Time Precision Needed for Celestial

[El Pinguino]

Some may find this interesting, start at page 23.
https://ndhadeliver.natlib.govt.nz/d...pid=IE39275029
For time keeping see page 31
'As noted earlier rating the chronometer can
be performed by comparing the longitude of a
known position with the longitude obtained by a
time sight but a number of other methods exist.
In particular the timing of the disappearance
and/or reappearance of stars in lunar
occultations provides a means of determining
CE. This method was also used by Worsley and
Reginald James, expedition physicist, to rate the
chronometers.'
and
'On Elephant Island on 24 April 1916, the
morning of departure of the James Caird for
South Georgia weather conditions permitted a
time sight to be made. Worsley (1916b) wrote ‘I
jubilantly welcomed this opportune appearance
of the sun for without it we should have been
placed in a still more difficult & dangerous
position making S. Georgia than we were.’
and in (Worsley 1998: 101) ‘Immediately after
breakfast the sun came out obligingly. The first
sunny day with a clear enough horizon for rating
my chronometer.’ No noon sight for latitude was
possible.'
I'm pretty sure a 'time sight' is what we now call 'Long by Chron'.

[SeanPatrick]

Originally Posted by El Pinguino

I'm pretty sure a 'time sight' is what we now call 'Long by Chron'.

Indeed. That is what the tables are that I mentioned in this post in the "Where to buy a sextant." thread: time sight tables. Time difference = longitude. I could have made the tables easier to use by formatting the final output as longitude. But, I instead chose to make the output the time difference and include the extra step of converting the time to longitude.

Why did I do this? Because I'm a sadist.

It also makes the tables feel more "old-timey" ... so there is that, too.

Originally Posted by El Pinguino

(Quoting the Records of the Canterbury Museum) [...] the timing of the disappearance and/or reappearance of stars in lunar occultations provides a means of determining CE.

Yes, but I think the accuracy required for this task is beyond what one could reasonably accomplish with a sextant. Especially aboard ship. The same could be said about the occultation of the Galilean Moons.

[bdgWesternMass]

This maybe the strangest way to get accurate time for celestial navigation.

https://nelsonslog.wordpress.com/202...an-ntp-server/

[s/v Jedi]

Originally Posted by bdgWesternMass

This maybe the strangest way to get accurate time for celestial navigation.

https://nelsonslog.wordpress.com/202...an-ntp-server/

You can’t do microsecond precision celestial navigation that I am aware of. For what I know about it, using thE time from your GNSS sensor is more than enough precision.

It is cool to see how dishy simply does this and outperforms big time servers

[Dockhead]

Originally Posted by s/v Jedi

You can’t do microsecond precision celestial navigation that I am aware of. For what I know about it, using thE time from your GNSS sensor is more than enough precision.

It is cool to see how dishy simply does this and outperforms big time servers

Sure, but again, if you've got data from your GNSS sensor then you've got your position, too, making the celestial unnecessary, even if it is still fun.

But to your point -- I don't see how our celestial processes can use sub-one second precision time even if you have it, so you're certainly right there.

I guess an electronic device could be devised which would make the actual observations and record them with more precision, particularly in regards to timing, than the human eye and hand could do. That would be a very useful device, wouldn't it.

It's a shame that despite various efforts to revive it, that hyperbolic radio navigation has now been shut down everywhere. Apparently there is still some movement towards implementing eLORAN, particularly in Europe (with EUROFIX) and in China. I guess the Ukraine war and all the attendant GPS disruption interest in these system will increase.

When I was a boy cruising with my father, before GPS, he used LORAN-C and had a LORAN set on his boat right up until he sold it not that many years ago. He never taught me how to use it. We already had GPS by the time I started doing any navigation myself.

[s/v Jedi]

Originally Posted by Dockhead

Sure, but again, if you've got data from your GNSS sensor then you've got your position, too, making the celestial unnecessary, even if it is still fun.

But to your point -- I don't see how our celestial processes can use sub-one second precision time even if you have it, so you're certainly right there.

I guess an electronic device could be devised which would make the actual observations and record them with more precision, particularly in regards to timing, than the human eye and hand could do. That would be a very useful device, wouldn't it.

It's a shame that despite various efforts to revive it, that hyperbolic radio navigation has now been shut down everywhere. Apparently there is still some movement towards implementing eLORAN, particularly in Europe (with EUROFIX) and in China. I guess the Ukraine war and all the attendant GPS disruption interest in these system will increase.

When I was a boy cruising with my father, before GPS, he used LORAN-C and had a LORAN set on his boat right up until he sold it not that many years ago. He never taught me how to use it. We already had GPS by the time I started doing any navigation myself.

I got the feeling that an app could be developed that allows you to point your phone to the sky and it will tell you where you are within one mike or so…

[Jammer]

I came across some old notes of mine on this.

The best all-mechanical pocket watches for CN are probably the Hamilton 992B series. Over half a million were made. Production was 1940-1969. This watch is widely regarded as the pinnacle of mechanical pocket watch development prior to the quartz era. This was a 21 jewel watch with a 52 hour mainspring and Elinvar balance spring (for temperature stability). When new these could be regulated to within about 10 seconds a week. These were cased by the manufacturer; the #15 stainless steel case is probably the most practical of the late cases that were made in significant numbers.

Note that the 992 (1903-1940) and 992B are completely different models.

My notes on the closely related Hamilton 4992B "G.C.T." - Military 24 hour version of the 992b with a black dial and sweep second hand. It is stem set with a hacking movement. GCT was Greenwich Civil Time, the predecessor to GMT. About 150,000 made. Some had matching rubber holders, used in aircraft, that could be attached to a surface with a screw. Most of these were made 1941-1944 with sources varying on whether there were actually any later production runs. In U.S. military aviation use through the 1960s.


The 992B is widely available for sale from dealers for between $500-$1000. It is hard to find model 4992B watches in good conditon since nearly all of them were in actual service in military aviation for twenty years and had a hard life. The few minty ones are expensive.

[Adelie]

Originally Posted by Dockhead

My best sailing friend studied astrophysics as an underground and is a master of celestial nav. I'm hoping to not be too busy the next time we cross an ocean together to take his classes.

I think with the widespread jamming and disruption of GPS we are seeing these days, these skills are more relevant than they were before. I note that the US Navy has re-introduced celestial nav.

Meanwhile I was thinking about how to solve the time problem in a robust way, so that you can do celestial even after a bad lightning strike etc. which wipes out your electronics.

I was discouraged to find out that mechanical chronometers are exorbitantly expensive to service and adjust.

So I'm looking at quartz marine chronometers, like the Seiko QM-10 and QM-11, which is not that expensive and readily available.

It has electronics so would be vulnerable to EMP but maybe could simply be kept in a metal box.

The accuracy looks like 60 seconds a year, but I don't know how crystal aging affects that -- these are decades old by now.

I also don't know if that's accurate enough -- 5 seconds a month?

There are HAQ watches available which are accurate to 10 seconds a year or even 5 seconds a year, which I guess could likewise be kept in a metal box.

Or maybe it's not all that important if you reset it to GPS time once a week.

With that process, maybe even a mechanical one would be OK.

Views?

As I understand it the USNavy never stopped performing CelNav, they stopped teaching it at the Academy. Then they realized there was a whole range of threats to GPS (Jamming, hacking, ASat weapons, CMEs, ....), so they started teaching it again at the Academy; they realized their officers needed to know enough to effectively manage the enlisted actually doing to work.

I wish they hadn't shut down the LORAN system. It was costing $35M/yr in 2003(?) to maintain. It would take better part of $1B to rebuild at this point. And it would be worth it because it is a lot harder to jam and you can't shoot them down. Water under the bridge.

Accuracy, precision and error rate are 3 different things.

Precision is about how close a group of measurements are around their average. If you are only taking 1 reading then you want as high a precision as possible. With electronic and mechanical watches, you can more or less ignore precision, you need precision around 1sec and the second to second precision wanders by milliseconds or less. You ability to read the timepiece to 1sec is way worse than the precision of the timepiece itself.

Accuracy is about how close the average of a group of measurements is to the target/reality. Even if you have a wide spread of data, with averaging the result is likely to be close to reality with some caveats.

Error rate is the rate at which the measurements drifts from reality. This is the important value. For timepieces if you know the watch error on a given date and know the error rate and the error rate is consistent you can determine time at some future date without reference to outside sources.

The key here is knowing the error rate and that the error rate is consistent over time (weeks).
How do you measure the error rate? By checking error over an extended period (1-3mo).
So for each timepiece check the error every 7 or 10 days. I prefer 10d as it makes the math easier.
I would reset the watch at the start of checking the error rate and not reset it again until you check the error rate again.
At the start of checking the error rate reset the timepiece and error is 0 on that date.
Every 10d check the error. Then calculate the error (+/-sec) since the previous check AND since the first check. And calculate the error rate +/-sec/day) since last check and first check.

Having a low error rate is nice because there is less math to do to get corrected time, but it doesn't matter whether the error rate is 60s/d or 2s/d. A consistent 60.0s/d is actually better than the 2s/d if it varies from 1.8s/d to 2.2s/d because in 10d you know the correction will be 600s for the 60.0s/d whereas the correction for 2s/d might be 18s or 22s.

So how consistent do we need the error rate to be?
Near the equator 4sec time error results in 1nm longitudinal error. Further up or down the planet the position error is less. Let's say you want to keep longitude error due to time error to 1nm. That's 4sec.
What's the longest you will need to travel to reach some place where you can confirm time or replace damaged electronics? Lets say 30days. So over 30d you want to make sure all the 10d errors don't vary by more than 1sec each.
So if the 10d errors are 24s, 20s, 22s, 18s, 20s, 21s, ... you have a problem, the error rate is going up and down inconsistently.
If the 10d errors are 601s, 600s, 601s, 601s, 600s, ... the corrections are larger but the rate of error is consistent over time.

If I were going offshore and wanted CelNave as a backup I would take 5 cheap Casio watches. Their error rates are usually consistent enough to count as Chronometers (that's the distinction between regular watches and chronometers, consistent error rate, not the magnitude of the rate.) Quartz oscillators are mostly but not completely temperature independent. The one on your wrist may be more consistent than the ones in the sextant box. Apparently error rate is not dependent on battery condition/voltage.
Use a Rite-in-the-Rain notebook to track error for each watch over time and keep it updated every 7-10d.

I also have a Molinja pocket watch I got off of Ebay for $35 and couple mechanical watches of my father's from the 1970s. I have not checked their error rates. If I ever get serious about offshore I will check their rates.
Mechanical watches need to be wound daily. Error rate varies with position, temperature, vibration and spring tension. I would store them in the sextant box to minimize both vibration and temperature variation, with the box in the the same position every day and the watches in the same position in the box every time (cut a slot in foam for each watch).
When the watch is first wound the instantaneous error rate is different than just before winding it 24hr later. However by winding it every day at the same time, the average DAILY error rate is the same from day so from day to day you can treat it as consistent. I would suggest that if you are 1/2 late winding one day winding it 1/2hr early the next may correct the net error. I have no info to back this up though.
Automatically wound watches are a special case. They have a weight in them that is jostled by hand and arm movements. The jostling of the weight winds the watch. I have no info on whether error rate is consistent or not. Spring tension is usually maintained near full winding and being worn means temperature variation is minimized. On average position is consistent thru the day so the only question is how much vibration is affecting error rate and how good the movement is.


Once you leave the dock sailing, once a day at 1100-1130 wind the mechanical watches. Once per week shoot 1 line of position of the sun. (using a sextant is not exactly a perishable skill but it does degrade) and update the errors for each watch against an external source (GPS, cell phone, broadcast time on shortwave) and check that the error rates remain consistent.

Dave Burch has a nice writeup on determining error rate: https://davidburchnavigation.blogspo...atch-rate.html
To accurately keep track of elapsed days since starting an error record I would suggest a scheduling wheel: https://www.amazon.com/Scheduling-Wh...ab3c66184794d1

If accurate time is lost and needs to be reestablished without outside reference there are 2 methods described in “Self-contained celestial navigation with H.O. 208” by John S Letcher. This can be found as a PDF for free and bought for a modest amount.

Here are some posts I've made on CelNav and related topics
https://www.cruisersforum.com/forums...ml#post2850485
https://www.cruisersforum.com/forums...ml#post2852228
https://www.cruisersforum.com/forums...ml#post2723304
https://www.cruisersforum.com/forums...ml#post2733360
https://www.cruisersforum.com/forums...ml#post2922991
https://www.cruisersforum.com/forums...ic-219989.html

[Uricanejack]

Going back to the OP question.
Accuracy required +- 1 second if you can. My best method of achieving this. As 2nd or 3rd Mate we took the time for each other. at the chronometer.
Otherwise on my own it required counting seconds until i got to the chronometer clearly less accurate. Close enough. Personally i never used a stop watch or a watch to time it.
Just one and, two and, ect.

If you want a Chronometer. A Rolex Datejust is still marketed as a certified chronometer.
Marketing aside. A Rolex is an automatic self winding watch worn on the wrist.
A Chronometer is kept in a glass topped box by the chart table to stay at as near constant or rather a slowly changing temperature and wound at the same time daily to keep the error change or rate as constant as possible. So in reality the Rolex doesn't quite meet requirements. My Dads is an early 70s it keeps quite good time. I don't wear it. I find my Seiko Automatic pretty good. I gave one to both my sons. they like them. you don't have to remember to change batteries.
If you really want to go mechanical probably as good as you are going to get today. A few seconds per day for rate?
Traditionally you didn't reset a chronometer.

WWV is still going. The BBC used to be reliable, Unfortunately even before i quite deepsea they started using satellite's to relay the signal which has a slight delay.
The Russians may or may not be still sending out pravda on air I don't know.
Voice of America not sure ither.
WWV can be received on a decent short wave pretty much anywhere. At some point day or night. Most likely without a sunrise or sunset between you and Colorado.

A good quartz clock should do the job well enough. Check the rate daily and record it.
BTW it will vary with temp. So keep a thermometer with the clock and record the temp.

As for ultimate accuracy. I was pretty good, doing it every day at sea for about 10 years. I would claim on a good day +- 2 miles ish.
On a really good day I might beat that. On not such a good day? I was lucky if i could get a position line.
Or maybe it was practice? Accuracy better than a DR.

It's a running fix so you are kidding yourself if you think you can claim better than +- 2 miles.
I was doing it every day from a large ship, with an accurate gyro compass and a doppler log.

The skill or art, comes into play when conditions are not ideal.
Getting a observation when its 6 or 7 or even 8 8ths cloud cover. When you get brief glimpse of the sun. or the horizon is barely visible.
I've called the Capt, to come up in the morning. So I could go down to the main deck to get a horizon. If I took a radio, He took the time for me.
Sometimes we would go days without a good position. Particularly WNA. or NPac.
Getting a position in the Trades no bother.

The Mate would always insist his stars were much more accurate then my Noon. Even so Noon to Noon was what we used for all the paperwork. While I would never openly admit it. The Mate was right. My noon was only as good as the log. as a running fix.
His stars were much better.
Only if he ran them other wise he was full of BS as well.
His run was much shorter.

As 3rd mate I would do stars when I had the opportunity. So not every day. Occasionally in the morning quite often in the evening. I often relieved the Mate during twilight for his evening meal.

I used Deca Loran and Omega back in the day. Deca was the best but coverage was limited and only coastal. Loran cover the whole NA ocean sort of but not really. Omega the military version might have worked. The civilian version we had could just about tell us reliably which ocean we were in.

Deca and or Loran were never particularly accurate or reliable positions but the error was nearly constant over the course of a day so it could give you a good run. If you took a loran fix with your obs and later at noon.

Before I quite DS Sat Nav was becoming common. Even so we always did our observations quite often using Sat Nav Dr.
Never sailed offshore with GPS. but Today my morning time check is taken by GPS.
My sextant pulled out on rare occasions. To show a cadet how it works.
My accuracy declined to +- 4 or 5 miles on a good day.

In order to be accurate,
You need good accurate time.
A good quality well adjusted sextant.
The sextants error checked.
A clear sky.
A clear horizon.
A good accurate compass. QM or Autopilot
A good accurate Log.
And lots of practice.

[Adelie]

Originally Posted by SeanPatrick

Just remember: Joshua Slocum famously circumnavigated with celestial and "a cheap tin clock".


And you can always recover UT (GMT) by shooting a lunar.

Slocum used Lunars which require significantly better precision and accuracy from the practitioner.

[Adelie]

Originally Posted by Uricanejack

Anyone can get a latitude with a lodestone or an astrolabe if you want more accuracy.
For Longitude you will need one of Harrison”s fancy watches.

Or one of several versions of Lunars.

[Adelie]

Originally Posted by waterman46

I would guess they didn't use much celestial nav due to the places they sailed, where it's almost always cloudy, and had no good charts. And they might not have known much about the ocean currents in those areas either. No electronic depth sounders. Those guys had balls of steel. I suppose they would simply go slow and keep a good watch.

Worsely used CelNav to fix the island the crew sheltered on.

He then navigated a lifeboat 800mi to South Georgia Island (a 20x100mi target) seeking help. They landed to boats, trekked over some mountains to a whaling station, which got them back to civilization. There they organized a ship to return to the island with the crew.

[Adelie]

Originally Posted by SeanPatrick

Indeed. That is what the tables are that I mentioned in this post in the "Where to buy a sextant." thread: time sight tables. Time difference = longitude. I could have made the tables easier to use by formatting the final output as longitude. But, I instead chose to make the output the time difference and include the extra step of converting the time to longitude.

Why did I do this? Because I'm a sadist.

It also makes the tables feel more "old-timey" ... so there is that, too.




Yes, but I think the accuracy required for this task is beyond what one could reasonably accomplish with a sextant. Especially aboard ship. The same could be said about the occultation of the Galilean Moons.

Occultation of stars by the moon and the moons of Jupiter are on the of the few means of reestablishing time without reference to civilization. Both require almanacs that may be hard to acquire.

In order to observe the Jovian moons you need to be ashore with a telescope. If you are ashore you probably know your location more accurately and can reestablish time by working a sun or star sight backwards.

[SeanPatrick]

Originally Posted by Adelie

Slocum used Lunars which require significantly better precision and accuracy from the practitioner.

Slocum shot one lunar during his entire voyage.

[El Pinguino]

This and several other threads seem to be wondering what to do if GPS is down, what to do if WWV is gone or your radio is cooked , what to do if all your instruments are toasted, in fact the whole prepper nightmare.
Simple really, starting with time. Keep a half decent watch ( in a tin if you must ).
I rather like the digital Casio F91W-1 as you wont be muddling up your minutes.
I've been watching my ADEC ( Citizen's budget brand) for a week or two and the rate is under 1 second a day. Bought a Timex Expedition last week for the clearer dial, rate is also under one second.
You chart plotter is toast so best you have a few paper charts if only planning charts for whatever ocean you are sailing on.
Print out the daily almanac pages for the passage you are making.
Carry sight reduction tables and also a copy of Nories, not only for the tables but also for the list of ports in the back .
Unsure of your chronometer error and want to check it?
Having found your way to wherever simply take a series of sights from a known position.
You could work these out 'long by chron' like Frank Worsley did on Elephant Island ( first piccy for the workings ) or just take a series of sights of bodies on the Prime Vertical and use Mc St H to establish the intercepts and from there the chron error and rate ( by converting Long into Time).
All pretty simple really.
Oh, you will also need a sextant and to know how to use it.... and a 2B pencil.