Epever 6415AN doesn't respect my settings

[jnrhome]

I've had this controller since 9/2019 and recently upgraded my battery bank from sealed lead acid to LiFePo4. The changeover was a bit nerve-wracking, but it all went fine after a couple of restarts.
Now, i've chosen my operating voltages, but the controller is ignoring them. I wanted the battery to top out at 90% or 53.60v so set that as boost voltage. Float is at 53.55v and boost reconnect at 53.50v. Equalisation charge is set to 54.00v and charge limit voltage to 55.00v. Boost duration is set to 30 and equalisation duration is also 30. I have the model with no load outputs so other settings are moot. I've charged for three days to get up to 53.60v. The voltage is continuing to rise, currently at 54.17v and stays in boost. Has anyone else seen this behavior? Not sure what to do...

 

[jnrhome]

Since my first post, I have charged well past my boost settings, but not quite up to the voltage limit so have only seen boost charge mode. This morning, at 53.12v, very early in the charge for the day, my epever suddenly switched to equalisation charge. Definitely below where it should be to EQ the battery. My settings are entered in the spreadsheet on the left and my real time software capture to the right. Makes no sense...

 

[JMH1783]

Looks like you are using the software in your screen shot. Did you use it to set the parameters? I tried using the MT50 and it was a bit frustrating. I did find that once I switched to User defined I had to set the overall voltage (24v for me) instead of auto or it wouldn't stick. The first time I did it it decided to set the voltage to 12v and reset all the charge parameters accordingly. It wasn't until I set the voltage that things seemed to work. Also you don't need equalization that is for lead acid batteries. If you search there are a couple of good posts setting the parameters. I think your reconnect is also too high. Hope this helps. Good luck.

John

 

[jnrhome]

a little more than four hours later, still in EQ though it's limited to 30 minutes in the settings. I will plug in the MT50 and manually set the voltage. In the switch from sealed lead acid, i found advice to use the auto setting to clear the 24v overvoltage warning when i switched to 48v lifepo4. I did use the software to set the parameters after the transition, but will go back and check with the MT50 the voltage setting. Auto as a voltage option doesn't seem to exist in the software. It is reporting 48V but will see what the MT50 says.

 

[jnrhome]

Just double checked with the MT50. I did set it to 48v after i cleared the overvolt warning when i switched from the 24v battery. Additionally, the MT50 shows all the settings i programmed in the software. Just to be sure, i set and saved 48v in the MT50. No apparent change in status tho. More than four hours in EQ mode... makes no sense to me
I'm only using EQ mode as a reference voltage at the moment. As I tweak the settings, and get things dialed in, i will eliminate EQ as it would only be to raise the voltage into the more vertical part of the charge curve, thereby fine tuning the cell balance, which has behaved excellently from the beginning. I top balanced at full voltage and am using a 2A active balancer to monitor my individual cells. I haven't needed to balance them at all and have only briefly seen 6mv delta since i fired them up. It's usually 1 or 2mv delta in normal operation so have left the balancer in standby.

 

[jnrhome]

all day yesterday the controller stayed in EQ mode. It charged normally though and seemed to experience a normal cycle. This morning, as soon as the controller began sensing input voltage, it started in EQ mode. Very strange status display with otherwise completely normal charge behavior.

 

[JMH1783]

That happens. It wasn't done. Depends on how much sun you are getting. Below is a graph from my system from 2019 showing the controller status by day. Note the green bars. In August it took a week to get out of equalization. When I did this graph I realized I had EQ still on and turned it off.

 

[jnrhome]

Ahh...
I'm letting it ride for now. I need to understand what decisions are being made by the controller and as long as it isn't risking overcharge, I can't see a danger.
BTW, What device/software generated that graph for you?
My controller only graphs 30 minutes of generation even though there are some statistics available from the day I turned it on.

 

[JMH1783]

I bought the logger from EPever. I exported the data and imported into LibreOffice Calc (like Excel but free) and built a pivot chart. I'm not running loggers on my controllers anymore. The first one had the battery holder shear off when I went to replace the battery. A second one just died on me for no discernable reason. Finally I realized that the data was good but it only told one side of the story. I wanted to know what was happening on the consumption side as well so I started building my own logger for the entire system.

 

[jnrhome]

Good thinking. I am also data intensive and want the entire story. The epever software is a tease for that. I like your output a lot. Wish I could extract every drop of data and organize it in a similar way...

 

[JMH1783]

Here's what I'm doing now. It isn't done yet. My first and only graph so far is consumption by hour. Note the blue is inverter power and the brown is building power. I forced this transition for testing purposes so the time of day doesn't make sense. Note I use my power for constant loads so it has a constant draw with a few spikes.



The second one is the summary by hour for today. Note we got some sun around noon and the arrays kicked into high gear. The summary also tracks consumption by hour and the net per hour so you can see if you are positive or negative. On the right is the amp hours to give me an idea of what the battery might be getting minus the efficiency loss of course. I'm going to add some graphs but haven't gotten there yet.



If you're curious the data collection is being done on a Raspberry Pi. I wrote a program in Python (not my favorite language but I like C++ less) to do the data collection and store it in MariaDB (MySQL). It also drives a console view. See picture below. The data analysis stuff is done in PHP as web pages pulling data from the database.

Note the building voltage is high in the screen shot below. I have to tune the conversion to make it match my multi-meter. One of the steepest learning curves was getting good quality consistent readings from all the meters.



I've learned a lot from this thing even while building it. For example I learned there was something wrong with my 7.2K (14.4K total) AGM battery bank. I also learned that the transfer switch had basically killed the batteries. On cloudy days it transitioned back to inverter power any time the sun came out and kept the battery bank at a low state of charge. My fault. Live and learn.

 

[jnrhome]

Excellent setup! I use raspberry pi's in conjunction with snmp to monitor network functions, so am familiar with data collection via snmp. They are ideal for simple data collection and analysis, not to mention super affordable. I need to learn some python though as solar doesn't use snmp and I have to figure out how to extract output data collection from existing feeds. No matter what turnkey options I've tried, they all fall woefully short in data display, not to mention comprehensive data analysis.
My generation is fairly variable. I have a 1,080w, 4 panel array set up in 2S2P for 48v nominal operation. I live in the U.S. Virgin Islands and get great sun most of the time, but have a surprising amount of small clouds that keeps the generation graph dancing quite often.
Since I upgraded the battery bank, I've bought 2Kw more panels and will be installing them next weekend, bringing the array up to 2S5P. I've been surprised at the existing panels generation since the upgrade. They have increased in total production by about 20% with the new battery rising from 4k/day on sealed lead acid to 5.3k/day on lifepo4. They just don't produce enough to carry my steady load AND get these monster batteries fully filled. I suspect the 3k array will get the job done.
My consumption is very even as I use my system to power a small data center, a WISP (Wireless internet service provider), and other constant draw items. That makes observing and calculation of load much easier as I have less than 10 watts delta from low to high consumption across the day/night.
I have to read and calculate all these values manually, which has become cumbersome over time. It may be the catalyst to dive in to Python and finally automate data collection.

 

[jnrhome]

Had to work today so ended up leaving the office for a time. When i returned, my voltage had spiked to 58.77v with two cells in excess of 4v each. The meter had barely moved for days, no more than 6mv delta for almost a week, each day getting no higher than 54.77 and no lower than 53.03. I leave for 4 HOURS and the voltage suddenly runs up above max, above every setting i entered, above lifepo4 defaults, dangerously high. Now i'm worried that i've damaged at least two of the cells by charging them past 4v each. Unbelievable!
I shut down the solar array and fired up the balancer. They are now safely under 3.65v per cell with a current 113mv delta and i've resumed charging but am well past the hot part of the day. I've also zeroed the EQ duration to get the system away from EQ charge. It's now in float and trickling 8 watts. It's overcast as it has been off and on this afternoon but WTF!!!
Clearly the controller has ignored my settings and gone off the reservation, waiting patiently for me to be distracted so it could try and destroy my new battery. Should I take it out back and "Office Space" it?

 

[JMH1783]

Excellent setup! I use raspberry pi's in conjunction with snmp to monitor network functions, so am familiar with data collection via snmp. They are ideal for simple data collection and analysis, not to mention super affordable. I need to learn some python though as solar doesn't use snmp and I have to figure out how to extract output data collection from existing feeds. No matter what turnkey options I've tried, they all fall woefully short in data display, not to mention comprehensive data analysis.
My generation is fairly variable. I have a 1,080w, 4 panel array set up in 2S2P for 48v nominal operation. I live in the U.S. Virgin Islands and get great sun most of the time, but have a surprising amount of small clouds that keeps the generation graph dancing quite often.
Since I upgraded the battery bank, I've bought 2Kw more panels and will be installing them next weekend, bringing the array up to 2S5P. I've been surprised at the existing panels generation since the upgrade. They have increased in total production by about 20% with the new battery rising from 4k/day on sealed lead acid to 5.3k/day on lifepo4. They just don't produce enough to carry my steady load AND get these monster batteries fully filled. I suspect the 3k array will get the job done.
My consumption is very even as I use my system to power a small data center, a WISP (Wireless internet service provider), and other constant draw items. That makes observing and calculation of load much easier as I have less than 10 watts delta from low to high consumption across the day/night.
I have to read and calculate all these values manually, which has become cumbersome over time. It may be the catalyst to dive in to Python and finally automate data collection.

That sounds like a great project for solar. I envy your sunshine. Here in PA things are pretty dark in comparison. My solar output daily average for December, January and February is typically about 0.8 times capacity! I also like the LiFePO4s. I think they absorb charge better for some reason. I'll probably add another bank to mine. I'm using the aluminum wrapped 200ah cells 8S1P.

I was a developer at the beginning of my career: business software, process control, e-commerce. I haven't been for a long time. I picked up Python when I moved from the ESP32 as my controller to Raspberry Pi. I had too many problems with the ESP32s. Its much more complete a language than it used to be when I looked at it years ago. I was able to write my program using threads and a semaphore so I was happy with that.

 

[JMH1783]

Had to work today so ended up leaving the office for a time. When i returned, my voltage had spiked to 58.77v with two cells in excess of 4v each. The meter had barely moved for days, no more than 6mv delta for almost a week, each day getting no higher than 54.77 and no lower than 53.03. I leave for 4 HOURS and the voltage suddenly runs up above max, above every setting i entered, above lifepo4 defaults, dangerously high. Now i'm worried that i've damaged at least two of the cells by charging them past 4v each. Unbelievable!
I shut down the solar array and fired up the balancer. They are now safely under 3.65v per cell with a current 113mv delta and i've resumed charging but am well past the hot part of the day. I've also zeroed the EQ duration to get the system away from EQ charge. It's now in float and trickling 8 watts. It's overcast as it has been off and on this afternoon but WTF!!!
Clearly the controller has ignored my settings and gone off the reservation, waiting patiently for me to be distracted so it could try and destroy my new battery. Should I take it out back and "Office Space" it?

You have the opposite problem I'm having. Mine cut out before fully charging the batteries. I added another hundred watts of load to make it just over 200 watts continuous and it basically ignored it. I rebooted both controllers they came back a bit. I increased the charge limit from 28.2 to 28.4 and they briefly pushed the voltage to 28.4 but then spent the rest of the day today losing ground. As of now the bank is down to 26.5 volts.

I think we both need to take our controllers outside and go full Office Space on them...music and all.

 

[jnrhome]

wish i had been here to see it. I have no idea what actually happened so no way to adjust or anticipate...
we need a REALLY long jumper between our systems to even everything out

 

[JMH1783]

wish i had been here to see it. I have no idea what actually happened so no way to adjust or anticipate...
we need a REALLY long jumper between our systems to even everything out

What you are seeing is one of the things that worries me about lithium. They are a lot more sensitive than lead acid. Your batteries might be alright. I'd definitely replace that controller. I'm just not sure what to do if I go to replace mine. Anything you are leaning towards? I'm not sure whether to go cheap or go more expensive.

 

[jnrhome]

I'm taking it up with epever first. I mostly like this controller, especially at the price point. There may be some guidance the factory can provide that will resolve the problem. If not, i'd likely bite the bullet and go victron. As overpriced as they appear, people seem to be quite happy with them.

 

[jnrhome]

I contacted epever tech support by email and they responded in a couple hours. They attached several relevant documents which i reviewed. Based on their advice and documentation, i reconfigured my controller. I think what happened was that i used the sealed/gel/flooded control parameter screen to configure in user mode, based on youtube advice (duh!) . The attached documentation says to use the control parameter for LiBattery screen instead of the lead acid control parameter screen. Now i will let it run under close supervision for a couple cycles and see if things behave. I've attached the pertinent docs so perhaps this won't happen to another user.

 

[jnrhome]

well it seems to work much better when you configure it properly. My settings are now being respected and i'm back to fine tuning.
The lesson here is to check with the manufacturer before consulting You Tube...