question

If you did that, you'd need to tell the BMV that you have a 3000A/50mV shunt, not a 333A/50mV shunt. You have applied the maths for resistors in parallel, the current carrying capacity of the three 1000A shunts in parallel goes up to 3000A. In other words, you'll need 3000A amps to flow to get 50mV drop across the parallel trio.
Also, getting equal current sharing amongst the three separate parallel shunts is going to be tricky. You'll only be connecting the BMV to one of the three, so it will almost certainly be inaccurate.
Having said that I'm sure I saw reference to a 6000A/50mV shunt somewhere. However, this introduces lower resolution measurements. Now, 1A is only bing represented by 8.3μV (16.6μV for 1A with a 3000A/50mV shunt). I don't know how many bits the ADC in the BMV is, nor do I know what the ADC reference voltage is, but I wouldn't be surprised if you loose your decimal places when using a 6000A/50mV shunt. The single decimal place you get for Amps when using 500A/50mV shunt is being represented by just 10μV
Those great big shunts are really only designed for massive installations where 2000A is a daily average and nobody cares about decimal places.

I understand where you are coming from regarding minimising losses, but really, with a 500A/50mV shunt with 500A going through it you are only burning 25W.
I'm not sure about the size of your system. Do you get up in the hundreds of Amps at all?

It's really a case of sacrificing (low) losses for resolution.
I hope I made sense there! :)

EDIT: Yes the Analogue to Digital Converter is inside the BMV, not the shunt. I'm pretty sure there are no amplifiers on the shunts either.

Nice to stir up some healthy dialogue.

@ peterpolz

They are 3 physical banks, but they are a single electrical bank.

Can you monitor 3 separate banks? I don't know.

I'd settle for that even, but I don't think you can. I don't think the ESS accepts that.

Leastways I have never seen a 3 battery bank example on the WWW.

So, I believe one would need a single BMV.

Open to comment on this point though.

@ Wkirby

Thank you for your reply.

I think I was clear in my original post that while the primaries of the shunt are in parallel the secondaries would be in series. I am aware that I want to operate in the accurate resolution "zone".

I've considered it further.

We are in fact both incorrect, as the shunt would neither be 3000A/50mV nor 333A /50MV

It would still remain a 1000A/50mV.

An Amp generates X mV regardless of which shunt it flows through, all these voltages are summated in series.

300A =15mV through a single shunt, but through 3 shunts the current would be thirded, a 100A through each shunt = 5mV +5mV +5mV =15mV. Same thing.

The driver for this is to overcome the physical limitations of my cabling, losses would be identical.

"EDIT: Yes the Analogue to Digital Converter is inside the BMV, not the shunt. I'm pretty sure there are no amplifiers on the shunts either. "

If you are correct I have my answer: it can be done.

@Phil_Gavin Yes i do can do that. I simpel read out values of each BMV over modbus from venus, recalcuate them and set on over SOC by modbus as corrected VE.Bus SOC, This was handled by one connected arduino to lan. That's the Geekway :)

If you are not able to do that, people simple take one BMV for each bank an one common for all.

@Peter Polz

"handled by one connected arduino" - I want to avoid this if there is a more elegant solution.

"people simple take one BMV for each bank an one common for all"

Again it is electrically only one battery bank, that single battery bank is physically arranged in 3 groups. I only care about the overall SOC of the entire electrical bank, but if I could monitor one common BMV for all, I wouldn't be asking this question.

Where is the Problem? Split up connection after Busbar .. connect all 3 connections to one Shunt and connect shunt to another Busbar.. Problem solved for you. If you dont want to monitor each bank seperatly why do you want to take 3 shunts? may you show us some Pictures of your installation?

Also again it is an absolutly fatal installation mistake if you don't monitor each parallel connected bank in an big system with parallel conneced banks, thats not the way you should handle such a sytems, becaus your overall SOC will be completly wrong if one bank make problems and as a consequence you are not able to monitor that one bank fails

Here is a diagram of what I hope is correct in my understanding of what you have in mind.

Top left: The basic illustration of measuring voltage over a shunt, I'm sure you know this.

Bottom left: An illustration of how I understand your thinking. Seems like it should work.

Bottom right: Closer scrutiny reveals that the sense wire connections are effectively bypassing two of the shunts.

Does this make sense (pun, sorry)?
I'd actually be wiling to conduct an experiment to provide a practical demonstration of this.

Thanks, but as per my original statement:

"The long and the short of it is that there is no real bottle neck of cabling that is suitable to add a single shunt."

remains true.

The number of cables(12 pairs), their sizes, and their separate routes precludes this option.

I don't want to diverted off on a system layout/design tangent. I have the layout I have.

Yes, things can go wrong with parallel strings, but the budget dictates that those bridges get crossed, if and when they arise.

@WKirby

Yes, something along the lines of your proposed setup is what I had in mind.

Interested to know your findings.

One thing to think about is that the individual banks may contribute current at rapidly oscillating levels. That's what I observe on my parallel strings, when I put clamp ammeters on them individually while the whole pack is under a steady load.

The computer (BMV) samples the shunt at regular intervals and sums the results to estimate net energy. If there is any sample-and-hold going on at the sense PCBs, I think you could get a very inaccurate sum if they have a timing issue.

On the other hand, if they are just passing through the (analog) value, I would guess that would be okay. From memory, those PCBs just have a few discrete analog components on them, but no digital ICs. So probably what you're proposing would work.

I'm having a hard time imagining a layout where you don't have everything bussed up somewhere. Is that not the case? At that point, is there just no physical room to install a single shunt? I know you are tired of people asking you to do something you've already ruled out, but your response in that area seems to be focused on all the 24 wires running everywhere, when really it is the wiring after the bus-up point that is in question.

@ben

There are 12 x 16mm2 cable to the positive and negatives of the batteries respectively.

The MPPT's and the inverter cabling is also on these busbars.

If am to go to trouble adapting the cabling I would need to mount two more mini busbars with a shunt between them, because the smaller shunt wont accommodate that many connections.

Then if I am prepared do that then I may as well mount the 6000A/50mV shunt which will accommodate that number of connections and cut the fins to bring it down to an acceptable resolution.

I could re-calibrate with a DC clip-on ammeter and set the BMV accordingly.