One month later...
Rec 24-mar-2008 21:49
Geez, it's been already one month since I last wrote here. Actually, I have been updating the site -- there
are new sections, new translations and alas, even new content
But that's just boring housekeeping. Sorry, I've been extra-busy with some projects at work. You'll hear
about it soon -- there's really cool stuff coming out of it.
I thought some of you might like to know my experiences with the power glitch monitor.
Well, it has faithfully recorded a quite a few power interruptions, listed below:
| Date/Time || Duration |
| 2008-03-02 21:40:06 || 0:00:07 |
| 2008-03-03 12:47:22 || 0:09:03 |
| 2008-03-12 01:53:45 || 0:00:07 |
| 2008-03-12 01:53:53 || 0:43:46 |
| 2008-03-12 02:37:39 || 0:00:07 |
| 2008-03-12 03:58:22 || 0:00:07 |
| 2008-03-12 03:58:31 || 0:48:11 |
| 2008-03-12 04:46:42 || 0:00:07 |
| 2008-03-12 04:46:50 || 0:27:31 |
| 2008-03-13 04:51:22 || 0:02:27 |
| TOTAL || 2:11:33 |
So we had slightly more than two hours of power interruptions in a month. Almost all of it was in a single night (and a particularly sultry one -- the lack of proper air conditioning was severely felt).
Since we had 29 days or 696 hours since I started measuring, this means the average power availability here is only about 99.7%. Interestingly enough, I have said elsewhere in this site
that the annualized avaliability here is >99.95. Will I have to bite my own tongue? Let's see how this statistic evolves over time.
Attentive readers may have noticed I rounded the durations to seconds, while the device generates them with hundreds of seconds. This is because I did this in kind of a hurry and it takes a bit of spreadsheet gymnastics to make Excel properly handle fractions of seconds.
Even though I have these nice results above to report, the power glitch monitor firmware still needs considerable tweaking. In fact, it recorded 81 events, not just the ten shown above.
The others seems to have been caused by very brief undervoltages -- the duration counter showed 0 or 1 hundreds of a second. Some of them were caused by me triggering a small spark when
plugging some appliance. Interestingly enough, the UPS that serves my computer didn't click on those occasions. So perhaps the measurement technique I'm using should be made a bit more robust
so as to become less prone to false positives.
Most of the time the undervoltage beeps agree quite nicely with my UPS clicking. But sometimes the UPS clicks without the glitch monitor beeping; sometimes its just the opposite. Probably my UPS monitors instantaneous power, while the glitch monitor is just measuring the voltage at each wave peak.
So, as I said, improvements are in order. As soon as I coding and testing those, I'll post the updated firmware here, along with my usage experience. At that time I expect to have the schematics drawn on Eagle
as well. Stay tuned.