Normally I would agree with you and I think it would look nice to mount all three right next to each other. However:



So I'm looking at mounting on every other stud which will really make things sprawl.



I'm looking at 24 panels per inverter and I bump up against the 600V limit with more than eight panels/string. Hence the 3x8 or 4x6 design.


Did a little sniffing on the serial port and it looks like a very straightforward query/response system. It appears that period tab ends each datagram which makes parsing it very pleasant. I noticed GT-View appears to do a continuous poll. I'm still figuring out some of the fields.

Is there any reason a poll every 60 seconds would be less accurate? Also, does the deratelimit of 3250 mean it will produce over 3KW if the temp is low enough?

Thanks,
Sean


Here's a sample:

Thats a know error in the manual , they are DESIGNED to be flush mounted , sidebyside , then you can use the boss boxes as the race way. So PV in one side , AC out another , no messy disconnects , ect







That's only an issue with the Sanyo panels , and for them you would need to fuse the strings , I would suggest an inline fuses , mounted in the BOS compartment




The GT limits based on input power .. so 3250 * .94 = 3055 limit seen on AC side




Unless the power stays exactly the same , you miss any variations between your sample time

GT-View allows you to log at any rate you want , if disk usages is your concern

The 15.9" width makes perfect sense now. But that would have been nice to know _before_ I ran conduit, wires and drilled all the holes in the wall. I really hope Xantrex updates the manual.

Anything else they got wrong?



For long term trending I'm curious what rate you would recommend? Yesterday my inverter was operational for over 9 hours. At 10 second intervals I would hit the 65,535 row limit in excel for monthly trends. Minute intervals fits nicely but I'm wondering if I'd lose any useful detail?

Sean

6 samples per minute * 60 minutes * 9 hours = 3240 samples

Excel limits to 32k not 64k and yes , at one second interval , you can go over that limit

at 2 second logging you would not hit the limit , even with 20 hour days

I'm aware I can easily fit a day of logging into one spreadsheet. But for some tasks it is possible to have too much data. I'd prefer to deal with 12 spreadsheets/graphs instead of 365. I was mainly planning on comparing trends to make sure all strings/inverters were producing.

My main question still remains, for system management, what benefit is there to logging at shorter intervals?

I just checked my copy of Excel 2003 and it goes to cell IV65536 which is 255 columns x 65536 rows. Open Office limits to IV32000. What version are you running?

Sean

I have found that a lot of data can be lost at longer sample intervals. You need fairly short sampling intervals to catch dynamic response of the MPPT to things like cloud edges and bird in flight shadows. I once recorded the array collapse of the old SunTie with a single bird flying in front of the array.

Now, most of the time, I record data once a minute, but it is not a single sample but the average of 60 one-second samples. I also record the max and min of each one minute period. If the max minus min is over a threshold percentage of the average, I record the whole 60-second data set, the previous 60-second data set and the following 60-second dataset to a separate file.

I use Excel 2003, It’s the graph functions that limit to 32K, not the cells, sorry



If you’re looking for nothing more than daily totals, log as slow as you like, the totals are provided by the GT itself. FishBowl/Suntie is different as only the once a second reading are available and logging precision then comes into play

I think what you want is already provided by GT-View, just use the daily totals file, which is different than the daily log files. Daily totals logs both the total power, peak AC power, time online of the inverter and when interconnected with other GT3.0's of the entire system

The high precision logging is mostly a left over from development, I used 1 second logging to analysis the mppt tracking.

wd8cdh thanks for your input. It helps a lot. A simple perl script can collapse the data into per minute pretty easily but quantifying 'interesting' data and leaving it alone will be a little harder.

Next question. I've played with my collected data and the voltage, current and power numbers all appear to be seperately collected, or at least I the calculations don't line up unless you take into account some serious rounding. Since V * I = P why the difference?

Sean

I looked at a log file and its usally better than 1% on the difference. V&I is read 1k/sec and there is averaging and calabration offsets applied. V&I are not read exactlty at the same time , you would have to have parrallel Analog to digital converters for that , so I can see how there might be some variance

I asked the software engineer why the difference , his response was :



Since both AC and DC reported power values are averaged, the reported efficiency is pretty close and accurately represents over time (1 second) the operating power levels. Volts and Amps are snap-shots and reflect Sub-Second what the GT is doing to maintain and track the array. (The GT mppt tracking is a the 1K/sec rate)

Now that we are in 2005, I have a GT3.0/NEC2005 question.

690.35(G) The inverter or charge controllers used in systems with ungrounded photovoltaic source and output circuits shall be listed for the purpose.

Is the GT listed for ungrounded operation and if so, where would I find proof?

Sean