By continuing to use this site, you agree to our use of cookies. Find out more
Forum sponsored by:
Forum sponsored by Allendale Until July 27th

Gents impulse clock

How does it work

All Topics | Latest Posts

Search for:  in Thread Title in  
SillyOldDuffer16/06/2019 14:26:54
4405 forum posts
957 photos
Posted by Michael Gilligan on 15/06/2019 13:42:10:
Posted by SillyOldDuffer on 15/06/2019 09:57:56:
.
I've got a stalled Arduino 'Clock Analyser' project due to not solving the accurate clock problem! [ ... ]

.

Dave

I think you have just made an excellent 'business case' justification for the GPSDO

Using GPS to 'discipline' an [already good] oven-stabiised Crystal Oscillator, is probably the most cost-effective way you will ever find of obtaining a reliable reference signal.

MichaelG.

Alas, Business Case fails on cost. I found a few GPSDO modules for sale at about £100 each. Would be possible to make one cheaper except a design goal was a high-end Clock Analyser that anyone could replicate with cheap readily available modules. Plug and play rather than grown-up electronics.

Wondering now about using a Wireless technology such as Bluetooth to link a GPS unit in the clear to the analyser-part indoors next to the pendulum. Unfortunately, although the radio link is quick ( say 0.1µS delay over 30 metres), encoding and decoding Bluetooth messages isn't. And a wireless link is more cost and complexity.

Very tempted to steal Robert's collection of Rubidium Oscillators, drool...

John's drunkards walk comment has me thinking too. It reinforces my original notion that a pendulum analyser needs to be able to see both micro-variations in time keeping and subtle long term effects.

My test pendulum may be misleadingly good in the sense it's not a realistic mechanical clock, hence me not seeing a drunkards walk during testing. It's a steel bob on a carbon-fibre thread energised by an optically triggered weak electromagnet. I suppose because the mechanical and temperature effects on the pendulum are low, it keeps remarkably good time. Better than 20ppm when running well and I got one run of 8ppm. This with the bob dangling from a wobbly Meccano gantry.

So good was the test pendulums performance, I've toyed with the idea of a properly engineered rigid frame mounted inside a length of 4" plastic drainpipe from which most of the air has been removed. I think a simple arrangement like that would outperform all but the best mechanical clocks, essentially by dodging the friction and impulse problems that make building accurate mechanical clocks seriously challenging.

Not quite as attractive as a well-made brass clock. I doubt anyone would admire 18" of old sewage pipe perched on my mantelpiece...

Dave

Michael Gilligan16/06/2019 14:53:09
avatar
13253 forum posts
578 photos
Posted by SillyOldDuffer on 16/06/2019 14:26:54:
Posted by Michael Gilligan on 15/06/2019 13:42:10:
Posted by SillyOldDuffer on 15/06/2019 09:57:56:
.
I've got a stalled Arduino 'Clock Analyser' project due to not solving the accurate clock problem! [ ... ]

.

Dave

I think you have just made an excellent 'business case' justification for the GPSDO

Using GPS to 'discipline' an [already good] oven-stabiised Crystal Oscillator, is probably the most cost-effective way you will ever find of obtaining a reliable reference signal.

MichaelG.

Alas, Business Case fails on cost. I found a few GPSDO modules for sale at about £100 each.

.

Point taken ... although I'm pretty sure that my [incomplete] collection of components includes all the clever bits, and has cost me much less than £100 so far.

... It might, however, be embarrassing to check how much, and how long it has taken to accumulate them blush

If it's any help; I have just rediscovered James Miller's simplified design: **LINK**

http://www.jrmiller.demon.co.uk/projects/ministd/frqstd0.htm

MichaelG.

Robert Atkinson 216/06/2019 15:27:06
avatar
265 forum posts
17 photos
Posted by Michael Gilligan on 15/06/2019 22:25:54:
Posted by Robert Atkinson 2 on 15/06/2019 19:24:05:

For clock setting you don't need a full GPSDO. Just get ... etc.

.

With the greatest respect, Robert

[and a man with 5 GPSDO's, 4 Rubidium atomic frequency standards, HP 5370B etc. surely deserves that]

My interest is more to do with studying the stability of a pendulum than setting its rate

and I would have more confidence in my results if the crystal oscillator was 'disciplined'

Please feel free to put me right [publicly or privately] if I have this wrong ... I am a mere dabbler.

MichaelG.

Any decent OCXO will be accurate to 1 part per million in a year (<1x10^6, HP's 10811A is more than 10 times better than that) so that's 31 seconds in a year worst case. Over shorter time scales eg days it's typically a thousand times better so 0.001 seconds per day. How good do you need?

Robert G8RPI.

Robert Atkinson 216/06/2019 15:31:05
avatar
265 forum posts
17 photos

For a lowcost time interval counter with logging on a pc for checking clocks etc see

http://www.ko4bb.com/doku2015/doku.php?id=precision_timingictic

Robert G8RPI

John Haine16/06/2019 15:53:15
2543 forum posts
132 photos

Hmmm...that link doesn't work very well with an emoji in it Robert! Please could you try again?

Michael Gilligan16/06/2019 16:03:41
avatar
13253 forum posts
578 photos
Posted by Robert Atkinson 2 on 16/06/2019 15:27:06:

... How good do you need?

Robert G8RPI.

.

I honestly don't know, Robert ... that's the problem

Not that I aspire to replicating one : The Shortt observatory clocks were thought to be accurate to about one second per year, but recent tests suggested that one second in twelve years might be nearer the truth !!

What I'm really interested in though is the shorter-term stability, and response to external stimuli, of a somewhat more practical pendulum than that.

MichaelG.

SillyOldDuffer16/06/2019 17:11:50
4405 forum posts
957 photos
Posted by Michael Gilligan on 16/06/2019 16:03:41:
Posted by Robert Atkinson 2 on 16/06/2019 15:27:06:

... How good do you need?

Robert G8RPI.

.

I honestly don't know, Robert ... that's the problem

...

What I'm really interested in though is the shorter-term stability, and response to external stimuli, of a somewhat more practical pendulum than that.

MichaelG.

Michael and I are interested in the behaviour of individual pendulum swings where averaging hides useful detail. The easiest way (I think) is to compare the test pendulum with a much more accurate clock, and the more accurate the better.

With a pendulum clock, tens of microseconds are interesting, and picosecond resolution would be even more revealing. It's a bit like measuring length: to measure a thou accurately, you need an instrument that can get close to tenths.

Looking closely at the timing of a pendulum shows clock faults, for example I think I can detect the difference between a pendulum swinging true and one following an ellipse. On my test rig an elliptical swing has more jitter, I'm not sure why. I suspect it's due to a combination of the impulse being misplaced relative to the bob, torsion in the rod, and maybe bending at the pivot.

Dave

Michael Gilligan16/06/2019 17:41:09
avatar
13253 forum posts
578 photos

Thanks, Dave ... We're definitely on the same wavelength

Incidentally, my Jupiter GPS module was a few more £s than I remembered blush

The OCXO though was £16 from ebay [new old stock in 2007]

... probably totally obsolete by now !!

ocxo.jpg

MichaelG.

Bazyle16/06/2019 18:46:27
avatar
4585 forum posts
185 photos

As mentioned by Robert I can't see how the actual time, from GPS, matters when you can get a local clock that has high stability in the time period you are measuring, and if it is drifting you can measure over an extended period and compensate.

I think you would find it more instructive to be able to measure the actual physical position of the pendulum hundreds of times per swing to compare swings to see the difference made by the application of the impulse, especially when mechanically impulsed.

Michael Gilligan16/06/2019 19:47:27
avatar
13253 forum posts
578 photos
Posted by Bazyle on 16/06/2019 18:46:27:

As mentioned by Robert I can't see how the actual time, from GPS, matters when you can get a local clock that has high stability in the time period you are measuring, and if it is drifting you can measure over an extended period and compensate.

.

But ... the 'actual time' from GPS is not being used. The principle is to make use of the timebase [thoughtfully provided to make GPS function] to fine-tune the local oscillator.

As with metrology ... The more accurate and stable the reference, the easier it is to obtain precise measurements.

MichaelG.

Robert Atkinson 216/06/2019 21:47:44
avatar
265 forum posts
17 photos

Michael and I are interested in the behaviour of individual pendulum swings where averaging hides useful detail. The easiest way (I think) is to compare the test pendulum with a much more accurate clock, and the more accurate the better.

With a pendulum clock, tens of microseconds are interesting, and picosecond resolution would be even more revealing. It's a bit like measuring length: to measure a thou accurately, you need an instrument that can get close to tenths.

Looking closely at the timing of a pendulum shows clock faults, for example I think I can detect the difference between a pendulum swinging true and one following an ellipse. On my test rig an elliptical swing has more jitter, I'm not sure why. I suspect it's due to a combination of the impulse being misplaced relative to the bob, torsion in the rod, and maybe bending at the pivot.

Dave

The PICTIC I linked to (corrupted by smiley) "http://www.ko4bb.com/doku2015/doku.php?id=precision_timingictic" Google KO4BB (O not zero) pictic to find it

is ideal for what you want. run it from a 10MHz OCXO and a optocoupler obscured every pendulum swing and it will send the time of every swing to a PC via the serial port.

Robert G8RPI.

Robert Atkinson 216/06/2019 21:50:11
avatar
265 forum posts
17 photos

Deleted

 

 

Edited By Robert Atkinson 2 on 16/06/2019 21:50:34

Michael Gilligan16/06/2019 21:57:08
avatar
13253 forum posts
578 photos

< I will try that again >

Edited By Michael Gilligan on 16/06/2019 21:59:01

Michael Gilligan16/06/2019 22:00:10
avatar
13253 forum posts
578 photos

< sorry, don't know what this forum is playing at >

Edited By Michael Gilligan on 16/06/2019 22:01:46

Michael Gilligan16/06/2019 22:12:10
avatar
13253 forum posts
578 photos

Thanks, Robert yes

Irritating little blighters, those smiley things !!

The link to PICTIC is on this page: **LINK**

http://www.ko4bb.com/doku2015/doku.php

... under 'Precision Timing'

MichaelG.

Michael Gilligan16/06/2019 22:30:02
avatar
13253 forum posts
578 photos

Posted by Robert Atkinson 2 on 16/06/2019 21:47:44:

... run it from a 10MHz OCXO and a optocoupler obscured every pendulum swing and it will send the time of every swing to a PC via the serial port.

Robert G8RPI.

.

I think that's where we are probably on different interpretations, Robert

... I anticipate needing to use several measuring points; being more interested in the perturbations within a swing than in its overall duration

MichaelG.

Robert Atkinson 217/06/2019 07:31:40
avatar
265 forum posts
17 photos
Posted by Michael Gilligan on 16/06/2019 22:30:02:

Posted by Robert Atkinson 2 on 16/06/2019 21:47:44:

... run it from a 10MHz OCXO and a optocoupler obscured every pendulum swing and it will send the time of every swing to a PC via the serial port.

Robert G8RPI.

.

I think that's where we are probably on different interpretations, Robert

... I anticipate needing to use several measuring points; being more interested in the perturbations within a swing than in its overall duration

MichaelG.

that's more of a challenge as to how you capture the data without affecting the pendulum. A high frame rate camera and software would be one option. A more interesting one would be line CCD or diode array sensor like those used in document scanners sensing the shadow of the pendulum rod. This would give better resolution, easily 1024 positions, and require less processing. The sensor and lens from a flat bed scanner would probably work as is. Disconnect the sensor from the mechanical drive (initially leave that is as is for home and end sensors) and just try the scanning software. The image should be a sine wave.

I just realised I made a mistake on the OCXO accuracy. 31.5 seconds per year is 1x10^-6 if it all went out in the same direction at the beginning of the year. A more typical 1x10^-9 average is 0.03 seconds in a year.

Robert G8RPI.

Michael Gilligan17/06/2019 07:49:35
avatar
13253 forum posts
578 photos

Thanks, Robert .... I think we're on the same page now. yes

Probably best to pause for thought and let 'Plasma' have his thread back.

MichaelG.

All Topics | Latest Posts

Please login to post a reply.

Magazine Locator

Want the latest issue of Model Engineer or Model Engineers' Workshop? Use our magazine locator links to find your nearest stockist!

Find Model Engineer & Model Engineers' Workshop

Latest Forum Posts
Support Our Partners
Warco
TRANSWAVE Converters
ChesterUK
Meridienne Sept 2019
Eccentric July 5 2018
Eccentric Engineering
Ausee.com.au
emcomachinetools
Sarik
Allendale Electronics
Subscription Offer

Latest "For Sale" Ads
Latest "Wanted" Ads
Get In Touch!

Do you want to contact the Model Engineer and Model Engineers' Workshop team?

You can contact us by phone, mail or email about the magazines including becoming a contributor, submitting reader's letters or making queries about articles. You can also get in touch about this website, advertising or other general issues.

Click THIS LINK for full contact details.

For subscription issues please see THIS LINK.

Digital Back Issues

Social Media online

'Like' us on Facebook
Follow us on Facebook

Follow us on Twitter
 Twitter Logo

Pin us on Pinterest