keeping Arduino timekeeping and clock synced up
It turns out there's a handy Arduino library for time. And it will
ingest GPS or NTP, etc., as well as run off the internal clock.
One strategy, then, is:
Set the "clock" in the Arduino
then, periodically (once a minute or hour)
look up the date and time
calculate "rate"
set tick rate for external clock driver
Then you have a thing which generates 1pps ticks at the desired rate.
Right now, I have a little interrupt loop that runs once a
second(adjusted by rate) do it.
One could also calculate "solar time" (as UTC + EOT offset) repeatedly
(after all, there's nothing else for the Arduino to do) and whenever the
seconds changes, send a tick to the clock.
This technique doesn't try to keep the integral of ticks aligned with
"UTC + EOT offset", though. If the tick rate were slightly off
(roundoff errors in the math, most likely), then there will be some
relative drift.
So, periodically, one would need to reset both the analog clock AND the
Arduino clock to bring them back to proper alignment.
I suppose that periodically, one could compare "number of ticks sent"
with "UTC + EOT offset" and try to compensate (by dropping ticks or
adding them).
So, periodically, one would need to reset both the analog clock AND the
Arduino clock to bring them back to proper alignment.
I suppose that periodically, one could compare "number of ticks sent"
with "UTC + EOT offset" and try to compensate (by dropping ticks or
adding them).
And then you'd have a GPSDST (GPS disciplined solar time) clock...
While you're at it, add a rotary switch and allow the (JPL) user to select which planet's solar time they want to display. Since there are now only 8 planets you can also do it with a 3-bit configuration switch (now you know the real reason Pluto was demoted).
If your project works ok for the earth clock, the next step is a jaw-dropping array of 8 (9) clocks in a JPL lobby showing the differently ticking solar time for each planet. Use 24h clocks for best results. They can be had from www.clockkit.com, an excellent source of DIY quartz clock parts.
/tvb
www.leapsecond.com
On 1/19/14 1:51 PM, Tom Van Baak wrote:
So, periodically, one would need to reset both the analog clock AND the
Arduino clock to bring them back to proper alignment.
I suppose that periodically, one could compare "number of ticks sent"
with "UTC + EOT offset" and try to compensate (by dropping ticks or
adding them).
And then you'd have a GPSDST (GPS disciplined solar time) clock...
While you're at it, add a rotary switch and allow the (JPL) user to select which planet's solar time they want to display. Since there are now only 8 planets you can also do it with a 3-bit configuration switch (now you know the real reason Pluto was demoted).
If your project works ok for the earth clock, the next step is a jaw-dropping array of 8 (9) clocks in a JPL lobby showing the differently ticking solar time for each planet. Use 24h clocks for best results. They can be had from www.clockkit.com, an excellent source of DIY quartz clock parts.
You have divined my ultimate goal..Display local solar time for every
lander, for instance.
Except at work, I'd get a bunch of 3325Bs (since we have pallet loads of
them around), driven from the maser based reference, and have a PC
sending GPIB commands to them to adjust the rates.
Handy to have a source for 24hr clock movements. Last time, I bought a
couple MFJ 24 hour clocks at the local Ham Radio Outlet..
To be honest, one of the interesting challenges is dealing with power
failures in these kinds of systems. The Arduino is not a low power
device..(at least not in the AA battery for 2 years sense).
So, do you run the whole thing off 12V (which is what I'm going to do)
and a float charged battery OR do you do something clever like detect
when power is failing and save it in NV storage, then when you come back
up, you send a bunch of clock ticks real fast to catch up.
Our lobby isn't that big at JPL.. I'll probably hang them down the hall
outside my office or something. I had the Mars clock outside my office
(with a 50 ft coax to the signal generator under my desk) for about a
year before I had to take it down in the annual "clean everything up"
festival.
So, do you run the whole thing off 12V (which is what I'm going to do)
and a float charged battery OR do you do something clever like detect
when power is failing and save it in NV storage, then when you come back
up, you send a bunch of clock ticks real fast to catch up.
Use a high-res Arduino web cam facing the wall of clocks and write OHR (Optical clock-Hand Recognition) code. That way its a single robust solution for setting it the first time, restarting it on power failure, resyncing after replacing a failed clock, or self-correcting after any mechanical glitch. You can also use the same video feed to show off the project live on the JPL web site.
/tvb
On 1/19/14 4:10 PM, Tom Van Baak wrote:
So, do you run the whole thing off 12V (which is what I'm going to do)
and a float charged battery OR do you do something clever like detect
when power is failing and save it in NV storage, then when you come back
up, you send a bunch of clock ticks real fast to catch up.
Use a high-res Arduino web cam facing the wall of clocks and write OHR (Optical clock-Hand Recognition) code. That way its a single robust solution for setting it the first time, restarting it on power failure, resyncing after replacing a failed clock, or self-correcting after any mechanical glitch. You can also use the same video feed to show off the project live on the JPL web site.
No, a low res cam with a robotic arm that moves it in front of each
clock in turn..
That is what is called "scope creep"..
I'll be happy if I get ONE clock running reasonably..
I've got the Arduino code running that does the EOT, once a second (but
using delay(1000) not the ISR), calculates the rate estimate, and
accepts a sync command over the (emulated) serial port to set the time
and date.
Curse the folks who develop "processing" because the current version
supports Mac OSX 10.7 and later, but not 10.6, which I am using, so I
don't have the nifty "click here to sync the Arduino" routine that's
provided as an example with the Arduino Time library.
Next I have to integrate the code I've got now with the other sketch
that does the ISR off the hardware timer.
#include <Time.h>
// Solar clock to drive mechanical mechanism
// Jim Lux, 19 Jan 2014
#include <math.h>
#define TIME_MSG_LEN 11 // time sync to PC is HEADER followed by Unix
time_t as ten ASCII digits
#define TIME_HEADER 'T' // Header tag for serial time sync message
#define TIME_REQUEST 7 // ASCII bell character requests a time sync
message
// T1262347200 - sample sync message
const double refclk=31376.6; //16 MHz/510?
const int clkpin1=6; // pins going to external clock
const int clkpin2= 7;
int dd, hh; //current day and hour
boolean UpdateClockFlag; // tells loop() that an interrupt has occurred
void setup(){
pinMode(clkpin1,INPUT); // set pins as inputs High Z for now.
pinMode(clkpin2,INPUT);
Serial.begin(9600);
delay(1000);
UpdateClockFlag = false;
}
void loop(){
int DOY;
double e1,e2;
double secsperday,ratedelta;
time_t t;
t = now(); // get the time
if(Serial.available() )
{
processSyncMessage();
}
delay(1000); // hack, til we get ISR timer running
UpdateClockFlag = true; // hack
if (UpdateClockFlag) {
if(timeStatus() == timeNotSet)
Serial.println("waiting for sync message");
else {
DOY = DayWeekNumber(year(),month(),day(),weekday());
hh = hour();
e1 = eot(DOY,hh); // EOT in minutes
e2 = eot(DOY,hh+1);
secsperday = (e2-e1)*1440;
ratedelta = secsperday*1.E6/86400; //ppm for now,
// but we'll change to
divisor later
Serial.print(ratedelta); Serial.print(" ");
digitalClockDisplay();
// code in here to update interrupt divisor, etc.
};
UpdateClockFlag = false;
}
}
// equation of time code from Tom Van Baak
//http://www.leapsecond.com/tools/eot1.c
double eot(int day,int hour){
double Pi = 4 * atan(1);
double y = (2 * Pi / 365.0) * (day - 1 + (hour - 12) / 24.0);
double eqtime = 229.18 *
( 0.000075
+ 0.001868 * cos(y)
- 0.032077 * sin(y)
- 0.014615 * cos(2*y)
- 0.040849 * sin(2*y)
);
return(eqtime);
}
void processSyncMessage() {
// if time sync available from serial port, update time and return true
while(Serial.available() >= TIME_MSG_LEN ){ // time message
consists of header & 10 ASCII digits
char c = Serial.read() ;
Serial.print(c);
if( c == TIME_HEADER ) {
time_t pctime = 0;
for(int i=0; i < TIME_MSG_LEN -1; i++){
c = Serial.read();
if( c >= '0' && c <= '9'){
pctime = (10 * pctime) + (c - '0') ; // convert digits to a
number
}
}
setTime(pctime); // Sync Arduino clock to the time received on
the serial port
}
}
}
void digitalClockDisplay(){
// digital clock display of the time
Serial.print(hour());
printDigits(minute());
printDigits(second());
Serial.print(" ");
Serial.print(day());
Serial.print(" ");
Serial.print(month());
Serial.print(" ");
Serial.print(year());
Serial.println();
}
void printDigits(int digits){
// utility function for digital clock display: prints preceding colon
and leading 0
Serial.print(":");
if(digits < 10)
Serial.print('0');
Serial.print(digits);
}
int DayWeekNumber(unsigned int y, unsigned int m, unsigned int d,
unsigned int w){
int days[]={0,31,59,90,120,151,181,212,243,273,304,334}; // Number
of days at the beginning of the month in a not leap year.
//Start to calculate the number of day
int DOY;
if (m==1 || m==2){
DOY = days[(m-1)]+d; //for any type of year, it
calculate the number of days for January or february
} // Now, try to calculate for the other months
else if ((y % 4 == 0 && y % 100 != 0) || y % 400 == 0){ //those are
the conditions to have a leap year
DOY = days[(m-1)]+d+1; // if leap year, calculate in the same
way but increasing one day
}
else { //if not a leap year, calculate
in the normal way, such as January or February
DOY = days[(m-1)]+d;
}
return DOY;
// Now start to calculate Week number
// if (w==0){
// WN = (DOY-7+10)/7; //if it is sunday (time library
returns 0)
// }
// else{
// WN = (DOY-w+10)/7; // for the other days of week
// }
}
Hi
Rather than calculating the ppm offset, calculate the number of ticks until you drop (or add) one tick. Your output pps can only be offset from the input pps by an integer number of ticks anyway. The next decision would be - how big a tick can you get away with? For a wall clock, 100 ms is in the ball park of begin to big. 10 ms is small enough that it’s below what a wall clock needs.
Bob
On Jan 19, 2014, at 7:38 PM, Jim Lux jimlux@earthlink.net wrote:
On 1/19/14 4:10 PM, Tom Van Baak wrote:
So, do you run the whole thing off 12V (which is what I'm going to do)
and a float charged battery OR do you do something clever like detect
when power is failing and save it in NV storage, then when you come back
up, you send a bunch of clock ticks real fast to catch up.
Use a high-res Arduino web cam facing the wall of clocks and write OHR (Optical clock-Hand Recognition) code. That way its a single robust solution for setting it the first time, restarting it on power failure, resyncing after replacing a failed clock, or self-correcting after any mechanical glitch. You can also use the same video feed to show off the project live on the JPL web site.
No, a low res cam with a robotic arm that moves it in front of each clock in turn..
That is what is called "scope creep"..
I'll be happy if I get ONE clock running reasonably..
I've got the Arduino code running that does the EOT, once a second (but using delay(1000) not the ISR), calculates the rate estimate, and accepts a sync command over the (emulated) serial port to set the time and date.
Curse the folks who develop "processing" because the current version supports Mac OSX 10.7 and later, but not 10.6, which I am using, so I don't have the nifty "click here to sync the Arduino" routine that's provided as an example with the Arduino Time library.
Next I have to integrate the code I've got now with the other sketch that does the ISR off the hardware timer.
#include <Time.h>
// Solar clock to drive mechanical mechanism
// Jim Lux, 19 Jan 2014
#include <math.h>
#define TIME_MSG_LEN 11 // time sync to PC is HEADER followed by Unix time_t as ten ASCII digits
#define TIME_HEADER 'T' // Header tag for serial time sync message
#define TIME_REQUEST 7 // ASCII bell character requests a time sync message
// T1262347200 - sample sync message
const double refclk=31376.6; //16 MHz/510?
const int clkpin1=6; // pins going to external clock
const int clkpin2= 7;
int dd, hh; //current day and hour
boolean UpdateClockFlag; // tells loop() that an interrupt has occurred
void setup(){
pinMode(clkpin1,INPUT); // set pins as inputs High Z for now.
pinMode(clkpin2,INPUT);
Serial.begin(9600);
delay(1000);
UpdateClockFlag = false;
}
void loop(){
int DOY;
double e1,e2;
double secsperday,ratedelta;
time_t t;
t = now(); // get the time
if(Serial.available() )
{
processSyncMessage();
}
delay(1000); // hack, til we get ISR timer running
UpdateClockFlag = true; // hack
if (UpdateClockFlag) {
if(timeStatus() == timeNotSet)
Serial.println("waiting for sync message");
else {
DOY = DayWeekNumber(year(),month(),day(),weekday());
hh = hour();
e1 = eot(DOY,hh); // EOT in minutes
e2 = eot(DOY,hh+1);
secsperday = (e2-e1)*1440;
ratedelta = secsperday*1.E6/86400; //ppm for now,
// but we'll change to divisor later
Serial.print(ratedelta); Serial.print(" ");
digitalClockDisplay();
// code in here to update interrupt divisor, etc.
};
UpdateClockFlag = false;
}
}
// equation of time code from Tom Van Baak
//http://www.leapsecond.com/tools/eot1.c
double eot(int day,int hour){
double Pi = 4 * atan(1);
double y = (2 * Pi / 365.0) * (day - 1 + (hour - 12) / 24.0);
double eqtime = 229.18 *
( 0.000075
+ 0.001868 * cos(y)
- 0.032077 * sin(y)
- 0.014615 * cos(2*y)
- 0.040849 * sin(2*y)
);
return(eqtime);
}
void processSyncMessage() {
// if time sync available from serial port, update time and return true
while(Serial.available() >= TIME_MSG_LEN ){ // time message consists of header & 10 ASCII digits
char c = Serial.read() ;
Serial.print(c);
if( c == TIME_HEADER ) {
time_t pctime = 0;
for(int i=0; i < TIME_MSG_LEN -1; i++){
c = Serial.read();
if( c >= '0' && c <= '9'){
pctime = (10 * pctime) + (c - '0') ; // convert digits to a number
}
}
setTime(pctime); // Sync Arduino clock to the time received on the serial port
}
}
}
void digitalClockDisplay(){
// digital clock display of the time
Serial.print(hour());
printDigits(minute());
printDigits(second());
Serial.print(" ");
Serial.print(day());
Serial.print(" ");
Serial.print(month());
Serial.print(" ");
Serial.print(year());
Serial.println();
}
void printDigits(int digits){
// utility function for digital clock display: prints preceding colon and leading 0
Serial.print(":");
if(digits < 10)
Serial.print('0');
Serial.print(digits);
}
int DayWeekNumber(unsigned int y, unsigned int m, unsigned int d, unsigned int w){
int days[]={0,31,59,90,120,151,181,212,243,273,304,334}; // Number of days at the beginning of the month in a not leap year.
//Start to calculate the number of day
int DOY;
if (m==1 || m==2){
DOY = days[(m-1)]+d; //for any type of year, it calculate the number of days for January or february
} // Now, try to calculate for the other months
else if ((y % 4 == 0 && y % 100 != 0) || y % 400 == 0){ //those are the conditions to have a leap year
DOY = days[(m-1)]+d+1; // if leap year, calculate in the same way but increasing one day
}
else { //if not a leap year, calculate in the normal way, such as January or February
DOY = days[(m-1)]+d;
}
return DOY;
// Now start to calculate Week number
// if (w==0){
// WN = (DOY-7+10)/7; //if it is sunday (time library returns 0)
// }
// else{
// WN = (DOY-w+10)/7; // for the other days of week
// }
}
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
From: Jim Lux
[]
To be honest, one of the interesting challenges is dealing with power
failures in these kinds of systems. The Arduino is not a low power
device..(at least not in the AA battery for 2 years sense).
[]
Jim,
For low-power, long life, battery operation, you might take a look at the
JeeNode. Very similar to Arduino, but he's had the thing running on
batteries for years - he's done a lot of work to minimise power consumption.
http://jeelabs.net/projects/hardware/wiki/JeeNode
"3 years on one set of batteries"
http://jeelabs.org/2013/09/08/3-years-on-one-set-of-batteries/
Cheers,
David
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@blueyonder.co.uk
On 1/19/14 1:51 PM, Tom Van Baak wrote:
Use 24h clocks for
best results. They can be had from www.clockkit.com, an excellent
source of DIY quartz clock parts.
I couldn't find 24hr movements on the clockkit.com site.. where are they?
On 1/20/14 10:06 AM, Jim Lux wrote:
On 1/19/14 1:51 PM, Tom Van Baak wrote:
Use 24h clocks for
best results. They can be had from www.clockkit.com, an excellent
source of DIY quartz clock parts.
I couldn't find 24hr movements on the clockkit.com site.. where are they?
Hi
So really it boils down to :
http://www.klockit.com/products/dept-379__sku-bbbii.html
Since that already has a full driver on it (battery / oscillator / chip) - you will need to do some surgery to get directly at the stepper motor. (or am I missing something?)
Bob
On Jan 20, 2014, at 1:17 PM, Jim Lux jimlux@earthlink.net wrote:
On 1/20/14 10:06 AM, Jim Lux wrote:
On 1/19/14 1:51 PM, Tom Van Baak wrote:
Use 24h clocks for
best results. They can be had from www.clockkit.com, an excellent
source of DIY quartz clock parts.
I couldn't find 24hr movements on the clockkit.com site.. where are they?
http://www.klockit.com/depts/SpecialtyClockMovements/dept-379.html
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.