Build a a 3->2 balancer across those 3 belts on the left. These 2 outputs, together, will be your feed of 2000.
Split a belt off of each of the other 4, and merge them together.
Split that into 2, and merge each of those into one of the 2 belts of the 2000 line.
The long version:
In my experience (and this may well be a product of my build style, and may bot quite be applicable to you) belt balancing is not an issue that needs to be addressed. Belt capacity and throttling leads naturally to self-balancing.
Example: in the Steelworks build I did recently, I had two different sets of machines needing to consume Steel Ingot: Constructors for Steel Pipe and Constructors for Steel Beam. Currently, the factory is clocked for a max belt of Mk1, since I can’t really afford the power for more, right now. My foundries produce 90 Steel Ingot/min, so that means I need 2 belts worth of Steel Ingot transport. I ended up building 12 Foundries, in 2 groups of 6, each outputting to a separate Mk1 manifold, so I’ve got 2 belts of 45/min each.
On the consumption side, I need 21.068/min to make Steel Pipe and 68.932/min to make Steel Beam. So, I still need 2 belts of bandwidth for Steel Beam, but only 1 for Steel Pipe. To make this happen, I split each of the Steel Ingot lines into 2, let 1 from each split go onward to Steel Beam, and took the other from each split and merged them, for Steel Pipe. All dumb splitters.
Now, you might say "well, that just gives you 45/min going to both Steel Pipe and Steel Beam, you just made it that the two separate Foundry lines can now mix. But this doesn’t account for (what I call) “back-pressure”.
The magic of back-pressure is that it makes dumb splitting (splitting evenly, instead of with ratios) irrelevant. With my setup, the Steel Pipe constructors only need 21.068/min but are getting 45/min. However, as long as the clock rates are set correctly on those machines, they will only consume 21.068/min, and the extra 23.932/min will eventually back up, and flow over to Steel Beam anyway.
Essentially, it’s the balancer vs manifold debate. Upside is, you don’t have to deal with crazy balancing calculations, when you’ve got odd ratios like you have. Downside is, the system doesn’t reach full efficiency until it “primes” up all the back-pressure.
The one caveat to back-pressure is if you have “re-combination” farther down the line. In my example, if I were going to take those Steel Beams and Steel Pipes and combine them together in some other recipe, then it might be impossible to build up enough back-pressure to balance everything out. Sort of like having an air-bubble in the system. You can solve this by either manually priming the back-pressure, or just swapping to a Smart Splitter with an Overflow output, in the right spot, to allow the “bubble” to bleed out.
TL;DR, I would do the following:
The long version:
In my experience (and this may well be a product of my build style, and may bot quite be applicable to you) belt balancing is not an issue that needs to be addressed. Belt capacity and throttling leads naturally to self-balancing.
Example: in the Steelworks build I did recently, I had two different sets of machines needing to consume Steel Ingot: Constructors for Steel Pipe and Constructors for Steel Beam. Currently, the factory is clocked for a max belt of Mk1, since I can’t really afford the power for more, right now. My foundries produce 90 Steel Ingot/min, so that means I need 2 belts worth of Steel Ingot transport. I ended up building 12 Foundries, in 2 groups of 6, each outputting to a separate Mk1 manifold, so I’ve got 2 belts of 45/min each.
On the consumption side, I need 21.068/min to make Steel Pipe and 68.932/min to make Steel Beam. So, I still need 2 belts of bandwidth for Steel Beam, but only 1 for Steel Pipe. To make this happen, I split each of the Steel Ingot lines into 2, let 1 from each split go onward to Steel Beam, and took the other from each split and merged them, for Steel Pipe. All dumb splitters.
Now, you might say "well, that just gives you 45/min going to both Steel Pipe and Steel Beam, you just made it that the two separate Foundry lines can now mix. But this doesn’t account for (what I call) “back-pressure”.
The magic of back-pressure is that it makes dumb splitting (splitting evenly, instead of with ratios) irrelevant. With my setup, the Steel Pipe constructors only need 21.068/min but are getting 45/min. However, as long as the clock rates are set correctly on those machines, they will only consume 21.068/min, and the extra 23.932/min will eventually back up, and flow over to Steel Beam anyway.
Essentially, it’s the balancer vs manifold debate. Upside is, you don’t have to deal with crazy balancing calculations, when you’ve got odd ratios like you have. Downside is, the system doesn’t reach full efficiency until it “primes” up all the back-pressure.
The one caveat to back-pressure is if you have “re-combination” farther down the line. In my example, if I were going to take those Steel Beams and Steel Pipes and combine them together in some other recipe, then it might be impossible to build up enough back-pressure to balance everything out. Sort of like having an air-bubble in the system. You can solve this by either manually priming the back-pressure, or just swapping to a Smart Splitter with an Overflow output, in the right spot, to allow the “bubble” to bleed out.