i think you could find a 16 to 8 balancer more easily, and then plug 3 of the outputs back as inputs alongside the 13 other lane.
16 to 8 is easier to begin with than 13 to 5 if you want to do it from scratch, and you can make it with 2x 8=>4 made out of 2x 4=>2
i would start by those 2x 4=>2 and try to make it a 1x 8=>4, ( it needs minimum 2 additionnal splitters to mix 1 lane from each of the 4=2 with a lane from the other 4=>2 ).
then repeat the process with 2 copies of that 8=>4 to make the 16=>8, ( using minimum 4 splitters)
and plug 3 output as inputs . try to minimize footprint every now and then
The simplest 4=>2 using only 3 splitters doesn't work well if all input comes from only the first 2 or the last 2 input. Similarly merging the 2 copies of 4=>2 using only 2 splitters is not "throughput unlimited", the expected result is potentially 4 lane of output, which is not possible if all the inputs comes from only 1 of the 4=>2. You may need to keep that in mind if you plan to progressively connect inputs lanes, it's better to alternate lanes when it's less than 50% of input connected when doing it this way with the minimum number of splitter.
Re: Balancer 13 to 5
Posted: Sat Sep 09, 2023 9:20 am
by ZuzMeduz
mmmPI wrote: Fri Sep 08, 2023 10:01 pm
i think you could find a 16 to 8 balancer more easily, and then plug 3 of the outputs back as inputs alongside the 13 other lane.
16 to 8 is easier to begin with than 13 to 5 if you want to do it from scratch, and you can make it with 2x 8=>4 made out of 2x 4=>2
i would start by those 2x 4=>2 and try to make it a 1x 8=>4, ( it needs minimum 2 additionnal splitters to mix 1 lane from each of the 4=2 with a lane from the other 4=>2 ).
then repeat the process with 2 copies of that 8=>4 to make the 16=>8, ( using minimum 4 splitters)
and plug 3 output as inputs . try to minimize footprint every now and then
The simplest 4=>2 using only 3 splitters doesn't work well if all input comes from only the first 2 or the last 2 input. Similarly merging the 2 copies of 4=>2 using only 2 splitters is not "throughput unlimited", the expected result is potentially 4 lane of output, which is not possible if all the inputs comes from only 1 of the 4=>2. You may need to keep that in mind if you plan to progressively connect inputs lanes, it's better to alternate lanes when it's less than 50% of input connected when doing it this way with the minimum number of splitter.
Yeah... The Problem is that I only recently started playing factorio and did not yet even touched topic of balancers, so I only understood only part of what you wrote. There is nothing to do, have to go deeper into the subject
Re: Balancer 13 to 5
Posted: Sat Sep 09, 2023 5:52 pm
by ZuzMeduz
I made a balancer by combining some other balancers together. Not the most elegant solution, but it works. For anyone interested in what it was needed for, here is the blueprint with balancer in it:
You might want to rearrange your smelter array to better fit output requirements instead of squeezing an uneven number of input belts to an uneven number of output belts. Balancers work best with powers of two as input or output belts, because with splitters you're multiplying belts with 2 or dividing by 2. And there is the belt speed of 1, 2 or 3. (15, 30, 45 per second).
You output red belts (30/s) and funnel them into blue belts (45/s). One blue belt is 2/3 of a red belt, and a red belt is 1/2 of a yellow belt. You can merge 3 yellow belts into 1 blue belt without balancing - just merge. You can merge 3 red belts into 2 blue belts without balancing - just merge.
In case of your smelters, make longer production lines and try to create or consume full belts. 13 beaconed furnaces in a line output exactly 45/s - one blue belt. They can be fed also by one blue belt full of ore. 4 of these make a nice consumer for an ore unloading station and a nice output for a 4 belt bus start or plate loading station. If you don't want to empty the unloading station that fast, use red belts and 2/3 of 13 furnaces: 8 or 9 furnaces. But 5 is not enough for a red belt and too much for a yellow belt.
Consider a smelter design like this (it's not my design; I found it here on the forum):
If you insist on merging 13 red belts to 5 blue belts, consider this (I used a 9-to-5 balancer from Raynquist's balancer book and in front of them, I just merged 3 red to 2 blue belts, which works because of the 2:3 capacity ratio):
Tertius wrote: Sat Sep 09, 2023 8:13 pm
You might want to rearrange your smelter array to better fit output requirements instead of squeezing an uneven number of input belts to an uneven number of output belts. Balancers work best with powers of two as input or output belts, because with splitters you're multiplying belts with 2 or dividing by 2. And there is the belt speed of 1, 2 or 3. (15, 30, 45 per second).
You output red belts (30/s) and funnel them into blue belts (45/s). One blue belt is 2/3 of a red belt, and a red belt is 1/2 of a yellow belt. You can merge 3 yellow belts into 1 blue belt without balancing - just merge. You can merge 3 red belts into 2 blue belts without balancing - just merge.
In case of your smelters, make longer production lines and try to create or consume full belts. 13 beaconed furnaces in a line output exactly 45/s - one blue belt. They can be fed also by one blue belt full of ore. 4 of these make a nice consumer for an ore unloading station and a nice output for a 4 belt bus start or plate loading station. If you don't want to empty the unloading station that fast, use red belts and 2/3 of 13 furnaces: 8 or 9 furnaces. But 5 is not enough for a red belt and too much for a yellow belt.
Consider a smelter design like this (it's not my design; I found it here on the forum):
If you insist on merging 13 red belts to 5 blue belts, consider this (I used a 9-to-5 balancer from Raynquist's balancer book and in front of them, I just merged 3 red to 2 blue belts, which works because of the 2:3 capacity ratio):
Thanks for the tips. I saw a similar smelter in a video, but it was ineffective with gaps at the output, I decided to make my own, but apparently everything had already been invented before me
Re: Balancer 13 to 5
Posted: Sat Sep 09, 2023 9:17 pm
by mmmPI
ZuzMeduz wrote: Sat Sep 09, 2023 9:20 am
Yeah... The Problem is that I only recently started playing factorio and did not yet even touched topic of balancers, so I only understood only part of what you wrote. There is nothing to do, have to go deeper into the subject
I'm sorry, had you mention it in your initial post i would have not jumped at it this way.
I saw your balancer and it looks nice but also overly complex , most importantly it works !
Still i'll try to clarify with pictures the previous ramble, maybe it could still help :
42duplication.png (123.87 KiB) Viewed 2836 times
from the left piece i called 4=>2 you can make the middle one, 8=>4, using 2 additionnal splitters and connecting them
and from the middle piece i made the one on the right using the same process, although at this point it required 4 additionnal splitters.
What i called additionnal splitters are those added that were not present if it was only a duplication process.
Then from the 16=>8 one can make a 13=>5 if 3 of the outputs are looping as inputs which looks like this :
plugbackin.png (189.48 KiB) Viewed 2836 times
at this point it seem arbitrary to choose which output are made to loop and where they are connected; It's also interesting i found to look at it at this point while feeding it some, all , 1, no lane of input, while counting the material at the output.
From those test i added priority splitters input and compressed it for this result :
Now there are different types of balancers for different purposes, this one is not "throughput unlimited" , meaning that's it's not because you have 5 full lanes of inputs out of 13 that you are guaranteed to have 5 lanes of outputs.
This comes from the fact that it is made from the simplest 4=>2 that cannot output 3 lanes (obviously) so once it's pieced up with another one, as in the middle piece in the first picture, then if you only give 4 lane of input in the same side, you will have a bottleneck. The end result will be "balanced" though the throughput is limited.
I wished to alert you that i think yours suffers the same problem to a lesser degree and despite what seems to be attempts at solving it :
bottleneck.png (277.9 KiB) Viewed 2836 times
looking at this part , it can have 3 input lane, but only 2 output, meaning that if only the 5 rightmost lane serves as input, you will not have 5 full lane as output, it will be 5 balanced lane, but with a bottleneck of the same kind as the one i proposed. (mine is worse though)
This maybe explains why i had to use priority on the splitters, and what is the drawback of it, because if i discard some of the outputs, and that they are stuck in the bottleneck, it would unbalanced the result or come to a stop due a loop being filled up and thus immobile. Using the priority for the inputs in the splitters make sure that this doesn't happen but it means that not all the input lanes are consumed equally, it's only the output that is balanced which now that i think of it ( and reading Tertius message), may not be what you were looking for in a balancer, those are difficult. x)
Hey it's not because you don't "need" it that you can't use it if you like it or if it fit your factory or whatever x)
Balancers are notoriously difficult to make. I tried running some more test and i found out that if for whatever reason the right most output lane is blocked, then the middle output lane is receiving less material on your balancer not sure if it matters :
cut.png (148.36 KiB) Viewed 2814 times
I used such blueprint in the test, without the /editor mode it's just a belt connected to a combinator to count the material passing, the infinity chest void material to make testing easier.
Hm, what is even your use case where you need a 13 to 5 balancer?
In general, balancers should be used only in specific cases. In many cases they are not necessary or even hurtful.
Examples where you need balancers is offloading from a train with several cars, so that each car is emptied equally. But in that case you should stick to train lengths with the power of 2 for easy balancers, and anything longer than 8 cars is kind of crazy anyway.
A case where you don't need balancers is on the bus. Use priority splitters instead to ensure that belts branching off are full.
Re: Balancer 13 to 5
Posted: Mon Sep 11, 2023 8:24 am
by coppercoil
An use-case is important because there also are input-balanced and output-balanced concepts.
Re: Balancer 13 to 5
Posted: Mon Sep 11, 2023 12:20 pm
by ZuzMeduz
Premu wrote: Sun Sep 10, 2023 9:33 pm
Hm, what is even your use case where you need a 13 to 5 balancer?
In general, balancers should be used only in specific cases. In many cases they are not necessary or even hurtful.
Examples where you need balancers is offloading from a train with several cars, so that each car is emptied equally. But in that case you should stick to train lengths with the power of 2 for easy balancers, and anything longer than 8 cars is kind of crazy anyway.
A case where you don't need balancers is on the bus. Use priority splitters instead to ensure that belts branching off are full.
. try to minimize footprint every now and then