direct charging LiFePO4 from alternator, or How I Learned to Stop Worrying and Love the Relay

(Apologies to Dr. Strangelove)

Note: this is a narrative, an overview. If you want data see this article on the RVwiki.

my venerable FLA bank

In 2018 I bought a set of 6v 200Ah Flooded golf cart batts for $200. Charged them mainly by solar with some contribution from the alternator via a Battery Doctor VSR. When the relay engaged they’d take a 42A from the alternator and then current would start tapering in a reasonably-linear fashion. That intitial acceptance (0.19C) was close to the theoretical acceptance rate of FLA (0.2C).

They lasted just over 1,100 cycles and would have lasted longer if I’d know what I was doing from the beginning. :-/

replacement with 100Ah of LFP

When the end was near I ordered a 100Ah LFP from Rebel Batteries. It came in and I connected it to the system and waited for all the van loads to come back online.

will I need a DC-DC charger?

My initial instinct was that I was going to need a DC-DC charger to limit the lithium’s insatiable demand for current. I was thinking something in the 20A-40A range.

Then I remembered some things that might make a test run with the existing setup a relatively low-risk affair:

  1. the alternator is rated 180A; even if current spikes the alt should handle it
  2. the BMS had a 100A limit, which should shut down the party if things got crazy
  3. and there was already an 80A fuse on the POS wire in the existing relay-charging setup

That’s it; I’m going for it.

the test

The LiFePO4 was resting at 13.1v. The phone on the dash was showing the BMS data, my right hand was on the ignition, and my left hand on the relay cutoff switch. There may have been sweating and slightly elevated heart rate at this point…

I started the van. After a few seconds the chassis came up to the 13.4v cut-in voltage and the relay clicked into service and…

The Lithium bank accepted ~29A from the alternator. Not exactly dramatic..

29A into a 100Ah battery is 0.29C (duhhh), well within the safe limit of charging current. How could this be? Don’t get me wrong: I was happy with the result, I just didn’t understand it.

daily use

It was pretty much the same ever after: about 0.2C in middling states of charge, more (up to 0.35C) in the lower knee, less (≤0.1C) in the upper knee. But most of the time somewhere around 0.2C. Ho hum.

addendum - a week of data

see this blog entry

some quirks

I did observe some quirks and gotchas over time.

VSR cut-out

The LFP’s cut-out point was too high for the VSR; it would stay connected until bank voltage dropped ≤13.2v. Not really an issue, and something I’d seen before on the FLA bank that floated at 13.8v. I’d already installed the cut-out switch to break the connection if I wanted to.

bank voltage and current acceptance

Current acceptance dropped off markedly when solar charging was present. I later realized this had to do with solar raising bank voltage, affecting the I=V/R relationship between alternator and the battery bank.

The bank voltage / current thing also affected alternator charging any other time bank voltage was increased, like after a long drive. When the bank is ~13.8v I’ve seen as little as 0.08C acceptance. Nowadays I generally disable the relay at ~13.7v. I bought a HVD to automate this but haven’t set it up yet.

This isn’t a function of LFP, but now I can see acceptance info from the phone while driving.

Example: I recently started a road trip before sunup. The alternator was contributing 0.3C (45A). As the sun came up the acceptance rate remained around 0.3C, with solar replacing alternator as bank SoC and voltage increased. By the time the bank was ~full (1030), the alternator was contributing 0.17C and the solar 0.13C.

the current taper

LiFePO4’s taper didn’t look like flooded lead’s taper:

  • FLA voltage would quickly rise to (near) alternator voltage then start a linear taper over the next few hours.
  • LFP would quickly come off the bottom knee, stabilize at a very gradual taper around 0.2C then suddenly fall off near the upper knee.

I later realize these taper shapes were the inverse of each chemistry’s charging voltage curve; I=V/R strikes again.

down the rabbit hole

This experience was so unexpected that I couldn’t stop thinking about it. I started googling, reading forums, and watching YT vids to see what happened when other people tried it.

why does it work?

It seemed to work for nearly everyone who tried it. The biggest common factor with successes, as far as I can tell, is using the chassis as the NEG return in the circuit. This introduces so much resistance that current acceptance can be quite modest. I=V/R again.

when does it NOT work?

Lithium’s thirst for current can be seen when the resitance is removed by running a fat NEG cable back to the starter battery. This is exactly how the three most famous failures were set up:

  1. Morton’s on the Move
  2. that famous Victron video (they also ran the alts at/below idle RPM)
  3. that less-famous Sterling video (they also added external loads)

why the misinformation?

So why do many people continue to repeat the canard that “lithium will take all your alternator can give, and will do so for hours”? IOW, why do people bullshit?

With influencers and marketers it’s pretty simple to suss out: they are uninformed and have financial stake in selling more expensive parts.

When there is no money to be made the motivations are less obvious. On this topic an clueful poster said:

if you want to make an informed decision, measure the current and know if there’s anything to mitigate

But that’s just it; I don’t think people want to make informed decisions. I think they want to be told what to do by the slingers of easy answers.


Relay charging is not the solution to every problem. DC-DC is preferred or required if:

  • you want a stable charge rate
  • the alternator is undersized for the bank
  • the alternator is “smart”

Be aware that installing DC-DC to protect the alternator is not a silver bullet: the charger still has to be sized to the bank and alternator rating.