LiFePO4 charging - the overcharging question

[ptourin]

I did my most extensive charging run this evening. I discharged with about a 50A load until I was down to 61% SOC - at which point I was reading 12.73 volts - the sag surprised me. I removed the load and the voltage came up to 13.00 after 10 minutes.

I've attached a graph of the charge cycle. My earlier runs were similar. I notice a few things:

* The charging current started high but immediately started to drop. The voltage also started to climb quickly.

* By 20 minutes the rates of change had slowed down. Then the current drop and voltage rise were pretty steady up until about 80 minutes.

* At 80 minutes I hit the knee - fast voltage rise and current drop. At that point I was at 13.7 volts and 41.8 amps. That's below my present monitor definition of 100% SOC (13.4 volts, down to 10 amps, hold for 3 minutes). Not too good! It appears that even though the current drops as the voltage climbs during the "CC" charging phase, it's not going to drop below 10A until the voltage goes up to over 14.3. I guess the charger is working more or less like it's supposed to work, and I need to decide if I believe it's not a good idea to charge up to near 14.6 volts.

* At 100 minutes I was logging values (manually!) every 2 minutes instead of every 10, and having trouble keeping up. So Booster's point about rapid change and time delays as you get near full charge is very well taken.

* The monitor read 100% SOC at 102 minutes - voltage was 14.18 and current was 19.20. So I guess the monitor was calculating SOC on the fly rather than looking at my definition (as set in the monitor params). I'm not sure how to get the monitor to use my definition - there's something I haven't figured out about it. There must be a way to sync the monitor to real conditions.

So this was pretty educational. I can see why you were talking about adjusting the charger to a lower max voltage, if we do believe that charging to 14.6 isn't a good idea. I want to find out more about possibly doing that! At present, I either accept charging to over 14V or stopping charging before the current drops.

[booster]

Originally Posted by ptourin

I did my most extensive charging run this evening. I discharged with about a 50A load until I was down to 61% SOC - at which point I was reading 12.73 volts - the sag surprised me. I removed the load and the voltage came up to 13.00 after 10 minutes.

That is an interesting curve, and looks like most of the published ones, with the exception of the 80 minute voltage which looks off of smooth profile. The rest is near perfectly smooth. It also appears that the Smartbatteries may need toward the higher end of voltage range that Powerstream talked about when looking at the knee in the voltage curve (although that is were the distortion is). Did you happen to notice what the voltage and current were when the SOC meter got to 100%?

[ptourin]

I also wondered about the 80 minute data - my guess is that I wrote it down incorrectly. The monitor read 100% SOC at 102 minutes - voltage was 14.18 and current was 19.20A, still rather high. As above - I'm not sure if that 100% reading came from extrapolation, since I'm pretty sure my definition values for 100% were 13.4v, 10A and 3 minutes - I wasn't down to 10A when the monitor went to 100%.

This run behaved like the last one in terms of voltage drop after charging stopped. When I turned the charger off, the voltage was around 14.5 and current was dropping by the second. After 10 minutes rest the voltage was 13.93; after 40 minutes it was 13.68; this morning after 10 hours it was still 13.56 (DC switch off, so the ~0.30A load I'm seeing must be from the monitor).

So we're into another question area. What does that 13.56 mean? Can I have overcharged and forced the resting voltage higher than the 13.4 specified by SmartBattery? Up till now, I have been visualizing the published resting full voltage as an absolute value relating to the battery chemistry and manufacture - that if I end up were below it, I didn't fully charge; and that if I overcharged, I might hurt the battery but the resting voltage would still end up at 13.4. Now I'm wondering if it isn't a small range of voltage rather than a specific value - so if I push the charging hard, the rest voltage ends up a bit higher. Another possibility is that either the LiFePO4's in general or my brand in specific take a long time to coast down to a stable voltage - we'll see what it's at when I get home this evening.

[booster]

Don't know about the 13.56v, but it may still go down a bit. It could also just be the cells they use, as Powerstream mentioned some differences between manufactures. Will be interesting to see where it lands if it changes any. It may be a total coincidence, but it is interesting that you are just under 13.6v, which is what the 3.4v per cell would be, and that seemed to be about the threshold for getting in the high 90's% state of charge on the Powerstream chart.

[ptourin]

I just looked at the SmartBattery site, which adds to the confusion <g>...
They now say that cell voltage is 3.2 and nominal voltage is 12.8 - that's a change.
They say that their BPS high voltage cutoff is 15.8 or 16 - both numbers on the same page. They still say that they have internal balancing, with no explanation.

They now say that most chargers will work - for multi-mode chargers they recommend Bulk 14.4V, Absorb 14.6V and float at 13.6V. That's a float that's higher than what they used to publish as the resting full cell voltage of 13.4

Also: "Most alternators charge around 14.4 - 14.8V This will sufficiently charge a Smart Battery under normal circumstances."

Also: "In some cases if your battery charger does not reach 14.4V - 14.6V during charging it may not fully top off the Smart Battery." Obviously they don't believe that high charge voltages will damage their batteries...

It's enough to drive ya crazy...

[booster]

Didn't they also used to say that they could be taken to dead without issue?

[ptourin]

They still say it - I think that's because they assume their BPS will open the internal relay when the voltage goes down to 8V.

BTW - Dave Dykehouse, the Progressive Dynamics lithium charger rep, says that the 14.6V is fixed - he says there's no internal pot in either the lith or the non-lith 9100 series chargers any more. Says they used to have a pot but don't now.

[booster]

Originally Posted by ptourin

They still say it - I think that's because they assume their BPS will open the internal relay when the voltage goes down to 8V.

BTW - Dave Dykehouse, the Progressive Dynamics lithium charger rep, says that the 14.6V is fixed - he says there's no internal pot in either the lith or the non-lith 9100 series chargers any more. Says they used to have a pot but don't now.

That's a bummer, sure would be nice to be able to tweak it a bit.

[ptourin]

Update - the batteries were still at 13.56 at 5:30 this evening. So apparently that's the steady no-load voltage they dropped down to.

[booster]

Originally Posted by ptourin

Update - the batteries were still at 13.56 at 5:30 this evening. So apparently that's the steady no-load voltage they dropped down to.

I wonder is Smartbattery lowered the full charge voltage spec because their batteries are aging quickly and losing voltage? It makes little sense.

When you talked to the PD rep, did you happen to ask him if they had done any testing or research about charging at lower voltages?

[ptourin]

SB may have lowered the full charge voltage, but they still recommend high charging voltages - "alternators charge[ing] around 14.4 - 14.8V ... will sufficiently charge a Smart Battery under normal circumstances." !!

I didn't ask Dave if they'd researched lower charging voltages, but when I asked if the lith charger line had an internal pot for adjusting the max charge voltage so that you could adjust it lower, he first didn't understand what I was asking, then replied in a puzzled voice, "But why would you want to lower the charging voltage?" So it sounds like they're firmly in the 14.6v camp.

[booster]

I would guess that the high charge voltage would still work the same as always with the battery going south, except the rested voltage would drop further, and you would have reduced capacity.

Not surprising that PD is still with the 14.6v, as most still are. PD is still saying that you don't need to turn the charger off when the batteries are full, too, I think, and shutting off in a true CC/CV mode is getting to be very common for lithium these days. The lower voltage thing is certainly not anywhere near mainstream, from what I have seen, as the only places I have seen anything about it is on electric vehicle sites, the Cruisers forum, and the Powerstream testing, so I doubt we will see manufacturers jumping to change existing products, especially when they have a product that would do the lower voltage already at 13.6v like PD does.

[ptourin]

Well back to the drawing board with the MB wiring. I lifted the single wire on the D+ terminal - my understanding was that this wire went to the isolator relay and closed it when the engine was running and there was voltage on the D+ terminal. No such luck - whether the wire is on or off the D+ terminal, the alternator still charges the house batteries when I start the engine.

So apparently that wire does not go directly to the isolator. I'll have to pull the plug to the isolator and try again - there are 2 wires to that plug - I assume 1 is ground and the other is somehow +12 when the engine is running, but it ain't the one I expected!

[ptourin]

I figured out how to open the isolator and turn off alternator charging, and this morning we had a 2-hour drive, so I took the RV and watched. Started at about 90% SOC. Alternator charging looked very much like charging with the charger, but the charging voltage was somewhat lower. Two things interested me:

1. The charger and the alternator behaved in a similar way, and neither looked like those CCCV curves we've been looking at. In both cases, voltage at the battery rose while current dropped, and in both cases, I could see when I got to the shoulder of the curve and the voltage started to rise quickly while the current dropped quickly.

I've defined 100% SOC as more than 13.4 volts and less than 10% current (20A). In the case of the alternator, the voltage started at about 13.3, climbed past 13.4 fairly soon. Charge current started way up, at over 70A, but quickly dropped to around 50A and kept dropping. I'd say that as I approached the shoulder, I was seeing maybe 13.8 or 13.9V when I hit 20A and my monitor said 100%. At that point things were changing quickly, and I opened the isolator and ended the run.

2. Charging was much slower than expected. Say I started at 90% SOC - so I needed to replace about 20Ah. I thought this would happen quite quickly - maybe 30 minutes. But it took a good deal over an hour. So at the beginning when I was seeing 70-50A charging current at the battery, something is a bit funny. I'm not sure why real life charging should be slower than expected - if the monitor says it's seeing 50A and I need to replace 20 Ah, I'd not expect it to take over 60 minutes.

[booster]

That is very interesting, as it appears like your lithiums are behaving more like lead acid, at least up until the knee in the curve.

The part that does seem somewhat like we had guessed was that under 14 volts would probably as high as you would ever need to go in the charge curve. Perhaps a 13.8v charger would be totally adequate, and maybe safer for the batteries, if one wanted to get to the charge level you are achieving.

[ptourin]

What's not clear about that last paragraph of yours is whether you push more charge in by using a higher charge voltage and cutting off at a low current setpoint. I should be able to do a simple test on that: charge with alternator until the current drops to 20A - voltage should be around 14.1 or a whisker less. Let it sit with no load for a day and see what the resting voltage is. Then do the same with the charger - when it gets to 20A the voltage should be up around 14.6. Then see how the system behaves under load after each charge scenario. I should be able to at least get a rough idea. I'm not sure how this will turn out until I get more chance to play with it - for example, I'm not sure whether the charger voltage will have maxed out yet when I get down to 20A. All very intriguing and surprising just at this moment... I'll have a chance to play soon - next friday we're heading to the MD eastern shore for 11 days and I'll get some peace and quiet to watch how it behaves.

[booster]

I think, if I am understanding what is happening in the charge cycle, is that the voltage climbs, current drops, which means the charger is not maxed out or is losing capacity do to heat. If it is not maxed out and is just climbing to your (pick a number) stopping point of 13.8 volts and you shut it off, there is no chance for higher voltage to push in more or faster. If you let it charge all the way to 14.6 volts, it certainly will put in more charge (but not much per the dogleg) but that is what you are trying to avoid by stopping early.

From what you have seen, if you don't go totally full, the voltage keeps climbing as the amps drop, so your voltage setting is essentially not doing anything at all as you are way past it before you hit the amps setting. All batteries have some charging variations based on temperature and stuff, so trying to hit a particular rising voltage and dropping amps point simultaneously could be pretty tough under the varying conditions. That is why I think (and hope) you could get the same results, in close to the same time, by stopping the voltage climb and then going by the amps. If you can, it should be considerably more accurate and repeatable than the double moving moving target scenario.

I agree that the only way you will know how the various methods work is to do the charge cycle and let them rest to stable voltage, which seems to be an acceptable method of determining lithium state of charge. Unfortunately, you wouldn't have the option of leveling off the voltage at any of the potential lower voltages with PD fixed 14.6v charger, and going only by amps after that.

[ptourin]

Yeah, I agree about the voltage setpoint. I define my 100% SOC as reaching 13.4v and dropping current to C/10, or 20A in my case - I do that because I want the system to think that under non-charging conditions, I'm at 100% when my voltage is at least 13.4, since that's the supposed rest full charge voltage according to SmartBattery. But it's totally clear that the charge cycle is going to take the voltage above 13.4 before the current drops to 20A. This is true for both alternator and charger, though the voltage is a bit different when I hit 20A.

So now I need to decide what my charging cutoff should be. I can stop charging based on an SOC setpoint, a voltage setpoint or both. If I use voltage, it's a bit hard to guess just what voltage. If I choose SOC, I'm really choosing 20A because I know I'll be over 13.4V. If I wanted to try to force the highest possible charge, and battery life be damned <g>..., I could choose both and set the voltage setpoint to 14.1 - I know that both the alternator and the charger will get up that high sooner or later.

But enough armchair guesswork for now - we'll see how things work next week - I'll try to get Jean to drive so I can watch the alternator charge process and take readings so I can chart things. I can't datalog yet off the Victron - it's possible, but their app is a Windows app and I carry a Mac on the road.

It's occurred to me that when you set up like this, with a charge cutoff, you're not using the charger to provide house power even when you're on shore power - you're draining the batteries until they ask for the next charge cycle - not like what we do with flooded cells and their chargers. I assume that this is OK because of the longer cycle life of the Lith batteries. Is this the way you see it?

[booster]

Originally Posted by ptourin

It's occurred to me that when you set up like this, with a charge cutoff, you're not using the charger to provide house power even when you're on shore power - you're draining the batteries until they ask for the next charge cycle - not like what we do with flooded cells and their chargers. I assume that this is OK because of the longer cycle life of the Lith batteries. Is this the way you see it?

That seems to be one of the negative side tradeoffs of having lithium, if you want to keep the voltage off them once charged. It is certainly not desirable for most folks. One partial solution is to use a CC/CV charger that does a full cutoff, but them does a recharge based on battery voltage. You could set the voltage to a point that it would recharge before the batteries got very low. In practice this works better on lead acid batteries than lithium because the lithium don't drop much voltage, but it could still work. Another possibility, that I haven't been able to get any answers on yet is if setting a float voltage lower than your cutoff, would be able to power the coach, once you got that low, without the strain on the lithiums do to constant charge. It has been very hard to get anything firm out of the battery manufacturers or the solar charging folks. I think you could also set a low voltage alarm on your Victron at a fairly high level, maybe 13.2 volts, and then just cycle the shore power on and off to start a new charge cycle.

[ptourin]

No time to post a graph now, but I wired up my charging circuit using the Victron monitor to control the charger. I started at 70% and 13.3 volts. I immediately discovered that the highest I can set the SOC relay clear value is 99%, not 100%. My last reading before I got there was 13.75V and 24.7 amps. So I terms of how I set the Victron to measure 100% SOC, I was above my 13.4V setpoint but not down to 20A. This seems pretty nice - voltage not too high, but high enough to get me close to fully charged. I was surprised that the voltage didn't rise higher - I'll post a graph later.

In any event - the control system seems to work reasonably well - now I just have to figure out wire routing so I can finish the job up - no idea yet how to get wires from A to B to C...