Page 1 of 1

N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Tue Dec 27, 2016 10:30 pm
by punchcrosser
Making beltbalacers... pfff... time to google some pictures and recreate them... :cry:
But, if you can live with a nearly perfect balancer, look no further your, your prays have been answered :D

Screenshot (1).png
Screenshot (1).png (1.03 MiB) Viewed 6973 times

This is my example of an 8 way 8 balancer, i will use it to explain this bad boy, but the logic behind it will work for as many input/output lanes you want 8-)

The reason why its not perfect is because a belt reader can read 6,7 or 8 items on a full belt (it counts items that are half way on the belt to)
If a full belt could only contain 6 or 8 items... :cry:

Step 1

Determine how many output an intput lines you want. In my example we have 8 input lanes and 8 outputs
Place as many splitters in the lanes until the toplane can feed the bottem lane and vica versa

Step 2

Connect two next to eachother belts togther in the same lane (in the example the red circuit )
The one closet to the splitter needs to be enabled when the item is less 6
The other belt reads the content on hold mode
This way i ensure only 6 items will ever be on the belt
Do this in all lanes

Step 3

Connect all output belts together with all readerbelts (green circuit)
The outputbelts should be enabled when item is bigger than 6 times the number of outputlanes - 1 (in my example its set to > 47) (why not = 48... i'll explain :D )

There you go, this is so easy :lol:



Pro's

Easy and compact


Negatives

All the problems are originated from the full belt that can contain 6,7 or 8 items otherwise the design could even be more simple

Lets explain > 47 and not = 48
When one ore more belts fill up it pushes the readerbelt with a value that it > 6 and the system comes to a halt

Another downfall is the troughput, its much lower than a normal balacer because i cutt the belt when 6 items are on it.
My guess is, you get half the throughput of a normal balancer

This system does not work with half lane belts simply because the reader will see only 3 items but if you understand the logic you know you can easly solve this one ;)


Sorry for my bad english
If you can improve the system or you just like the post, please comment (maybe devs will see the post and if possible change the 6,7 or 8 problem :D )
Thx for your attention :D

Re: N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Tue Dec 27, 2016 11:04 pm
by Yoyobuae
punchcrosser wrote:All the problems are originated from the full belt that can contain 6,7 or 8 items otherwise the design could even be more simple
If I'm understanding right that issue could be solved by reading from 9 belt tiles together. Added together 9 belt tiles contain EXACTLY 64 items. Try it with a completely full, moving belt and you get a constant 64 signal read from the 9 belt tiles.

Re: N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Wed Dec 28, 2016 4:16 pm
by punchcrosser
Thx Yoyobuae :D this has solved a lot of problems :D but new ones has risen :x at least the system is stable

Doing the 64 belt tile counter has made the perfect balancer but the troughput is no good. For some reason i lose compression in the counterbelts when the outputbelts open.
Screenshot (3).png
Screenshot (3).png (980.5 KiB) Viewed 6911 times
I could partialy solve this problem by placing a delayline in red belt.
Screenshot (5).png
Screenshot (5).png (941.45 KiB) Viewed 6911 times
Screenshot (4).png
Screenshot (4).png (1.27 MiB) Viewed 6911 times
Now you still have a perfect balancer but with the max throughput of red belt.

For my purpose it was not good enough, so i tweaked more...
Screenshot (6).png
Screenshot (6).png (1.34 MiB) Viewed 6911 times
to count in the compression loss i set the outputgate > 495 (counterblok = 512 - (2 items compression loss a belt times the amount of output belts) - 1 )
putting the output gate 3 tiles away from the counter yiels better results
using a decider combinator to activate the output gate also yields better results

so when the balancer goes overdrive (counterblock hits 496 and more) you create an max unbalance per belt of 496/512*100 = a bit more than 3%
for my purpose 3% is nothing :-), i can live with that :-)
below 496 the balacer works again perfect

pro's

easy to set up and easy to expand on it

negatives

a bit bulky for low lane balacing but past 8 lines it starts to shine
shifting items from top to bottem lane with splitters takes time so there is a small amount of compressionloss in extreme conditions


for 4 lane balancing i do see me use my design but starting from 8 and up this will be my go to solution :-)

Re: N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Sun Jan 22, 2017 3:24 pm
by huliosh
what if only first belt is filled and the rest are empty?

Re: N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Tue Feb 21, 2017 5:02 pm
by punchcrosser
splitters move it to the belt next to it.
it does not work with a single lane belt, it always needs one belt with two lanes to work

Re: N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Fri Feb 24, 2017 4:09 am
by Avezo
I'm not 100% sure it would provide proper throughput, but something like this shouls be enough in place of that splitter block:

Image

Re: N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Fri Feb 24, 2017 4:00 pm
by Yoyobuae
Avezo wrote:I'm not 100% sure it would provide proper throughput...
It doesn't:
http://imgur.com/a/hpeJZ

It's limited to one belt throughput when belts are shifted.

Re: N WAY N EASY, nearly perfect BELTBALANCER with circuit logic

Posted: Fri Feb 24, 2017 5:51 pm
by kann_
I posted this a while ago on reddit.
Image

It was the same principle, but I didn't understand the belt mechanics good enough to find the solution. Good job.

If I understand it correctly the last problem is lane balance. That one might be quite annoying to tackle... the dirty solution I used was the reflow belt with lane balancer in the gif.