om*g lol i cannot believe what i am observing here on my XP system. last night i synchronized my WinXP clock with that well-known
Win-dialog function, using the "time.nist.gov" server from the drop-down menu. successful. And i was assuming that that clock would be 100% equal to big ben (or my smartphone). so i started my mc3k Refresh program at 22:00:00 o'clock (10p.m. sharp!).
I left my PC running over night. Now, only 12 hours later, at 10a.m., i am seeing that my WinXP clock is off by almost 3minutes, ahead!, compared with my smartphone!!!
I can't believe it, because if this continued, it would mean that after 24 hours the WinXP clock would show 22:06 o'clock instead of big ben's 22:00 o'clock. Omg, how inexact is the WinXP clock? Or is it my PC hardware? LOL!!
So i did again the
Win-dialog function and watched what happened: Indeed, the WinXP clock was automatically set back by 3mins, now again matching the clock on my smartphone. OM*G!
Wow, that was surprising. I am sure that none of you ever came up with the idea to clock/time/stop your WindowsPC clock running for 1-2 days straight with an external stopwatch/chronograph. I had always thought that my PC clock is nearly as exact like a Swatch or big ben!! By far not! 3min within 12 hours is insane! And hard to believe. And maybe you cannot reproduce this unbelievable result with your brand new Win10 system and brand new hardware. My hardware is from 2010 hh.
Usually at night, i turn off my WinPC completely. The next day, when i boot up the machine, i am guessing that the PC automatically synchronizes time with the time.nist.gov server, and that's how the time stays ~exact for the day being. But the longer the WinPC is running, 10, 24, 30hrs straight, without reboot, you'll see how off the PC clock gets compared to actual time like big ben.
Now the good news. The timestamp on dex log and the lcd timer do represent big ben timing. "The three" (dex, lcd, big ben) were synchronous at 10p.m. yesterday. Right now, the LiIon Refresh program is still running, taking a "snap shot" of the three timers gives:
snapshot said:
pc (right now synchronized with time.nist.gov server): 10:35:00
dex: 2016-03-09 10:35:02.xxx
lcd: TOTAL: 0†12:35:07
So there you have it. Compared to big ben, the mc3k lcd clock runs 7sec faster every ~12hrs (big ben hours). And the time stamp in the dex log is still exact within 1-2sec (also faster!) after 12 big ben hours. After another 12 big ben hours, the dex log should be 3-4sec ahead, and the lcd clock 13-15sec ahead.
In summary, all three timers do slowly drift away from each other. The lcd clock is ~14sec off after 24 big ben hours. I'll accept it. The MSRP of a Swatch is 50,00€. We can't expect a Swiss watch to be ticking in the MC3000.
Apart from the hilarious WinXP clock performance, the other interesting point we're learning in this post is that DEX uses its own "independent" timer and which is more truthful to big ben than the lcd clock. Logically the dex log will look 5sec short (after 12 big ben hours, if Finished) compared with the lcd Finish time.
The next question would be: how does PCLS do the clocking? Does it copy over lcd time, pc time, or does it have its own timer like dex? From my impression, pcls time is very much a copy of lcd time, maybe ±1sec. The *.CSV-log is always few seconds short though, missing a few rows before Finish. That's why i had suggested to make use of C.RESTING and D.RESTING, when doing drop dead serious test runs ("C>", "D>"): If some last rows of the resting routine are missing in the CSV-log, then no prob.
All in all, we have five timers, don't we?
- Big Ben / radio controlled clock
- WinPC clock
- dex log
- lcd timer
- pcls timer
Summary:
In a perfect world, all 5 would be running synchronously and showing the exact same progress of time. In reality, even Swiss watches need to be re-synchronized after a few months. Among the 5 listed, the worse is the WinPC clock (on my system, at least!). Luckily, neither dex log nor pcls (nor the lcd timer) use the WinPC clock as basis/reference. The mc3k internal clock introduces an error of ~7sec (per 12 big ben hours). The dex log introduces less of such an error, but the dex log might look "incomplete", when you look at the logged time period; Excel gives you the chance though to re-calculate mAh and compare it with the logged mAh value. The truthful value would lie in between i guess. Also note that this post was about logging for straight 12-24 hours. That's when the effect of "diverging timers" begins to play a (minor) role. If you're logging a Discharge for 6 hours, i wouldn't care about a displayed mAh value being "3.5sec off the truth".
1000mA*(3.5/3600)h = 0.97222222222mAh
I think we're good here. DEX does a great job with the timing and logging. PCLS needs to be fixed with respect to the missing 3-10sec in the CSV-log. And i am accepting that MC3000 does not have a 50€ Swatch inside as clocker. :laughing: