Charging LiFePO₄ Cells with a Maplin N27GG Bench Powersupply

An Eccentric Anomaly: Ed Davies's Blog

Introduction

As previously mentioned, I have 10 of these 20 Ah LiFePO₄ cells. They've been in storage for a year and I was getting a bit concerned that their self-discharge might be reaching the point where they'd damage themselves by going under their minimum voltage so I fetched them and gave some thought as to how to charge them.

When I was playing with them before I used the 12 volt gel cell charger which I previously used for glider batteries, basically as a constant current source. This is not very satisfactory as the charger gets quite hot in the near-shorted state charging a single 3.3 volt cell and you have to watch things very carefully to make sure you don't overcharge the cell.

That was for the initial conditioning charge (when you first get these cells you need to charge them to 4 volts first time, then charge to some value a bit less than this subsequently) and for additional playing. Most of the time I was using them on my little solar panels with a Morningstar TriStar MPPT controller.

Inspired by Outtasight I got a Maplin N27GG bench power supply from Maplin in Blackpool while down that way on a family visit. It's in some ways quite cute but in others quite horrible. The first one I had blew up, moderately spectacularly.

Ideally you charge these batteries (and many others) using a constant current/constant voltage profile. That is you charge them at a constant current until the voltage reaches a certain limit then charge at decreased current keeping the voltage constant until the current has dropped to a point where you consider the battery charged.

(For lead-acid batteries you tend to also want an absorption phase at low current and, once a month or so, an equilization charge at higher current. These are generally not good things to do to LiFePO₄s or other flavours of lithium batteries.)

The profile I've been going for is to charge these 20 Ah cells at 4 amps until they reach about 3.6 volts then back off the current until it drops well below 100 mA, typically around 50 mA at which point I stop the charge.

A nice thing about a reasonably clever bench power supply like this one is that you can set both the voltage and current limits and leave it to get on with the charge. So long as you don't leave it too long it should sort itself out nicely.

Given that a fairly full cell will have a voltage around 3.34 volts the gap between well charged and full is quite small. This means that any voltage drops in the leads or across the battery fuse (yep, don't like wires running away from batteries without a fuse on them) are significant. With 4 amps through it the little in-line fuse holder and fuse I've been using has a voltage drop of about 0.2 volts, it seems, so a resistance of 0.05 ohms which seems a) plausible and b) quite sufficient to make the power supply back off the charge current much too early making the charge process long and tedious.

Therefore, another nice thing about this supply (and why I bought it rather than a cheaper otherwise-similar one) is that it has remote-sense connections. These are separate terminals (on the back of the supply) which connect directly to the load (cell, in this case) and carry only a tiny current so don't have any significant voltage drop which allow the supply to measure the voltage it's actually delivering and up its output to compensate for the drops. Charge controllers used between batteries and PV panels or wind turbines typically have similar remote-sense leads.

Problems

I charged the first two cells using the bench supply just fine, though a bit slowly and having to do without the in-line fuse. For the third cell I tried using the remote sense function. I first tried with the normal charge configuration then dabbing the remote sense leads briefly which increased the charge current nicely which was encouraging. I then connected up the remote sense leads properly and switched things on at which point the power supply made a loud popping noise and emitted a small cloud of light-grey smoke which was less encouraging.

I took the supply back to the Maplin branch in Inverness. After a bit of testing they replaced it without any undue hassle but expressing puzzlement as they say they're usually quite happy with these supplies. They were well aware of their use as battery chargers and only checked that the battery wasn't connected to anything else when the problem occurred (it wasn't).

However, a private email conversation with another user of these supplies indicates that they seem to have two potentially serious problems:

  • If the remote sense is connected but the outputs are not then the supply will increase its output voltage progressively in an attempt to ramp the remotely sensed voltage up to the set value. At least by default, there's no limit on this other than when capacitors in the supply explode.
  • There are capacitors and other circuitry on the output side of the supply downstream of the output on/off switch which can be damaged by higher external voltages (e.g., 24 V) if they're connected while the supply output is switched off.

I think it was the first of these problems which got me; I don't recall the exact sequence but I may well have switched on the supply output before connecting it to the cell.

The supply does have an “upper voltage limit” (UVL) function which might protect against this problem. However, I rather suspect not because a) if it did then it would be silly not to set it by default to a state where it gave such protection and b) the setting on my second supply (never checked on the first) is 39.82 V - a sufficiently small margin over the nominal maximum output voltage of 36 V that I can't believe the capacitors would actually be protected by this. Whatever, I'm not going to risk a £100 power supply testing to see if a lower UVL would allow the remote sense to be safely connected without the output.

Procedure

Using a supply with this problem for battery charging is intrinsically risky. I have adopted a procedure for doing it which, I think, keeps the risk to a minimum but still, if the battery fuse blows or there are other unexpected disconnections, things could still go wrong. Here's what I do:

  • Make all of the negative connections.
  • Connect the in-line fuse to the cell positive terminal.
  • Make sure that the positive leads from the supply output and for the remote sense are free and not connecting to anything.
  • Switch on the power supply, check it's on the correct output range and the voltage and current limits are set as required.
  • Switch on the output of the power supply.
  • Connect the positive output from the supply to the main battery fuse. At this point the cell starts charging though typically at reduced current.
  • Connect the positive remote-sense lead to the cell positive terminal at which point the cell starts charging at the set current.

Care needs to be taken not to jog the battery fuse connection while making the positive remote sense connection.

By the way, the remote sense leads also have a fuse in the positive lead. At the moment it's a 5A one; I'll replace it with a 250 mA or so one at some point.

The disconnection sequence is the reverse of the connection sequence.

So far I've done 12 cell charges using this sequence with no further fireworks.

Self Discharge

Though I was only expecting to store the cells for a couple of months (rather than a year and counting) I thought I left them reasonably well charged but it seems not.

Of the ten cells, the first eight all took about 8 amp-hours to charge (ranging from 7.86 to 8.2) with data for one cell (the first I charged on the replacement supply) missing because I was concentrating on getting things done safely using the remote sense without worrying about logging as well, but it took 4 amps for nearly two hours then tailed off.

Based on those figures I was coming to the conclusion that these cells would lose approaching half their charge in a year, which is not unreasonable.

However, the last two cells only took a tiny amount of charge (0.30 and 0.31 amp-hours).

I therefore suspect that the first eight were the cells I was using with the solar controller and that they weren't left in anything like a fully charged state for whatever reason. The last two cells, presumably, were the ones I'd charged with the gel charger and had, also presumably, been left fully charged. If so, their self-discharge is impressively small.