Page 2 of 2

Re: Beautiful Sushi Belts

Posted: Fri Jan 31, 2020 3:18 am
by Olacken
disentius wrote:
Thu Jan 30, 2020 10:33 pm
...
Yes but it's far from circuit free

Edit: Yeah My comment is kind of irelevant since it was a response but still I do check this forum to see beautifull circuit free sushi belt and was sadend after waiting for the loading of the image to see that there were in fact discusting circuit hanging in (Couldn't tell if they are green or red)

Re: Beautiful Sushi Belts

Posted: Fri Jan 31, 2020 7:17 am
by wobbycarly
Olacken wrote:
Fri Jan 31, 2020 3:18 am
<snip>discusting circuit <snip>
:D :D :D :D

Re: Beautiful Sushi Belts

Posted: Fri Jan 31, 2020 7:59 am
by Olacken
Please don't keep track of that because I may or may not be a very heavy circuit user

Re: Beautiful Sushi Belts

Posted: Thu Apr 09, 2020 4:12 am
by Skellitor301
xaetral wrote:
Sun Sep 08, 2019 4:30 pm
new design!

2700 SPM Sushi Belt:

-snip-
Any updates on this design that fixes the issues?

Re: Beautiful Sushi Belts

Posted: Sat Jun 27, 2020 10:32 am
by xaetral
Skellitor301 wrote:
Thu Apr 09, 2020 4:12 am
Any updates on this design that fixes the issues?
I am sorry I am not on the forum often (at all), I also have been playing less factorio these days (I really need to continue my big mod).

To answer your question I have found what was causing this issue, it's the sushi-to-belt modules at the top, they are nice and compact but they need to be in perfect sync otherwise the items get stuck for a fraction of a second each time, and that delay accumulates because there's no buffer space between the filtered ouputs and the main recycled belts.
So this cause the entire sushi to stop, which then stops the recycled belts, which then makes the sushi stops a bit more, etc
At the end the entire system is locked.

You can try with a bit of modded faster belts at the end of the sushi right before the recyclers, I'm sure this would fix the issue (not a very appealing solution tho I admit)

So I haven't found a solution yet, but the smaller architectures are working flawlessly because they are not this much space constrained.

Re: Beautiful Sushi Belts

Posted: Sat Jun 27, 2020 10:43 am
by xaetral
Olacken wrote:
Fri Jan 31, 2020 3:18 am
Yes but it's far from circuit free
If you are interested in the reason why I posted this in the circuit-free designs it's because it really is circuit-free, especially compared to the original sushi belt design.

This design requires full belts on the inputs, it may be ensured by using circuits but it is not mandatory.
As you can see people have been rambling about how to do it without circuits, it's perfectly doable tho it obviously increases the size of the footprint.

I hope you'll find your way into the holy sushi :p

Re: Beautiful Sushi Belts

Posted: Fri Jul 17, 2020 6:19 am
by yagaodirac
You need a 1/7 limiter for every sci pack. In that way, if the input keeps stable, the tail filter is even not needed any more.
I am planning to redo the tutorial in English. But not decided yet. It takes time.

Re: Beautiful Sushi Belts

Posted: Fri Jul 17, 2020 12:58 pm
by Nosferatu
Looks like I never posted my final build:
This one consumes 255 science/min - so 4 of them gives you a nice 1k science/min base ;)
Circuit-free and congestion free
ScienceShushi.png
ScienceShushi.png (1.28 MiB) Viewed 454 times

Re: Beautiful Sushi Belts

Posted: Sun Jul 19, 2020 11:18 pm
by foamy
May as well drop this in here as well, I guess:

Image

Bottom is with full inputs; top is without. Both work.

Re: Beautiful Sushi Belts

Posted: Mon Jul 20, 2020 12:06 am
by Impatient
I use a sushi belt for the lab complex. But instead of creating a compressed mix (and therefore running into problems if there aren't enough of a type of flask), I use combinators to count how many flasks of each type are on the belt.

I use a memory cell with EACH and all inserters putting on the belt count up, all inserters taking from the belt count down the number. A const combinator holds the target number. If the count is below target, the respective inserter is activated.

And that's it. no problems forever after. just needs 1 belt, but all inserters need to be hooked to the two counting wire networks.

While skipping through this thread, I found the idea to use some other item type as placeholder, if a type of flask is not available and the player does not want to use the circuit network, pretty inventive. well thought! :thumbsup:

Re: Beautiful Sushi Belts

Posted: Mon Jul 20, 2020 12:23 am
by foamy
Impatient wrote:
Mon Jul 20, 2020 12:06 am
I use a sushi belt for the lab complex. But instead of creating a compressed mix (and therefore running into problems if there aren't enough of a type of flask), I use combinators to count how many flasks of each type are on the belt.

I use a memory cell with EACH and all inserters putting on the belt count up, all inserters taking from the belt count down the number. A const combinator holds the target number. If the count is below target, the respective inserter is activated.

And that's it. no problems forever after. just needs 1 belt, but all inserters need to be hooked to the two counting wire networks.

While skipping through this thread, I found the idea to use some other item type as placeholder, if a type of flask is not available and the player does not want to use the circuit network, pretty inventive. well thought! :thumbsup:
I used to use that kind of circuitry -- though with a belt feeds on instead -- until I ran across DaveMcW's implementation that only required circuitry on the feed. If it's a loop, you can use what's coming back to tell you what you need to add, after all. At that point all you have to do is wire up as many downstream tiles (plus the tile you feed onto) as you need to account for the proportions, set them to read-hold, and feed that back to a belt segment set to enable if it gets less than X on the desired signal. Structure the lanes so that only one lane is being fed on, and that that lane feeds in on the downstream side of the tile it joins, and you get a very simple, very compact, way to generate sushi belts for whatever purpose is desired.

However, this is the circuit free forum :p



Regarding the possibility of starved inputs (or missing ones), my layout can handle those. The mechanism of action is to produce uncompressed belts that are then mixed. This acts to provide a cap on the maximum amount of material that can be inserted onto the sushi belt, and any returns from that sushi belt are given absolute priority in being re-injected into the system. So if the input can't keep up, all that happens is less of that material is added to the belt; the other stuff can't cannibalize the free space because they're all throughput limited.

Re: Beautiful Sushi Belts

Posted: Tue Jul 21, 2020 8:05 pm
by Nosferatu
This one also works by decompressing every input to 1/4 (or 1/8 in case of space science) and then mixing it.
Leftovers have priority over the new stuff when coming back so it never backs up.
ScienceShushi_1lane.png
ScienceShushi_1lane.png (1.17 MiB) Viewed 338 times


edit- just looked at your version again foamy. Never realized before what fish filters are for. Maybe i can compress it a little more if I use them :?

Re: Beautiful Sushi Belts

Posted: Wed Jul 22, 2020 5:08 am
by foamy
Nosferatu wrote:
Tue Jul 21, 2020 8:05 pm
edit- just looked at your version again foamy. Never realized before what fish filters are for. Maybe i can compress it a little more if I use them :?
Maybe! I admire your spaghetti twists. Mine just bruteforces it in straight lines.

I think it looks pretty, though. Here it is, now fully implemented as the lab system for a 924 science/m setup:

Image




(In principle, the lab lines could be extended out, so that these four sushi belts would exactly eat the one blue belt of each science being fed it. That would require 41 labs per row, provided the last lab had an extra beacon. I didn't do that because I also wanted to test the splitter block tiling for pulling the sciences from the central bus)

Re: Beautiful Sushi Belts

Posted: Fri Jul 24, 2020 6:02 pm
by Nosferatu
While I was thinking how to compress my setup even more I came to a kind of obvious result.
We are decompressing everything to 1/4 after the sorter - the other to 3/4 are sent back to the feed.
And we do this for each sushi.
But why not make one sorter/decompression area that feeds two sushi lanes?

Here it is: Two sushi lanes fed by one sorter spaghetti:
Science2Shushi.png
Science2Shushi.png (1.81 MiB) Viewed 246 times
This one eats 520 science/min

PS: Actually it's 1 and a half sorter. It turned out that one sorter can not handle the backflow of two sushi in case science stopps completely. So I had to build a few pre-sorter too ease the pressure.


Re: Beautiful Sushi Belts

Posted: Sat Jul 25, 2020 6:30 am
by foamy
I came to the same conclusion -- you can see that it produces two belts worth of sushi.

The ideal would be to take all seven belts and produce eight belt's worth of sushi out of them directly. Then you don't need to decompress them at all -- as long as the sushis all properly return to the input feed you're all set.

This calls for some experimentation...

Re: Beautiful Sushi Belts

Posted: Mon Jul 27, 2020 2:11 am
by foamy
Hrm.

When fed fully compressed belts, splitters exhibit odd behaviours and don't mix as expected.

Image

In the 8:8 balancer, note how what the splitters are doing is a belt swap, not a belt mix. This kind of kiboshes things, at least with a simple plan like this. One option might be to decompress to 50%, and then balance things over 14 belts instead of 7...

Re: Beautiful Sushi Belts

Posted: Thu Jul 30, 2020 1:53 pm
by xaetral
Well yeah, things can become weird when the feedback is rigid (by that I mean no space to backup/expand unlike the first designs I posted).

BTW I managed to make my 2700spm sushi belt work.
I just... built 4 times my first dual sushi setup...
Yup, no need to overcomplicate things, if it works why would you want to do it another way :D