everything worked fine for three years , up to this morning ...when I try to start the engine and the battery wasn't providing enough current to the starter motor. after few minutes of generator , the engine started and I will replace the battery ASAP...

...but I kept thinking about why it happened, and I think I have an explanation.

yesterday it was very hot and I ran the aircon from batteries. I stopped the aircon around 40% about, but it went down overnight to 12% (usually , I never go below 30%).

in my system , the lithium battery is charged by the same chargers (the masscombi  and the solar panels) charging the agm, and I never thought what this means: they are subject to the same voltages, therefore they are in parallel!

let's forget for a second about the charger , because that's not what was happening last night.

I checked all the relevant voltages of the discharge cycles and we have three configurations.

lfp soc > 15% (temp. dependent). the service battery has a voltage higher than a fully charged agm battery, so it will  trickle charge the agm batteries all the time. the agm batteries therefore never "float", and this reduces the lifetime. I somehow knew it ...but accepted it. this is the normal situation...

SOC around 15%: the voltage of the lfp start going below the fully charged AGM battery , and the agm battery starts charging the lfp battery .. that's 10x bigger , so it's voltage starts dropping and you have an equilibrium for a while...but the agm starts emptying fast ...

SOC below 15%: the lfp voltage drops faster than the almost linear Agm voltage drop , but essentially the AGM battery empties into the Lfp battery than has a larger capacity. now the engine may not start anymore  !!! 

conclusion: I never explored this regime , but it's clearly not good. the agm battery can be fully discharged by the service battery when the  SOC goes  below 15%, and that's not what you want.

solution: I will install either a battery charger (like the mastervolt mac plus 12/12-50) or a converter (like the mastervolt magic 12/12-20) to charge the agm batteries...I prefer mastervolt because all components (batteries , mppt, inverter) are mastervolt and I have a czone monitor .

in both devices  , I can implement floating for the agm batteries , improving their lifetime and avoiding them to be discharged into the Lfp service battery.

lesson learnt , I hope...

addendum: studying the masscombi 12/2200-100 manual I discovered that there is a secondary 5A charging port...but in my opinion the boat schematics says it's not used , and the battery is attached directly to the main 12V main bus . Am I right or wrong ?

Paralleling battery chemistries always seemed like a very bad idea to me but there people our there evangelical about paralleling Pb batteries with LiFePO4 to mitigate for BMS going into shutdown when real answer is to have a BMS/monitoring system that alerts user to a potential shutdown condition (as now required by ABYC .. if you like in North America) before the shutdown occurs.

Wonder how your Pb/LiFePO4 batteries parallel. Is that through a battery combiner or a function within your charger? Some (most) auto-voltage sensing battery combiners need upgrading to LiFePO4 compatible models otherwise engine/service parallel at too low a voltage.

Sounds like DC-DC converter to charge Pb from LiFePO4 is the answer. 

Then , isolating the battery with a dc/dc converter  would it mean that the  alternator stops charging the starter battery directly at high current ? in that case , is 5A trickle charge enough ? 

I have a generator, so these are in consideration of that. Without one, you may go in another direction.

Option 1 (easiest and cheaper of the two)

Alternator from engine is connected to AGM start battery.

DC to DC charger (30-50 amp limited) is connected to house bank.

Original battery combiner is removed.

In this setup, the DC to DC limits the house bank from over drawing the alternator and thus keeps it from overheating. The DC to DC also only operates when the engine is running so that the house bank can’t drain it down to zero. The alternator is protected from a BMS shutdown by virtue of also being connected to the AGM start battery which dampens the shock. The drawback is that my 125 amp alternator is not being used to its full capability in charging the house bank while running. My generator capability backs that up when needed. Also, this eliminates the charging of the AGM start battery on shore or generator power because the DC to DC is one way to the house bank. This limitation can be overcome with a small battery charger which is wired to the shore/generator only output from the inverter/charger. Alternately, you could just wire one solar panel to the start battery only to keep it topped off. I have been told that using the small trickle charger output of the inverter/charger would not be advisable in this setup because when the DC to DC is running, there is a circular current path with that trickle charger - still a little fuzzy on that. This trickle charger is fine to use with the standard all AGM setup but is unnecessary because of the battery combiner which does the trick.

Option 2. (More complicated and more expensive but probably the best setup)

Remove the battery combiner.

Upgrade to a bigger alternator or keep the same one and install an alternator regulator like a Wakespeed, Sterling or the new Arco Zeus (this one is designed to be used with the bigger alternator for faster charging at idle engine speeds).

In this installation, the alternator charges the house bank and the alternator is kept from overheating by the regulator and you can use the full capability of the alternator for charging the bigger house bank directly. A DC to DC is then used to charge the AGM start battery. The alternator is protected from a BMS shutdown by the BMS itself which shuts off the alternator. All the batteries get charged anytime with either shore, gen or engine running.

I don’t run AC’s off of the batteries, so that’s not a problem for me. My bow thruster AGM is charged from a small DC to DC off of the house bank which would just stay that way.

From what I have learned, the biggest problem with just keeping the battery combiner and not using a DC to DC between the alternator and lithium bank is that the large lithium house bank can easily overwhelm the alternator because of its very high charge capability. In essence, it can take everything the alternator can give which will cause an over temp and destroy the alternator.

Disclaimer: I am not a marine electrician or even an expert, only an internet researcher (yikes!)

Take a look at the MasterVolt ChargeMate. If you have this installed instead of the battery combiner, it will prevent the house bank from draining the AGM start battery. The drawback is that you won’t have charging of the AGM start battery unless the engine is running. You could solve that by installing a small battery charger which runs off of the short/gen only output of your inverter charger.

“ This looks like a right setup ( the one with 3 DCDCs and all charges to service), one only has to worry about shutting off the alternator if service Li bank gets overcharged and BMS wants to turn off all charging sources, I think smart alternators are capable doing it.”

 In the setup which we talk here the alternator and all other charging sources are directly connected to service bank.For alternator one can lower the voltage of the combiner as Arcadia said (not sure if I can reprogram the victron combiner for this) or just remove the combiner. At the same time DCDC charger is inserted to charge engine battery same as DCDC chargers do for thrusters. I think its a nicer setup which alllows proper charging profiles for all components. I have to think more if there is any downside going that way but its easy to hook everything this way, just need to buy another DCDC from Victron.

this could change my analysis. let's assume it works correctly: the lithium bank went down in voltage , the starter battery was disconnected from the bus...and my battery is "tired", e.g. subject to sulfation because I don't float it (my settings are for lithium). then it didn't have the crank capacity that it has when the two buses are connected...thanks to the lithium!

now I also understand why Leon insists on moving the alternator to the service battery. the battery combiner may separate the alternator from the service battery, making recharging the service battery when the service battery goes down impossible with the engine. still , in my boat solar panels and generator will charge it.

in practice, what does it mean to move the alternator onto the service battery: deploy 200A cables from the engine room to the service battery?

For the Victron Dc-Dc, the manuals say to mount close to but not above the battery (which is a lead acid thing with potential off-gassing). Heat is an issue. The smaller 30 amp units apparently get very hot and many people in vans and boats installed ventilation fans. The newer 50 amp units don’t get as hot but the manual says it can get up to 90C and to mount on a surface that can withstand that temperature.

I think it works fine with AGM’s on both sides of it. There is built in delays for connect and disconnect. It engages at 13.2v and disconnects at either 12.8 after 60 seconds or 11.8 after 4 seconds. I would also think, that with the alternator running and putting out good amperage, even the discharged lithium bank wouldn’t pull the voltage down below 12.8 for a solid minute, but I could be mistaken on that. In any case, this is really an AGM device. Most lithium designs these days don’t incorporate a battery combiner at all. I’ve even seen setups where people completely separate their AGM start battery from their lithium service bank just to simplify things.

I think the big stumbling blocks with adding lithium to our previous all-AGM boats is the different chemistries/ charge profiles and the fact that the alternators were designed for lead/acid batteries - they don’t have lithium charge profiles built in and they don’t have an installed shutoff that the new lithium BMS can activate when needed to safely cut off the charge under any circumstances. So, we need to start jury-rigging things by adding or subtracting devices to overcome the stumbling blocks. I’m still struggling with the possibilities with an eye for keeping it as simple as possible while still covering all bases with the minimum of modifications and money spent. It’s a challenge!

I think one can just replace the combiner with a solenoid, something like Lewmar el.winch contactor would have a right amperage for this. In this case as long as BMS allows charging sources, the contactor is closed, and the alternator charges the service bank. When BMS signals to stop charging, the contactor will just open. 

you can see the manual here : it's configured in automatic mode.

https://images.mastervolt.nl/files/ManualChargeMate2502_090205.pdf" rel="nofollow - https://images.mastervolt.nl/files/ManualChargeMate2502_090205.pdf

It seems it has histeresis and delay features preventing flipping on and off, and very interestingly one could both wire a manual switch , overriding the off position allowing to start on the service battery , and even an indicator /czone information to know what's happening without taking the sofaaway ...

I checked all the cables , and I confirm whatever had been said above by 32mike, Marsella and Arcadia.

in my system I have two additions: 2x mppt, on the service bus , charging the service battery, and another battery that's used to start the generator.

in particular , the agm generator service battery is completely in parallel of the engine battery. there is no battery splitter. I was puzzled by this : when I had the problem I couldn't start the engine , but I could start the generator. this is probably simply because the crank requirements are lighter. then , I realise I have an old agm battery in parallel to a new one : I'm sure it's a bad idea , the new battery may be continuously discharging into the old one ...

going back to my original problem , I start suspecting that the sequence of events  was:  the service battery went below 13v, it disconnected from the engine battery and the generator battery. then the engine battery discharged partially even if fed by the generator battery (higher internal resistance because of lower CCA) ...and both couldn't provide the cranking current. when I started the generator , the splitter went on and the 10kWh lithium batteries covered the engine battery problem...

now I need surely to replace the engine battery , it shows some sulfation signs on the contacts . 

..and now I know that I can overwrite the battery splitter and use the service batteries to start the engine!

as a last piece of information , I even have 2x masterbus controlled isolators (ml-rbs from ocean systems) on the lithium batteries: they are programmable with masteradjust, and I currently use it for last resort discharge below 10% and thermal limit. it rarely happens , and I think this is what is described here as a abyc requirement.

ultimately , I could redesign the system ...but admittedly it worked flawlessly for 4 years , up to the moment the engine battery got old (the battery was used for 5 years ) so you could understand my hesitation: don't break what works ! 

I know the current limitations now:

1) agm batteries , generator and engine , aren't optimally using ideal bulk/abs/float timings and voltages...but this deserves another thread, because the differences been lfp and agm settings are not very big.

2) the two agm batteries are in parallel without a split. is this what I want/like ? the two batteries aren't identical in capacity...

3) potentially and theoretically, the lithium batteries could disconnect from the alternator when the engine is running creating load jumps that may affect the alternator. admittedly , this should happen only on very rare situations, and the BMS should anyway have priority because of safety