Timing

From: Bob helen mcclure at aol dot com
Date: Fri, 18 Jun 2004 21:53:10 EDT
Subject: Time marking for Dataq

Hi John and Chris,

I have attached a schematic and photo of my once-per-minute time marker generator. The time source is the second hand of a WWVB controlled wall clock. The second hand has an aluminum foil reflector glued near its tip end. The sensor block is black-painted wood, with two angled holes pointing at the space where the second hand is on each minute. A phototransistor and amplifying npn transistor is mounted in one hole, and an LED is mounted in the other. The block is held in place on the clock face with double-sided tape and secured with a string and rubber band girdling the clock. Black electrical tape is placed around the block and on the clock face under the block to eliminate any interference from ambient light. The angled mounting holes are reduced in diameter at their bottom ends to limit the field of view of the phototransistor and LED.

The photransistor is in a metal case. If one uses one mounted in plastic, it will be necessary to paint most of it black or otherwise shield it from stray ambient light.

No power source is needed for the phototransistor circuit. The voltage coming out of the digital input terminal of the Dataq A/D is all that is needed. The LED of course requires DC power with a series resistor to limit the current to a proper value. I use 5 volts, and a series resistance of 3K. One must experiment with block placement and LED current to get a timing pulse above logic threshold when high, and below threshold when the second hand crosses the minute. I parallel the digital terminal of my DI-154 with the Channel 4 terminal and view that channel at a high sampling rate to view the timing pulse when I do the block alignment.

My clock aligns itself with WWVB at 2am, 10am, 2pm, and 10pm. Then, the timing may be off for several minutes.

Regards,

Bob McClure




 

Follow-up Discussions

In a message dated 6/19/2004, Chris At Upw writes:

I wondered why you did not use one of the readily available photo reflector switch modules? I seem to remember that the HP one even has integral lenses. Accurate modules are made for reading bar code strips

Hi Chris,

At the time, I was unaware of a good pickoff module that was within my budget. I built what I did for pennies out of parts on hand. What I started trying to do at first was photo-electrically pick up the lower right hand segment of the 10's of seconds digit on a WWVB-controlled LCD clock. That segment goes blank for 10 seconds on the 20 second mark of every minute. My seismo buddy Victor was successful at doing this, but on a clock with very large digits. I am not claiming that my solution is the best, only that it works for me.

My PC clock is absolutely weird. If the PC is shut down, it gains about 6 seconds per day. When the PC is running, it keeps losing seconds sporadically, depending on the processing burden being imposed on it. I do not know whom to ask for an explanation of how this happens. I use Windows Millenium Edition.

I cannot use it at all for seismic time keeping. I have a Visual Basic program I wrote which displays time and date. If you click on the time, the PC clock is advanced by one second from what it was reading at the time of the click. If you click on the date, the PC clock is reset to the time displayed at the click, which means it gets retarded by whatever time elapsed since update to that second. It is great for making fine adjustments to the PC clock.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

An earlier exchange on this topic

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Hi John,

Two old letters to the PSN List follow. One is mine, and the other is from Jonathon Peakall. His schemes are good, except that it is harder to determine which second you are on than which minute you are on.

Cheers,

Bob


Subj: Subject: Re: Atomic Time Source for data logging
From: Bob
Date: Sun, 23 Feb 2003

My nearby seismic buddy and I have established a local area tripartite network for the study of microseisms. We record for short periods simultaneously at three sites, and require accurate timing of the three records in order to calculate time differences of arrival. We use Dataq RS-154 and RS-194 digitizers which have a digital input for registering time marks on the records. We use WWVB controlled wall clocks for generating minute marks.

My buddy uses large LCD RadioShack clocks, and photo-optically looks at the lower right segment of the tens-of-seconds digit. This segment is black for all digits except for the digit "2," so every minute it disappears at 20 seconds past the minute for 10 seconds. His pickup circuit has a logic high output whenever the segment is black and low when it is absent. The Dataq recorder will make a mark at either of the transitions. He uses the high-to-low transitions, so gets minute marks 20 seconds past the minute.

I use an analog wall clock, brand name "Atomix," distributed by Cheney Instruments. I removed the plastic face, glued a .25-inch circle of aluminum foil reflector to the sweep second hand, and replaced the face. The pickoff is held in place by a rubber band and string, and is positioned on the clock face to detect the second hand when it is at the even minute. The pickoff circuitry is mounted in a wooden block, painted black. The light source is a red LED and the detector is a phototransistor, coupled to a medium-gain NPN transistor. There are some potential problems with this scheme. For one thing, the clock goes through an adjustment procedure every few hours to synchronize with WWVB. It starts running fast and then slows down to fall into alignment with the radio signal. Except for the few minutes when this is going on, the clock is very accurate. Another potential problem is that the second hand does not simply jump from one second to the next, but overshoots before settling down. The pickoff must be carefully positioned to avoid sensing the bounce from second 59 and well as the dwell at second zero. I purchased two of these clocks from HeartlandAmerica, and one of them had too much bounce to be useful.

We have tested both marking schemes at the same site on a single digitizer, and have determined that there is no detectable differences in our timing (except for the fixed 20-second offset.)

With analog wall clocks, it is also possible to generate one-second marks by inductively picking up the clock's stepper motor's current pulse.

If anyone is interested in trying the above schemes, I will be happy to give further details.

Bob McClure
----
Subject: Re: Subject: Re: Atomic Time Source for data logging
From: "Jonathan Peakall" <jpeakall at mad labs dot info>
Date: Sun, 23 Feb 2003

Bob, All,

Following Bob's ideas and with his help, I made basically the same system. The only difference was I used an photogate of the style that is used in floppy drives, and painted the second hand white. Works like a champ and is cheap. I put the gate on an arm that pivots on the "hands holder", and clamps on the edge of the face for easy adjustment.

Later however, when I was trying to get the actual time and date from the unit, I realized how easy it was to pull a direct 1PPS from the unit. I then got rid of the photogate and used that to make marks via the DATAQ 194-RS. This is much better, no bounce. I eliminated the actual clock mechanism, and now the battery lasts forever. Mind you, it isn't very useful as a clock anymore! Bob, if you are interested in details let me know. It's very simple surgery.

The main reason for going to all the trouble I did for atomic time is I wanted to be able to make more than minute marks on my log, and I have a number of projects that can use an accurate time base. Since I will have at least 4 devices using the timebase, it was a reasonably economical solution as well, I have a total of less than US$100 into the atomic time broadcast system. A clock chip for all my projects would have been about the same or a little less, but then I don't have to use any buttons and I/O pins used (except one I/O pin) or use program space on time setting routines.

Another reason is the cheap clock I was using didn't link very often to WWVB. This is due to my poor reception. The module I am using now is much better.

Now all I have to do is weatherize my seismic sensor and get back to work on it. The school year is crazy for me, so I hope to get it all set up this summer, if I don't have to move.

Regards,

Jonathan
www.madlabs.info

BACK