As promised, here is Part 2 of our Lithium install saga. (skip to Conclusion to miss the technical bullshit)
First, I want to start this post off with a small disclaimer. There are many ways to implement these batteries, but this is the way we chose to implement them. That doesn’t mean we think other people’s installs are wrong. It just means that this was right for us after learning our needs from almost 3 years of cruising experience in the Caribbean. Our goal was a robust, moderate, automated system on a budget that functioned in the tropics (the last point is really important). So far it has proved to be exactly that.
Alright…. So the first controversial detail about our install. We have completely removed our alternators from the charging circuit. Yes, both our Volvo engines only charge their respective lead acid starting batteries. Shocker, I know (no really). However removing the alternators from the system simplified things immensely. I didn’t have to worry about buying high dollar alternators that can take the abuse lithium batteries dole out. There was no extra external regulator that I needed to wire up (another potential failure point). I didn’t have to try and figure out how to put a serpentine belt on old Volvo MD2030s. I didn’t have to deal with the batteries robbing valuable horse power from my primary source of propulsion (there is no free HP people). I don’t have extra wear on my expensive main motor. Most importantly, my math said I just didn’t need the power! Nope, I just left the engines “as is” with 70 year old technology and focused on the other charging sources.
However, during this decision I was not blind to the fact that I was removing redundancy. By making this move, I was not only removing a redundant source of charging, but I was loosing the ability to boost the motors from the house bank if required. I was not comfortable with that so I added some extra components to keep that functionality in an emergency situation. I essentially added BIG solenoids between the FLA batts and the lithium house bank. Those solenoids are controlled from momentary switches at the helm. This not only allows me to boost the motors from the lithium bank, but also charge the lithium in an emergency from the alternators. Here is a sketch of the installed config.
(YES, DANGER DANGER as I would be running those alternators WIDE open if I connected them to the lithium bank for charging. However, this functionality is only for emergencies and I would closely monitor things. The switches are momentary and cannot be accidentally switched to this mode with intelligent intervention.)
Now the second controversial part of our install is the house charger. I had at least 3 experts tell me I needed to replace my house charger as it did not have lithium mode. However, in their defense that recommendation was based on the fact that my chosen BMS could not control the house charger. Since this was a 2 boat buck (2K USD) touch, I decided more research would be in order. The research showed, that for once I had actually lucked out. Although our boat is equipped with a relatively unknown brand of house charger, I was able to find documentation for it. Its an Italian made house charger that has some unique charging profiles (really no bullshit, Italians actually make more than pasta and espresso makers) . Turns out Italians actually knew how to charge GEL batteries properly. As luck would have it, the real charging profile for GEL batteries is pretty much ideal for Lithium. THIS IS WHERE IT GETS CONTROVERSAL. DO NOT ASSUME THAT A GEL SETTING IS FINE FOR LiFePO4 batteries. Crazy as it sounds, manufacturers never agreed on a charging profile for GEL batteries. Not all manufactures have the same settings.
Again in our case, we got lucky and the Italians had gone conservative. They picked 13.8V with a %10 tail current for the GEL charge profile….. Well slap my ass and call my Travy.. That happens to be an absolute perfect profile for our new batteries. However, even though it is a perfect match for our batteries, that doesn’t mean we allow it run unattended. The house charger is run on the rare occasions that we don’t get enough sun to cover our usage. Since its powered by our Honda 2000, its never run without close monitoring.
So with the alternators and house charger figured out, that just leaves the solar arrays. Long time readers of our blog will remember that I have configured our solar panels into 4 different banks. This has not only allowed me to reduce shading issues, but provides us built in redundancy. In the event that one solar controller fails, then we just reduce our usage and limp on the remaining 3 banks until that controller can be replaced. Of course the downside to this configuration is the fact that I have 4 solar controllers that I needed to purchase. Although I had a good experience with the Chinese solar controllers we started out using, I decided to look at another brand after a few died (due to the lightening strike). I eventually settled on Victron Bluesolar MPPT controllers. They have fully adjustable settings, a great warranty and most importantly had been used successfully by others. Unfortunately, even after all the research this is the one area that I have had the most issues with. While I generally like Victron gear, it has become clear that they are probably not the best choice for a lithium setup.
The big problem is Victron never designed their MPPT controllers with an external voltage sensor. This means that the voltage is actually sensed at the terminals on the controller itself. Anyone familiar with electronics will understand how this can be an issue. You see, a wire (even big wire) is essentially a resistor. This means that if you are flowing current down a wire you will read a higher voltage at the source then the drain (the wire resists the flow and wastes some to heat). How much difference there is depends on wire size, wire length, quality of connections and the amount of current flowing. This difference leads the controllers to “think” the voltage is higher than it actually is and then cut off charging early. On top of this, it appears that the controllers also have a software bug where they sample the voltage without averaging which leads to even more early cutouts. The software bug means that operating big loads like a windlass or electric winch will often trick the controller into the next mode (there is a really long explanation for this that gets very technical and requires a oscilloscope. Hint – flyback voltage). Finally to make matters worse, Victron realized this issue and released a solution that was obviously rushed and poorly designed. Instead of releasing a hard wired voltage sensor they released a Bluetooth one. Not only does it mean older generation of equipment require a second device (Bluetooth dongle), but the voltage sensor has reportedly terrible range (like 3 feet) making it almost useless.
These issues really left us in a pickle as there were days when the controllers went into float with the batteries still able to accept 80-100ah. For the reasons I mentioned above we decided against the Bluetooth voltage sensor. So that really left us with only 2 options.
1. Replace all the controllers with another brand that supported external voltage sensors. (sigh, spendy spendy)
2. Rewire the boat in a way that would reduce the issue as much as possible. Then tweak the settings in the controllers to try and “work around” the issue.
So in the end we decided to try option #2 and see if we could get an acceptable solution. The following 2 diagrams will show our original configuration and the changes we made.
While the changes didn’t completely eliminate the issue, they definitely helped immensely. The voltage difference between the solar controllers and the battery terminals went from 0.3-0.35V down to 0.1V. This change completely eliminated the early dropouts in the middle of day with 1/2 charged batteries. However, the controllers were still dropping to float before the batteries hit %100 (my %100). To help offset this, I played with the “Absorption Time” setting on the controllers so they would hold that voltage (13.8V) until the batteries got there and the current tapered. (as the current tapers the resistance losses become so low that there is essentially no voltage difference between the 2 points). Its been a couple of months since these changes and things are working well.
Now with all the charging sources sorted, I needed a way to actually monitor the bank. Unlike DIY systems with fancy BMSs, ours is very simple and doesn’t have a central monitor. However, monitoring was still required with our old FLA bank so we already owned a Victron Battery Monitor 700. Although it is capable of monitoring a LiFePO4 bank, the exact settings are different from system to system. Here is where Internet forums and T1 Terry came to the rescue. Terry has posted some starting settings on multiple forums that were gold. From there I was able to slightly tweak the settings over a few weeks until I was very confident it was reporting our proper state of charge. Here are the values I ended up with:
Charged Votage 13.8V
Tail current %4.00
Charged Detection Time 1m
Peukert exponent 1.02
Charge efficiency factor 97%
Current threshold 0.05A
time-to-go averaging period 3m
So after 2 LONG posts of the technical mumbo jumbo, what is our conclusion about a DIY LiFePO4 system?
We absolutely love it. HOWEVER….. There are some things I would like to say.
- While LiFePO4 are better than the other battery technologies out there, they are not magic. They will not magically give you oodles of power if your whole system is undersized.
- Be very careful what you read on the Internet as a ton of it is sponsored by both sides (and thus skewed). Treat anything posted by T1Terry as gospel. IGNORE anything posted by John61CT.
- DIY is not for everyone. Ensure you have the proper skills and knowledge before running this path.
- Try to source cells from North America. Don’t bother importing from China.
- You will need to forget everything you know about batteries and living off-grid on them. LiFePO4 are completely different.
Further to point #5, making this change required us to change how we manage and use our power. We no longer care about getting back to %100 charged and frankly it has become my job to ensure we don’t. I talked about about tweaking the controllers so we can actually get to %100, but that really only happens when we are off the boat all day. If I’m aboard, I do my absolute best to ensure we stay in BULK with the batteries around %80 state of charge. If our controllers go to FLOAT, then we are just throwing away power that I could have used. Instead, we charge laptops, external batteries, tablets, Kindles and heat water with that excess power. I will admit that it still feels crazy as its so different from our old FLA bank. However, its damn nice to just plug the laptop in when it goes dead at night.