Sushi thread

[halcyon wing]

I'm considering reworking my sushi bus to maintan an average-seen-over-X minutes. I haven't thought too much about the math, but I imagine each sighting would add some multiple to the tracker, and then I would do something like * 999 / 1000 each tick. This would eliminate the need for multiple banks of memory cells and would make item availability more mixed instead of having large surges.

That ought to make the feeding more consistent.

[neat idol]

I have 4 combinator average throughput counter I use for example with sushi

Seen here: https://media.discordapp.net/attachments/139677590393716737/999726748725940295/unknown.png

It takes ~50 seconds for items to make full loop, so my rolling average decay is set to -1/3000

[halcyon wing]

4 combinators! That's way simpler than just one of my memory cells.

[neat idol]

I can share blueprint if you're interested

[neat idol]

here's the example setup (left one is rolling average over 10 seconds, right one is over 100 seconds):
https://katiska.dy.fi/n/temp/factorio/blueprints/circuitry/rolling-average-pulse-counter/example.mp4

[neat idol]

blueprints are in same folder: https://katiska.dy.fi/n/temp/factorio/blueprints/circuitry/rolling-average-pulse-counter/

[halcyon wing]

Thanks.

[clear latch]

Made a sushi construction timer to compare segment build speeds. I'm going through a lot of revisions and am done eyeballing the EvoGUI Play time in the corner lol. Unlike Ghost Scanner the Recursive Blueprints+ scanner does not need to be in a logistic network and can scan ghosts across multiple isolated logistic segments aka sushi rails. Changing one of its coordinates a token amount every tick forces it to output continuously to start the timer, also kills UPS unfortunately. Paste Signals lets me "print" the output to CCs

[clear latch]

I decided to sacrifice map view upgradeability for shorter belt loops. Tracks are cutout from CodeGreen's 4-way. It took about 1m to build a straight and 2m for a 4-way, not bad for 5-6 active inserters (1/2 to 3/4 were redundant to cover all directions). Overall flow mattered less than specific bottlenecks. Some of the things I tried: 2x overall flowrate = no reduction beyond a certain point, 4x bots/m => -1/2 build time, 2x medium poles + signals (often the last to build) => -10s

[glass needle]

Just started an eight high ribbon world. Wondering if sushi can save me from the hell I've put myself in?

[timid cosmos]

if you're planning to launch a rocket you might want to restart

rocket silo is 9x9

damn that was 2 hours ago

hope he didn't get too far

but yeah, sushi would be great in a ribbon world, especially builder sushi I imagine.

production sushi also would be good i guess since it is usually compact

[glass needle]

I know, I'm going to edit the silo in

I just want to one up that guy on reddit who did it in 9 high

[timid cosmos]

ah

well what Foeglas posted above, i think thats sushi designed to build rails for them. The same concept could be applied to a ribbon world for building outposts and defenses

[glass needle]

What's the easiest way to make a belt that combines coal and stone onto it 50:50?

Like this but it doesn't break

[distant gorge]

Slap a circuit wire on the two belt halves before the splitter and do “enable if Everything == 4”

Or just do what you’re doing and add a recycle loop

[neat idol]

It's pretty hard to have robust sushi without a loop

[glass needle]

I don't have enough space for a loop unfortunately

[plain hare]

I usually say not that you need not a loop but a guaranteed destination

A loop can suffice because it is a guaranteed destination… just of continuous flow down the belt

So yeah loops super helpful but you could say use a buffer of stone and coal at the other end and send a signal down along the belt to tell the stuff loading to turn on / off based on buffer capacity.

[normal gulch]

why cant stone be one lane coal be the other? im blind

[plain hare]

I’ve been thinking about how I wanna do my sushi belt for this bus base to accomplish low buffer. And I’m considering doing request based with some buffers by production lines. Like you can store a few units on the belt before the production line and send some more down the sushi belt when it’s less than 50. And then it would naturally surge up to 200 or so as the buffer on that sushi bus belt comes in. Dunno which would be better that or flow limiters with items always moving and preset amounts. I think flow limiters would look more organized and controlled but there might be a beautiful chaos to the randomness of that request belt.

[neat idol]

finally some prod 3 modules

[normal gulch]

wheres the sushi??

[neat idol]

I thought I had factorio chat selected, and not this subthread! Nothing to sushi here, please disperse!

[pulsar lance]

How were you going to take stuff from the main bus? Filter splitters?

[plain hare]

Filter splitters or filter inserters

At lower throughput an inserter might be easy

[supple shore]

sushi for beginners

[north perch]

reviving

@steep cloak Did you post your recent sushi idea stuff in here btw?

[steep cloak]

no

I will today though

[north perch]

cool, I didn't actually catch it

[limpid cloud]

What's the easiest way to do this then?

Can't get them even

[steep cloak]

yeah I wanted to actually figure it out and get it working before doing something official

[north perch]

I probably should have asked why you wanted sushi in the first place; you want a single belt feeding all the labs?

[limpid cloud]

Yeah

[north perch]

ah, well then https://www.reddit.com/r/factorio/comments/7siu7j/safe_rainbow_sushi_is_easy_now/

note the use of both filters and input priorities favoring recirculation over the fresh input

hmm although as noted in the comments that design has issues with resilience and needed some hacks https://i.imgur.com/CSjbx6b.jpg

this appears to be more robust https://www.reddit.com/r/factorio/comments/bzel0u/science_sushi_loop/

[limpid cloud]

This is starting to seem like too much effort to have 1 belt

[north perch]

hehe, it can be easier to use multiple belts creatively with long inserters (and undergrounds), yes

[steep cloak]

I have a better way

[plain hare]

For labs there is definitely a better way 🙂

[steep cloak]

like this

[normal gulch]

just about to comment that theres an easier way now

[pulsar lance]

In worst case active measurement is also easy (glances at the next picture)

[plain hare]

https://cdn.discordapp.com/attachments/419527276875481111/1004819025441865818/unknown.png If you want to replicate take red or green circuit wire, choose 1, wire every belt to its neighbors. For the belt settings use measure belt contents, hold. Now you have a giant rotating measured chest. Take your wire and connect it to belts / inserter loading the belt. Enable them when (specific color of) science < X (X being desired count on the belt) .
One belt / inserter per science loading the loop

[pulsar lance]

This via madzuri like style or via clock?

[steep cloak]

clock

you don't even need all of those combinators

since they're same ratioed

just one DC clock to 7 and belts output on signal = 0

that image was just from me experimenting with a different circuit, it just happened to work for science at the time

[hardy fable]

because he's doing it wrong, he should filter out the iron at the mixer output, not during the recycling

it's not new stuff

[limpid cloud]

What are those square things?

[normal gulch]

?

[plucky isle]

Belts connected by circuits

[normal gulch]

read what putin said above that image

[tropic shuttle]

no /blueprint in here?

[normal gulch]

no bot

@clear wedge

oh it does work now!

[tropic shuttle]

Here's a badly thought out proof of concept I put over in #blueprint-designing to show sushi done with inserter ratios and no other logic

[plain hare]

Ratio sushi! (Inputs 10 gears for every 2 belts).

[tropic shuttle]

could use another wire and condition, to disable the gear inserter if the belt chest is <2

[tropic shuttle]

my actual use case was a 1:1 ratio, so even simpler

[naive rivet]

Doing sushi in satisfactory is absolutely horrible

It's a huge pain to do lol

I'll stick to factorio for that from now on

[tropic shuttle]

now you have me wanting to see if I can do sushi in autonauts or factory town

[hardy fable]

is it really? shouldn't be different than factorio: a mixer, a recycler and a prioritiser

[naive rivet]

There are no circuits

It's a pain to figure out the right ratio each time

[pulsar lance]

An approach like this does not need the „correct ratio“ though

[naive rivet]

Often times it does

You need some items way more than others

It's just generally annoying to do imo and doesn't save on anything

There's no punishment for overproducing either way

[hardy fable]

my sushi belts don't have circuits either

[naive rivet]

I know you don't need circuits for sushi belts

But there's no point in them

[hardy fable]

you're exagerating

you don't like it but that doesn't mean it is useless

and if you're mentionning the ability to handle empty belt it can be done without circuits as well

[naive rivet]

It is useless though

Like give me some good reasons to build them except for the sake of building them and saving like half a foundation of space

[pulsar lance]

I mean Sushi can simplify designs, lower the buffer or can make builds smaller…. With or without circuits (though some of those applications are appeareantly easier with circuits)

[naive rivet]

Buffers don't matter at all, if anything they're actually good

It doesn't simplify designs compared to just running belts on the sides of machines

And space is plentiful either way

[plain hare]

Sounds kinda like the conversations we had about sushi belts in factorio a year ago

[naive rivet]

There is no enemy/pollution/resource runout

[pulsar lance]

Buffers= setup price & hiding missed inputs untill it is to late

I mean you are still running belts on the sides of machines, just filled with the appropriate materials/ratios and much less inserters and belts needed

If you are going that way, why bother using anything else than T1 assemblers, T1 belts, splitters, T1 inserters everywhere?

Ohh and that obviously excludes modules… why use them when you can just build more of everything… you have the space available

[naive rivet]

There is no price to having buffers

And there are no modules or inserters

[pulsar lance]

And there is always the opportunity cost of not using the materials for something else
Though This is about satisfactory now?

[naive rivet]

If anything you're just using more materials for smart splitters

[naive rivet]

It has been about SF this whole time

[naive rivet]

.

[hardy fable]

I get it even less, that is also true for circuit-based sushi

[steep cloak]

satisfactory has infinite resources and practically infinite space, and there is no threat from expanding and no downside to spending resources

[naive rivet]

satisfactory

^^

[steep cloak]

there is no problem for sushi to solve in satisfactory

[naive rivet]

Thank you

You managed to explain it much clearer than me

[hardy fable]

there are two topics here that are mixing together badly

1. how difficult it is
2. how useful it is

yes it's useless, regarless of the technology used

no it's not hard since circuitless sushi exists

[pulsar lance]

Sorry, I did not completely realised that and hence Argumented badly. I have never myself played satisfactory (and my computer probably wouldn’t be able to)

[naive rivet]

It's not hard, just annoying like I have said multiple times

[pulsar lance]

=> conclusion: factorio is better than satisfactory

[hardy fable]

always has been

can't wait for official mod support

[naive rivet]

From satisfactory?

They aren't adding that

At least afaik, they've said they don't want to get in the way of SML peeps

[hardy fable]

for now™️

[naive rivet]

Their statement is they won't

[hardy fable]

where did they say that?

[naive rivet]

In a q&a video

Though I guess they might add it later yeah

[hardy fable]

yeah later, once the game is fully 1.0 stable and all

[pulsar lance]

=> hire the community

[hushed atlas]

i've never used a sushi belt for anything.

when is this useful? surely if you don't have the exact correct distribution of resources, or your input falters, bad things could happen?

[plain hare]

The trick is to make the sushi belt so no bad things happen! Otherwise it’s just spaghetti!

Sushi belts excel at low throughout multi item transfer

Usually occurring around space constraints, low tech, or construction systems, or low buffer!

Sometimes just for art tho.

[hushed atlas]

The trick is to make the sushi belt so no bad things happen!
i don't know how you'd do this.

[plain hare]

There are many many ways!

[hushed atlas]

resources in factorio aren't really infinite, so they could run out and then your sushi belt could have wrong proportions.

[plain hare]

Ahhh but you simply don’t allow this to happen! (Not the resources running out but the belt filling up with wrong proportions)

Sushi 101: https://cdn.discordapp.com/attachments/419527276875481111/1004819025441865818/unknown.png If you want to replicate take red or green circuit wire, choose 1, wire every belt to its neighbors. For the belt settings use measure belt contents, hold. Now you have a giant rotating measured chest. Take your wire and connect it to belts / inserter loading the belt. Enable them when (specific color of) science < X (X being desired count on the belt) .
One belt / inserter per science loading the loop

Here’s a basic one.

Anyway this is but one method

Dubbed ‘active measurement sushi’

There are many others that similarly never break, no matter what you do!

Aside from normal stuff that breaks production lines… like breaking it. Letting bugs in. Rotating belts. Changing circuit conditions.

[hushed atlas]

i guess this makes sense.

does this scale very well? you'd have to sort of count how many belts you have, and this might be unperformant.

[plain hare]

Yeah, this sushi method in particular excels for early game as it requires no combinators. And it’s easy to learn for sushi newbies.

And no scale / performance issues with measuring lots of belts Early game.

[neat idol]

one of my favourite types of sushi is to limit input with a clock to set requested ratios:
https://katiska.cc/n/temp/factorio/sushi/spider-sushi.mp4

all items are looped and removed from belt before being reinserted

[plain hare]

^^ I like this method too, more in the mid game tho. It produces a more stable sushi belt with even spaced items, it’s less performance intensive, easier to change on the fly. Can use underground belts and splitters with no worries.

Usually called a ‘rate limited sushi belt’

Oh and the sushi loop itself can be any size with the rate limits

A more advanced approach!

[plain hare]

@steep cloak I think you and warbaque have convergent design on the rate limited sushi belt with the modulo operator

Though warbaque has done this for a long time, I just didn’t understand it until now

[steep cloak]

I love warbaque's videos

I think we had a conversation before about it, it was pretty much the same thing but warbaque's was manually set and you had to do it yourself

whereas I designed mine to be automatically calculated and dynamically controlled

[neat idol]

https://katiska.cc/n/temp/factorio/sushi/spider-sushi-2.mp4
does this work better?

[steep cloak]

yes, that one worked

the core is the same, modulo clock with carry, but the purpose is different

[plain hare]

Tbh it seems the exact same

The pictured video is just example

[steep cloak]

I haven't seen anything more advanced from warbaque

not to diminish their design - it's the exact same mechanism

I just have a small addon to it that allows for like a tiny amount more control

[plain hare]

Which add on?

[steep cloak]

my entire design is the addon

inside my design is the same modulo clock that warbaque uses

[plain hare]

You’ll have to be more clear.

[steep cloak]

mine has extra stuff to calculate the total for you, specify exact items per time unit, and protect against zero signal specifically for dynamic control of the ratio through circuit

[neat idol]

for most purposes the draw is quite static, and if we dynamically adjust ratios I've found that that adds unnecessary latency to system since you need to wait until items are looping back before you can calculate how much they are being used

[steep cloak]

I have several iterations of the design, the most advanced one being the one I'm talking about

[neat idol]

because of that I dropped all dynamic designs and just setup static ratios in blueprint, fire and forget 🙂

[plain hare]

I mean, none of that is really, the core of the design… those are just like added safety features tbh so that the modulo style can be used for construction sushi belts

[steep cloak]

it can be used for whatever you want

I've had some ideas on how to use it in basewide resource distribution

I'm not trying to sell my design or anything, it's slightly more combinators for a couple extra features

like you could do it with a single arithmetic combinator if you wanted a static ratio on a full belt

and then add a dummy signal for spaces

[neat idol]

I like to use 2 arithmetic combinators so I can have simpler enabled rules for belts and keep all belts disabled until there's power

one of the fail states that sushi has is low power

[steep cloak]

yeah I was going to ask what was the second combinator doing

[neat idol]

https://katiska.cc/n/temp/factorio/blueprints/sushi/sushi-timer-2.mp4
for example here (around 2:25) you can see how sushi doesn't start until you add power, and all belts use [item] < 0 as enable condition

sure all ratios are 1/7 in this but functionality doesn't change

[hardy fable]

https://i.imgur.com/36Vq3oO.png

it scales for sure :D

on the topic of usefulness it's quite energy efficient because you only need one inserter for your many items (given the throughput needed is low enough) which makes it perfect for labs

[hushed atlas]

maybe i should make a mod with a recipe with 13 inputs, to force people to use a sushi belt.

wait, no, people could just use a cargo wagon or chest or requester chest or whatever.

hmm.

[hardy fable]

can't use trains in beaconed setups

but yeah requester chests are an option, tho very power hungry

[hushed atlas]

not with that attitude.

i'm thinking of just using the cargo wagon, not the train.

[hardy fable]

same

I mean the most efficient beacon setup allows you only 2 tiles on the side of the assembly machines

you can place a wagon but no inserter to move stuff from/to

[hushed atlas]

you can also just use chests for this accursedly.

[hardy fable]

yup

[steep cloak]

@neat idol ah I see you just made it subtract instead

you could get rid of it if you made the belt condition =7 probably

[neat idol]

That wouldn't work with arbitrary ratios

[steep cloak]

well yes but for the 1:1 x7 of science it would

you technically don't even need the modulo clock

just dot = any number 1-6

[plain hare]

Not quite! You get like 2.83 slots empty slots per machine on average

[neat idol]

Yeah, I know. My first version used simple C<7 clock. But since it doesn't play too nice with arbitrary ratios, I ditched it in favour of modulo clock.

https://katiska.cc/n/temp/factorio/sushi/Factorio 1.1.8 2021-01-08 23-13-07.mp4

Here you can see how spider parts are injected in bursts, which can cause jamming when ratio is close to 1

[steep cloak]

well yes, I meant for 1:1 all the way down

[tropic shuttle]

I'm thinking I should get around to making a roundup post on reddit/forum, maybe even on the wiki

[plain hare]

Roundup as in post your way to make a sushi belt?

[tropic shuttle]

don't you need circuit tech to get colored wires?

[steep cloak]

blueprints are free

[tropic shuttle]

no, post all the ways

or at least a representative sample of the broadest categories of ways

[steep cloak]

funny you mention that, I'm currently writing an alt f4 article on it lol

like literally I have it in front of me

[tropic shuttle]

woo!

better you than me, on multiple levels

[steep cloak]

oof, don't sell yourself short

[tropic shuttle]

are you looking for input? I'd be happy to contribute brainstorming

[steep cloak]

uhh yeah at some point, I'm trying to get a general structure down first before getting into specifics

[tropic shuttle]

cool

[glass needle]

Can you get sushi to work with undergrounds?

It seems incredibly hard since you can't connect them to the circuit network for some reason

[plucky isle]

There are plenty of other methods that do not require you to connect all belts. Keeping track of all in- and outputs for example

[glass needle]

oh right

Isn't that complicated

[plucky isle]

Not really. Instead of reading the amount of items in the system directly from a belt you just have to read it from a combinator serving as a memory cell

[glass needle]

oh okay I guess that's not too bad

Maybe I should have done that instead of having my system break every 2 seconds?

I ultimately just limited everything to one on a belt and trusted in the undergrounds to hold the necessary number of items

Somehow that worked...

[plain hare]

There’s a lot of methods to allow underground usage

Memory cell is one example! Rate limits another

[hardy fable]

https://i.imgur.com/F6vL1A9.png

[glass needle]

How does that work? There are no circuits

[timid cosmos]

Circuitry isn't the only road to sushi, here's an old design that uses splitters to limit output

[steep cloak]

splitters take evenly from their inputs

[timid cosmos]

I wanna say the design Xaetral posted can break with 7 sciences

max 6 items I think due to using yellow belts as the limiter

[steep cloak]

so the bottom four are taking equally from the four, and the top three are limited by one yellow belt per lane, which is one third of a blue belt

that is 7 sciences

only the top 3 are yellow belt limited

[timid cosmos]

Thats what I'm saying I'm pretty sure the design in that image will break

or not distribute evenly

[steep cloak]

well it's not perfectly even but it does work

it's slightly more yellow, purple, and space science than the other four

but it's still a loop with input priority

[timid cosmos]

My gut tells me there's jamming potential as one lane gets too full of 1 type of science

[steep cloak]

I'm testing it now, I can't get the four science lane to jam up

even input starving the sciences doesn't do it, only after re-adding all of them it did jam

so unless you have a long period of time where you input starve the loop and then reintroduce it and all of the labs were off and the science doesn't make it to the labs, then it will jam

[timid cosmos]

that scenario isn't uncommon, so i wouldn't trust it

[steep cloak]

yeah it's worth considering

there is a one combinator circuit that turns off all belts instantly if one is lacking

[foggy junco]

Not really, but it's very easy to break

[glass needle]

Right, but doesn't this require all inputs to be full?

Oh I just read the rest

[foggy junco]

That one is flow limiting sushi

[glass needle]

Wow I would have expected it jam easier

[foggy junco]

If you control the maximum throughput of each science, it will never clog

Even if one input is left empty and then added back

E.g. 3 yellow belts can merge into a single blue and never clog as long as the blue one is priority fed back into the inputs

Now if you wanted to do that with 7 sciences id rather recommend a flow limiter using circuits

Since breaking a belth into 7ths isnt beautiful

one sec i can show you a circuit based one

it's not nearly as complicated as you'd think

[glass needle]

Uses circuits not allowed

jkjk

[foggy junco]

a single combinator

[hardy fable]

you can add a tiny circuit to lock the belt when one input is missing, it's 100% reliable

there's also the rate-limited version that is also 100% reliable without circuit

[foggy junco]

No need to though if its flow limited via any method, is there

Oh wait

Only the top is flow limited

I see

[hardy fable]

there's also the third option to fill the empty gaps with another item

[foggy junco]

But wont that mean the bottom 4 sciences are a bottleneck?

[hardy fable]

I'm not sure what you mean but yes, this is 45/8 science per second, not 45/7

that can be overcomed with that third option by filling the gap with the sushi itself but that makes the output patern pseudo-random thus less beautiful

[foggy junco]

As in, there's 1/6th of a belt of white, yellow and purple but only 1/8th of grey, blue, green, red

[neat idol]

This type of sushi works great with undergrounds: https://katiska.cc/n/temp/factorio/sushi/spider-sushi-2.mp4

[glass needle]

Ah so is it calculating the percent of things on the belt instead?

Sorry might have trouble figuring out what’s going on there

[neat idol]

Yes, belt might have for example 123/777 or 15.83% of one item, and all unused items are looped back and fed again to input at priority

So if items are not consumed, no new items are added

Ratio limiter works by limiting input belt speed (e.g. 123/777)

[neat idol]

https://katiska.cc/n/temp/factorio/blueprints/sushi/sushi-timer-2.mp4
example with science

Blueprints if you want to play around with them
https://katiska.cc/n/temp/factorio/blueprints/sushi/sushi_timer.txt
https://katiska.cc/n/temp/factorio/sushi/spider-sushi.txt

Another type of sushi that kinda works with undergrounds and unsaturated belts is rolling average counter:
https://katiska.cc/n/temp/factorio/sushi/sushi-average.mp4

If any item drops under threshold, just add more items. But that's more finicky and sensitive to latency.

[glass needle]

So many sushi varieties :0

[plain hare]

We haven’t even found them all yet!

Or good uses for some of them!

[hardy fable]

half a belt is 3/4 and the other is 4/4
there are 7 items on 8 places

[foggy junco]

take any random box on your belt that contains one green+blue+red+grey science and you'll see you always have one extra science of either purple, white or yellow

[hardy fable]

no

one less

[foggy junco]

in that box there's one extra purple science

[hardy fable]

wait

[foggy junco]

overall there's 33% more white, purple and yellow science than the rest

[hardy fable]

oh nvm

I'm mixing stuff up

omg, when the original designer can even remember how it works

yeah u right

but using a placeholder item instead of T1 belt limiting doesn't have that issue

[foggy junco]

i mean, in the end it doesn't really matter since they're on the same side and don't interact with the opposite side

but still worth noting in case suddenly one side looks emptier even though you're producing / consuming the same amount of each science

[hardy fable]

yeah but it increase the buffer size

also that extra 1/8 can be filled with another sushi belt so you can use all the belt throughput

[near patrol]

I heard you guys like sushi, how about a modular sushi crafting logistics system:

[normal gulch]

I feel like you can the number of combinators @near patrol

[near patrol]

prob

but it works anyways

[steep cloak]

I also feel like you can the number of combinators

[ashen rampart]

sooshi

figured I'd try and apply the belt speed-based sushi method to DSP

[naive rivet]

Looks kinda similar to SF in terms of recipes?

Are those water bottles + copper ore turned into copper ingot?

And empty barrels returned on the back

[ashen rampart]

hydrogen/coal/refined oil turned into more refined oil

[naive rivet]

👍 kinda like the diluted fuel recipe

[quasi kelp]

mmmMmMmMMMmmmm

[steel scarab]

whats sushi? sushi belting? or are we actually posting sushi

[plucky isle]

Sushi belts: putting more than one item type on a belt lane

[steel scarab]

i know what sushi belting is, but i want to know if we can post sushi

[steep cloak]

probably not

[desert tide]

you can put sushi in #nomtopic

[neat idol]

Maybe it should have its own sushi subthread

[proven perch]

Oh clap I just wanted to make a mall

[near patrol]

same difference

[oblique crater]

My first sushi attempt. 3 items on the same belt in equal quantities maintained. working on making it much smaller

[steep cloak]

looks like there's twice as much copper as everything else

but you've got the right idea

[timid cosmos]

If you wanna do splitter sushi, you'll need some more loops and proper input priorities

this was my try at splitter sushi, I don't think it's the best at all but its the only working example I have on hand

[oblique crater]

you're right. standby for a fixed build

[oblique crater]

incorrect! this can be done in a SIGNIFICANTLY smaller space. standby for my mk2, now flawless

[steep cloak]

it doesn't look like they're doing splitter only, some of those belts are hooked up to what looks like a rate limiter clock

although only on the one side

[oblique crater]

me? or what?

[steep cloak]

yeah, what's the circuit wire for

[oblique crater]

just a simple clock, yes

[steep cloak]

but it's not hooked up to the output of the splitter, meaning the stuff that gets recycled isn't getting limited

[oblique crater]

exaclty, which is what i'm working on now

like I said, my first time messing with sushi belts

in 5k hours I typically think of them as unnecessary and unsightly monstrosities

that being said a recent encounter has peaked my interest, and now I need to see if I can make the same exact thing in like 1/100th the space.

[steep cloak]

I will say I have done exactly the same method you have recently, but I want you to discover it yourself

so much so that I've made loopless sushi

it just gives the machines the right amount of resources and never backs up

[oblique crater]

i appreciate that 😛 Im having a blast so far

flawless. under any load, or no load there's no way it messes up.

[timid cosmos]

Circuit sushi is much smaller, I thought you were going for some kind of splitter design

[oblique crater]

i don't really know/care what design im "going for" yet tbh.

[steep cloak]

._.

I just tried to blueprint the loopless thing and it didn't work

[oblique crater]

I just want a belt with any number of items on it - without like a thousand square miles of real estate occupied with the setup lol

this thing I came up with here is still a tad cumbersome imo

[steep cloak]

blueprint bot is not working for me

[oblique crater]

Here's a "split belt" for you, that's not 1/2 and 1/2 😄

15:1 iron/pro units

super compact, and usable.

[steep cloak]

that is weird

[oblique crater]

can change the ratio of the split to anything pretty easily! even like 1:1000 if you deside

working on a 3 item build now

[steep cloak]

good luck with that

3 gets hard when the ratio is weird

[oblique crater]

3 items. They can be any ratio with 3 more combinators. For equal it's easy.

[steep cloak]

it's interesting watching you follow a very similar approach to how I did it

[oblique crater]

5 items. Working on a more compact design.. standby

[steep cloak]

I assume you're setting the number in each belt, and the total is in the decider?

[oblique crater]

just a simple clock that resets at x ticks, where x=number of items. belts set to enable at the first tick only. for unequal number 3 more deciders are needed for each unique item after 2 - so that would get to be a huge logic system if you have 20+ items for example.

that being said, with bitshift it would become more compact again after about 7-8 items

[steep cloak]

using this system, you can set the numbers in the belts to the ratio you want on the belt and have the clock reset after the total of those items

[oblique crater]

I'm working on something cool rn

[steep cloak]

only problem is when you have multiple things releasing in batches it can clog while everything tries to fit onto the same belt and it doesn't empty before the next cycle

[steep cloak]

with this method, I should specify

[oblique crater]

yes ofc, that's why you only use multiples of the # of belts, or divide by slightly more to have a 'loose' belt

instead of enabling 3 belts 1/3 of the time, enable them each 32/100ths of the time

the eye doesn't know the difference

[steep cloak]

well no you can set the numbers to be like 3:2:1:1 and the clock to 7 that way the first belt is open for 3 ticks, second is 2, and the rest

that way it's a 3:2:1:1 ratio on the belt

which works for low and similar numbers because of belt buffer

[oblique crater]

you said exactly what I just said lol

[steep cloak]

lol yeah I guess I misunderstood

the 32/100 would be not great because it would release items for 32 ticks and then be off for the 68 remaining, and when you get more item types it tries to force the entire belt's worth of items right at the beginning of the clock and it backs up really easily

it's also just not balanced

[oblique crater]

11 items.

not sure if this is possible in a smaller area

[steep cloak]

it is possible, but I wouldn't worry about it

[oblique crater]

22 items. can really be any amount. pretty sure i found the most compact tile. works for any amount of items flawlessly

[steep cloak]

you're doing great, but I wouldn't go throwing around terms like "most compact" so easily :P

[oblique crater]

consider it a challenge then haha

[steep cloak]

can you adjust the ratio?

[oblique crater]

let's see it 😛

[oblique crater]

yes

[steep cloak]

dynamically?

like through circuit network and not manually

[oblique crater]

it's possible, but I haven't messed with that no

set it, and it'll maintain a certain % of each item

[steep cloak]

try that next

[oblique crater]

but it would be cumbersome to change "on the fly"

[steep cloak]

I wouldn't call it cumbersome

you're still only using one combinator for the clock

[oblique crater]

no need for more than one with equal setups

[steep cloak]

well yeah, but I mean for changing ratio on the fly

[oblique crater]

would also need buffer chests and likely a different belt setup for that

[steep cloak]

mhm

I'm trying not to spoil your fun, but if at any point you want to see my stuff I can post it

[hardy fable]

I'm pretty sure mine is more compact

https://i.imgur.com/xULcZS7.png

[steep cloak]

you can make it one more compact

instead of having the third row of splitters two tiles above, make it one tile above and sideload in before the splitter on the left side while the right input side is connected directly to the previous one instead, and make them all filtered so nothing extra gets through

[oblique crater]

yes but mine can do any amount and any ratio

yours is limited to 2x multiples

[hardy fable]

not quite

as long as the sum equals a power of 2 I can do it

and with a recursive feedback I can get any ratio

[oblique crater]

That’s exactly what I’m talking about lol…

[oblique crater]

Show me 11

[hardy fable]

sure, 16-5

actually 8 + 2 + 1 with 4 + 1 as feedback

[oblique crater]

Like I said, show me 11. This is words and numbers lol.

[hardy fable]

I was at work, can't boot up the game there

there you go, 11

[oblique crater]

Rather than just showing me a balancer, I was more referring to the complete build with other factors considered

[hardy fable]

aren't you asking a bit much 😄

[oblique crater]

Symmetry, expandability, etc

[oblique crater]

You’re the one who said yours is better lol

[hardy fable]

a splitter sushi is a balancer anyway

[oblique crater]

They’re not exactly purposed for the same thing anyways but still

[hardy fable]

no, I said I believe than mine is more compact

[oblique crater]

But you’re not comparing the same thing…

Different capabilities

[hardy fable]

dunno, I assumed we were talking about sushi belts XD

yeah ofc each one of them has different characteristics

[oblique crater]

Yes but there’s more to it than just that

Show me a smaller design that can do any amount of inputs would be a good comparison

I’m here to learn fyi, as I’ve never experimented even once with sushi in my 5k hours

[hardy fable]

depends on the real application, I've spent much of my time on science sushi

so 1:1:1:1:1:1:1 ratio is sure easier to figure out than the ratio for some recipes

[oblique crater]

Yes definitely. With 3 more combinators I can do any ratios basically. But then again with combinators any array of splitters likely can

2:1:5:2:1:9:7:20:4 whatever.

[oblique crater]

What’s this doing?

[hardy fable]

electric motors

duh

[oblique crater]

Well yes lmao

I’d like to think you know what im asking

[hardy fable]

2:1 engine and circuit, no big deal

[oblique crater]

Loopback somewhere?

[hardy fable]

yup, right here

I made the one for flying robot frames as well

but this one and the one for engines are a bit tricky to make them work

you need some sort of flattened belt loop, it's kinda weird

[steep cloak]

let's see if the blueprint bot works today

wtf

[oblique crater]

so this kind of build just perfectly consumes the exact number of items, and no need for a loopback?

i assume it'd only work with <9 items then?

[steep cloak]

there's still a loop, it just doesn't go all the way back

this is what it looks like to not have a loop

exact ratio, exact amount on the belt, input underflow detection and output overflow detection

[hardy fable]

absolutely

[hardy fable]

why?

[steep cloak]

what happens when you run low on one of the resources

or the output backs up

[hardy fable]

well, it's not the same kind of loop
since it's just used to account for the timing windows of the inserters only

[steep cloak]

but there's still some sort of a loop?

[hardy fable]

ah you mean the recursive mixer?

[steep cloak]

cause this is a loop

[hardy fable]

yeah it's a loop in the sense that the belts go round

[steep cloak]

but not to the input splitters, yeah

[hardy fable]

but they rather meant the loop in a more abstract manner, the recycler going back into the mixer

[steep cloak]

my point is that it's really hard to do fully loopless sushi

in fact I don't think I've shown this to @plain hare or @normal gulch

[oblique crater]

Full. Loopless. Sushi. Hmm

Apparently loopless is two words. Learned that today also 🤣

[steep cloak]

lol

[oblique crater]

Ratios would have to be spot on. And output couldn’t back up either

[steep cloak]

mhm

[steep cloak]

that's what this does

[oblique crater]

That actually seems simple enough tbh. I’d have to pick apart your combinators but I’m sure after a few eons I could figure it out 🤣

I spend 4/7ths of my time in Factorio in test worlds anyways lol.

[steep cloak]

it is not simple lmao

[oblique crater]

Well for something as simple as 2-3 ingredients to start it seems simple, unless I’m missing something fundamental about the whole thing

[steep cloak]

sure, but to make something that works universally for any number of inputs

that's what this is

[oblique crater]

Because belts can contain only 8 items, so in the case of a belt stopping perhaps one of the items an assembler needs wouldn’t be present on the belt so things would stop

But again I haven’t played with this method at all so I’m not qualified to answer properly yet lol

[steep cloak]

the simple version of this circuit can fill a belt with any ratio of items, dynamically controllable on the fly

[oblique crater]

By dynamically, are we talking about:

If x is being consumed faster than y, then x will become more frequent on the belt over the subsequent loops? Or something else

[hardy fable]

like this one? :p
#930147588160782436 message

[steep cloak]

dynamically as in you can control the ratio on the belts with the circuit network while it's running

[hardy fable]

unless you have quite some chance on the recipe there's no way to get around messy inserter timings

[oblique crater]

So you have to mess with a few combinators?

[steep cloak]

and that works flawlessly, under every circumstance?

[normal gulch]

wont that design break if input is not there?

[steep cloak]

if whoever is using this circuit comes up with a circuit to control the ratio of items on the belt, then this will handle it

[hardy fable]

lmao no, you can have 1000 different items on the belt if you want to
it's just that this belt won't be able to feed more than 45/1000 recipe crafts per second

[oblique crater]

I’m talking about loopless, and I could do without the “lmao”

[steep cloak]

xaetral I'm like 99% sure your design will break if there's not enough of one input or the output backs up

[oblique crater]

Also I’ve stated I’m learning something new lol

[steep cloak]

under ideal conditions, it runs fine, but not in a real base when shortages occur

[hardy fable]

yup

[steep cloak]

I find that very hard to believe

[normal gulch]

btw, can I have the string of this? probably would be helpful at some point

[steep cloak]

oh my god

blueprint bot broken

[hardy fable]

yeah but I mean we discussed about this a thousand times already, my sushi works flawlessly as long as you ensure this point
but you can use a small combinator circuit to do so

[steep cloak]

oh sure, it works as long as you prevent the thing that makes it break from happening :P

but in it's current state it doesn't

which is what I'm trying to get at

[hardy fable]

it's very specific to that recipe to be honest
first there are only two items, and then since I put all the engines on the bottom lane, once the inserter first get rid of the lil bit of circuits within that lane you get a full lane of each, like this:

that's why it works

[steep cloak]

bot appears to be broken in threads

#bot-stuff message

[oblique crater]

This is fucking magic. When you initially brought up dynamically controlled I said it would need buffers and such. I swear my brain was thinking of something very similar to this. Let me work out the circuitry today and see if I can replicate without you telling me the logic 😀

[hardy fable]

I'm also talking about the recyclerless, and that lmao wasn't supposed to be rude, the fact that you could have so much more items than that is quite funny when you just think about it

[oblique crater]

Perhaps I don’t know the difference then. In my mind I’m seeing the “end” of the loopless belt, and it doesn’t contain the item the last assembler needs. ie: limit of 8

I’m also fully aware that you can have many more items on the belt, as the initial conversation starter was my 22 item build

[steep cloak]

@normal gulch forgot to mention, only turn off the red combinator when the rate limit belts are full

[oblique crater]

At some ratios this would break due to not enough buffer correct?

Hmm. Maybe not tho 🤷‍♂️

[steep cloak]

nope

[oblique crater]

No ratio, and no amount of time or consumption imbalances will break it?

[steep cloak]

nothing breaks it

it doesn't release resources in batches

it's evenly spaced over time

[oblique crater]

What about length of loop? Would need to increase buffers then?

[steep cloak]

nope

there's enough buffer with the output and input sides that you could just route it right back up to the start

[oblique crater]

What about a 0 ratio? Still fine?

[steep cloak]

what do you mean 0 ratio

like one thing gets set to 0

[oblique crater]

2:6:10:0

Yes

[steep cloak]

then it wouldn't show up on the belt as you cannot have signals with a value of 0

still full

[hardy fable]

ah I see, that's why I used the specific term "recyclerless" to refer to the absence of a feeback that goes all the way to the making of the sushi
but what you mention is a good example of why I'm doing those flattened half loops (btw with 3 inserters you can have access to 3 belts thus 24 different items in theory)

[steep cloak]

you can see the top where it started changing the ratio

[oblique crater]

Wow

Even a single item is fine?

(Disregarding inserter throughput)

[steep cloak]

yeah disregarding throughput, but this is still the highest throughput for a yellow belt with one inserter

as long as one resource doesn't exceed 50% of the ratio the belt is full, as far as throughput limitations go

if you actually do need more, you can add another silo of the same resource and change up the ratio to compensate, which I think I have a circuit for but misplaced at the moment

[oblique crater]

Absolute fire. Certainly a better design doesn’t exist for this

You mentioned your loopless sushi, can I see another one of those builds you made?

[steep cloak]

sorry I missed this, that's the only build I've actually done with loopless, more as a proof of concept

[oblique crater]

well after a few hours messing around myself, I'm ready to see the inside of your combinators lol

[steep cloak]

are you familiar with the modulo operation

[oblique crater]

I made a version that was in fact dynamic, but it required changing the variables in several combinators, not at all convenient

[oblique crater]

just the remainder of a division correct?

[steep cloak]

yeah pretty much

there are two types of basic clocks, decider count and arithmetic modulo

[oblique crater]

i've never heard of the second

[steep cloak]

blueprint bot appears to be broken in threads currently, one moment

ah whatever I don't need it for something this simple

this will cycle once every 60 ticks, or one second

CC is dot = 1, of course

[normal gulch]

@clear wedge I think if you ping it it joins the thread, so it should be fine now?

[steep cloak]

because when dot reaches 60, 60 / 60 = 0 R 0

nah it was already in the thread

I did some testing earlier and it's an issue with the bot itself

someone is working on it I think

for now, links to bot stuff I guess

[oblique crater]

ok.

so im assuming this is a much more diverse clock?

[steep cloak]

yes

let's say I want to count up by 7

7, 14, 21, 28, 35, 42, 49, 56, and then it goes to 3

[oblique crater]

following.

[steep cloak]

so it carries over leftover from the previous clock cycle

then 10, 17, so on and you get the idea

[oblique crater]

yes

[steep cloak]

I'll replace 60 with a signal [T] for total

let's say I want a belt to be filled evenly with three items

the ratio is 1:1:1

[oblique crater]

3 items total

[steep cloak]

if we add up each number of the ratio, yeah it's 3

so [T] becomes 3

and we supply the ratio numbers to the modulo clock

each % [T]
output each

[oblique crater]

wow

[steep cloak]

now you get all 3 signals to cycle every 3 ticks

[oblique crater]

i was going down a much different path

[steep cloak]

you can get an output when the input signal is greater than T

so hook the input ratio and the clock output to another decider, and that's pretty much it

each >= [T]
output each 1

I forgot to mention how getting the total works

[oblique crater]

just add everything and convert to T

[steep cloak]

it's a single arithmetic combinator, each + 0 output T

[oblique crater]

or *1 output T

[steep cloak]

yeah pretty much

#bot-stuff message

here's the example circuit

you can put any number of signals in the combinator with any value and it will work, provided you don't make the numbers so big you run into integer overflow

[steep cloak]

I just realized that this image has an extra combinator, you only need the 3 for this to work

[oblique crater]

wow, i can see it all work now

very simple, and also works way better lol

[steep cloak]

I didn't come up with this as easily as I explained it to you, it took me a while

so don't feel bad for not thinking of it lol

[oblique crater]

it's utterly insane watching this work ngl. flawless

[steep cloak]

all because of the modulo clock

it felt amazing running into that

[oblique crater]

what do you have your buffers set to ? like 50?

[steep cloak]

on what, the chests?

[oblique crater]

yeah

[steep cloak]

oh I didn't really think too hard, I think they were like 100 or something

that can be changed pretty easily as well, you could have it be a signal and set it via CC instead

that way you don't have to set each individual belt condition

like enable when resource < [A] and set [A] to 100 in a constant

[oblique crater]

some low belt compression under certain load it's seems, have you discovered that too?

[steep cloak]

what does the input look like

the belts should enable when resource > 0

[oblique crater]

oh nvm, it's just the 1/2 belt max per resource

[steep cloak]

ah yeah

67% on that specific design

[oblique crater]

yea

[steep cloak]

that was meant to be for a larger amount of smaller ratios

oh, also if you give the ratio a signal that doesn't have an output, it will essentially be "gap" on the belt

[oblique crater]

wouldn't it be 83.3% for that specific one?

[steep cloak]

so if the ratio is 2 iron, 1 coal, and 1 fish, and there's no fish belt, 1/4 of the belt will be empty

[oblique crater]

50% + 1/3 (33%)

[steep cloak]

I was just thinking 1/3 is 33% and 2/3 is 67%

and 67% is over 50%

[steep cloak]

anyway with more combinators I abused this property so you could control the ratio of items to gap on the belt with a different combinator

that's how this design works, and how the belt is so empty

I calculated the items/minute that the build uses and put that in as the input rate, so the machines get exactly the amount they need to function

[oblique crater]

pray you don't get low on power lol

[steep cloak]

well you can probably detect that with accumulators being low

there's a shutoff combinator that freezes the clock if 🟥 > 0

that's what I use to turn it off if there's not enough input resources or the output backs up

I've been meaning to do a full sushi base at some point, combining a bunch of different variations of this circuit and a bunch of inputs

I'd need to make a different version of this that outputs some signals for use in global control maybe

[half phoenix]

sushi furnaces? 🤔 https://www.youtube.com/watch?v=XiMJ156RZWs

[dark grove]

hello sushi chat! Im going to work on this here i hope its relevant

idea: train that provides the exact number of raw resources to make x science into a sushi belt and some magic

[steep cloak]

the sushi part is

not sure how you're going to go about exact numbers of things

[dark grove]

i think i have an idea

[vestal jasper]

this looks nice

[dark grove]

im gonna try to work on The Line now

really

this is not ok moment

i cannot deal with excess

200 almost works but

i am not using any prod mods

itll only make it worse

if i get rid of space science its better slightly

but iron is still annoying

18 science gives workable and cool looking numbers

hmmm the oil isnt good

[normal gulch]

so do multiples of 18?

[dark grove]

possibly

bruh

that means the clean number will be 468

maybe i can just not deal with the oil

[steep cloak]

nice gears

[dark grove]

very nice

i need just barely 3 train wagons

this is terrible

its 84 stacks

[dark grove]

YES it WORKS

the inserter limiting thing works

sushi

[vestal jasper]

good work mate

[vestal jasper]

ı just like saying mate ım not from uk

[oblique crater]

Not really sushi in the traditional sense, and you may need some logic on either the train/inserters or a sorting area otherwise it’ll back up.

An omni smelter is something I’ve worked on in the past, but again that’s getting away from sushi and delving into combinator territory.

[steep cloak]

it's sushi

look at the output of that balancer

that's going to be all of the resources

it's just not controlled sushi

[dark grove]

Won’t back up, it’s calculated

[oblique crater]

Can I see the rest of the build?

[dark grove]

It doesn’t exist yet

[dark grove]

furnaces

ive made the circuitry work

now i have to figure it out for two items

a test run

its perfect

issue is that it doesnt deal well with recipes with more than 2 items

[dark grove]

figured out a 3 item way but it forces me to set it to stack limit 1 unless i can do some circuit magic to circumvent that

i can just increase this for more items

[dark grove]

more

[dark grove]

huh its done

[naive rivet]

Why does it need to be stack size 1?

[oblique crater]

To keep an exact count of items, otherwise inserters can move a random amount of items up to their stack limit

[dark grove]

It turns out my whole implementation of this is flawed

I will have to do some circuit magic

[wraith badger]

I think you're the only other person I've seen try to do loopless sushi using circuits.

Here's my version that might hopefully might be some inspo

[steep cloak]

I have

#bot-stuff message

the bot was broken when I sent these so that's why it's in bot-stuff

[wraith badger]

Do you not face assemblers starving each other?

I guess there were enough assemblers in the line or circuits are limiting input throughput to prevent the problem

[steep cloak]

the circuit is perfect mostly

if you want to test out that blueprint, wait until the belts stop moving and then flip the top combinator

[wraith badger]

I recommend testing it at max speed for a period of time. Cuz i guarantee that assemblers will start starving each other

[steep cloak]

they don't

I had it on max speed for a while when I had it originally

[wraith badger]

I had to fix it by creating a counter for each assembler and syncing the assemblers

Regardless nice 3 item input build. Maybe I'm wrong and 3 items dont face the assemblers starving each other as much w/ 2 items only especially for blue circuits

[steep cloak]

it might be the recipe I picked, but theoretically this works for any number of inputs

[wraith badger]

yea i think w/ blue circuits w/ 20 greeb vs 2 red i think i faced a rare issue

[steep cloak]

I don't know what the circuit you used is, but mine is a rate limiter that is set to a calculated ratio at a calculated rate, and it freezes the clock if there's not enough input or the output gets buffered

I'd have to try blue circuits with my design to see if that's a unique issue or not

[daring ivy]

https://tenor.com/view/sushi-roll-food-japan-rolled-sushi-big-roll-gif-16104626

[timid cosmos]

when you try to eat that it'd just be like sushi casserole

which sounds alright

[tacit cedar]

behold, item ingredient shortage

my first sushi attempt

[tacit cedar]

definitely needs another sushi belt of input, and it's not stable at all

in theory, 3 sushi belts into 2 assemblers should be a full blue belt of output (5.4k/m)

[foggy junco]

You might wanna try using fast inserters instead of stack filter ones?

At least thats what i figured out would work best with the science sushi

By the way, you're filling the belt with 1/3rd of each ressource, right?

Instead of the big splitter flow controllers you could also just use yellow belts to flow limit

[tacit cedar]

oh yellow belts is big brain, but i still fundamentally need another belt to supply everything
i gave up on that design anyway

[tacit cedar]

interesting

[steep cloak]

technically you wouldn't need yellow belts on it anyway since splitters input balance

there's one or two reasons yellow belts are technically better if you're having a shortage on one thing but too lazy to get into it

[tacit cedar]

I was worried that if one input stopped, the loop might get clogged with the other input

[steep cloak]

it shouldn't clog as long as you have input priority on the loop side

[tacit cedar]

what im thinking is
if only one side has input, it will fill the belt completely, right? with just a splitter
so when it has to go back to only filling half of the belt, there would be nowhere for the excess to go

[steep cloak]

I guess if it backed up more than halfway around the loop yeah

I'm used to there being buffers between loop input and output

[tacit cedar]

like this

[tacit cedar]

ya the yellow belt thing makes it a lot easier

but what if I sushi the sushi?

it actually works lmao

oh it stopped moving, green circuits got clogged

I just had to break the other two loops, and leave only the big one

sushi can be sushi'd

oh wait that's not right, guess the belt trick doesnt like it when it's backed up

but I guess neither would splitter sushi

this is hard

im not smart enough for this

[steep cloak]

if you're trying to get 6 belts of input, make each one use one side of the belt instead of both, then lane balance after combining 3

[tacit cedar]

lane balance?
one side of the belt is smart tho

I mostly just wanted to see what happened if I did splitter sushi, but all the inputs were sushi

[steep cloak]

btw those loops have output priority and not input priority

[tacit cedar]

o lol

[hardy fable]

something I found out is that mixing two sushi belts (or more) never ever works

not because I didn't try hard enough for sure

what I tried to do was to have 8 science sushi belts, sorting them out into the same 7 belts then reuse that as usual with a priority

but weirdly enough, this is unstable at all times

anyway, glad to see a fellow poyo enjoyer doing splitter sushi

[tacit cedar]

I figure it would work as long as none of the sushi belts ever get backed up, they all have to be moving at full speed

[tacit cedar]

putting this at the bottom to reference it later

idk I think the belt looks kinda ugly, I think I prefer splitter sushi

[normal gulch]

this ^ is why I want to develop a 3rd type of sushi, basically the idea is it reads how much is on the belt and how much needs to be put on / taken off. It should be compact like this wothout having the ugly yellow boxes

[tacit cedar]

beautiful

only a little big

[tacit cedar]

I would assume you can just use a number to figure out how many of each item are on the belt without needing to read the entire belt
just add 1 for each item that enters the belt, and remove 1 for each item that leaves
you'd have to wire every input and output though

[normal gulch]

oh theres more compact options than these limiters!

[tacit cedar]

limiters?
I just took a 1-7 balancer, and routed 6 back to the beginning
I did take it apart a little bit so it was less big, though

[normal gulch]

yeah they are usually called rate limiters

theres new ones ( within the last year ) that use circuits to limit the belts insatead of splitters which gives much more accuracy

[tacit cedar]

here's the balancer on top of my custom™ rate limiter

the splitters seem to be accurate as long as it doesn't get backed up

and a proper sushi belt shouldnt get backed up

[normal gulch]

Yeah its accurate but big when you want mroe than just 1/7th or 1/8th of a belt

[tacit cedar]

oh you mean like 1/100? yeah I could see that being an issue

I still enjoy the simplicity of splitter sushi

[normal gulch]

Yeah its simple but I found it ineffective for construction belt sushi

where you have a belt that makes more factory

[tacit cedar]

is there a way to make this better?
like have the sciences in order, and matched up (red on one side; red on the other)

I hope the answer isn't "yes, but you have to use circuits"

[normal gulch]

maybe using 1/8th of the belt instead? I havent found a fix for that

[tacit cedar]

I guess matching it up would mess up whenever it goes around a turn

but having them in order feels like it should be doable

[normal gulch]

this is what the new limiters look like and how compact

[tacit cedar]

what is the dot on the splitter?

[normal gulch]

thats just to not waste the 2 science packs that would sit there forever

[tacit cedar]

what does it do?

like how does it save the science

[normal gulch]

copy tool iirc

if I didnt have it there, there would be 2 science packs waiting on the end of the belt

[tacit cedar]

oh, you prioritize something impossible

I didnt know what I was looking at lol

[normal gulch]

Yeah its a lot of optimizations lol

and since I mentioned construction sushi.. all this is built off of a mall belt

and it would be nearly impossible with the splitter sushi I did before ( for sushi rails mk1, this is mk4 now )

[tacit cedar]

I never got deep into circuits so I can't really read what's going on
I can automate cracking, but I haven't actually placed any combinators

[normal gulch]

Yeah.. ill link a convo where codegreen ( the creator of the circuit ) explained it

[tacit cedar]

codegreen is watching, I can see him through my discord-colored window

[normal gulch]

#combinators message

[tacit cedar]

why are you using more combinators than in the explanation?

[steep cloak]

they control the output rate of the belt as a whole

[normal gulch]

im doing like 90% of a belt instead of a full belt

[steep cloak]

also I think there's a redundant combinator in there that was an artifact from a previous circuit

the one with the can be removed iirc

[normal gulch]

probably, but its a lot better than what it was before with a clock for each item

[tacit cedar]

if you don't want a full belt, you can just but a blank signal in the input
like he says in the explanation

[normal gulch]

Yeah but if I add more items on the belt id need to adjust the blank #

[steep cloak]

you'd also have to calculate out how to multiply everything if you wanted an exact rate on the belt

if you wanted 14 items per second, you can do that by setting a signal to 14 instead of having to calculator everything

and it's automatically adjusted if the ratio changes

[tacit cedar]

how do you put the copy tool or whatever into the filter for the splitter

[normal gulch]

0eNp1j8EOgjAQRP9lzsUEBMT+ijEGcdVNStu0i5EQ/t0WL148zmbmzc6Cq5nIB7YCvYAHZyP0aUHkh+1NvsnsCRosNELB9mNW9z5KEb1hEQpYFdje6A1drmcFssLC9AVtYr7Yabwmpy7/IBS8iynlbO7MpLJqFGboot7XuyZV3DjQ8HXUCnc2OacxOD8XPsGoEOdMIrlJ/CSXtMqF1J08gR9PQf5tm6F/Viu8KMSNWnVlfThWh7Zturbr1vUDcA9faw==

[steep cloak]

magic

[normal gulch]

heres a string for one, made by codegreen

[tacit cedar]

I will cherish it forever

I'm trying to follow along on the #combinators link you sent, but it's hard to see where all the wires are
blueprint string no work either

[normal gulch]

The string should work..

[tacit cedar]

it says unknown version or something like that

[normal gulch]

My laptop decided to turn off when it’s on I’ll see if I can reshare it

[tacit cedar]

maybe? idk

the middle one is cycling quickly, which I assume means it's working

the belts aren't getting the signal though
and I'm pretty sure I did the wiring right

oh I didn't set up the last combinator
I don't know what it's supposed to be, though

[steep cloak]

why isn't the blueprint working

[tacit cedar]

oh it worked that time
🤔

looks happy

it doesn't want to get in nice r/g/b/b/p/y/s order tho

full screenshot because I might as well

[tacit cedar]

do you have special stack sizes on the inserters?
my version seems to back up slightly every once in a while

oh is that what the gaps are for?

[normal gulch]

yeah.. theres another version thats probably better

an underground belt goes thru the limiters in other versions so its max thro ughput is increased

[tacit cedar]

it probably only happens when you use the full belt like I am

it'll probably just be fine if I use both sides of the belt

hm, I put a lane balancer to try and make it look better, but it does waste some science

[normal gulch]

that wasted science isnt bad compared to the modules

[tacit cedar]

oh wait I just realized I dont need a lane balancer if it comes in on one lane anyway

i can just move it to the other side

with a single splitter

very nice

[steep cloak]

if you're doing something where everything is output the same amount, you only need one combinator

[foggy junco]

Thats a lot of combinators though

[normal gulch]

Yeah it is

I made a new version last night that has 3 combinator and stacks with beacon lines, not perfect yet unfortunately

#blueprint-designing message

[foggy junco]

nice

but are you sure about stack inserters?

they usually caused problems in beaconed sushi setups for me

[normal gulch]

Oh interesting.. I’ll downgrade to blue and see what happens

I’m also trying to save a tile or two and make them only input on one side

[tacit cedar]

cleaner mini version

and tiled (2.7 spm)

I would assume a military science-less version would only be 6 lab rows

something like this

[normal gulch]

Nice

Does it consume all 2700 spm

[tacit cedar]

probably, it’s still a full belt of each science

[normal gulch]

I still want to see if I can get my version more compact

I think using fast inserters as someone said is the key

[normal gulch]

What is this?

Rn everyone is using a 3 combinator version instead

[tacit cedar]

I'm using 4:
constant -> arithmetic -> arithmetic -> decider

[normal gulch]

Yeah I didn’t count the constant combinator

[tacit cedar]

ok I was wondering if that was it

[steep cloak]

each % number of inputs hooked to itself

[normal gulch]

whats the initial input to the combinator?

oh I think I figured it out; ty

[tacit cedar]

I can't figure it out lol

[normal gulch]

this is for 6 science

err

0eNqdU9FqwzAM/BfB3tyyuEua+VfKCE6iroLEDo5SVkr+fXJCuzC6dd2LQbZ0pzvJZyibAbtAjsGcgSrvejC7M/T07mwT7/jUIRggxhYUONvGyAbiQ4tM1arybUnOsg8wKiBX4weYZFR3MSIXW8e3EfT4pgAdExPOLU3BqXBDW2IQil+BFHS+l1rvIr/gbTd6nSo4gVmlerNOhUnqOPimKPFgjyRFkrmnhjH84MGRAg9yc2WeM1a1Z5jwhmhjstQw0zisYit9xEriEbBeaiKJtGRSqAbiKUykdowufpOt78zghvDNRfbzJLumMLczc94w4Qu5kOearr3vKfRc/NkXtNUhGtNjhCkucwKTKfAdBjt3AU9S6Qfuhoexx3/7qxfeXvxWoB8ZjizotNFm8YkUHGV9ZnPz5GX7qrdZluZZno/jJ9vrKn4=

heres the raw string

[tacit cedar]

what are the belts set to?

[tacit cedar]

ty

[tacit cedar]

https://www.reddit.com/r/factorio/comments/10cs35c/no_long_inserter_challenge_four_ingredients_one/

[hardy fable]

wtf, why the stack inserters

[tacit cedar]

deadlock resistance, apparently

[hardy fable]

that's interesting, I'll try that out

but it's literally the same as putting a yellow belt right before the sorter

oh, maybe not

[foggy junco]

how does that prevent deadlocks though? if the belt is filled up with one resource because of a lack of the other resources, it won't fix itself

i do not see any sort of flow limiting mechanism

[plain hare]

Yeah I had a good chuckle reading that post

[hardy fable]

yeah it would work if the limiting is done after the priority merging, which is exactly the idea of rate limited sushi

[tacit cedar]

what on earth are "Sushi Rails"?

[normal gulch]

#bot-stuff message

sushi rails are basically rails that have roboports so they can expand with just you placing blueprints from the sky

and mk3 had some support for addons like mining, some intermediates but the mk4 which hasnt been tested is SO much better

[plain hare]

https://cdn.discordapp.com/attachments/139677590393716737/1050516462848192532/image.png

[normal gulch]

it has support for more intermediates, has turning on / off stuff going onto the sushi belt.. adjusting the rates of items that go on the belt so it doesnt flood it with rails or something

so many combinators that do the magic

[plain hare]

The reason you use sushi belt is so that each track section can have its own roboport network and still have access to central logistics

This lets you do things like turn your train tracks and not worry about bots flying across an enormous abyss without roboports

[normal gulch]

"mall on wheels"

its also really fast, I dont have the numbers but it can be less than a minute per block ( 96x96 tiles ) iirc

[plain hare]

There’s a little holy trinity of logistics thing going on I find very satisfying. A belt of robots (and stuff for robots to use!) to build train tracks.

It’s also a way to purge bugs with crazy speed. Automatically expanding artillery stations.

Back when We tested sushi castle mk1 I was astounded by how many resources it can consume

Just 1 little belt full of construction materials… if you eat it full speed you can easily have costs of 20+ blue belts of iron

https://cdn.discordapp.com/attachments/139677590393716737/888416365977948261/unknown.png

Belt goes out full of shit… broken rocks, trees, and airlifted fish come back…

[tacit cedar]

I wonder how effective it would be with higher tier belts from mods

[plain hare]

Probably even better assuming inserters don’t have trouble missing items

That being said a blue belt (if full) is 162,000 items per hour

So in the context of construction… yeah that does a lot of work at blue belt speed already

[normal gulch]

Without modules or beacons.. those will greatly inflate the cost 📈

[tacit cedar]

is this the general way to loop things?
using codegreen's rate limiter as reference

[plain hare]

Looks right

[normal gulch]

You’d want a chest buffer in most cases where you change the amount if items that get put on the belt but besides that good

[foggy junco]

especially want that buffer chest so it doesn't break when you change flow rates and stuff

[half phoenix]

https://www.reddit.com/r/factorio/comments/10fldya/i_made_a_mod_that_adds_a_recipe_that_requires_1/

y'all got this

[hardy fable]

guys I did it

I managed to make my fluid filter so that its functioning does not depend on build order anymore

it just works

you still have to fill the tanks to set the filter tho, so it's not quite 100% blueprintable yet

I mean at least you can ask bots to build it without having to deconstruct anything later, that's already nice

[normal gulch]

That’s great! Can i get a blueprint of that?

[hardy fable]

sure, I'll post it in #bot-stuff so it doesn't clutter the channel too much

actually the real blueprint should have all the horizontal pumps reversed

so that you don't set the same filter for all tanks

[mystic heath]

that's really cool. I was wondering how viable that was. Can you then use a single pipe for your sushi and have drain -> switch -> fill timed loop?

[hardy fable]

I'm not fully sure to get what you mean but I can tell you that it allows me to run a refinery without any downtime using only 2 pipes

[mystic heath]

that's pretty much exactly what I mean :) thanks

[hardy fable]

ah yeah I get it now, you mean switching between filling a pipe and draining it

yeah that's exactly what I'm doing on the input

it's quite messy

[foggy junco]

Got any fixes for the 0.0 liquid remaining in pipes, clogging them of you sushi output from advanced refineries?

[hardy fable]

yes, that's like the first thing I discovered

to empty pipes the rule is to have no loop and no storage tank

and IIRC one refinery on the sushi pipe is ok but multiple ones doesn't work

the successful usage I found for this are

1. universal fluid unloading train station
2. using any available fluid for flamethrower turrets

[hardy fable]

so many belts, so many items 😵‍💫

all of this to save 150kW while there are 170MW of beacons sitting right next to it

maximum efficiency

[mystic heath]

every little bit counts. some day you might need that 150kW 😄

[tacit cedar]

fish

ゆ

[sullen gale]

I think I did it. My first sushi belt.

@plain hare Thanks for your help!

[plain hare]

Tip: higher tier belts are cheaper here. You have direct control over the buffer and you could halve everything with a red belt, the reduction in buffer of expensive product probably pays for the belts

[sullen gale]

I'm not sure I understand. All that the higher tier belts do is move items faster; I'm not maxing out the "belt chest".

I guess it would be more responsive though; faster to respond to setting the requester chest if the items are withdrawn/added faster...

[plain hare]

If you move items faster you can put less items on top of belts to get them to the destination in time!

[sullen gale]

ok, after some thought, you're right 🙂

I solve this kind of crap all day for work monday through friday, so OF COURSE I would get up saturday morning, sit down at my PC with a cup of coffee, and solve the same kinds of problems, but this time for free lol

Ok so i've never done a sushi belt before and this is kinda neat, I might have just dove down a rabbit hole havent i

i have my like 8th mechanical keyboard, or at least the supplies to build it, next to me as well, i might have a problem lol

[plain hare]

Sushi supreme

On a different note. Did you combine both memory cell and active measurement?

[sullen gale]

I've been doing a funky thing with combinators for things I can't build myself that either need to go into space (for space assemblers to use) or for direct insertion into assemblers on nauvis, where I have a combinator of the ingredients for one, and a combinator for the AMOUNT that I want made, and this setup requests all of the required ingredients. I might start doing a giant sushi belt with this kind of thing, defining how many I want of each item, lol.

[plain hare]

I like that

I’m just over here fudging it with intermediate counts by the thousands xD

[sullen gale]

Yes? I have a masters in computer engineering, I should probably have a better answer for you ><. I have a combinator adding 0 to everything and then outputting that, i think that thats the memory cell. And then I measure everything on the belt, and add/subtract everything that inserters touch.

[sullen gale]

I can just set that combinator for the space science to "2000" and then it auto loads my rocket with stuff to build 2k space science. It's awesome.

[plain hare]

If you measure everything on the belt already… do you need to also remember it in a memory cell?

[sullen gale]

well now, that's an excellent question, lol.

[plain hare]

I’m just remarking you have 2 viable control system on the same belt

You might notice if you leave it long enough the memory cell might have an oopsie. It was my first method of sushi belt making, but I mostly don’t use it anymore. It’s very vulnerable to disturbances making the count wrong.. low power, mistake while editing, player presses f, that kind of thing

[sullen gale]

I am using the results to set whats in the requester chest. I suppose I could disregard items that are being handled by inserters or actively in the machines (which would result from disconnecting the inserters), which would result in a bit more being requested and a bit more on the belt. Hmm. Will think on this.

Would you recommend only counting the contents of the "belt chest"? And not dealing with the inserters?

[plain hare]

Personally yes it’s more reliable which I think is more important than the extreme accuracy of memory cell (when it works)

[sullen gale]

That seems reasonable. It would account for accidental player pressing F and such.

[plain hare]

Memory cell has the advantage of working with splitters and undergrounds well. Active sushi can’t wire those to read them

[sullen gale]

So what I'm hearing you say is: I either measure whats IN the "chest", or I measure inputs and outputs (effectively measuring contents), but right now I'm measuring BOTH, and although I'm properly combining those measurements, it is more prone to error because it has more components and therefore more that can go wrong. I should pick one solution and stick with that, rather than two melded together.

[plain hare]

But there are other sushi methods if splitters and underground usage are important I would suggest first… usually

[sullen gale]

With this, you mean "measuring the inputs/outputs via inserters" and "not measuring the active contents of the belt", correct? Just to be sure I'm following.

[plain hare]

Yes, but more specifically I think the memory cell itself is the only thing super prone to error here

Okay so high level sushi belt analysis rn is there are basically 2 main variations in sushi control… and under each of those many variants. Both memory cell and active measurement are tracking exact count of items on the belt. The other big way you can do it is measuring rates on the belt

I have only 100 iron on this loop vs I have 1 iron per second on this loop

[sullen gale]

So to go from my "memory cell (measure inserter input/output)" plus "reading the whole belt" thing to just a "reading the whole belt" thing, I should remove the wiring from the inserters that insert to the assemblers as it doesn't matter what's pulled off the belt, as only the total measurement of the belt's contents matter. I would want to leave the conditions on the OUTPUT inserters (that place sub-components back onto the sushi belt) and keep those limited to stuff like "only insert if rocket fuel tank < 10" so that I don't overload the belt. I think.

[sullen gale]

well that's an interesting approach. I think I'll worry about that later though, lol

[plain hare]

Perfect

[sullen gale]

Thank you. This has all been extremely helpful. I was going to try to get to a moon in SE to get cryonite to try and unlock logistics today, but I may instead just mess with sushi belts all day lol

[plain hare]

Hell yeah

I need to setup a cryo outpost today too

What are the odds