question

Paul Beskeen avatar image
Paul Beskeen asked

Parallel Multiplus II installation and config

Due to lack of UK G99 certification I have had to swap out the originally supplied and installed 10kv Multiplus II for two 5kv units - which will need to be connected in parallel. I would like to to verify that my game plan for the conversion looks plausible. This system is designed for time shifting electricity, so its currently batteries only, no solar. The Multiplus IIs are 48/5000/70-50 units with very similar recent serial number/dates.

The DC side of the system consists of 6x Pylontech US5000s connected to a Lynx Distributor. These are split into three sets of two batteries each connected via a Pylontech long cable set to the Lynx and 125A fuses. The DC output from the Lynx to each Multiplus II is via duplicate sets of equal length 70mm2 cables. The positive connections are each made via a Marsen 200A fused isolator. Neutrals are directly connected.

The existing AC In and AC Out cabling is 10mm2 60A rated cables. The plan here is to connect each of these cables to a small consumer unit, one for AC In and one for AC Out. These will each hold 2x 32A MCBs and connect to the Multiplus IIs via equal lengths of 4mm2 twin and earth. Cable length for 10mm2 cabling is about 5 meters and for the 4mm2 will be about 80cm.

Hopefully this wiring setup will satisfy the requirements for equal length/resistance AC and DC connections, and individually fused AC connections. I would expect to cut the twin and earth to mm level precision - would that likely provide acceptable tolerance? Should I extend the length or, given such a short length, use 2.5mm2?

I'm expecting that I can config the system before I finally connect A/C side (its a parallel not 3 phase system). Also is it vital to have the BMS CAN BUS connected direct to inverter, or can that be left on the disconnected CerboGX during initial config?

Somewhat confusingly there seem to be several approaches to configuring the parallel system - VictronConnect, VE.Bus Quick Configure or VE.Bus System Configurator. I will need to update fw, install the ESS assistant and config each inverter separately. VictronConnect doesn't seem fully featured enough yet - so probably take the QuickConfigure/VEConfigure route...

I'm expecting that VEConfigure "AC input current limit", "charge current" and any other current defns will be half the required totals. Any other settings that should I be wary of for a parallel system? I also expect that ESS/battery settings done via CerboGX will not require halving.

Appreciate any comments/suggestions.

Paul.

Multiplus-IIESSmultiplus in parallel
2 comments
2 |3000

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Alexandra avatar image Alexandra ♦ commented ·

@Paul Beskeen

Easiest way to program a parallel system is with victron connect (that is the initial config and master/follower assignment)

You will need systen configurator and then ve config for ESS assistant.

Do yourself a favour and run over the training videos on victron professional. And the few short ones on youtube and the blog.

This knowlege is a fundamental basic, so you need to have a good grip on it.

The victron pylon guide is useful for the individual unit programming.

0 Likes 0 ·
Paul Beskeen avatar image Paul Beskeen Alexandra ♦ commented ·
Thanks, yes I have gone through the professional videos. The key question I have is the AC side cabling - will it be within the required tolerance to balance the two inverters?
0 Likes 0 ·
2 Answers
johnone avatar image
johnone answered ·

Hi Paul, some thoughts:

1. If it's a UK project, the MP 48/5000 doesn't yet have G99 certification - this is pending and turning into a saga (see various threads on here!).

2. Are you able to say what the end-use is? Especially re loading? The 27kWh batteries might give you around 24kWh per day. But, some loads can be direct grid use during time shift period (and thus don't suffer from 'round trip' efficiency problem). Is it constant loads you're supplying or small base-load and large occasional load? This may affect cable sizing.

3. re cable sizing, if it's a 230-50v AC grid and 48v DC then the 70mm DC side / 4mm AC side cable sizing might need reconsidering? i.e. crudely, you'd expect the DC cabling to need to be around 5x the AC sizes? (not 17x - 70mm and 4mm).

4. If it's a domestic project, 2x MP 48/5000 will create noise. MP II serial numbers need to be later that HQ2225 to be the new lower fan noise models. The units need to be rubber-mount isolated from building structure to minimise hum.

Hope that helps.

1 comment
2 |3000

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Paul Beskeen avatar image Paul Beskeen commented ·

Thanks John,

1) yes I have swapped certification that is months/years away for the 10kv, for "hopefully" weeks away for the 5kv - but clearly this is a debacle that has let UK Victron customers down very badly. DNO application going ahead with external company handling it - UK Power Networks are a bit more accommodating than most DNOs, so there is a small hope that they will accept the Victron paperwork as is, but this is probably wishful thinking.

2) The end-use is simply timeshifting of the entire household. 24kWh is about what I expect to get - this should cover the background use and higher loads from oven, electric fires, washing during the day - though we do push washing, car and hot water to the cheap rate slot wherever possible. We are on oil so cheap rate elec is still cheaper for hot water and individual room heating rather than whole household. Sizing was calculated from looking at usage recorded by Zappi CT.

3) 70mm2 is the Victron recommendation, the 4mm2 is pretty short and to both of the inverters - my main consideration here is that it should be able to cope with 32A and not be over-speced to (hopefully) be within tolerance to balance two inverters in parallel. The 4mm2 is really my main worry trying to determine if this will be within acceptable tolerance.

4) Its in a home workshop/office, so I can cope with a bit of noise, but its not ideal! I wasn't sure about the exact date of the fan fix, I have HQ2234, so thanks for that - I will look at rubber mounting!

Cheers, Paul.

0 Likes 0 ·
rata1 avatar image
rata1 answered ·

Hi.

I have 2 x 5k MPII installed (under the previous G99) as the 10k was not an option due to the G99. Would have made things so much simpler...I also have 6 x US5000.

I connect mine directly to the grid via a changeover and MCB's on the AC_in and both AC_out. I only use AC_Out1 (connected to the CU) and reserved AC_Out2 for later use. I strongly recommend a change over. I have had to take both of mine off the wall several times and the changeover made it less of a PITA than it could have been. I use two of these (https://www.consumerunitworld.co.uk/fusebox-sco125swb-125a-changeover-switch-and-busbar-5647-p.asp) that allow bypassing of the MP2's completely and connecting a genset if I need to ;) All signed off by a friendly electrician.

I use 70mm DC cable from each MP2 to 3 x Lynx distributors which is a 4m run. Each +ve DC has its own isolator on mine. You mention a single 200A one in your plan. 70mm is fine for upto 5m in the manual. I use 25mm from the grid isolator to the AC_in 100A isolator and which feeds 2 x 50A MCBs. From there I take 2 x 10mm cables to each of the MP2's. The AC_outs are the same - 10mm from each of the MP2s to MCBs that go to another 100A isolator each which feeds the changeover in 25mm. Cut to the mm - very important! Cables by the way are 10mm tri rated not T&E on mine. So much easier to work with, look better IMO but need to be protected it not double insulated.

When doing the config remove the VE cable from the Cerbo leaving it powered and the Pylon BMS cable remains connected to the Cerbo. If it isn't the batteries will turn off. So you have PC - MK3 - MP2 (m) - MP2 (s) nothing else. I use(d) VE.Bus System Configurator - just like the excellent training video referenced above already.

The current limit is half for each MP. I have an 80A fuse in my DNO cutout so set mine to 38A each. Set the charger correctly for Pylon batteries. I set mine to 600Ah.

I see you don't have solar and want to time shift. Currently I do that as there is no sun to speak of, and I can charge from 10% SoC to 100% within the Economy7 window easily as well as heat my water. I set mine to pull 5kw if charging only (5kw is just under 90A so nice and gentle for 6 x US5000) and I add an additional 3kW for water in the last 2hrs as well via MQTT otherwise the battery charge slows down meaning low SoC the next gloomy day. I don't use the scheduled charge.

I run my house all day from 07:30 to 00:30 on the batteries with minimal solar top up. I easily pull loads in excess of 8kw in the day when my wife is in washing, drying and cake baking mode. 6 x US5000 in 17hs is plenty for me in the winter and only once I have gone to 10% SoC and that was at 00:15 so no big deal at peak prices.

As johnone mentioned noise, here is my input - I did a lot of research about noise. It was going to be a show stopper but I went for mounting them on a 18mm ply with 6mm dense rubber sheet glued to the front and back. Each MP2 is on its own peice so not only isolated from the wall but from each other. Works well but I have no benchmark. I just did it at install rather than do it again after the noise problem. They are also in a small room which helps. Not my ideal or planned location but glad they are. When I asked the questions on here I was told not to put them in my original intended location - 100% correct.

I originally had the early MP2's but one had a fault so was replaced. The replacement was incompatible with the remaining original in a parallel pair (way out of balance) so ended up getting that one replaced too. All works fine now (apart from an annoying grid feedback issue) and the fan noise is better. The main take-away in my experience is the units must be as close as possible in manufacturing date and you can't have one earlier than HQ2225 with a later one. Which is OK for your two.

Hope that helps.



3 comments
2 |3000

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Paul Beskeen avatar image Paul Beskeen commented ·

Thanks so much for the detailed response.

Our installation spec sounds quite similar. My AC Out1 is however only connected to my office/workshop and indeed that portion is connected to the sub CU by a Live Electrical changeover switch (which looks pretty much identical to the one you point at).

The grid connection is to the main CU via 60A MCBs with 10mm2 SWA. For this reason I will limit A/C to 60A. The house grid connection is a 100A DNO fuse. I currently have an ET112 connected, in theory, via zigbee, but having a 'mare getting that working. Looks like I will have to swap to CAT5 and RS485/USB converter.

I do have an isolator on each of the two 70mm2 +ve cables, sorry if my description was misleading. The 70mm runs are only 1m in total each.

As you have 2x 50A MCB->10mm2->MP2 it sounds hopeful that 2x 32A MCB->80cm of 4mm2 ->MP2 (4mm2 is rated to cope with 32A) should be within tolerance to balance the two inverters - how long are your 10mm2 runs though?

As I have already mounted the inverters on wall via 18mm of OSB, I'll first see how much an issue the noise is - and then, inevitably perhaps, try sound isolation... I'll not do final fix down of the cabling until I am confident one way or the other. :-/ I assume your approach relies on the glue strength of the rubber/plywood interface, or do the mounting screws go all the way through to the base wall?

Thanks for the config comments, will start on that now.

Good to hear your system is working well for you - especially as the setup looks so similar to mine :-)

Can you elaborate on your grid feeback issue?

Regards, Paul.

0 Likes 0 ·
rata1 avatar image rata1 Paul Beskeen commented ·

Hello,

My 10mm cables are all the same - 2.6m IIRC.

The 4mm cable will be rated differently as to how it is fixed and how it is made. That is, single core, T&E,clipped direct, bundled with others, in conduit etc so the de-rating factor needs to be taken into account,

Have a look here: http://www.cable-ratings.co.uk/ if you don't know. Your electrician who will do the final connect will know...:)

Would 6mm be a better choice? I used to have a 60A main fuse and still used 10mm as any upgrades I did would mean I never had to replace. I restricted the inverters to 28A each. Then I had a smart meter and the fuse upgraded to 80A. No rewiring needed and I changed the inverters to 38A each. It is unlikely that I would ever get a 100A fuse as the service head would likely need "work" but I would still be fine with 50A per inverter on 10mm. Like to future proof ;)

There is a mention of using overly thick or short cables in the Wiring Unlimited in relation to parallel and 3P systems - so I would see what happens and make it longer if needed.


I use the internal CT sensors in my MPs as they are sat between the grid and the loads so don't use/need the ETxxx sensor.

My issue is that I bleed back between 600 W/h to 1kW/h from the batteries some days but only pull 10's of W/h from it (when not charging). This is without sun and export disabled. A kW/h is a fair bit really when it should be going back to the batteries.

I think it is the way the pair fight with each other and don't react quickly enough but posted a question here to get an answer as different configs / ext CT produce different results. My MP2's are balanced to a few percent which is important. It seems to work fine on a single MP2 from what I read so to me it is SW...

I won't care in the sunny months ahead but my OCD means I have to fix it ;)

0 Likes 0 ·
Paul Beskeen avatar image Paul Beskeen rata1 commented ·

Hopefully I wont suffer too much from that - although there will always be some overshoot as the inverter tries to balance the load. Finally got my zigbee ET112 connection working. Turned out the installer had broken the RS485 wiring on the ET112 to Zigbee connection. Hmm, I wonder if it is worth trying to get the sensor working at 19200 rather than 9600 baud - an experiment to perhaps try once I have it all up and running reliably!

Paul

0 Likes 0 ·