals are not synchronized. If you divide the 10MHz output of the atomic clock by 12500 to obtain an 800Hz signal and then divide 800Hz by 400 to obtain 2Hz, the desired signal could be generated by one NAND gate
when your friend takes back his cesium clock will you run it from your GPS clock? Because my casio does better than 17sec a day. I have a Heathkit GC-1000 but I'm not lending it, lol.
@@KD5NJR I have no idea. I assume it had an application when it was designed. From what I've read it's not that accurate compared to what's available now. Also mine doesn't have the RS232 accessory.
I'd assume the modulator also introduces some jitter because the phase relationship between the two signals is free. Could you use the low part to reset the phase/oscillator in the modulator so it always starts the same?
@@CuriousMarc Right, I just imagined that if you are able to reduce the jitter by n orders of magnitude, then the time needed to achieve the same precision is reduced by the same amount. But sure for demonstration purposes it's not that important. Thanks for your entertaining and educational videos. They motivate me to take deeper dives into topics and also get my grey cells going ❤️
The more I watch this video series, the more I get enamored by the bare functionality of this thing. So much in fact, that I kinda want to build one for myself! Would it be possible to get some dimensions of the clock case so that I could model and 3D print myself a replica?
I've been wondering why the Soyuz clock module external clock signal is a tone, given added complexity and phase noise? One thought I had was the clock module could be clocked via a radio link, rather than from a local source. It might not have been used in this application (although it might have been useful as a "last resort"), but it could be useful in other "master clock" applications.
I would suggest the 400Hz was chosen to place it in the middle of the passband for a typical voice circuit from the era. It would work equally well on a telephone line, AM, sideband or narrowband FM transmission, and may well have been a more widespread standard.
There were at least three: a separate electromechanical “digital” unit, an internal Patek Philippe clock face display (drool), and an internal LED display.
@@CuriousMarc OK... I'm looking at the panel of the cesium thing, and I see what looks like a watch dial with 'C FIELD' label... is that not a clock face? I guess all you'd need would be a simple 1-second mechanical movement driven by the 1PPS pulse.
@@scowell No that’s the multi-turn pot adjustment for the C-field. Which we have not done yet. Hopefully we can do it when the shelter in place is lifted.
Also note: each one of those instruments has it's own clock source. Each one has its own drift which changes over time and temperature. You cant afford instruments had will give you accuracy out to 6 or 9 decimal places.
The instruments were referenced to éther the Cesium or the GPSDO. Gives 11 decimals. But even if the universal counter was used on its internal source (which it wasn’t), it has the high stability option good to 9 decimals. It was calibrated in the previous video and was actually better than 10 decimals.
Great video Marc, thank you! But would you rather I sent you a watch? I have several lying around that might do the trick in very little space as compared to that tower of stuff you are using LOL!
can you tell me the name of the song in the beginning of the video please? i know the melody from a hiphop track but now im curious where the original came from. thanks
And at the same time take out the 800Hz from the divider chain and use a gate to send that to the output, so you have phase-coherent bursts. Should be doable with a couple of 40xx CMOS chips, then some line driver for the output.
The PLL adds phase noise, particularly at low frequency. But no drift. The clock will not drift more than the Cesium. That said you are right, dividing the clock would have been the more correct way to do it. Indeed, HP made a divider specifically for this Cesium source, but regrettably, we don’t have it...
@Rob: That would certainly help a lot, but the simple clock recovery mechanism of the Soviet clock is never going to be phase accurate enough, by a far cry. And then I bet the Soyuz timing source did not do that either, it just used the 28V 800 Hz from the ship supply. Because of course, this short term phase noise does not matter a bit for the purpose of indicating seconds to astronauts. The only important thing is long term stability.
@@CuriousMarc It also appears they are not really interested in subsecond accuracy as the clock cannot be set really correctly anyway. Using 2PPS instead of 1PPS does not make that better either, in case of 1PPS you could at least know that the seconds tick would be synchronous with the reference. Of course indicating the seconds does not require better accuracy but the alarmclock+relay function potentially could control something like an engine burn which could require some more accuracy. In practice it proably doesn't, for the typical Soyuz mission.
Ok but now you have the clock frequency accurately synced to the atomic standard, but what about the clock phase? There is nothing that makes the seconds tickover at the right time...
@@rkan2 the 5334b universal counter has an internal oscillator it uses as a reference for period, its delivering data out of gpib to the computer, so the measurement can only be as good as that ovenized crystal oscillator
The Allan Deviation of the output frequencies of the 3325A/B is going down proportionally from 20 MHz to low frequencies, so @ 1sec time constant, an output of 1Hz got 1E-4 stability left only. You better build a divider for that purpose, either with TTL counters, or more easily with the PICDIV, that is a PIC controller, programmed as a divider state machine. See here: www.leapsecond.com/pic/picdiv.htm
That is absolutely right. A gated divider would be the correct way to do it. That said it does not affect the clock accuracy. Just makes short term measurements very difficult and somewhat irrelevant.
The counter cannot do reverse measure method ? Very strange. Usually you do not measure the slow input signal but use it as gate for the precision 10mhz ref clock.
Tens of thousands of dollars in equipment to make an obsolete clock be ludicrously accurate. You sir, are my hero.
I love that you actually use vintage computers and vintage test equipment in your projects.
Every time Marc turned around there was a new HP machine. I totally lost it when he panned the camera to the HP85...
Yeah, this quickly turned from "restoring a soyuz clock" to "recreating a fully equipped HP research lab from the 1970s".
Marc dreams in RPN.
@@scowell If I remember correctly it also went from "oh look, it's a NASA cartridge" to "we restored an entire AGC to drive our homemade Apollo mission simulator" :)
I don't have even words to describe how much I appreciate this channel other than #yolo
I once worked at a naval air station.
They had "labs " for civilian engineers.
They did something to shift the frequency from 60hz to up and down by about 15% . It killed most of the equipment that I was responsible for and caused havoc.
It took days before I figured out what was going on. No one checks the frequency of the mains going into various equipment rooms, no one.
I eventually did.
Wow, what a difference a few Hertz makes.
Air to ground, ground to ground, navigation beakins and weather station.
All took a dump over less than 15/30% total frequency deviation.
You have neat toys, I can't do it anymore but, I can appreciate.
Thx.
Twenty fouuuur little cycles
Hmmm.
Can't think of words for Brought the Sun and the Flowers
Отличная работа, никогда не перестаете удивлять! Хорошо что хоть кто-то помнит про миссию Союз-Апполон, такое ощущение что сейчас все игнорируют это огромное достижение космонавтики.
Thanks! Translation by courtesy of Google translate: "Great job, never cease to amaze! It is good that at least someone remembers the Soyuz-Apollo mission, it seems that now everyone is ignoring this huge achievement of astronautics."
My mom did translation of a lot of the in-flight docs from Russian to English. Always found it fascinating that she could she could translate the docs but pronouncing Russian was always really tough for her. My dad worked for Boeing out at Johnson Space Center during that period. He told us that the US had to build the docking module since the Soviet Soyuz spacecraft couldn't handle the necessary precision docking required. Always wondered the details on that. But, I was a young kid at the time.
@@68Polara500 The Apollo service module was grossly over powered - the original specification had it launching from the lunar surface.
That gave a lot of reserve power that was never used, to change orbit.
Soyuz needed more precise launch parameters to dock in orbit due to limited delta v.
@@68Polara500 The story with the docking module is actually pretty strange and amusing. It had to be designed as a female to female adapter, because traditionally a docking system would be male > female, and neither country wanted to be female. (I suspect both were designed to be "male" from day 1 regardless). Supposedly the soviet engineers on the project started calling it "Hermaphrodite", and this term had to be explained to a high ranking military official that happened to visit.
I would imagine soyz could handle the precision given that they achieved totally automatic docking quite a bit earlier, back in 67.
Полёт Союз-Аполлон был в 1975 году, а в этих часах на радиодеталях 1983 год. В 1973 году в СССР уже производились электронные наручные часы. Спрашивается, зачем отправлять в космос (где каждый грамм на вес золота) часы с массой утюга и энергопотреблением утюга?
Don't you just love it, all those machines working together. It totally validates the ownership of every device. See, i knew i needed it!
When I was a kid I very much wanted an HP-85, I had an HP-47c and spent so much time with it you might call it indecent. I wrote such crazy program’s for it, I was always maxing out the memory. Mostly celestial navigation and complex matrix arithmetic. Those were the days. Charlie
In 1983 as a newly minted engineer, I designed a piece of test equipment at GE Aerospace that had an embedded 8085uP. It was managed by an HP 85, via the GPIB interface. I had a little difficulty handling the noise on the GPIB handshaking lines - but some seasoned engineers helped me solve it. Great memories, seeing that HP 85 in use! Great video and analysis, as usual.
But you measured the Russian internal quartz oscillator accuaracy here in earth' gravity. We now need a measurement of it in zero G.
Assuming you are serious, there is indeed a correction for gravity on Quartz oscillators. But it's minuscule. For my HP oven Quartz oscillator for example, it's on the order of 25 mHz at 10 Mhz. You can measure it (and I did!) by simply flipping the instrument over. The normal Quartz oscillators are not stable enough to allow a meaningful measurement. But assuming you actually could, I imagine on the order of 2.5 mHz at 1 MHz. Not going to get your 17 seconds back I am afraid... I willing to test in orbit if someone pays for my ticket :-)
I was wondering the same in the last video 😋 Also the orientation makes a difference if I recall correctly. I assume it's half-joking half-serious :)
I got to see the atomic clock at gold Stone in California. Back in 1983-84. My mother was dating a guy that worked there, who also would send and receive messages to the voyagers. I was told..."At the time" the atomic clock there was the second most accurate clock in the world. 1 millionth of a second off a year or 1 second every million years! I think that's how I remember it? I was only 13 or 14 yrs old. I am still friends with this guy today. Watching you guys work on this, brings back old memories!☺️👍
The time you could still send messages to the voyagers and wait for the reply (after a cup of coffee or a lunch)...
@@Rob2 yeah, it would take about two hours to get there & about 2 hours to get back is what my buddy told me.
@@ronniepirtlejr2606 Today it is nearly a day to get there plus the same time to get back, so nearly two days before you get a reply...
I've been enjoying your series on this Soyuz Clock very much, thank you for all the effort you've put into this project and for sharing it with us!
This is honestly some of the best content I have seen all year on TH-cam. Your clock got me subscribed and I have enjoyed the entire journey so far. Looking forward to the next video.
Bless you Marc. This is the content that will see me through this isolation! And the hp-85 thrown in as well! Thank You!
Absolutely excellent, it was a pleasure to watch the whole series of Soyuz clock videos. It's great you made the clock fully operational as opposed to it being an interesting item sitting in a glass display box.
I love your videos - all of them, quirky and brilliant - Thanks so much !
Tintin, Frenchiness? OMG, Belgians will be pissed. Tintin is Belgian not French.
I know, I know, I was born in Belgium. A touch of "francophonie" I should have said. And of course that means the Belgians made it first to the Moon! RIP Hergé.
@@CuriousMarc belgian here too, I second that.
Though I meet a lot of people not knowing where Belgium is or that it is a country.
Out of curiosity, where were you born in Belgium?
@@CuriousMarc note that the French reached the moon first by that logic, 85 years earlier, they just didn't land.
It's ok. Belgium isn't a real country.
😜 😉
@@randomelectronicsanddispla1765 Jules Verne and Meliés ftw
We need that equipment stack and the clock on a T-shirt please.
With an egg strapped to the top of the clock on a "timer" and being cooked by activating a Spark gap microwave generator with the "German" sign in the background!
@@digitalshackonthelane And the Tintin figures!
Your HP8116 amplitude modulation is not phase locked to the gating signal. The 800hz being gated by the cesium (via the 3325) starts at a random point on the sine wave, this has got to be a big source of the noise into the soyuz clock, because its integrator is going to have very different time delay based on the starting point of the 250ms wide sine pulse waveform. If you used an arbitrary waveform source, like an HP33250 (if I am remembering right) you could have the cesium 2hz waveform act as a start and stop command for a sine burst that always starts at the same point on the sine wave. Or could you use a gated 800 square wave input into the soyuz clock? and generate it from a higher frequency clock that is locked to the cesium as well, perhaps using two 3325s, one at 2hz and one at 800hz both driven by the 5MHZ cesium reference?. I do love the HP85, I used one of those back in the day until the belts and rollers disintegrated in the tape drive. Maybe I'll dig into some old equipment now that I am stuck at home. Thanks for your enjoyable videos.
Yes, you can clearly see that's one of the biggest contributor to the clock recovery signal jitter. Which is all fine of course. Astronauts are not sensitive to their seconds signal phase noise...
@@CuriousMarc Well, if you had a pair of 3325A's you could get around the problem. Clock both off the 5 MHz and use one to gate the other.
Of course a TTL or HCMOS divider chain off the 5MHz to generate both the 800 Hz and 1 Hz would pretty much eliminate the phase noise completely. I can't recall if HP made some programmable counters in that era that could be used as programmable frequency dividers, but I think they or Tektronix did.
They did one just for that very application, to divide the Cesium and generate very low phase noise clock timing. That’s the unit you see on top of the flying clock rack. But we don’t have it...
@The Harbinger - I wondered the same about locking...but in layman's terms.
Love the use of the HP 'terminal' 🙂👍👍
wow, that printing computer tho. you whipped those results up out of nowhere
omg! all my electronics engineering / hobby stuff could fit in a suitcase. you you have ..a whole building 😨😨😨
Love getting this engineering fix! 👍
I really glad that clock was end up Your desk instead some random collector - as You was able to create so much interesting content based on that! have watched all series an totally enjoyed!!!
It's amazingly accurate, it's a bit pricey, but it's worth it
Great video! I watched twice and learned a lot more the second time around.
This readout on the HP 85 reminds me of the paper strip machines used for timing mechanical watches - Vibrografs, or newer timegrapher machines.
Certainly an interesting way to implement the external clocking. I guess to make it resistant to vibrations?
Next episode?? What more could there be??
That Soyouz clock keeps on giving 😉
Niiice. Now all we need is the rest of the spaceship. :3
Marc, it would be cool if you could take a micrometer to that clock so that someone can make an accurate computer model that can be 3D printed or CNCd. Or just to further document this piece of engineering history. Love the videos!
I love your sense of humor, and the soviet memes.
продолжайте работать товарищ
Fascinating. You get a lot jitter but at the end of the story, the overall precision is the atomic clock time
Fascinating!
Perfection! Well done
I don’t know exactly what’s going on but hurrah for science
This is a fun series to follow.
Such a devices! Reminded me of soviet flight computer I saw three days ago on classifieds. New one, with instruction, made in 80s I suppose.
I hope part #7 will be synchronizing to WWVB. Perhaps GPS.
The clock is running at an accurate 1 Hz. Now the clock needs to be fed the current time.
You totally need to make a rack-mount case for that, with the switches mounted, and connectors broken out to the front etc. :D
Any knowledge about the type of clock/time source used onboard the Soyuz?
Not much, except a passing mention that maybe it received signals from Earth encoded in the TV channel.
@@CuriousMarc ok, so local time source, other than the built-in crystal. Makes sense.
I think you should verify the HP calculations by punching the raw data onto paper tape, convert that to cards, write a IBM 1401 Fortran program to re-do the calculations and then run that on the IBM 1401 computer.
Would you mind posting that basic program? I am curious what something like that would look like with the HPIB.
I was a kid watching the Apollo-Soyuz handshake, too.
Excellent!
Marc, maybe you can help me fix the flashing 12:00 on my VCR?
It only requires a small fusion of chernobyl and three mile island meltdown power supply! Amazing project and walkthrough!
Did I see the clock jump 4-5 seconds as Marc was switching between signal sources @ 5:30? I didnt expect that.
Maybe there is space-time warpage in my lab. Or it could be poor video editing too. Who knows. I vote for space time warpage.
@@CuriousMarc Aha. Slight audio distortion across the rift! Thanks Marc
Right on, Marc... that'll get keep that clock ticking nice and steady. :-)
For your next feat, synchronize it with UTC!
You'd need a large project box to put that lot in.
I am not sure about that the external clock source is designed to be feed from some reference on Soyuz board. I am thinking about possibility that theese 2Hz pulses came from some radio receiver, which could lock the clock to some radio clock station on the earth. Using A1 modulated signal to interconnect it with some clock reference inside soyuz simply makes no sense and looks too complicated to me. And on top of that. If I am right and this was connected to some radio equiptment instead of some local reference, it opens the posibility, that there was more commands than "increase ½second", e.g. "first second in minute" or something similar. I do not know how much this was reverse-engineered, so I do not know if this is possible, or not.
Fun measurement setup, but while it would have made made no difference on the long term graphs, you could probably get substantially better single-period stability if something other than the 8116A was used to generate the tone burst. Since it's using an integrator to sense the start of the burst, the phase the burst starts at is critical to eliminate leading edge timing jitter, and the 8116A can't be locked to an external reference! You could see the phase of the beginning of the burst drift on your scope captures, and if you used an instrument that either started the burst on the trigger (rather than just gating a free running oscillator) or which was reference locked to the same atomic source, you could make sure that every burst started with the same phase, so that the integrator's rise would be much more precisely controlled (I am assuming that the soyuz clock isn't measuring the time between the beginning and end of the integrator and is instead timing based off of the rising or falling edge of the integrated burst signal).
Would be interesting to see an 11 or 12 digit/s counter thrown at it too, though not period appropriate, to really flesh out the noise floor measurement for the synthesizer, but I suppose I am a bit of an amateur time nut.
15:17 Trust me, I'm an engineer! I'll never not make that association. lol
A clock with atomic precision but no time display...now there's a clock to my taste!
At least there is no false advertising: just says Cesium Frequency Standard on it.
So did the Soyuz program have equivalent frequency generator soviet equipment onboard? or did they just use the internal clock? is there a way to send a radio signal to sync to Moscow time? How did the Soviets provide an appropriate level of precision to allow the clock to function in a useful manner? If I understood correctly, some of the inaccuracy derives from the 60hz grid frequency being slightly inaccurate over short periods of time, is that correct?
Was the burst oscillator phase locked to the 2 Hz clock? If not I think you would get much less phase noise from the integrator if it was.
Woo! Awesome!
I've been watching this series and I am curious. Mind you, I'm not an electrical engineer, but I have an interest in accurate time bases for clocks. How would what you have done here, in the long term, compare to the accuracy of using the AC power grid frequency as a time base (I'm in the US where it's 60 Hz.) I think it's interesting as I have AC motor powered clocks that, as long as the power is there, that I nearly never have to set.
Very enjoyable series, by the way, and thank you for presenting it!
That is true, because the power grid engineers actually have a clock similar to your in their control room besides a very accurate clock (probably synced to standard time via Internet or GPS). When they control the line frequency (by ordering more or less power into the grid from the power stations) they have two targets:
1. In the short run (seconds), keep the frequency as close to 60 Hz as possible
2. In the long run (hours), make sure that the power grid clock is not fast or slow as compared to the true clock
The second target is there to cater for the dwindling amount of line synced clocks like yours still in existence.
I assume the whole problem is that the 800Hz and Hz signals are not synchronized. If you divide the 10MHz output of the atomic clock by 12500 to obtain an 800Hz signal and then divide 800Hz by 400 to obtain 2Hz, the desired signal could be generated by one NAND gate
How what is the listing of your program for the HP-85?
when your friend takes back his cesium clock will you run it from your GPS clock? Because my casio does better than 17sec a day. I have a Heathkit GC-1000 but I'm not lending it, lol.
do you know why the jack on the back of the Heathkit outputs 3.6 Mhz and not 1 Hz ?
@@KD5NJR I have no idea. I assume it had an application when it was designed. From what I've read it's not that accurate compared to what's available now. Also mine doesn't have the RS232 accessory.
Thank you! Next episode:"Hold on, but can it brew some coffee?". 🤣
Snowy (wuff) concurs!!
i am almost in ussr once upon time we had radio corrected clocks maybe they had too.besides they always manage to land in Kazakh step....
I'd assume the modulator also introduces some jitter because the phase relationship between the two signals is free. Could you use the low part to reset the phase/oscillator in the modulator so it always starts the same?
You could, but none of that matters. The phase noise does not impact accuracy at all. It just makes short term measurements somewhat irrelevant.
@@CuriousMarc Right, I just imagined that if you are able to reduce the jitter by n orders of magnitude, then the time needed to achieve the same precision is reduced by the same amount.
But sure for demonstration purposes it's not that important. Thanks for your entertaining and educational videos.
They motivate me to take deeper dives into topics and also get my grey cells going ❤️
The more I watch this video series, the more I get enamored by the bare functionality of this thing. So much in fact, that I kinda want to build one for myself! Would it be possible to get some dimensions of the clock case so that I could model and 3D print myself a replica?
Yes it’s on my to do list.
@@CuriousMarc Awesome!
Good day!
I've been wondering why the Soyuz clock module external clock signal is a tone, given added complexity and phase noise? One thought I had was the clock module could be clocked via a radio link, rather than from a local source. It might not have been used in this application (although it might have been useful as a "last resort"), but it could be useful in other "master clock" applications.
It enabled them to put the external signal through a transformer to give Galvanic isolation.
@@cambridgemart2075 I think there has to be more to it than that - a pulse would still work through a transformer.
Not reliably. Any switching noise nearby would get picked up. So integrated burst is the correct way to go.
I would suggest the 400Hz was chosen to place it in the middle of the passband for a typical voice circuit from the era. It would work equally well on a telephone line, AM, sideband or narrowband FM transmission, and may well have been a more widespread standard.
Is there not a mechanical clock display available on the HP? Patek-Philippe? The most accurate PP ever made?
There were at least three: a separate electromechanical “digital” unit, an internal Patek Philippe clock face display (drool), and an internal LED display.
@@CuriousMarc OK... I'm looking at the panel of the cesium thing, and I see what looks like a watch dial with 'C FIELD' label... is that not a clock face? I guess all you'd need would be a simple 1-second mechanical movement driven by the 1PPS pulse.
@@scowell No that’s the multi-turn pot adjustment for the C-field. Which we have not done yet. Hopefully we can do it when the shelter in place is lifted.
Also note: each one of those instruments has it's own clock source. Each one has its own drift which changes over time and temperature. You cant afford instruments had will give you accuracy out to 6 or 9 decimal places.
The instruments were referenced to éther the Cesium or the GPSDO. Gives 11 decimals. But even if the universal counter was used on its internal source (which it wasn’t), it has the high stability option good to 9 decimals. It was calibrated in the previous video and was actually better than 10 decimals.
Great video Marc, thank you! But would you rather I sent you a watch? I have several lying around that might do the trick in very little space as compared to that tower of stuff you are using LOL!
can you tell me the name of the song in the beginning of the video please? i know the melody from a hiphop track but now im curious where the original came from.
thanks
This is Полюшко-поле. th-cam.com/video/IFgfrCiy6fA/w-d-xo.html
Maybe you could find a proper Soviet clock in that old abandoned Buran shuttle?
And remember kids! Phase noise matters!
Thank you!
I’m pretty sure dividing down the atomic clock to 2Hz would be a lot more accurate, PLL systems have drift and jitter.
And at the same time take out the 800Hz from the divider chain and use a gate to send that to the output, so you have phase-coherent bursts.
Should be doable with a couple of 40xx CMOS chips, then some line driver for the output.
The PLL adds phase noise, particularly at low frequency. But no drift. The clock will not drift more than the Cesium. That said you are right, dividing the clock would have been the more correct way to do it. Indeed, HP made a divider specifically for this Cesium source, but regrettably, we don’t have it...
@Rob: That would certainly help a lot, but the simple clock recovery mechanism of the Soviet clock is never going to be phase accurate enough, by a far cry. And then I bet the Soyuz timing source did not do that either, it just used the 28V 800 Hz from the ship supply. Because of course, this short term phase noise does not matter a bit for the purpose of indicating seconds to astronauts. The only important thing is long term stability.
@@CuriousMarc It also appears they are not really interested in subsecond accuracy as the clock cannot be set really correctly anyway.
Using 2PPS instead of 1PPS does not make that better either, in case of 1PPS you could at least know that the seconds tick would be synchronous with the reference.
Of course indicating the seconds does not require better accuracy but the alarmclock+relay function potentially could control something like an engine burn which could require some more accuracy. In practice it proably doesn't, for the typical Soyuz mission.
@@Rob2 Oh yes, you are right, forgot about that. We don't even know within 0.5 seconds when it will tick!
I have to play "Tintin on the moon" after seeing this
YES
It readjusted the time by a few seconds jumped forward when reconnected to the external o
No, that’s just my poor video editing. It switches back and forth glitch free.
17 seconds a DAY? _Ouch._ Methinks the clock crystal has seen better days.
Darn, I all but had a HP85, now all that is left is the CRT
Unfortunate.
Egg timer? It probably dissipates enough to cook the egg too!
In a past video they mentioned the V*A... Was it like 5W?
@@rkan2 For one piece of equipment, maybe. For all that they were using here, no way.
Any more equipment and Marc probably has to call the power company before he switches has lab on ;-)
So what would have been the retail valve of your International Egg Timer Alarm Clock?
Isn't 1E6 seconds on the order of like 11.6 days?
Oh cool, a HP 85, how old is this video? TODAY?! Niiiice
Ok but now you have the clock frequency accurately synced to the atomic standard, but what about the clock phase?
There is nothing that makes the seconds tickover at the right time...
good to know I'll be able to keep my cesium pulsed soyuz clock accurate for a decade without worry
Can you help set the time on my Casio watch
Truly excellent Marc. Although is that not German marking's on the V2 rocket, bringing is us, Europeans in, or even us, the Scots.
It's an American-Russian association supervised by an excellent French manager 😉
what is generating the reference being used by the frequency counter???
The cesium tube made by agilent?
@@rkan2 the 5334b universal counter has an internal oscillator it uses as a reference for period, its delivering data out of gpib to the computer, so the measurement can only be as good as that ovenized crystal oscillator
the cesium clock is about 5 digits more accurate than the oven oscillator in the 5334b
the current setup is saying its 1s gained/lost roughly every 2 years which is in the realm of what a high performance crystal oscillator can give you
he can't measure the accuracy of this system properly without a more accurate clock source
Lovely stuff. Love that Soviet simple design ❤️
Can you clear the timer without stopping it first? Seems not that relevant, but I have been wondering about this since the last episode :D
No you can’t. I demonstrate it somewhere in the last video I think.
@@CuriousMarc Oh, I must have overlooked that somehow... Thanks for the answer!
Does the internal Clock consider relativity?
Where is the snooze button?
The Allan Deviation of the output frequencies of the 3325A/B is going down proportionally from 20 MHz to low frequencies, so @ 1sec time constant, an output of 1Hz got 1E-4 stability left only.
You better build a divider for that purpose, either with TTL counters, or more easily with the PICDIV, that is a PIC controller, programmed as a divider state machine.
See here: www.leapsecond.com/pic/picdiv.htm
That is absolutely right. A gated divider would be the correct way to do it. That said it does not affect the clock accuracy. Just makes short term measurements very difficult and somewhat irrelevant.
Marc is the basic of that machine precise enough? remember there are all kinds of limits on some BASIC interpreters with math.
... and just like that, you've become a time-nut... :)
Not sure about that, but I sure am an old HP test equipment nut!
The counter cannot do reverse measure method ? Very strange. Usually you do not measure the slow input signal but use it as gate for the precision 10mhz ref clock.
Of course it does.
@@CuriousMarc then I do not understand why you get only four digits resolution at 8:35, and not at least six or seven...
@@thomasw6169 I have to admit I was confused too for a minute. It’s just because phase noise dominates the measurement.
@@CuriousMarc phase noise in the synthesizer ? Hm. Ok.
Yes it turns out this particular synthesizer is very noisy in the low end of the spectrum. A simple divider would not have had this problem.
Anyone think Marc is a frustrated HP traveling salesman?
You are having TOO much fun exercising some very expensive vintage equipment!