question

marekp avatar image

How to balance 2 MP-IIs in parallel

For last 3 months I trying to set-up 3 phase/parallel system out of one EasySolar-II and 5 MultiPlus-IIs.

First, in June 2020 I purchased ES-II and 2 MP-IIs and set-up my 3 phase system.

After couple months I realized that 2.4kW per phase is not enough for my house.

Decided to add 3 MP-IIs in parallel.

Asked my Victron supplier if there is a MP-II still compatible with my MP-II for parallel operation.

In turn he asked Victron the same question and got the positive answer.

"There is no problem to connect the actual produced unit in parallel to existing. The only thing is - firmware must be identic (which means that the units are identic also)."

At the beginning of 2021 I received 3 new MP-IIs.

To make this story short, after setting all up I have a problem with load balancing in all 3 parallel channels.

All wires on the DC and AC are equal lengths from the point they split to master and to slave.

On DC side wires are 35mm2 (around 50cm long from the split point) and on the AC side 6mm2 (around 60 cm long from the split point).

FWs are identical.

I am getting the following results when each phase is, one by one, loaded with 2.4 kW resistive load.

L1 Total 66A, Master 38A, Slave 28A

L2 Total 54A Master 32A, Slave 22A

L3 Total 55A, Master 32A, Slave 23A

Any suggestions how to make the currents equal?

The stickers on the boxes show the model numbers and part numbers of the first and the second batch.

old.jpg

new.jpg




Multiplus-IIparallel
new.jpg (211.0 KiB)
old.jpg (192.3 KiB)
2 comments
2 |3000 characters needed characters left characters exceeded

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

Can you post some photos of the install? Especially of how the breakers are all wired? Are you using bus bars in the DB or do you have wire jumpers?

marekp avatar image marekp shaneyake ·

Thank you @shaneyake,

Here are some picts to answer your questions.

phase-l2.jpg

view.jpg

phase-l2.jpg (219.9 KiB)
view.jpg (446.9 KiB)
1 Answer
Kevin Windrem avatar image
Kevin Windrem answered ·

AC input wiring should also match in length. (Obviously that won’t make a difference when inverting.)

As I understand it, in parallel operation, the inverters all run the same duty cycle so current differences will be due to different path resistances (AC load and DC).

Make sure all connections are tight and maybe even the same torque.

LONGER AC wiring might help.

If all else fails, you could adjust AC load wire lengths to empirically balance loads: increase lengths on the inverter with higher current.


6 comments
2 |3000 characters needed characters left characters exceeded

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

Thank you @Kevin Windrem,

I know that path resistance is important, but as I can control what is happening outside MP-II, I have no control what is inside of it.

My main question is what is the difference in internal resistance between hw rev 12 and hw rev 16.

All connections are tight and since the problem is identical in all 3 phases I would not think that I have here the random problem.

To me it is systemic.

I can play with the wire lengths, but having only 60cm to play with, does not give much confidence that I will get it fixed this way.

Do you know what is the internal resistance of MP-II?

Better question is what is, what is the difference in internal resistance between MP-II hw rev 12 and hw rev 16?

I can also rearrange my MP-IIs so only one phase have 2 different units but that would mean different hardware in one phase.

I agree with @Kevin Windrem here.

There is nothing that can be done about any internal variations to the resistance between the MultiPlus-IIs - but what you CAN do is increase the resistance on the AC output side to make those internal variations negligible.

Either by reducing the AC input and output wiring diameter (if safe to do so), or increasing the length.

There is some more detailed engineering in the Output Resistance section of this document - https://www.victronenergy.com/live/_media/ve.bus:rva_-_theory_on_wiring_large_systems.pdf

It is a common tendency to slightly oversize wiring in power systems - however for parallel Victron systems - this will lead to imbalance issues. So better to get the AC wiring size only just as big as you need and no bigger.

marekp avatar image marekp Guy Stewart (Victron Community Manager) ♦♦ ·

Hi @Guy Stewart (Victron Community Manager),

I also agree with @Kevin Windrem and I already tried to use AC wires with smaller diameter.

If you look at the results I posted, phase L1 had AC in and out wires 2.5mm2 when L2 and L3 6mm2.

As you can see the difference is maybe 1 or 2% on the imbalance values between L1 and L2 or L3..

The document you posted a link too I know and because I know it I asked a question about the difference in internal resistance between hw rev 12 and hw rec 16 of my MP-IIs.

This info will allow me to calculate the length of the AC cables that would make sense.

I would like to know if I have to make AC cables 10 meters long or 1 meter would be sufficient.

Do Victron have the info on the internal resistance of their different batches of production?

I do not thing the problem is related to the single unit since it is identical on all 3 phases.

Hi Marek,

"Do Victron have the info on the internal resistance of their different batches of production?"

I have asked, if this info is somewhere available, but unfortunately it is not.

BR


marekp avatar image marekp markus ♦♦ ·

Hi Markus,

I did some calculations based on the link posted by Guy.

On DC side.

Rin= (1 -η)*Pin / Iin² = (1 -η)*Vin/ Iin

Rin for MP-II comes out to be 50mOhm when eff is 95% and 94mOhm when is closer to reality at 90%.

DC wire

35mm2 has 0.48 mOhm/m and I have 0.5m of it in each branch.

On AC side.

Rout= ((1-η) / η)*Vout/ Iout (when P=3000VA Vout=230V Iout=13A)

Rout for MP-II comes out to be 931mOhm when eff is 95% and 1966mOhm when is closer to reality at 90%.

AC wires.

I have 0.58m in each branch on AC-in and 0.58m in AC-out

6mm2 has 2.8 mOhm/m

2.5mm2 has 6.7 mOhm/m

Do you see a problem on DC or AC side?

To have 2.5mm2 AC wire resistance close to Rout I would need close to 200m of it and it is ridiculous. :(

Where did I make mistake in my calculations?

marekp avatar image marekp Guy Stewart (Victron Community Manager) ♦♦ ·

Hi @Guy Stewart (Victron Community Manager),

After reading this:

There is nothing that can be done about any internal variations to the resistance between the MultiPlus-IIs - but what you CAN do is increase the resistance on the AC output side to make those internal variations negligible.

I took out the calculator and attempted to calculate the length of the wire on the AC-out to "make those internal variations negligible"

Rout= ((1-η) / η)*Vout/ Iout (when P=3000VA Vout=230V Iout=13A)

Rout for MP-II comes out to be 931mOhm when eff is 95%

and 1966mOhm when is closer to reality at 90%.

The currents ratio between Master and Slave is 22A/32A and that suggests the internal resistance ratio is the same.

You might say that outside wires are adding to that restance and you will be correct but.

The presentation, from the link you posted, suggests that main reason for the imbalance comes from AC-out wire difference in resistance between Master and Slave branches.

When the Rout is measured in Ohms the resistance of the AC-out wires is measured in miliOhms, there is no way those wires can have significant influence on the total resistance in the Master and Slave branches.

In this case, Rout difference between hw_rev_12 and hw_rev_16 is so big that there is no way to compensate that difference by making AC-out wires longer even when using their minimal diameter.

Do you still agree with the Victron statement that:

"There is no problem to connect the actual produced unit in parallel to existing. The only thing is - firmware must be identic (which means that the units are identic also)."

Only 2 or 3% difference in efficiency of old and new unit would be enough to cause this imbalance.

Or maybe I made a big error in my calculations.

Please tell me that I did.

Regards,

Marek