 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : (MM/DM) Unix time for data logging
| Author | Message | ||||
| DuinoMiteMegaAn Senior Member  Joined: 17/11/2011 Location: AustraliaPosts: 231 |
(MM/DM) Unix time for elapse time data logging. A quick search for any Unix algorithms for GW-Basic on the Internet brings up almost nothing. Time / date stamping of data should be done with a "elapse time" reference instead of a time / date reference. Unix time - the time in seconds from 1970 is the "almost" perfect choice. Note: In 2038 Unix 32 bit counter will rollover! Note: In the Maximite/DuinoMite there is a millisecond timer but the "long term" accuracy and rollover are in question? Note: I am planning to use a precision RTC that refreshes the PIC32 RTC for precise datalogging. Anyone on this board seen any GW-Basic algorithms for converting to/from a time/date stamp to Unix time? |
||||
ksdesigns Senior Member Joined: 25/06/2011 Location: United StatesPosts: 143 |
I have been working on a sleep command using the pic32 internal rtcc alarm interrupt you can set it to wake up as far as once a year . On the DuinoMite-Mega running with 12v in the current draw is about 80ma and in sleep mode, drops to about 8ma . Anyway if you have a external precision RTC you can use it to calibrate the internal RTCC i would like to see what the long term drift of the internal RTCC.. ken |
||||
| DuinoMiteMegaAn Senior Member Joined: 17/11/2011 Location: AustraliaPosts: 231 |
Anyway if you have a external precision RTC you can use it to calibrate the internal RTCC i would like to see what the long term drift of the internal RTCC..
The internal RTCC, I believe, uses the "32768 hz watch crystal" which is no better than +- 20 PPM. (Greater than -+10 minutes per year) If this "watch crystal" is exposed to anything other than ambient temperature, this +-10 minute error will increase/decreased dramatically. Of course, you could calibrate the RTCC, with expensive equipment, for better accuracy but what would the calibration temperature be when you are using it for data logging where the ambient temperatures varies a lot? You need a +- 2 ppm temperature controlled oscillator precision RTC - Maxim-IC DS3231 which is code compatible with the DS1307. The drift would only be +-1 minute per year under large temperature extremes! Having this precision RTC with the flexible power savings on the Duinomite would be the perfect data logging choice. 
Note: You need to use only a 3.3 VDC RTC (Not +5 VDC) for the DuinoMite Mega if you are using the Lipo power source for data logging! BTW ... Olimex could easily produce a DS3231 UEXT board. This SMD SOIC16 only chip is $7.0 USD (expensive) but it is well worth it.
There are, on the internet, two breakout boards containing the DS3231. One is from Macetech US (ChronoDot) and the other is from Love Electronics UK. Both of them cost about ~ $15 USD which I believe is outrageous! I got both of them working on the DuinoMite Mega. |
||||
| DuinoMiteMegaAn Senior Member Joined: 17/11/2011 Location: AustraliaPosts: 231 |
i would like to see what the long term drift of the internal RTCC..
Just take the board out of ambient temperature and you will see.
Of course, set the clock to a known good reference. Again, has anyone seen an Unix time algorithm in GW-Basic? |
||||
TassyJim Guru  Joined: 07/08/2011 Location: AustraliaPosts: 6097 |
|
||||
| DuinoMiteMegaAn Senior Member Joined: 17/11/2011 Location: AustraliaPosts: 231 |
Here is an unix algorithm, I got from the Internet and converted to MMBasic. I believe MMBasic forces an error of 16 seconds. What the author did was to shift the seconds base from 1970 to 2000. Then he added (SECONDS_FROM_1970_TO_2000) 946684800 seconds to make it unix time. Maybe we could shift the time base to 2010? Test code only - possible bugs! Note: Leap seconds not accounted for? 10 'Code by JeeLabs http://news.jeelabs.org/code/
20 'Released to the public domain! Enjoy! 30 ' - Thanks Jean C Wippler! 40 'RTClib.cpp was modified from a C++ lib. to MMBasic by DuinoMiteMegaAndy 50 'https://github.com/adafruit/RTClib 60 ' 70 '--------UNIX_R1.bas---------------------------- 80 '--------Unix time utilities for elapse time data logging -------------- 90 'Manual prompt input year/month/day/hours/minutes/seconds 100 'Generates Unix time (in seconds since year 1970) for data logging 110 ' 120 ' 130 'Author: Hacker - DuinoMiteMegaAndy 140 'Date: 27/11/2011 150 'Version: Rev. 1.0b 160 'MMBasic Firmware: 2.7 DuinoMite Mega 170 'Platform: Win XP/SP3 180 ' 190 ' 200 ' --------------Code Begin-------------------- 210 SECONDS_PER_DAY = 86400 220 SECONDS_FROM_1970_TO_2000 = 946684800 230 ' 240 ' 250 'Number of days since 2000/01/01, valid for 2001..2099 260 DATA 0,31,28,31,30,31,30,31,31,30,31,30,31 270 DIM days(13) '12 months 280 INPUT "Year yyyy"; Year 290 INPUT "Month 1-12"; Month 300 INPUT "Day 1-31"; Day 310 TempD = 0 320 IF year >= 2000 THEN year = year - 2000 330 FOR I=0 TO 12 'put data in days array 340 READ Days(I) 350 '? "days ",days(I) 'data in array 360 NEXT I 370 IF Month >= 1 OR Month <= 12 THEN GOSUB 530 380 '? " total days before leap year ",tempD 390 IF (Month > 2 AND Year MOD 4 = 0) THEN tempD = tempD + 1 'leap year? 400 '? " total days after leap year ",tempD 410 day_total = tempD 420 day_total = FIX(day_total + (365 * Year) + (Year + 3) / 4 - 1) 430 ? "Number of days since 2000/01/01 ",day_total 440 INPUT "Hour 0-23 ";Hour 450 INPUT "Minute 0-59 ";Minute 460 INPUT "Second 0-59 ";Second 470 Unix_Time = day_total * 24 471 ? " Unix Time hours ": ? FORMAT$(Unix_Time,"%10.1f") 472 Unix_Time = Unix_Time + hour 473 ? " Unix Time + hours ": ? FORMAT$(Unix_Time,"%10.1f") 474 Unix_Time = Unix_Time * 60 475 ? " Unix Time Minutes ": ? FORMAT$(Unix_Time,"%10.1f") 476 Unix_Time = Unix_Time + minutes 477 ? " Unix Time +Minutes ":? FORMAT$(Unix_Time,"%10.1f") 478 Unix_Time = Unix_Time * 60 479 ? " Unix Time Seconds ": ? FORMAT$(Unix_Time,"%10.1f") 480 Unix_Time = Unix_Time + seconds 481 ? " Unix Time +Seconds From Year 2000 on 1/1 ": ? FORMAT$(Unix_Time,"%10.1f") 482 Unix_Time = Unix_Time + 946684800 490 ? "Unix Time Seconds From Year 1970 on 1/1 ": ? FORMAT$(Unix_Time,"%10.1f") 500 ' 510 END 520 ' 530 FOR I=1 TO Month - 1 540 tempD = days(I) + tempD 550 '? "array add ",tempD 560 NEXT I 570 tempD = tempD + day 580 RETURN 590 ' 600 ' 610 END |
||||
| TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6097 |
A couple of typos: 450 INPUT "Minute 0-59 ";Minute 460 INPUT "Second 0-59 ";Second You Minute and second but later they gain an 's' 476 Unix_Time = Unix_Time + minutes Even with that fixed, try entering two times one minute apart and subtract the two 'unix times' to see how many seconds there are. You will get different answers depending on whether you use time since 200 or 1970. Neither will be 60 seconds. It is hard to get away from the lack of significant figures in 32bit numbers. You have to keep the numbers to a max of 7 digits. You CAN do integer maths but it is much easier to simply keep the whole days and the part days separate and forget about unix time. Jim VK7JH MMedit  MMBasic Help |
||||
| TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6097 |
Try this simple code: 100 bignumber=9999999 110 biggernumber=bignumber*100 120 nextnumber=biggernumber+1 130 print "Starting number ";format$(bignumber,"%11.0f") 140 print "Multiplied by 100 ";format$(biggernumber,"%11.0f") 150 print "adding one ";format$(nextnumber,"%11.0f") Jim VK7JH MMedit  MMBasic Help |
||||
| DuinoMiteMegaAn Senior Member Joined: 17/11/2011 Location: AustraliaPosts: 231 |
It is hard to get away from the lack of significant figures in 32bit numbers.
You have to keep the numbers to a max of 7 digits. 9,999,999 <--- Max for MMBasic PIC32 32 bit 
and forget about unix time.
I will just use a date/time stamp string for data logging and let PC Excel take care of the conversion. |
||||
| DuinoMiteMegaAn Senior Member Joined: 17/11/2011 Location: AustraliaPosts: 231 |
@TassyJim Explain this from my Unix simulator? Inputs
yr 2011 mon 12 day 15 Hr. 0 min 0 sec 0 Calculated Unix time from MMBasic (in seconds) 1323907200.0 12 /15 / 2011 <---------- Calculator on Internet - "Spot ON" Hour : Minute : Second 0 : 0 : 0 ================================================== Inputs yr 2011 mon 12 day 16 Hr. 0 min 0 sec 0 Calculated Unix time from MMBasic (in seconds) 1323993600.0 12 /16 / 2011 <---------- Calculator on Internet - "Spot ON" Hour : Minute : Second 0 : 0 : 0 =================================================== The moment I add any " RTC time offset" to the calculation the calculation is wrong - probably due to your 7 digit max. limit in MMBasic? So how does both date calculations (without time offset) compute "SPOT ON" when there are numbers larger than 7 digits in the calculation? BTW ... From the Olimex pdf manual "limits" The range of integers (whole numbers) than can be manipulated without loss of accuracy is ±16777100. |
||||
 Print this page |
