I never actually measured the difference in internal resistance of the ones we built and some we got with our EVE LF280K cells. Time to correct this …
The flexible busbars delivered with our cells have approximately these dimensions: 105mm * 20mm * 2mm. This should give us a surface area of 40mm2 and an ideal internal resistance of around 0.045mOhm (according to some online calculators).
When measuring this with our trusty and highly precise TR1035+ the result displayed is 0.08mOhm.
Pre-built busbars 105mm * 20mm * 2mm
The Klauke version which comes with tinned copper lugs gives us a reading of 0.06mOhm – despite the smaller cross-section of 35mm2. The ideal internal resistance of the copper wire could be expected to be around 0.051mOhm (according to some other online calculator), so we are much closer to the measured value.
Custom busbar 105mm * 35mm2
The display error of the TR1035+ is expected to be similar as the measured range of both busbars is quite similar. There is certainly quite an amount of uncertainty (not only due to the maximum resolution of 10uOhm) to it. But it is interesting to see that the resistance of the custom busbars seems to be lower despite its smaller cross-section – while being better insulated (not seen on the picture) and more flexible at the same time (as it can be bent in more directions).
Producing these custom busbars is certainly more expensive (a single lug costs around 1.17GBP or 1.27CHF) and involves more manual and time consuming labour. However, it is quite easy to double them and/or increase its cross-section to up to 70mm2 per wire (when using multi-stranded DIN EN 60228 conductors).
So, this is it for today with fascinating facts about busbars for LiFePO4 cells.
The other day, I realised that I never wrote about the case build of our 16s 48V batteries, as I did for the 8s case and the 4s case. So, here it is – and I am actually describing 2 revisions as we made some adjustments.
First, the total weight of the cells alone would be roughly 16 * 5.3kg ~ 85kg. This is way beyond what a single person can – or at least should – lift. So, I deciced to split the battery into 2 separate cell blocks of 8 cells each (similar as I did split the 8s battery in the Toyota HiAce). With this approach, I would be able to:
reuse the 8s design (including the RAKO boxes)
be able to move or lift half a battery (which weighs roughly 53kg)
This battery has a nominal capacity of 3.2V * 16 * 280Ah = 14'336Wh and can be charged or discharge with up to 140A ^= 7'168W. We currently have 2 of these batteries running on our 3-phase setup with 3 * Victron MultiPlus-II 48/5000/70-50.
So essentially, I built 2 8s batteries with a connection cable between cells 8 and 9. The main negative and the BMS would be in one box and the main positive with the DC breakers would be in the other box. To avoid confusion, in this setup I went for coloured Anderson SB175 housings, with
Red 2 * 35mm2 H07RN-F cable main positive
Grey 2 * 35mm2 H07RN-F cable main negative
Blue Interconnecting both blocks 2 * 35mm2 H07RN-F cable connecting from cell 9 positive to cell 8 negative
In all cases
16s Battery Connectors
To connect the cells to the BMS balancer cables I extended the balancer cables with 2.5mm2 wire via WAGO 221-2411 inline splicing connectors. I then measured the increased resistance of the additional cable length and adjusted the values in the BMS configuration for cells 1 to 9.
With these inline connectors I am now able to disconnect the blocks from each other so I can move them around independently, if needed.
On the BMS, I connected a USB RS-485 TTL adapater with a USB extension cable which leads to one of the USB ports of the Victron Cerbo GX. With the help of dbus-serialbattery and BatteryAggregator I can control the DVCC settings in Venus OS.
The rest of the build is, as I already mentioned, pretty much like the 8s build.
Revision 1
Here are some images of the completed build of revision 1.
16s Battery top view16s Battery Block 1 main negative with BMS16s Battery Block 2 main positive with DC breaker
Revision 2
These are the changes I am currently making for the next revision:
add additional connectors for the balancer cables to further facilitate the disconnection of both blocks;
use 16mm2 M6 Klauke DIN46235 compression cable lugs for the connection of the main negative (cell 16) to the B- of the BMS (only relevant to the older JK-BMS), to be able to disconnect and potentionally replace the BMS;
use a WAGO 35mm2 DIN rail connector in the main negative block on cell 1/9 for the outgoing cable;
use cable glands on the external connections; (this allows for easy disconnection and re-building the block as an 8s battery);
use ratchet straps for compressing and mounting the cells to enable easier maintainability of the cells;
use Anderson PowerPole PP180 connectors instead of SB175, so I can use mounting plates for the PP180 and do not have dangling cables on the outside of the case (these connectors are expensive and increase the price of the overall build by roughly 60GBP).
Now, that we got our Toyota HiAce we thought it might be a good idea to add more power to the vehicle: in form of an 8s EVE LF280K LiFePO4 battery and a Victron MultiPlus Compact 24/1600/40-16 inverter/charger. In the following, we describe our setup and the reason why we built it like this.
The Requirements
The sustained output power of the inverter must be over 1'200W.
Charging via AC via EVSE or generator must be possible.
Charging via alternator must be possible (but is not the norm).
Charging of 60% of the battery (from 20% – 80%) via AC should take less than 180min.
The installation should use the minimum amount of space possible.
We should be able to use our existing Eve LF280K cells, thus limiting the overall current to 140A.
As the vehicle will not have a diesel heater, it should be possible to run a 150W infrared heater for at least 3 * (4+2)h = 18h (^= 2'700Wh).
In addition, the battery should be able to run a refrigerator with an average power consumption of 50W for at least 72h ^= 3'600Wh (next to other power consumption).
Design Considerations
With a maximum current of 140A and a cable run length of 1.5m, we should plan with a cross section of at least 35mm2.
Basically, with Eve LF280K cells we have three choices regarding the battery size:
1* 4s (“12V”) Configuration 4 * 3.2V * 280Ah = 3'584Wh This would lead to a required nominal AC charge power of at least 716.8W/h and a charge current of at least 56A/h.
2* 4s (“12V”) Configuration 2* 4 * 3.2V * 280Ah = 7'168Wh This would lead to a required nominal AC charge power of at least 1'433.6W/h and a charge current of at least 112A/h.
1* 8s (“24V”) Configuration 8 * 3.2V * 280Ah = 7'168Wh This would lead to a required nominal AC charge power of at least 1'433.6W/h and a charge current of at least 56A/h.
The Victron MultiPlus Compact xx/1600VA inverter/charger provides enough sustained power output (while being smaller than the non-Compact edition). Depending on the voltage of the battery, this will slightly impact the amount of charge current.
To charge the battery via the alternator we would need a DC/DC converter that depends on the battery configuration as well (either 12-12 or 12-24). So, let’s have a look at the battery first.
1* 4s (“12V”) Configuration
The smallest, lightest and cheapest configuration. But capacity requirements regarding the fridge are only fulfilled, if there are no other loads. In addition, the discharge current is relatively high (scratching the maximum discharge rate of 0.5C).
2* 4s (“12V”) Configuration
More complex setup, as each battery needs a separate BMS, which leads to the need of an aggregator for both batteries to correctly report SoC and calculate CCL and DCL. In addition, more cabling and fusing is required (and probably to a large bus bar). Comes with the advantage of having a redundant battery in case a single battery fails. Most expensive configuration.
1* 8s (“24V”) Configuration
Custom battery build needed, as there is not enough space for a typical 2 * 4 cells setup behind he seats. But, only a single BMS and thus less wiring is needed. Comes with a slight disadvantage of not having native 12V from the battery. This is actually not an isse, as all our DC devices also accept 24V. Cells can better balance voltage differences across a single 8s bank.
The Setup
In the end, I decided for the 8s configuration, due to less complexity. Splitting the 8s configuration across two cell blocks seemed to be an acceptable compromise.
As a regular MultiPlus 24/1600/40-16 would not fulfill my AC charge requirements, I had to decide to either add a second MultiPlus or to add a dedicated charger. I opted for a Phoenix Smart IP43 Charger 24/25 instead of a second MultiPlus. The MultiPlus in parallel would always consume 10W though most of the time I would not need the output power. Whereas, the Phoenix would only need power, when connected to AC. And reconfiguring the MultiPlus every time I charge was not an option for me. And yes, I lose redundancy – but also save some money (Phoenix is much cheaper). So, in the end the nominal charge power is 40A + 25A = 65A, which lets me charge at 1'560W reaching 60% within 165min.
The HiAce comes with a 70A alternator, so I chose a Orion-Tr Smart 12/24-15 DC-DC Charger. With this charger, I could run the engine in standby and still have the car heater running. And this is probably the predominant use case (if charging via alternator at all).
For the DC bus bar I went for a Victron Lynx Distributor, so I could use and install MEGA fuses. Having a 1’000A bus bar seems certainly overkill, but a separate bus bar and fuse box that accepts 35mm2 cable and MEGA fuses would be not be much smaller.
I changed the existing AC inlet of the HiAce to Neutrik PowerCON True1 TOP (congrats to the marketing department, I am still amazed how this name rolls of the tongue) and installed 2 Siemens compact 16A CRCBOs (external AC in, internal AC out). I am aware that theoretically I could support more than 16A on the internal AC out (via PowerAssist). If ever needed, I can replace the RCBO with a 20A version.
I added a VE.Bus Smart Dongle to the MultiPlus and opted against a complete (Raspberry-based) GX installation. The reason, I keep a USBMK3 with me anyway (in case I need to reconfigure the MultiPlus) and still have (Bluetooth) access to the most important settings and information of the MultiPlus. With the GX, I would to be running a WiFi hotspot (and consuming more energy as well). The disadvanage of not being able to use DVCC with information from the BMS is clear to me and accepted.
I selected a B2A8S20P JK-BMS that has an integrated 2A balancer and an RS485, CAN and heat port. In case, I ever add a GX device, I am still able to connect them and use DVCC.
The Specs
Nominal power (“capacity”) 8 * 3.2V * 280Ah = 7'168Wh
Maximum discharge power 1’600VA (1'280W, capped by the inverter) with a maximum current of 80A/63A/55A (at 2.5V/3.2V/3.65V)
Maximum AC charge power 1'560W
AC Charging from 20% – 80% in 165min
Maximum DC charge power 360W
MultiPlus self-power consumption 10W
The Build
As mentioned before, due to space constraints I had to split the battery in 2 parts (with each having 4 cells). Instead of using utz RAKO boxes I used 12mm (sanded) plywood which I did not screw together but tied down with a banding/tensioning tool and a ratchet strap. With this setup, I can easily access und disassemble the cells if needed, while still having a sturdy case. Both cell blocks are connected with a (blue) Anderson SB175 connector.
The BMS itself is mounted to the side of one of the cases (I took extra care to use short screws, in order not to drill into the cell casing). I used M6 Weidmüller 35mm2 90° angled compression cable lug to get the wire away from the BMS and into the bus bar. All other compression cable lugs are DIN 46235 from Klauke (M6 35mm2 on the cells, and M8 16mm2/35mm2 on the bus bar).
The AC and DC wires are all Eland H07RN-F (except for the last two points):
Charger to bus bar, battery to bus bar: 35mm2
Cell block to cell block: 2 * 35mm2
Alternator to DC-DC converter, DC-DC converter to bus bar: 16mm2
External AC in to RCBO, RCBO to inverter/charger (both directions), RCBO to internal AC out: 3G2.5mm2
For the connection of the Inverter/charger to the bus bar, I used the Victron installed 25mm2 welding cables.
Images
The installation is barely visible behind the seatsView from the back with preliminary wiringConnection of cell blocks with SB175 connectors, cell block 2 and DC-DC converterLynx Distributor with cell block 1Inverter/charger with space for second charger and cell block 2 (left)
Note: the Phoenix charger is not visible on the images, as I am still waiting for it to be delivered.
Charging via EVSE
Conclusion
We now have more than 7'000Wh of additional energy without losing any storage space for roughly 2'850 CHF/2’500 GBP (parts without labour). We can survive an extended weekend of 72h without recharging while still being able to enjoy amenities as using a coffee machine, heating and refrigerator. In case of longer periods of usage, we can recharge at any EVSE, or via shore power. And in emergencies, we can also charge via our Honda EU10i or via the alternator of the vehicle.
The battery is placed directly over the engine which helps in cold weather conditions to easily warm up the batteries to a chargeable level.
Out of the box the MultiPlus comes preconfigured with a pair of 1m35mm2 (2AWG) welding cables (and as a side note: with unusually thin M8 cable lugs).
When connected to a 12V battery based on 4s Eve LF280K cells, the maximum current drawn can go beyond the recommended 0.5C rating – especially when the cell voltage decrease under the nominal 3.2V or the total cable length is longer than 1m. Using a larger inverter or smaller cells will make things even worse.
And when we look at Victron’s Recommended battery cables document, we see that they are recommending 50mm2 for currents up to 150A anyway (for cable lengths of up to 5m).
For a 1m cable the theoretical voltage drop is within their recommended range of 0.259V. But they explicity state that resistance leading to additional voltage drop due to contacts is not calculated into the recommended cables size.
Already a cable size of 2m will lead to over 3% and 0.3V voltage drop when the cell voltage is only 4* 2.6V = 10.4V (and by default raise a “Low Voltage Alarm” on the MultiPlus). And even a cell voltage of 4* 2.75V = 11V is close to 3% and over the recommended threshold (of course, calculation is based on full inverter load of 1600VA). Besides Victron explicitly recommends a voltage drop of under 2.5% in their Wiring Unlimited document.
So why is Victron fitting the inverters with only 35mm2 cables? Especially since they are using welding cables that are only rated up to 60°C. I do not know.
But I do know, how I can fit an additional 35mm2 pair into the inverter and minimise a potential heating problem.
Adding a second pair makes particular sense at least in my case, as I am using a JK-BMS and Eve cells that both come with two M6 terminals per connection point. So running two cable pairs to battery and BMS saves me from using a bulkier and stiffer 70mm2 cable that I would have to split at the BMS and main positive cell anyway. And with that, I can still use the Anderson SB175 connectors with regular housing and 1383 wire contacts and without having to resort to 2/0 housings and 1328G1 wire contacts.
The inverter comes with 30mm holes in the front panel where the supplied cable is fitted with an M25x15mm cable gland (side note: why are they using IP68 glands when the whole inverter is only rated at IP21). Eland H07RN-F 35mm2 cable has a diameter of 14.6mm, so actually two of these cables do not fit through the holes at once.
But as the cable lugs are actually that long that they stick out of the chassis the required diameter is 2* 12.5mm = 25mm which is just the size of the hole. When wrapped in heat shrink we need some more space. And certainly we want a little bit of head room, so the cables do not scratch against the metal when moving.
So, the M25 holes had to be enlarged slightly to make space for the double cable lugs as seen on the picture below. I used a Hilti GDG 6-A22 grinder for this. I covered the inverter to prevent metal splices and dust getting inside (board, circuity) of it. And I added extra insulation around the cable lugs to prevent them cutting into the metal.
Bottom side of MultiPlus with enlarged holes and extra cable insulation
Mounting the cables to the connection points is done with two Klauke M8 35mm2 DIN 46235 compression cable lugs (back to back).
Note: the compression cable lugs from Weidmüller will not fit, as their connection plate is too long.
Instead of the factory supplied washers, spring locks and nuts, I use M8 serrated washers and lock nuts. As the negative connection point (which is directly under the 250A MEGA fuse) is around 2mm higher than with only one cable lug, I added an additional (copper) washer under the fuse terminal to make more space.
I could cover the original bolts with insulation tape to prevent accidental contact with the chassis when squeezed (but this is something that could have happened even before).
Victron MultiPlus Compact 12/1600/70-16 with dual 35mm2 battery cables
Now we have a 2* 35mm2 = 70mm2 connection to our battery as seen below.
2* 35mm2 connection between battery and inverter
So the voltage drop over the whole cable (1.5m from invert to battery with 70mm2) should be around 114mV:
Voltage drop at different cell voltages
So as it seems, the main difference between the larger and the smaller cable is the power loss (17.22W vs 34.44W at full power or 15W vs 30W at 0.5C). So all in all we save nearly 0.84% battery capacity per cycle with the thicker cables (which is 5000Wh over the whole cell life) – probably less than we spend on the cables and lugs, the labour and the time to do the calculation and writing up the article …
The Caravan we got last year did not come with an inverter, so getting coffee in the morning or running a microwave was only possible when our main generator was running. And the installed battery for 12V support had a rather small capacity. This was clear to us from the beginning, as we eventually wanted to connect the Caravan to our EVE 280Ah cells.
But since we got our Starlink internet and our router did not seem to run easily on DC power, we needed -in addition to the temporary morning AC coffee spike – a more permanent AC solution.
Of course, first I updated the firmware of the inverter and configured it work with the battery:
Setting the AC input to 16A
Setting the battery type to LiFePO4
Setting the charge current to 70A (which is over the recommend amount of 50A, but see below for details)
As I did not want to connect a Cerbo GX to the system, I just used the VictronConnect App. Maybe I add a VE.Bus Smart dongle later on, or I connect some GX nevertheless. Who knows … Until now, it needs a wired connection to the inverter to see its status.
After powering on the generator, I confirmed everything was roughly working as expected. During the first run, the SOC was shown as 100% though the BMS of the battery internal saw it differently. In addition, the reported Amps and temperature were seen differently, as well. So, even that I set the inverter over the recommended maximum of 50A for the battery, the actual charge power was never much higher than the actual maximum).
This is what the inverter saw (100% SOC, 14.05V DC cell voltage, charging at 64A):
MultiPlus charging the Liontron battery via the generator
And this is, what the Liontron BMS reported (76% SOC, 13.8V DC cell voltage, charging at 55.5A):
The SOC as seen by the Liontron battery BMS
In the end, the BMS stopped charging when it thought its batteries were full. And the inverter did not complain. However, I noticed that the cells were really not in balance (with a delta of 200mV between the lowest and highest voltage).
Discharging was ok, as well. However, I soon realised that the 100A discharge current could not be achived in my setup. The inverter tried to draw power and the BMS cut off with a “Discharge over-current” (OCD). SO, still no coffee via our Nespresso machine (and no microwave either, for that matter).
So, what is the take away of all this?
It works and now, I can run the Internet all day.
All in all, it is a relatively simple and quick setup.
The Liontron battery does somehow not live up to its specs (and yes, I know the battery could be a size bigger for what I want to achieve; but I did not want to buy an additional battery for this temporary solution).
It is way cheaper and more flexible than to buy this “off the shelf”.
Maybe, I add a Victron SmartShunt to get a more accurate SOC reporting (as I do not see any other way to integrate the BMS with the inverter).
Charging of the battery is quite fast when running the generator.
Our Journey to Scotland 