nonlinear avatar image
nonlinear asked

Quattro parallel acceptable load deviation

Greetings Community,

I have a quattro parallel installation 2x8kVA running in ESS with a slight imbalance.

The system is newly commissioned and as far as possible Victron guidelines have been followed with respect to battery sizing, cabling sizing, layout etc. In general the system appears to be running well with the ESS assistant. I have two items of concern that may be interlinked:

1. Shortly after commissioning the local grid transformer failed and the system has been running continuously for weeks with no grid. Effectively off-grid. Is running with ESS in this way an issue?

2. There is an imbalance on the inverters. At varying loads (but not full load) I have measured a difference of about 8% between the two units' DC current, and roughly 8% between units AC currents. The lower loaded unit (the master) seemingly more noisy (hum/buzz) than the higher loaded unit. Is this difference sufficient to cause concern?

Other potentially useful information. The system has 800Ah of Pylontech UP5000 capacity, it has a Cerbo unit, 2x 150/75 SmartSolar MPPTs and (for indication only) a BMV measuring at the battery. Although the system is sized at 16kVA it will most likely never be fully loaded, the owner requested suffiecient headroom to ensure the system was not stressed.

I have searched the forums but have not found any definitive answers that state what level of imbalance is of concern and at what point noise should be concerning. Any help or guidance would be greatly appreciated.

Thank you.

MultiPlus Quattro Inverter ChargerESSmultiplus in parallel
2 |3000

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

2 Answers
nickdb avatar image
nickdb answered ·

It varies. 8% is ok. Have you measured under different loads?

My own ones (multis) are around 3% but many factors can affect it and it will never be perfectly symmetrical.

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.

nonlinear avatar image nonlinear commented ·
Thank you for the quick reply Nick. I finally got back there in the late afternoon to do some more measurements and I can confirm that under varying loads the deviation varies between 6 and 9%. The variation is more likely due to the accuracy of my clamp-meter than anything else. If I could achieve 3% that would be great and probably the best I could get given the accuracy of the meter. I may start by re-lugging the DC cables and seeing if that makes a difference. The cable lengths on both the AC side are within 1 or 2cm of each other while the cable lengths on the DC side are within 5 - 10cm of each other. Unfortunately the AC cables are quite short but feed straight onto a common busbar.

I also got around to viewing the ripple on each unit, both seem to toggle between 0.1 and 0.2VDC (at relatively low load though). Again, I could not find a definitive criteria for what is good but I do know that it is below alarm levels so I suppose it is relatively acceptable.

Thanks again for the input and I'll keep going and report back if I make progress.

0 Likes 0 ·
iand avatar image
iand answered ·

There's no kind of active load balancing between parallel units, they just have common PWM bridge drive signals, so the current imbalance depends on the resistance difference between cabling and the internal switching MOSFETs in the power pack.

These have unit-to-unit variation due to production tolerances in the silicon; even though there are multiple MOSFETs in parallel (6 rings a bell...) there will still be significant mismatch. I wouldn't be surprised if Victron use binning to try and match the resistances *within* a Quattro to minimise any residual DC current through the transformer, but there will still be an overall resistance variation from unit to unit.

Anything less than 10% would be reasonable, I would have thought.

2 |3000

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