PeterGutti Posted July 24, 2018 Share Posted July 24, 2018 (edited) I'm running two Axperts in parallel mode, monitoring with ICC 2.4.3. It happens rarely but regularly: The value of battery watts drops to -6'000'000!! Just for a millisecond. A normal curve looks like this (Axpert users know it) What is this? Is it a serious problem? A com problem with sql server? Above graphs come form ICC, so no communication involved. Other curves sometimes look similar, i.e. SOC. Edited July 24, 2018 by PeterGutti Clarify environment Quote Link to comment Share on other sites More sharing options...
Guest Posted July 24, 2018 Share Posted July 24, 2018 2 hours ago, PeterGutti said: What is this? Dud data coming through, having seen similar reading solar equipment data. If no-one else is seeing same, would look at changing the cable. If more people are seeing same, software must be altered to accommodate this. Quote Link to comment Share on other sites More sharing options...
___ Posted July 25, 2018 Share Posted July 25, 2018 (edited) Might be interesting to see the actual value logged. -6,553,500 ? :-) Values like 32768 and 65535 are to us computer people like 999999 is for your car odo :-) Edited July 25, 2018 by plonkster Atti2de 1 Quote Link to comment Share on other sites More sharing options...
Guest Posted July 25, 2018 Share Posted July 25, 2018 (edited) 3 hours ago, plonkster said: Might be interesting to see the actual value logged. -6,553,500 ? :-) Exact value should be in their MySQL database. From what I have seen, seeing as one does not go about storing the continuous data stream, but rather snapshots, is that when one is stored, it was quite accidentally. Now if that is a case of these unique data points being continuously there or they are truly once in a blue moon, we never bothered, we just coded for them, as it skews the stats a wee bit. Edited July 25, 2018 by Guest Quote Link to comment Share on other sites More sharing options...
___ Posted July 25, 2018 Share Posted July 25, 2018 10 minutes ago, The Terrible Triplett said: Now if that is a case of these unique data points being continuously there or they are truly once in a blue moon, we never bothered, we just coded for them, as it skews the stats a wee bit. I remember a problem I had with software once, where drawing the graphs took too long if there were too many data points. We figured out, eventually, that there is no point in plotting a million data points if the chart is only 1000 pixels wide on the screen. You can't plot more than 1000 points in a meaningful way in any case, especially if they are spaced closely together and just end up on top of each other. The naive solution? Pick 1000 equally spaced data points and plot those. But what if there was a spike and the spike happens to be between the picked points? Modified solution: Plot equally spaced points AND any data point that is outside the band by more than one standard deviation point. The graph plotter module was slower than the stats module (standard deviation involves square-roots so could be slow), so that is what we did :-) So not really on the topic here, but one could easily shave off spikes that are just clearly out of the realm of reality. Quote Link to comment Share on other sites More sharing options...
Guest Posted July 25, 2018 Share Posted July 25, 2018 1 minute ago, plonkster said: ... , but one could easily shave off spikes that are just clearly out of the realm of reality. We did not even store them, if the incoming value exceeds the inverter or array size (pre-loaded by the user), we simply ignored that data point. Quote Link to comment Share on other sites More sharing options...
PeterGutti Posted July 25, 2018 Author Share Posted July 25, 2018 Very interesting discussion. The hint of changing the cable (RS232-USB) seems logic to me. The following graph comes from emoncms (local, on same pi as ICC). The first graph a posted came form ICC GUI. On this graph one can read the absolute value: It's one single value, but it is damages the graph! Eventually one could modify emoncms by specifying the y-range limits: Max +/- 8000w. btw, the soc has the same effect: It goes down to 0.00. Quote Link to comment Share on other sites More sharing options...
Guest Posted July 25, 2018 Share Posted July 25, 2018 11 minutes ago, PeterGutti said: btw, the soc has the same effect: It goes down to 0.00. If the SOC and the inverter watts are in sync, one peaks the other goes to zero, have not seen that combination. 12 minutes ago, PeterGutti said: Eventually one could modify emoncms by specifying the y-range limits: Max +/- 8000w. That is the easiest, but would be nice to try and find what is the cause, dud data, Pi, cable, interference ... Quote Link to comment Share on other sites More sharing options...
PeterGutti Posted July 25, 2018 Author Share Posted July 25, 2018 failures of SOC and Batt.watts are not in synch: SOC occurs more often. Quote Link to comment Share on other sites More sharing options...
PeterGutti Posted July 25, 2018 Author Share Posted July 25, 2018 i just recognised that the operating temp of Axpert is up to 55°C.. Mine is currently running at 71° !! Here we have outside temps up to 32° right now. Also bad for the batteries. Should i cool the room? Quote Link to comment Share on other sites More sharing options...
PeterGutti Posted July 25, 2018 Author Share Posted July 25, 2018 Multigraph: battwatts vs soc Quote Link to comment Share on other sites More sharing options...
Guest Posted July 25, 2018 Share Posted July 25, 2018 (edited) 11 minutes ago, PeterGutti said: Here we have outside temps up to 32° right now. My controllers compensate during charging for the higher temp. 32 is not uncommon in SA. Mine in summer is regularly at 32deg in a "Braai room" (Meat Grilling Entertainment Room). Some people have their batteries in their garage, also hottish. My point, 32 is high, but not 40 or the new Africa record of 50.1 degrees Celsius. So, if your controller can compensate for higher temps during charging, not an issue now and then. If you can cool the the room, by all means. 5 minutes ago, PeterGutti said: Multigraph: battwatts vs soc BMV does that, perfectly normal. Edited July 25, 2018 by Guest Quote Link to comment Share on other sites More sharing options...
Chris Hobson Posted July 25, 2018 Share Posted July 25, 2018 If they are 14 minutes ago, PeterGutti said: Here we have outside temps up to 32° right now. Also bad for the batteries. Should i cool the room? If you have lead acids I would cool the room. Lead acid batteries ages twice as fast at 30°C as they do at 25°C. When I had AGMs I had a good excuse to install an aircon just above my desk. It is the only downside to having installed Lithiums. In fact now that I think of it I tell the missus I still need the aircon . We have indoor temperatures of 32°C + outside temps hit the upper 40s last summer. Quote Link to comment Share on other sites More sharing options...
PeterGutti Posted July 25, 2018 Author Share Posted July 25, 2018 I have OPzV batteries (lead-acid with gel). The manual says: The recommended operating temperature range for PowerSafe ® OPzV technology is -15°C to 40°C (Humidity <90%). Optimum life and performance is attained at +25°C. I won't have 40° in the battery cabinet. No need to cool the room. Quote Link to comment Share on other sites More sharing options...
PeterGutti Posted July 29, 2018 Author Share Posted July 29, 2018 Battery watts and SOC both com from BMV monitor, no? It's a USB cable with enlargement. May be the there is contact problem when it gets warmer..? Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.