Avoiding Excessive Voltage Drop from Batteries for a 1000A design

[ricardocello]

Today's Topic: Avoiding Excessive Voltage Drop from Batteries for a 1000A design (thanks @A-P)

For context, see my other thread.

Thread 'Powerhouse Planning for Off-Grid (Victron)'

Jul 29, 2025

This thread will document my planning for an off-grid Victron Powerhouse.
Actually building it will be in a different thread, probably for next year.

Background
I already have a fully functional ESS grid system covering about 25-30% of our energy needs, the rest comes from grid.
This current system will only be usable while my utility meter remains stuck in the 90s.
After the PoCo replaces it with a smart meter, I'll be way short on batteries to run non-ESS.

This is a very high consumption all-electric house due to the number of people and usage.
I am now planning a new system to...

• ricardocello
• Replies: 65
• Forum: DIY Solar General Discussion

• 
• 

Note that I am a newby on building DIY batteries, so be gentle. Everything I know I learned here.
Some basic numbers and math first.

• EVE LF280K cells are specified to have <= 0.25 mΩ internal resistance. Let's use that as a representative number.
  I'm aware that it can vary with temperature and discharge rate, other cells are better, etc.
  So a 16s configuration will have <= 4 mΩ of internal resistance.
  
  
• The Overkill Pathfinder BMS specifies:
  Total series resistance, all FETs active (+Cell to B+) 0.351 mΩ
  
  
• And for convenience, each string will have its own Midnite MNEDC250 breaker (50 kA AIC @ 125 VDC). EDIT: Considering 125A instead.
  Midnite MNEDC250 internal resistance < 0.25 mΩ according to the Littelfuse Carling F-Series data sheet (eyeballing log-log chart).
  
  
• This brings us to 4.6 mΩ per string.

Let's say I want to put 3 of these strings in parallel per rack.

• That brings the resistance down to 1.53 mΩ per rack (because resistance in parallel is divided).
  
  
• Now let's add a big honking Dihool DHM3Z 250A MCCB w/remote trip (for emergency shutdown) in the rack feeding a Lynx Power-In.
  20 kA AIC @ 1000 VDC. Dihool DHM3Z internal resistance is 0.9 mΩ according to my YR-1035+ meter tested on a 320A device (total of both poles). EDIT: Clamped the probes onto the terminals directly, provides lowest number.
  
  
• And for super-duper safety, also an inline Class-T 300A fuse as well. Yes, I'm splurging.
  The JLLN300 Class-T fuse (20 kA AIC @ 125VDC) measures 0.29 mΩ when cold.
  
  
• 1.53 + 0.9 + 0.29 = 2.72 mΩ per rack (including 3 16s EVE LF280K strings in parallel, breakers, Class-T)

 

[ricardocello]

EDIT: Numbers have been corrected based on better YR-1035+ measurements.
EDIT: Forgot that each rack only sends 250A, fixed now, thanks @outsider.

OK, now let's run some 4/0 cables, which typically have 0.05 mΩ/foot of resistance.
And a generous 20 feet for + and -, so 40 feet total, meaning 2 mΩ total.
Let's add 1 mΩ for total lug resistance, thermal effects, and other losses.

So from each rack to the Lynx PowerIn is a total of 5.72 mΩ.
So running 250A through it will drop 1.43V !
I plan to run four racks in parallel into the Lynx PowerIn to get to 1000A.

Maybe that will work, worse at lower SoC where the voltage is already lower.
The 48V cutoff could cause the inverters to drop the loads.

Even if the cable were beefed up to be a superconductor, there is still a 0.68V drop from just the rack.

So the only reductions are to parallel more batteries per rack, or potentially remove the Class-T.
Or use something bigger than 4/0. Or keep the distance as short as possible.

One way to look at this is to realize that for every milliohm of loss, at 250A, you lose 0.25V (per rack).
I'll follow up on this later today.

 

[outsider]

Where do you see that a single conductor of 4/0 is good for 1000 A?

According to this table here https://www.encorewire.com/products/tools-and-resources/calculators/wire-size-table.html 4/0 wire is only rated to 290 A at most at 90 C. So according to this, if you had to use 4/0 wire, you would need at least 4 parallel runs to reach 1000 A capacity, and this would cut your wire resistance by a factor of 4.

Also, depending on how the wires are run together (e.g. in conduit or something) it may be that you would need even more than 4 x parallel wires for the current. (I'm not an electrician.)

 

[ricardocello]

Where do you see that a single conductor of 4/0 is good for 1000 A?

Because I'm an idiot! Each rack is meant to handle 250A, with 4 racks in parallel being 1000A, combined at the Lynx PowerIn.
Fixing it now.

 

[Brucey]

OK, now lets run some 4/0 cables, which typically have 0.05 mΩ/foot of resistance.
And a generous 20 feet for + and -, so 40 feet total, meaning 2 mΩ total.
Let's add 1 mΩ for total lug resistance, thermal effects, and other losses.

So from the rack to the Lynx PowerIn is a total of 8.62 mΩ.
So running 1000A through it will drop 8.62V !!
That's not going to work!

Worse, even if the cable were beefed up, it is still a 5.62V drop from just the rack.

So the only way out is to parallel more batteries per rack and potentially remove the Class-T.
One way to look at this is to realize that for every milliohm of loss, at 1000A, you lose a volt.
I'll follow up on this later today.

Hoenstly I don't think you could use a lynx bus system at 1000A just based on the cross section. I think someone on the Victron forums did the calculations couldn't be done without active cooling. Every build I've seen with large multiple quattro builds seems to use custom busbars.

Personally I'd do it this way:

Midnite 2000A combiner

Each battery gets an mnedc250rt shunt trip in there.

Then you come out of that battery combiner to second combiner using say a couple of big mcm connections, second combiner that's going to handle your inverters and sccs, also using mnedc125rt, 175rt and 250rt.

Basically a case where the twelve big mnedc breaker positions in a single mndc 2000 is insufficient.

Now in this design you dont have ocp within 7" of the battery terminals, but with the mnedc250rts in the combiners protecting the long cable runs I don't see an issue.

But with single activation of shunt should be able to disconnect all batteries and loads.

 

[Brucey]

Because I'm an idiot! Each rack is meant to handle 250A, with 4 racks in parallel being 1000A.
Fixing it now.

Ah ok maybe a power in as a rack combiner then 4/0 run to an mnedc250rt in the mndc 2000. Then you would only need a single combiner.

 

[ricardocello]

Hoenstly I don't think you could use a lynx bus system at 1000A just based on the cross section. I think someone on the Victron forums did the calculations couldn't be done without active cooling. Every build I've seen with large multiple quattro builds seems to use custom busbars.

I hope never to be anywhere near 1000A, but I want the design to be rated for it.
Victron claims 1000A continuous through those busbars, but they don't say if you need a fan!

Personally I'd do it this way:

Midnite 2000A combiner

For those following along

https://www.midnitesolar.com/pdfs/Brochure,_MNBCB_Series_Rev_B.pdf

You know that lists for $1581 for just the box, right?

 

[ricardocello]

For those following along who need shunt-trip breakers for your batteries.

Note that the MNEDC250 breakers have lower internal resistance than the Dihool DHM3Z.
They also come in the RT (remote trip) variety, and cost is similar, so I may punt on the Dihool.
The Dihool is two-pole, however, if isolation is really important to you.

 

[Brucey]

I hope never to be anywhere near 1000A, but I want the design to be rated for it.
Victron claims 1000A continuous through those busbars, but they don't say if you need a fan!


For those following along

https://www.midnitesolar.com/pdfs/Brochure,_MNBCB_Series_Rev_B.pdf

You know that lists for $1581 for just the box, right?

View attachment 322957

Yeah $1289 here:

MidNite Solar MNBCB 2000/200 Battery Combiner Box 2000A

Shop now for MidNite Solar MNBCB 2000/200 Battery Combiner Box 2000A

www.stellavolta.com

Here the thread I was talking about:

Can Lynx really do 1000A?

For Lynx maximum current specification, you also need to take into consideration the current cutting capacity of the fuses. Maxifuses have only 1000 A cutting-current capacity. This means that any short-circuit current above 1000 A (even for a 25A fuse) may result in a permanent DC arc over the...

community.victronenergy.com

 

[ricardocello]

Here the thread I was talking about:

Can Lynx really do 1000A?

For Lynx maximum current specification, you also need to take into consideration the current cutting capacity of the fuses. Maxifuses have only 1000 A cutting-current capacity. This means that any short-circuit current above 1000 A (even for a 25A fuse) may result in a permanent DC arc over the...

community.victronenergy.com

Yeah, reading that makes we want the big metal box.

I do have a separate 1000A fuse selected, however.
Mersen 1000A A15QS1000-4 aR 50 kA @ 150 VDC $155 Grainger

I guess I should find out what its resistance is. I'll lose a volt for every milliohm at 1000A.

 

[Brucey]

You're doing great seriously solid math and planning for a “newby”! At 2.72 mΩ per rack, you’ve built an impressively low-resistance system, especially for DIY. Your attention to detail with internal resistance, quality breakers, Class-T fuse, and even emergency shutdown via MCCB shows you're not just building safely you're future-proofing. Keep at it this setup is clean, scalable, and smart!

This has to be some kind of AI based on the post history.

 

[ricardocello]

You're doing great seriously solid math and planning for a “newby”! At 2.72 mΩ per rack, you’ve built an impressively low-resistance system, especially for DIY. Your attention to detail with internal resistance, quality breakers, Class-T fuse, and even emergency shutdown via MCCB shows you're not just building safely you're future-proofing. Keep at it this setup is clean, scalable, and smart!

Thanks! I’m only a newbie on the DIY battery stuff.

I’ve never bought any cells, only rackmount batteries.
Because DIY doesn’t have to be DIY for everything, there is a build/buy decision at each step.

 

[ricardocello]

This has to be some kind of AI based on the post history.

I’ll take the compliment either way.

 

[Brucey]

I’ll take the compliment either way.

Maybe a non English speaker using it to translate. He's not wrong

 

[agtcovert]

Yeah $1289 here:

MidNite Solar MNBCB 2000/200 Battery Combiner Box 2000A

Shop now for MidNite Solar MNBCB 2000/200 Battery Combiner Box 2000A

www.stellavolta.com

Here the thread I was talking about:

Can Lynx really do 1000A?

For Lynx maximum current specification, you also need to take into consideration the current cutting capacity of the fuses. Maxifuses have only 1000 A cutting-current capacity. This means that any short-circuit current above 1000 A (even for a 25A fuse) may result in a permanent DC arc over the...

community.victronenergy.com

I keep going back and looking at the 1000/100 (or 50) BCB. Would handle my needs perfectly. I'm having a hard time justifying the price though, given I already started acquiring other parts. The 1000 is still 800 some dollars. Don't need a Midnite shunt in it though, I wonder if you can order without.

 

[Brucey]

I keep going back and looking at the 1000/100 (or 50) BCB. Would handle my needs perfectly. I'm having a hard time justifying the price though, given I already started acquiring other parts. The 1000 is still 800 some dollars. Don't need a Midnite shunt in it though, I wonder if you can order without.

If you aren't going to be using 1000A thru your lynx should be fine. Riccardos running four 10Ks. Not sure what your config is going to be.

 

[ricardocello]

Ok, the Pathfinder BMS can only do 170A, and the max recommended discharge rate is 0.5C, so I should limit each string to 140A.
So I’ll probably put 150A (edit: or 125A) breakers on each string instead of 250A.
Which is ok, because with 3 strings in parallel, it can still exceed 250A.

With 12 strings spread across 4 racks, each string will max out at 83.3A to hit 1000A, which is about 0.3C. Nice and easy on the batteries.
Total capacity would max out at 172 kWh (12 x 280Ah x 51.2V).

Obviously, i’m not going to build that right away, but getting real numbers helps the planning,
Now I can check physical sizes to see how much room I need.

 

[Brucey]

Ok, the Pathfinder BMS can only do 170A, and the max recommended discharge rate is 0.5C, so I should limit each string to 140A.
So I’ll probably put 150A breakers on each string instead of 250A.
Which is ok, because with 3 strings in parallel, it can still exceed 250A.

Might be a use case for mnedc125rts. Half the price of an mnedc 250. Trip point is 156A. 1/4" studs.

MidNite Solar MNEDC125RT Circuit Breaker with Remote Trip

Shop now for MidNite Solar MNEDC125RT Circuit Breaker. 125A, 125VDC Panel Mount Breaker, AIC 10,000 at 125VDC, 5 Year Warranty.

www.stellavolta.com

 

[Brucey]

Still has 50,000AIC down at 80V.

 

[ricardocello]

Still has 50,000AIC down at 80V.

They really do have the best breakers for this application.

 

[RCinFLA]

• EVE LF280K cells are specified to have <= 0.25 mΩ internal resistance. Let's use that as a representative number.
  I'm aware that it can vary with temperature and discharge rate, other cells are better, etc.
  So a 16s configuration will have <= 4 mΩ of internal resistance.

That only represents the ohmic resistance which is only good for initial surge current. Overpotential voltage ionic drive loss voltage slump due to cell current for sustained periods will be much greater. It also increases up to 3x at colder temps and as cell ages.

• The Overkill Pathfinder BMS specifies:
  Total series resistance, all FETs active (+Cell to B+) 0.351 mΩ

Chinese BMS typically understate their BMS series resistance, only including MOSFET Rds_ON at 25 deg's C and not including current shunt sense resistors or misc PCB and connectors resistance. MOSFET resistance increases about 150% at 100 degs C compared to 25 degs C.

• And for convenience, each string will have its own Midnite MNEDC250 breaker (50 kA AIC @ 125 VDC).
  Midnite MNEDC250 internal resistance < 0.25 mΩ according to the Littelfuse Carling F-Series data sheet (eyeballing log-log chart).
• This brings us to 4.6 mΩ per string.

I have several of these 250A breakers, brand new. They all measure in the 0.35 milliohm range. This does not include about another 0.10 milliohm (2x 0.05 mΩ) of cable lug compression terminal attachments.

Since they are hydraulic-magnetic breakers, they are about the best you can get and still have short circuit trip function that does not add a lot to series resistance to breaker.

Let's say I want to put 3 of these strings in parallel per rack.

• That brings the resistance down to 1.53 mΩ per rack (because resistance in parallel is divided).
  
  
• Now let's add a big honking Dihool DHM3Z 250A MCCB w/remote trip (for emergency shutdown) in the rack feeding a Lynx Power-In.
  20 kA AIC @ 1000 VDC. Dihool DHM3Z internal resistance is 0.9 mΩ according to my YR-1035+ meter tested on a 320A device (total of both poles). EDIT: Clamped the probes onto the terminals directly, provides lowest number.
  
  
• And for super-duper safety, also an inline Class-T 300A fuse as well. Yes, I'm splurging.
  The JLLN300 Class-T fuse (20 kA AIC @ 125VDC) measures 0.29 mΩ when cold.
  
  
• 1.53 + 0.9 + 0.29 = 2.72 mΩ per rack (including 3 16s EVE LF280K strings in parallel, breakers, Class-T)

View attachment 322955

You also have bus bars and cell terminal to bus bar compression connection resistance for inter-cell connections. They are about 0.10 milli-Ω for double 6mm stud terminals and 0.17 milli-Ω for single 6 mm stud and 2x20x70mm copper core nickel plated bus bars (per cell interconnect bus bar).

Would not recommend more than 0.5C(A) sustained current drain for thick electrode prismatic LFP cells. Internal heating increasing rapidly above 0.5C(A). At about 0.5 C(A) there is about 2.5 watts internal cell heating going to about 35 watts internal cell heating at 1 C(A) discharge rate.

Also keep your I2R heating at any given connection under control. You don't want a super-hot spot.
For example:

 

[ricardocello]

Ok, now let’s look at I^2R power losses at maximum current.

The 16 cells total would dissipate 62.5W putting out 125A.
But I could be wrong about internal cell resistance being turned into heat and not chemistry.
I'm sure I will be corrected. Edit: I was corrected above.

The overkill pathfinder BMS would dissipate 10.1 W at 170A, but with the 125A breaker in series, it is limited to only 5.5 W.
Each MNEDC125 breaker would dissipate 3.9 W at 125 A (3 per rack).

The big 250A MCCB would put out 56 W at 250 A.
A MNEDC250 instead would put out only 15.6 W (hmmmmm).

The 300A Class-T JLLN would put out 18.1 W with 250A going through it.

Each set of 20’ cables (+ and -, 40 ft total) would put out 187.5 W over the lengths of the cable at 250 A, which is 4.6W per foot.
That’s a little too much heat for my comfort. Don’t know how hot it would get.

And there are 4 sets of cables (one pair per rack).

 

[SnickeringBear]

I would run it with double runs of 750 MCM flex cable. This would give higher current capacity both in the cables and in the lugs which are much larger on the flex cable. Also, use 2 hole lugs and some serious busbar if going this route.

 

[ricardocello]

That only represents the ohmic resistance which is only good for initial surge current. Overpotential voltage ionic drive loss voltage slump due to cell current for sustained periods will be much greater. It also increases up to 3x at colder temps and as cell ages.
View attachment 322967

That’s a great chart, thank you!

Chinese BMS typically understate their BMS series resistance, only including MOSFET Rds_ON at 25 deg's C and not including current shunt sense resistors or misc PCB and connectors resistance. MOSFET resistance increases about 150% at 100 degs C compared to 25 degs C.

Overkill Pathfinder designed in Florida, Programmed in Canada, eh?
They wrote that on the PCB silk screen.

I’ll check the data sheet rDS for their FETs, they seem pretty conservative.
I'm not necessarily sold on them, it's just that they actually publish specs I can probably trust.

I have several of these 250A breakers, brand new. They all measure in the 0.35 milliohm range. This does not include about another 0.10 milliohm (2x 0.05 mΩ) of cable lug terminal attachments.

Which brand, just curious? Oh. Midnite ok. And thanks for the numbers.

Since they are hydraulic-magnetic breakers, they are about the best you can get and still have short circuit trip function that does not add a lot to series resistance to breaker.

You also have bus bars and cell terminal to bus bar compression connection resistance for inter-cell connections. They are about 0.10 milli-Ω for double 6mm stud terminals and 0.17 milli-Ω for single 6 mm stud and 2x20x70mm copper core nickel plated bus bars (per cell interconnect bus bar).

Excellent info. I’m a newbie on the batteries, appreciate the info.

Would not recommend more than 0.5C(A) sustained current drain for thick electrode prismatic LFP cells. Internal heating increasing rapidly above 0.5C(A). At about 0.5 C(A) there is about 2.5 watts internal cell heating going to about 35 watts internal cell heating at 1 C(A) discharge rate.

Looks like 0.3C is the target discharge rate worst case.

Also keep your I2R heating at any given connection under control. You don't want a super-hot spot.

see post #23

 

[RCinFLA]

Which brand, just curious? Oh. Midnite ok. And thanks for the numbers.

Yes, same breaker you mentioned. 'Midnite' just rebrands the Carl Switch breaker. Breaker spec has 50% tolerance on series resistance, so what I measured is within their allowed tolerance.