Page 1 of 1

Inline 4-4 Lane Balancer

Posted: Tue Mar 12, 2019 2:47 am
by farcast
4-4 balancer test.gif
4-4 balancer test.gif (1.05 MiB) Viewed 39564 times
Here is a perfect count input/output balanced 4 tile wide 4-4 lane balancer.
The general idea is to have half of the incoming items switch lanes, then balance the lane-switched items with the un-switched items. As far as I know, the only way to reliably switch the lane an item is on is to side-load onto the input of an underground belt.
Lane Balancers.png
Lane Balancers.png (448.85 KiB) Viewed 39564 times
Also included in the blueprint book are a shorter 4-4 lane balancer, and two variants of a 2-2 lane balancer.


Re: Inline 4-4 Lane Balancer

Posted: Wed Mar 20, 2019 5:08 pm
by Mango
Very nice. I think I'll use the second 4-4 a lot. Symmetricity is the property I'm looking for :)

Re: Inline 4-4 Lane Balancer

Posted: Thu Apr 25, 2019 10:40 am
by Xeridanus
Mango wrote:
Wed Mar 20, 2019 5:08 pm
Symmetricity
And Symmetry is the word you're looking for. ;)

Re: Inline 4-4 Lane Balancer

Posted: Mon Apr 29, 2019 3:17 pm
by Mango
:oops:

Re: Inline 4-4 Lane Balancer

Posted: Thu Feb 20, 2020 9:12 pm
by tamanous
At least the 4:4 balancers are sadly not input balanced. In addition, they perform different, as the shorter but wider version balances the items better than the longer version. When fed by 4 fully compressed belts, they just split the input without any balancing. Plus, when input varies, those balanced lines do randomly switch positions. See attached sandbox save file.

I've made a much bigger 4:4 balancer (14 x 22 = 2 cargo wagons + empty space in between), that balances input and output, works with and without compressed inputs and balances variation in input evenly. I would appreciate it if someone is willing to give it an overhaul.
4-4 balancer.jpg
4-4 balancer.jpg (1.69 MiB) Viewed 34814 times




edits:
- added some other splitters / balancers to savegame
- better screenshot

Re: Inline 4-4 Lane Balancer

Posted: Thu Mar 12, 2020 10:36 pm
by farcast
tamanous wrote:
Thu Feb 20, 2020 9:12 pm
At least the 4:4 balancers are sadly not input balanced. In addition, they perform different, as the shorter but wider version balances the items better than the longer version. When fed by 4 fully compressed belts, they just split the input without any balancing. Plus, when input varies, those balanced lines do randomly switch positions.
Ensuring every output draws from every input regardless of total activity is the property of a belt mixer. Balancers only need to ensure every output is given the same number of items regardless of where those items come from, and/or ensure every input is drawn from evenly regardless of where those items end up.

Regarding the difference between the two 4-4 balancers: the short one switches lanes before balancing the result, while the long one balances before lane-switching the result. As far as I know, [lane-switch -> 2-2 balance] & [2-2 balance -> lane-switch] should be functionally identical, so I'm not sure what test you did to differentiate the two.

They do have problems, though. If you treat every input lane as a separate belt, then they should be equivalent to the non-universal 8-8 belt balancer, and share all the same problems that it has. That includes not being input balanced if not all inputs are compressed.

Re: Inline 4-4 Lane Balancer

Posted: Fri Mar 13, 2020 7:37 pm
by tamanous
farcast wrote:
Thu Mar 12, 2020 10:36 pm
Ensuring every output draws from every input regardless of total activity is the property of a belt mixer. Balancers only need to ensure every output is given the same number of items regardless of where those items come from, and/or ensure every input is drawn from evenly regardless of where those items end up.
Yey, while further investigating this area, I found that - obviously - someone has entered that area before and the term mixer was used. I didn't want to edit my posts .. Thanks again for the clarification. Funny thing though, mixers (as far as I found) seem to balance both in- and output very nicely.