outback vs trace
System
Posts: 2,511 admin
I have replaced 2 stacked sw5548 inverters on a 12 kw battery backed up pv system with 4 fx3648 outback inverters. The trace inverters worked ok but thermally shutdown if the output was more than 9.5kw for more than a hour. Result is that the 4 outbacks are worse! With 11.5kw from the PV's the traces would output about 10.4 kw, the 4 outbacks top out at 9.5 KW. The output current waveform of the outbacks is not sinesoidal and they have present a huge amount of ripple current th the batteries. Whats going on here?
Comments
I have heard there is a heat issue when stacking 4 outbacks one over the other. I saw pictures of a outback system installed with Midnite solar's stuff they had a Not a Gutter and Epanel light's with this configuration all the inverters are spread out and there isn't one on top of the other.
On this configuration they where getting more power as the outbacks where staying cooler.
Maybe you can watch the Mate and let us know what the tempatures of the inverters are when it is derating?
The Outbacks are not derating. I am measuring 195 A @ 58.3 volts going in and only 9.5KW going out. The power output was measured by both the eguage software and the electricity meter to the utility. The VA, however was over 10KVA because the outbacks produce a powerfactor below 1. The old Traces were more efficient with their stepped sine wave.
so you are inputing 11.36kw into the battery bank and the outbacks are selling 9.5kw How much are the battery's using to float? and i assume you have made certain there are no house loads on? i would think the inverters themselves would be more efficient at selling then this?
11.36 Kw was what is going into the inverters, not the battery bank. The 9.5kw is what is coming out of the inverter and into the home (200w) and grid (9300w). Your observation about the inverter efficiency is my point exactly! 84% is bad! The old SW5548's were 90%+ at this same input.
Boulder,
Four OutBack GVFX3648 inverters should not be used for grid-tie applications. The maximum "stack" is two inverters for a 120/240 application, with one inverter per leg.
HTH,
Jim / crewzer
Yes I know that and I should have said so. The system is two seperate stacked 3648's, each fed by its own battery bank. The PV system is split 55:45 between them. The measurements I gave is the sum of both systems. The individual efficiencies are the same.
Your sell voltage is way to high ... it should be ~51-52V volts ... That's the cause of your ripple seen at the battery bank.
Get your sell voltage down are re-check your performance.
Also the higher the sell voltage on an Outback, the lower the efficiency
http://gosolarcalifornia.org/equipment/inverter_tests/summaries/Outback%20GVFX%203648.pdf
Looks like you will be ~88% pushing the inverters at 3kw ( its maximum even thought its claimed to be a 3600 watt unit )
Also You don't mention what your charge controllers are, but high ripple will cause Mppt controllers to have tracking issues.
I agree with Guppy - we usually set the sell voltage for those type systems at around 52.5 to 54, depending on the battery type and temperature.
At 58 volts you are nearly at the equalize voltage.
Yep... drop the "Sell" voltage. You might also want to visit the OutBack forum for more product-specific discussions. See: www.outbackpower.com/forum , and this topic in particular: http://www.outbackpower.com/forum/viewtopic.php?f=15&t=1455
The CEC test results were based on their test protocol, which, as I recall, includes a variety of temperature conditions (storage, environment, etc.). I also seem to recall that the protocol was changed after the GVFX3648 was tested. Anyway, the 3648 will deliver its full 3600 VA output at its rated 25 C (77 F) ambient temp.
Regards,
Jim / crewzer
Gosh, I didnt mention one more thing. That is that I am using a 54v battery bank. (nine 6v batteries in series for each bank). But the efficiency vs vin curve does shed some light here. Because most of the pv's are from older BP490 panels, with C40 charge controllers, the higher voltage battery bank matched the power curve better. I have one part of the array (3kw) coming from some seimens and an MX60 cc. The MX60 works fine although its output current ripple is huge (45 AAC on top of 50 ADC) to compensate for the inverter. The cb for that string kept popping until we upgraded it. I am almost wanting my sw5548's back.
Jim,
You may wish to read thru the test protocol,
http://gosolarcalifornia.org/equipment/documents/2004-11-22_TEST_PROTOCOL.PDF
its was written by Sandia not the CEC. It does not use different temperatures for the opertaional test but what the inverter under test specified as ambient for the listed power rating.
This has become the equalizer between what the RE manufactures claim and what using laboratory grade test equipment, witnessed by UL actually perform at. My testing of GridTie inverters was a behind the scenes with Xantrex that helped push this. At the time some manufactures ( PVPowered comes to mind ) were making claims of 97.6% efficiency, when in reality they were more like 92+ percent.
I have never heard that due to the GTFX performance the CEC made changes to any testing, do you have any links to substantiate those claims?. I have at my home a GTFX3048, and it cannot sustain is nameplate rating for more than about ~30 minutes, running it at ~2500 watts is the best my sample size of one the unit can maintain, its also the reason that the units became vented I believe.
That's going to be a real issue using the FX inverters ... they are most efficient at 40 volts, least at 60V. the Xantrex XW series appears to not have this issue much an operational variance based on input voltage
http://gosolarcalifornia.org/equipment/inverter_tests/summaries/Xantrex%20XW6048.pdf
If your really concerned about performance, you should be running a 48V battery pac and using Mppt Controllers, not pwm units. Whatever is being lost by the FX's can easily be gained by Mppt controllers better harvest
The AC ripple you reported is off the scale in will cause all sorts of issues, which leads me to a question, what is your battery bank size in aHr?
I haven't heard that either, and that's not what I said. My recollection is that the draft/original/revised CEC protocol, regardless of who wrote it, used a different temperature (for passive storage and/or active tests) than OutBack's 25 C (77 F) spec.
I also could have mis-read the 2004 draft protocol... However, I suspect that this strorage requirement may have affected the test results: A later version of the protocol cited (2005; still labeled "Draft", BTW), includes the comment below. It's not clear to me if the 150 min / 180 min tests were performed at 25 C (mfr ambient) or at 40 C:
OutBack's power specs for the "big" inverters date back to 2003 or earlier, when the units were submitted for ETL testing. These tests were conducted before the Sandia/CEC protocol was drafted/adopted.
The specs are revised for some applications. For example, OutBack's new SmartRE (tm) products use the GFX3048T and GVFX3648 inverters in enclosures with the FX80 charge controller and other equipment, and their power specs have been derated to 2,500 W and 3,000 W respectively.
Regards,
Jim / crewzer