Factorio Forums

More "splitters decompress belt when they shouldn't" stuff. The top two are known ways of taking off a quarter of a 4-belt bus, at least on paper. In practice I've never had them work reliably, as you can see in the save.

Comparing the second and third is more interesting. The second is getting back-pressure from the splitters (which means decompressing down the line). The third, despite being identical, does not. I assume build order caused initial item spacing to be different which created interference patterns in the splitter timings.

Creative-mode-fix is needed in this save, for simplicity.

Game version? 0.16.16 has just released with some splitter fixes.

This is 0.16.16. I've seen this sort of behaviour many versions back though (0.14? Earlier?)

Slightly smaller case using vanilla loaders and infinity chests:



The top left splitter does not preserve compression.

This should demonstrate some of the wonkiness with splitters. Way back in... 0.12? this method would output to both belts just fine. It was quite a while ago that it broke.

Ditto - still happening on my 6-lane setup I originally posted in the other bug report.

See viewtopic.php?p=333224#p333224 and viewtopic.php?p=333233#p333233 for a test case with save.

MadZuri wrote:This should demonstrate some of the wonkiness with splitters. Way back in... 0.12? this method would output to both belts just fine. It was quite a while ago that it broke.

Seams that internal buffer in blue splitter is too small to hold an item for 9 ticks where it should offload it to yellow belt so it forces blue input to back up sometimes when an item comes from yellow input
Probably increasing internal splitter buffer (delay) would help but yet again it will remove “express splitter bus” which noone uses anyways.

Just another sample (https://gfycat.com/BleakDeterminedBorer):

Another simpler example:

Simple 4-4 balancer decompresses the 4 fully saturated lanes.

socket wrote:Slightly smaller case using vanilla loaders and infinity chests:

https://i.imgur.com/9ErgC7N.jpg

The top left splitter does not preserve compression.

that's not a bug, thats 1.5 belts feeding into the top two belts at the end. I'm sure the bottom two inputs are running slower than max speed. wait- you weren't refering to the end compression.

Sifu wrote:Another simpler example:

Simple 4-4 balancer decompresses the 4 fully saturated lanes.

Can you post this save to the thread so it can hopefully be looked at Sifu?

golfmiketango wrote:

Sifu wrote:Another simpler example:

Simple 4-4 balancer decompresses the 4 fully saturated lanes.

Can you post this save to the thread so it can hopefully be looked at Sifu?

There you go. I have uploaded the save which has my simple 4-4 balancer decompressing.

socket wrote:Slightly smaller case using vanilla loaders and infinity chests:



The top left splitter does not preserve compression.

I was able to reproduce this bug using a similar test setup: a drawn-out 4-blue-belt balancer with vanilla loaders and infinity chests. Then I pulled up and replaced one of the belt segments that was not being compressed, and the behavior reverted to normal.

(Sorry, no pics or video because I couldn't reproduce the bug a second time.)

This should be fixed in 0.16.20

It's a lot better but not not entirely fixed. In particular the blue/yellow combo posted by madzuri still fails if you get the alignment right.

Priorities often fail as well but you may consider that to be a separate issue.