Mixing lithium with AGM's - bashing around some ideas.

[@Michael]

Quoting this from a few years ago (bold is mine):

Originally Posted by Mfturner

Firstly, I feel compelled to say: y'all be careful out there.

My years of EE work has no battery chemistry so what follows are speculations, moderate temperatures are assumed:

1. If I were to take a fully charged LA battery and short its terminals to a fully charged LFP battery on my driveway, I would expect the LFP to slowly dischare through the LA, probably current limited by the series resistance or acceptance of the LA. My guess is this leakage would need a disconnect to be acceptable, and a BMS might help keep you from totally depleting the LFP. Hopefully this would not get dangerously hot, but who knows.

2. If a load were presented to the paralled batteries while the LFP still had some juice, my money is on the LFP providing most (90%?) of the current until 5-10% SOC (or BMS cutoff) when the LA helps or takes over. You might wind up with a mostly discharged LFP and a partially discharged LA. I think this is what you want to take advantage of the LPF for most of your discharges.

3. If a partially discharged LA battery is connected directly to a full LFP, I agree that the LFP would charge the LA, I buy that. However, I bet that once charged you would need to disconnect them or the slow discharge would eat up what was left in the LFP. I think this is also what some of you want to try, but I don't think it will model the general way paralled batteries will work since I think the LFP would be drained by loads first.

4. I agree that the easiest way to recharge is with independent charging systems while the batteries are disconnected. It may be possible to charge them in parallel, where the bulk charging phase charges both batteries and the absorption only really charges the LA. My no-power solar controller might actually work in this application. But I still bet that without the charger or a disconnect, the LA slowly eats the LFP. I wonder if the disconnect could be a blocking diode, but a diode's Vt would mean the LA wouldn't help until the LFP was very low SOC, maybe less than 5%. Complicating things significantly might be a pmos switch for the disconnect controlled with a microcontroller watching both battery systems to control things, I bet Roadtrek/Hymer is doing something like this, out of reach of most diy'ers.

Here's my current installation:


I have this in place for a bit over a week now. My initial take is that the above from @Mfturner is really close to what I'm seeing. When the combiner is in line, most of the overnight battery drain is from the lithium. If the lithium has juice, the fridge stays running all night. If the lithium is discharged, the AGM is also.

I've started gathering better data so I can see what's happening to both batteries when combined with the diode combiner. I have Thornwave monitors in line with each. Yesterday I bought a Raspberry Pi, learned Python, and wrote a script to connect via Bluetooth and slurp up battery data from both monitors. That'll help a lot.

I'll know more in a few days.

For reference, my fridge draws 5.5A on a 50% duty cycle in moderate weather. Near or over 100% duty cycle in hot (>85deg weather if the camper is in full sun. I know that if I lower the cutoff on the low voltage disconnect I can get a full night out of the AGM's. I'll eventually make that cutoff programable.

[GallenH]

Great. Keep us posted. I still don't understand your use of the Victron combiner. Wouldn't the combiner have 2 inputs and 1 output?

[@Michael]

Originally Posted by GallenH

Great. Keep us posted. I still don't understand your use of the Victron combiner. Wouldn't the combiner have 2 inputs and 1 output?

I'm using it in a non-standard configuration per an idea of @Markopolo starting at this post. Taking advantage of the fixed .3V drop from input to output, put the lithium with its higher voltage on the input and the lower voltage AGM on the output, and let the AGM draw from the lithium as long as the lithium has a charge.

One down side so far is that the AGM's will always be charged at whatever the lithium charge profile is, minus .3v. So one probably would have to charge the banks separately with dedicated chargers or the AGM's will not be properly charged.

Another down (or up) side is that you'll always drain the lithium down faster than the AGM. From what I see with a day's worth of data, 80-90% of the current is drawn from the lithium & the balance from the AGM. Because of internal resistance?.

I'm not sure what will happen if I charge the Lithium directly at a high rate while the combiner is in line. The AGM's will try to stay .3v below the lithium, so I suspect the AGM's will absorb the charge at their maximum rate until they start to taper.

Originally Posted by @Michael

I've started gathering better data so I can see what's happening to both batteries when combined with the diode combiner. I have Thornwave monitors in line with each. Yesterday I bought a Raspberry Pi, learned Python, and wrote a script to connect via Bluetooth and slurp up battery data from both monitors. That'll help a lot.

Here's a chart of watt-hours for both batteries for roughly a day with the fridge running. The AGM supplies somewhere around 1/5 - 1/10 of the total current. Cool, cloudy day, so the solar barely kept up with the fridge.



I spent the day figuring out how to use as Azure Logic App to present a JSON API that lets me add rows to an excel spreadsheet hosted in OneDrive from a UNIX command line. That should allow me to create real-time charts without having to spend money. Should have the parts and pieces glued together soon. Never wrote an API before. Maybe should have just used RRDTool.

[@Michael]

A bit more data.

Seems as though as the system voltage drops, the AGM batteries supply a higher proportion of the total current to the load.

Here's data from last night:

• Fridge running, moderate overnight air temps (75f). Spikes in current are fridge cycling.
  
• Current is on the left axis. (negative numbers == current draw from batteries), voltage on the right.
  
• Voltage drops from 13.1 Li/12.8 AGM to 12.7/12.4.
  
• Li current drops (less negative) as voltage drops.
  
• AGM current rises (more negative) as voltage drops.

Last night I tied the two batteries directly together while charging with the 120v WFCO converter for an hour or so. The charge was roughly split half/half between the AGM and Lithium batteries. I.E the WFCO converter could put out 55A, of which about half went to each battery. I have no theory as to why.

[markopolo]

Really great that you can collect the data like that. Thanks for posting it.

[@Michael]

More charts, this time using a tool called 'rrdtool' & gathering data every 5 minutes.

As before, the lower the lithium voltage, the more of the load is picked up by the AGM's. That's probably a good property. The 'heartbeat' blips are the fridge running, and the big spikes are from being plugged in or because we got a bit of sunshine yesterday.

These two are from Sat am to Sun am. You can see on the righthand half that as voltage decreases, more of the current is supplied by the AGM's untill they supply about equal current. (Negative current == more current).





I also have it set up so that I can automatically load the data up into a OneDrive cloud-hosted excel spreadsheet. Not sure if that's going to be better than rrdtool or not. It's more complex, but has better charting.

[markopolo]

Is the lithium voltage reading before or after the combiner?

The AGM's seem to be more involved than I would have guessed (if I'm interpreting the data correctly). That might be at least partly caused by the added resistance due to the Combiner.

Prior to this data, I probably would have considered giving all solar output to the lithium but now I think it should go to the AGM's as they still need a high standard of care.

[@Michael]

Originally Posted by markopolo

Is the lithium voltage reading before or after the combiner?

The AGM's seem to be more involved than I would have guessed (if I'm interpreting the data correctly). That might be at least partly caused by the added resistance due to the Combiner.

Prior to this data, I probably would have considered giving all solar output to the lithium but now I think it should go to the AGM's as they still need a high standard of care.

The voltages are measured at or close to the respective batteries, so the lithium voltage is on the lithium side of the combiner.

The AGM's contribution is small when the systems voltages are higher & close to 50/50 at lower voltages. And as you indicate, one would still need to have a properly profiled charge on the AGM's somewhere in the system. I put my Victron solar charger on an SPDT A/B switch & created a couple of charge profiles that I can switch between.

I also drove around a bit and tracked how the alternator charged the batteries. If I switch the Load A/B to the 'Both' position, the alternator charges both lithium and AGM at roughly 50/50 when both are discharged. It appears as though I could start the engine, disconnect the combiner, flip the A/B to 'Both', and let the alternator charge both batteries at once without the Redarc B2B. Charging with the converter and the A/B on 'Both' works the same way.

Seems as though for a final charge with correct profile, I'll need the Redarc though.

A valid design would be to have back-to-back B2B chargers with proper profiles and a means of switching between them, so that either bank could charge the other with a correct profile. The lithium -> AGM B2B could be smaller - 10 amps or so. The converter could be directly on the lithium side.

@hbn7hj has a couple of interesting implementations.

[booster]

One thing I think would be good to consider is that the alternator charging will not always be the same in most vehicles, especially the newer ones. The charging control, even if it is in the alternator as an internal regulator, often includes charging controls that are similar, but vehicle use tailored, to the multistage charging we see in smart chargers.


It is certainly not unheard of to see 14.8 or even 15.0 volts at a first start and under 14v running later. It probably would be good to understand how your particular van reacts to extra, and variable, load of shore battery charging in real life, along with normal things like headlights, heater fans, engine cooling fans (if electric), underhood temps, etc.



My guess would be that for any kind of system like this that needs specific charge controls is that relying on the van alternator to supply what is needed on a consistent bases may not be very accurately done.

[@Michael]

Originally Posted by booster

One thing I think would be good to consider is that the alternator charging will not always be the same in most vehicles, especially the newer ones. The charging control, even if it is in the alternator as an internal regulator, often includes charging controls that are similar, but vehicle use tailored, to the multistage charging we see in smart chargers.


It is certainly not unheard of to see 14.8 or even 15.0 volts at a first start and under 14v running later. It probably would be good to understand how your particular van reacts to extra, and variable, load of shore battery charging in real life, along with normal things like headlights, heater fans, engine cooling fans (if electric), underhood temps, etc.

My guess would be that for any kind of system like this that needs specific charge controls is that relying on the van alternator to supply what is needed on a consistent bases may not be very accurately done.

Agree.

In my particular case, AFAIK the alternator on my 2017 Transit isn't all that smart. I've never seen more than 14.4V at the chassis battery, and because the wire from the chassis battery to the Precision Circuits BIM-160 Battery Isolation Manager is about 15' of 8 ga. I'm only seeing low or mid 13's voltage at the coach batteries when the lithium is charging. I'm going to make that a 10' 4ga. run soon, so I'll see a bit less drop in the future.

Also - the semi-intelligent Precision Circuits BIM-160 disconnects after one hour of engine run time, and it claims to prevent "Equalization cycles from Damaging Chassis Battery", implying that it has some sort of overvoltage disconnect.

I expect that if I use the alternator to directly charge the lithium, it'll be only to bulk charge untill the battery is close to full, then use the Redarc B2B to fully charge and float.

On my late model GM trucks, the alternator stops charging while driving, and system voltage drops to the high 12's pretty regularly. I assume that's to conserve fuel.

--Mike

[markopolo]

So the voltages shown are the actual voltages measured at or near the batteries. Are all house batteries connected to the combiner inputs or are the AGMs on the combiner output?

The flat discharge curve associated with lithium batteries is not there that I can see.

[@Michael]

Originally Posted by markopolo

So the voltages shown are the actual voltages measured at or near the batteries. Are all house batteries connected to the combiner inputs or are the AGMs on the combiner output?

The flat discharge curve associated with lithium batteries is not there that I can see.

The lithium is on one of the inputs, the AGM's are on the output.



I deliberately ran the batteries down before combining them, so AFAIK the lithium is showing the steeper voltage drop-off that occurs as it nears capacity limits. I.E. something like 8 hours into the the 10 amp curve on this Battleborn doc



This doc has more readable charge/discharge graphs. I'm assuming that most LiFePo4's are similar.

https://www.batteryspace.com/prod-specs/9055.pdf

Also -

• I assume that the lithium will follow a discharge/voltage curve that is influenced by being tied to the AGM via the combiner, not a curve that it would experience in isolation.
  
• My voltage and current monitors are not calibrated, so errors are likely creeping in. I don't have a means of calibrating, unfortunately.
  
• The monitors are on opposite sides of the vehicle, the two AGM's are on opposite sides of the vehicle, and I have only 6ga. wire connecting everything, so voltage differences due to voltage drop are possible.

Thanks for all the feedback on this. Helps me a lot.

--Mike

[hbn7hj]

How is this going? Do we have a hybrid system grandma can run that we can get the power out of both batteries without intervention? Same with charging?

[markopolo]

This system -> https://www.bos-ag.com/products/le300/ <- has been available since at least 2018 in Europe. It probably can't get easier than that.

For a DIY solution on an older van like mine it could be done like this for example:

GM CS series alternator
120A isolator
PD 45A converter

Alternator to isolator center post.
1st isolator outer post to chassis battery.
2nd isolator outer post to coach battery.
Coach battery to house loads.
Shore charger to Coach battery.
Solar controller to Coach battery.
DC to DC controller with LFP profile from coach battery to LFP battery triggered "on" at 13.6V and "off" at 13.5V via voltage sensing relay.
LFP battery to coach battery via Schottky diode.

[Bud]

Originally Posted by markopolo

This system -> https://www.bos-ag.com/products/le300/ <- has been available since at least 2018 in Europe. It probably can't get easier than that.

For a DIY solution on an older van like mine it could be done like this for example:

GM CS series alternator
120A isolator
PD 45A converter

Alternator to isolator center post.
1st isolator outer post to chassis battery.
2nd isolator outer post to coach battery.
Coach battery to house loads.
Shore charger to Coach battery.
Solar controller to Coach battery.
DC to DC controller with LFP profile from coach battery to LFP battery triggered "on" at 13.6V and "off" at 13.5V via voltage sensing relay.
LFP battery to coach battery via Schottky diode.

Marko, the advantage is cost savings vs all lithium?

Bud

[GeorgeRa]

Interesting product, for systems tuned for AGM there is no change required, just connect the unit.

I couldn’t find costs in US. In Australia it sells for AUD 786 > $564 https://www.buildsolar.com.au/produc...ension-battery.

So, for my system two units would be $1128 using existing Magnum inverter/charge and Morningstar Charge controller.

For cold weather it is relaying on an internal heater so there is time limit for cold storage.

[markopolo]

Originally Posted by GeorgeRa

Interesting product, for systems tuned for AGM there is no change required, just connect the unit.

I couldn’t find costs in US. In Australia it sells for AUD 786 > $564 https://www.buildsolar.com.au/produc...ension-battery.

So, for my system two units would be $1128 using existing Magnum inverter/charge and Morningstar Charge controller.

For cold weather it is relaying on an internal heater so there is time limit for cold storage.

I think each module is 25Ah so $22 per Ah ..........

[markopolo]

Originally Posted by Bud

Marko, the advantage is cost savings vs all lithium?

Bud

Yes. You get to utilize what you already have. Also, having a lead acid battery in the mix provides a needed 12V reference (battery) for a solar controller and inverter/charger. The lithium battery could disconnect itself due to over or under voltage or low or high temperature. You'd be able to turn on lights in the coach for example even if the lithium had turned itself off.

[GeorgeRa]

Originally Posted by markopolo

I think each module is 25Ah so $22 per Ah ..........

Indeed it is expensive. Changing from AGM to Li in my case I would likely had to replace/modify/add to Magnum and Morningstar for $????.

[hbn7hj]

For a DIY solution on an older van like mine it could be done like this for example:

GM CS series alternator
120A isolator
PD 45A converter

Alternator to isolator center post.
1st isolator outer post to chassis battery.
2nd isolator outer post to coach battery.
Coach battery to house loads.
Shore charger to Coach battery.
Solar controller to Coach battery.
DC to DC controller with LFP profile from coach battery to LFP battery triggered "on" at 13.6V and "off" at 13.5V via voltage sensing relay.
LFP battery to coach battery via Schottky diode.[/QUOTE]

Looks like it would be worth a try. I don’t need it but if anyone puts a system like that together let us know how well it works.