MX60 amps/watts output accuracy

Discussion about the MX60 Charge Controller

Moderators: OutBack, OutBack Moderator Team

Post Reply
dgd
Forum Member
Posts: 10
Joined: Wed Nov 17, 2010 1:39 pm
My RE system: Midnite Classic 150,250, Outback MX60, 20 140W pvs, 2Kw turbine, Midnite Clipper, Epanel, Trace SW3024E (1997), 1025Ah Century Yaesa FLA(1999), Arduino power monitor and web server. 48v 400Ah LiFeYPO4. Favourite toy: red 911 gt3, location: Titirangi, West Auckland, New Zealand

MX60 amps/watts output accuracy

Post by dgd » Wed Dec 02, 2015 12:49 pm

I have made an Arduino DUE based web server that connects directly to the mate (serial) port of an MX60 so that running data can be displayed in a web page
However I have run into an issue with the accracy of data being reported by the MX60.
It appears that the reported output current and hence watts of power being produced is anywhere from 3% to 14% overstated and runs about 10% over actual most of the time.
I have the MX60 -ve output connected to a Deltec 500A shunt which also has a Midnite Classic with WBjr connected to it so that a the DUE web server reports the amps output from the Classic and also the output amps to the battery from the Classic and MX60
When I disconnect input PVs from the Classic and just have power from the MX60 going to battery bank the current measured by the WBjr shows about 10% less current than the MX60 is reporting.
As a further experiment I just set up the MX60 connected to a separate shunt that had a Bogart TM-2020 battery monitor on it and sure enough the same disparity was again reported by the monitor.
Is this likely to be just my MX60 or is this inaccuracy common?
And how about the newer FM60, FM80, do they report output current more accurately?

dgd

User avatar
tallgirl
Forum Emperor
Posts: 4334
Joined: Wed Sep 26, 2007 9:59 am
My RE system: 16 Kyocera KC175GT (2,800 watts DC), MX-60, 2 OutBack GVFX3648, 8 GC2 batteries (6v @ 215AH),
Location: Austin, TX

Re: MX60 amps/watts output accuracy

Post by tallgirl » Thu Dec 03, 2015 2:39 pm

They aren't lab-grade equipment -- the various current and voltage sensors are notorious imprecise compared to more precise tools.

My solution was to map an FnDC shut to the charge controller output and report that. That does assume you've only got one charge controller, but it does help.
Julie in Texas

I ride bicycles. A lot.

dgd
Forum Member
Posts: 10
Joined: Wed Nov 17, 2010 1:39 pm
My RE system: Midnite Classic 150,250, Outback MX60, 20 140W pvs, 2Kw turbine, Midnite Clipper, Epanel, Trace SW3024E (1997), 1025Ah Century Yaesa FLA(1999), Arduino power monitor and web server. 48v 400Ah LiFeYPO4. Favourite toy: red 911 gt3, location: Titirangi, West Auckland, New Zealand

Re: MX60 amps/watts output accuracy

Post by dgd » Thu Dec 03, 2015 3:20 pm

Yes,I had sort of figured that out.
Although 10% inaccuracy did surprise me as both Midnite Classic and MS MPPT60 seem to be in the +/- 1 to 2% range of accuracy. So I assume that if I get a new FM80 and its amp output/wattage is more than a few percent inaccurate this is not considered a faulty unit I can get replaced under warranty?
My MX60 data is not worth processing for a web page report or live meter display, the only useful data seems to be the reporting of the charger state.
True current, voltage, Ah, Watts, KwHr would seem to be better measured and calculated externally. :???:

dgd

User avatar
tallgirl
Forum Emperor
Posts: 4334
Joined: Wed Sep 26, 2007 9:59 am
My RE system: 16 Kyocera KC175GT (2,800 watts DC), MX-60, 2 OutBack GVFX3648, 8 GC2 batteries (6v @ 215AH),
Location: Austin, TX

Re: MX60 amps/watts output accuracy

Post by tallgirl » Thu Dec 03, 2015 3:49 pm

I'm able to report both whatever my FM60 and FnDC are reporting separately, but this discussion has come up before and neither the CCs or FXs have "guaranteed" accuracy. The FnDC is supposed to be in the +/- 2% range, however.
Julie in Texas

I ride bicycles. A lot.

dgd
Forum Member
Posts: 10
Joined: Wed Nov 17, 2010 1:39 pm
My RE system: Midnite Classic 150,250, Outback MX60, 20 140W pvs, 2Kw turbine, Midnite Clipper, Epanel, Trace SW3024E (1997), 1025Ah Century Yaesa FLA(1999), Arduino power monitor and web server. 48v 400Ah LiFeYPO4. Favourite toy: red 911 gt3, location: Titirangi, West Auckland, New Zealand

Re: MX60 amps/watts output accuracy

Post by dgd » Fri Dec 04, 2015 8:14 pm

Thanks for the info.
So using an MX60/FM60/80 needs Mate3 and FNDC to get decent accuracy for reporting the running output current.
The MX60 voltmeter seems spot on so it would not be too difficult to attach a Deltec 100 or 200Amp shunt on -ve output
and get the Arduino to read a 12 or 16bit AtoD device to accurately get the output current reading.

What I can't see is if the Mate3/FNDC combination will use its Amps reading to update the FM controller so that watts produced/KwHr are more accurate and end amps for Absorb state termination can be set by the Mate3/FNDC.

I would like to get the Arduino DUE to, in effect, do what the Mate3/FNDC combination does, especially EndA control
(This would allow my Classic via the DUE to control the charge state of the FM controller but I suppose this depends on discovering if there is serial input via the FM's mate port that sets the FM's charge state)
I must investigate the Modbus registers available in the AXS to see if there is a register that can be set to control an FM's charge state, EndA etc.. :wink:

dgd

User avatar
tallgirl
Forum Emperor
Posts: 4334
Joined: Wed Sep 26, 2007 9:59 am
My RE system: 16 Kyocera KC175GT (2,800 watts DC), MX-60, 2 OutBack GVFX3648, 8 GC2 batteries (6v @ 215AH),
Location: Austin, TX

Re: MX60 amps/watts output accuracy

Post by tallgirl » Fri Dec 04, 2015 10:05 pm

Trying to measure the current through a shunt directly is pretty hard. Most of the ADCs in devices like an Arudino have a range of 0 to 5 (or 3.3) volts. A 10-bit ADC, like in the original ATmel based devices, has a resolution of 5 millivolts. 12-bits will give you about 1 millivolt, but when you consider that most shunts have a range of +/- 50 or 100 millivolts, 1 millivolt resolution is horrid.

That said, there are chips which can be used to amplify shunt output, if you're in the mood to make such a thing.
Julie in Texas

I ride bicycles. A lot.

dgd
Forum Member
Posts: 10
Joined: Wed Nov 17, 2010 1:39 pm
My RE system: Midnite Classic 150,250, Outback MX60, 20 140W pvs, 2Kw turbine, Midnite Clipper, Epanel, Trace SW3024E (1997), 1025Ah Century Yaesa FLA(1999), Arduino power monitor and web server. 48v 400Ah LiFeYPO4. Favourite toy: red 911 gt3, location: Titirangi, West Auckland, New Zealand

Re: MX60 amps/watts output accuracy

Post by dgd » Sat Dec 05, 2015 12:28 am

I agree with you that the DtoA converters with the Arduinos are not really suitable for decent amps resolution.
However, after completing several multi Midnite Classic installations I have a dozen or so MN WBjr devices in stock.
Some time ago I asked for tech info on these in the Midnite forum and brad.midnite answered with details of the interface, how to drive it and the async commands and data responses.
I have been connecting a WBjr to an Arduino and making progress with it connected to Deltec 500A and 100A shunts.

Although I have progressed with async comms between a DUE and the serial port on my MX60, it would be nice to get the details of what available async command-and-response-data pairs are available for the FM controllers (as I intend to replace my MX60 with an FM controller). Perhaps there is a brad.midnite equivalent at Outback who could provide this info. :-s
I am especially interested in commands to set the charge state, define Absorb stage EndA value, force Float and set/change AUX modes and values.
This would enable my MN Classic via DUE to work with the FM, implementing a version of 'follow-me' to more accurately control battery charging.

dgd

Post Reply