Factorio Forums

There is a simple circuit trick that can be played with belts and that I have found useful for a number of different jobs. I've not noticed it discussed anywhere here (but doubtless will now be pointed to a large number of previous posts). I call it a 'sparse belt'.

It's just a line of several belts hooked up into a circuit network, with all except the first one set to read the belt contents (in hold mode), and the first one enabled if 'Everything' is less than a threshold value. Typically I use 4 belts and a threshold of 2, but there is nothing magical about these numbers: Copy blueprint

Its behaviour differs from a plain belt in two ways. Firstly, if the belt is backed up, then instead of there being 4 items per lane per cell sitting on the belt, there are only (say) 2 every 4 cells. For example, here is a sparse belt feeding ammunition to a line of turrets: With a plain belt, there would be 10 cartridges in each turret, plus another 16 sitting on the belt; the sparse belt reduces that to 10+2. Since the ammunition is the great majority of the cost in this case, that reduces the overall cost of the wall by about a factor of 2. (Note that the turret inserters are placed at the end of each section of sparse belt, so that the 2 waiting cartridges are ready to hand.) This was the application that led me to the idea in the first place; I've also used the same approach to supply rocket or nuclear fuel to train stations.

Secondly, if the belt is not backed up, the throughput speed is reduced by the same factor as the backed-up storage. This might not immediately sound useful, but I've recently realised that it can be used to make a supply-controlled sushi belt: Here, the sparse belts are just 2 segments long, with a threshold of 1, so that the maximum possible throughput speed for each flavour is a quarter of the usual yellow belt rate (in practice, it's more like a fifth), and all seven fit comfortably in one lane of a red belt.

I came up with what I called a "Sparse Belt" too.
My solution was to limit what was placed on the belt: The circular (square) belt has a reader, when something gets detected, it sends that to the filter inserter which drops one item on the belt.
That goes in a loop around the walls of the base, and if it returns unused it goes back into the request box.

Along the belt there are robo-port stations: The CC has what items are required, the math just multiplies the chest contents by -1.
Both go to the filter inserter, which will keep the chest full of what the bot station needs.
It also keeps the bot levels topped off.

There are also a few artillery posts they just feed the turret as it has a max of 5 shells.
I use laser turrets (you can see one on the controlling loop.
Gun turrets should also work, however you'd probably want more magazines than I have artillery shells or some turrets would get starved for bullets.

I've done this in various ways, but for the application in the OP -- an ammo belt feeding turrets along a wall -- my solution was to build a looping belt. Periodically it would be interrupted by chests, which would be the feed points, delivered by whatever means chosen -- handloading, an ammo plant, trains, whatever. That chest outputs onto the belt with a baseline inserter, and is fed by a fast inserter. Thus the belt has an average item density limited to the throughput of a single yellow inserter, which is not particularly substantial but, running continuously, quite sufficient to supply gun turrets over a fairly long stretch.

foamy wrote: Fri Jun 19, 2020 10:14 pm for the application in the OP -- an ammo belt feeding turrets along a wall -- my solution was to build a looping belt.

That's exactly what I did until I came up with the method in the OP. I think the inserter-controlled loop has several disadvantages compared with the circuit-controlled version:

1. You need a loop of belt. If you are feeding ammo to defend a small fort, there is no problem having a loop around the inside, but the main base external wall is going to be interrupted by water (or cliffs, but that's just a local problem). To complete the loop, you must either continue the belt along the inner lake shore, or double back so that there is a belt going each way along the wall. Either way is wasteful (and I also found the second alternative getting in the way of power pole placement, but that's just my wall design).
2. The belt is always moving (and is not full), which can do bad things to UPI in a large base.
3. The inserter is always running, which is a small but continuous power drain.

Khagan wrote: Sun Jun 21, 2020 10:36 pm

foamy wrote: Fri Jun 19, 2020 10:14 pm for the application in the OP -- an ammo belt feeding turrets along a wall -- my solution was to build a looping belt.

That's exactly what I did until I came up with the method in the OP. I think the inserter-controlled loop has several disadvantages compared with the circuit-controlled version:

1. You need a loop of belt. If you are feeding ammo to defend a small fort, there is no problem having a loop around the inside, but the main base external wall is going to be interrupted by water (or cliffs, but that's just a local problem). To complete the loop, you must either continue the belt along the inner lake shore, or double back so that there is a belt going each way along the wall. Either way is wasteful (and I also found the second alternative getting in the way of power pole placement, but that's just my wall design).
2. The belt is always moving (and is not full), which can do bad things to UPI in a large base.
3. The inserter is always running, which is a small but continuous power drain.

The power drain of a couple of inserters on the belt is completely dwarfed by, like, the power drain of everything else ever in your factory, including the inserters driving the turrets themselves. If necessary you back break things into multiple individually smaller loops, too.

foamy wrote: Mon Jun 22, 2020 6:47 pm The power drain of a couple of inserters on the belt is completely dwarfed by, like, the power drain of everything else ever in your factory,

Absolutely. I put that point last because it is by far the least important.

foamy wrote: Mon Jun 22, 2020 6:47 pm including the inserters driving the turrets themselves.

Depends how many turrets in the loop. The active running cost of a pair of loop inserters (one fast for belt to chest, one standard for chest to belt -- and yes, that's what I used too) is 59 kW. You would need nearly 150 turrets in your loop segment for that to be matched by the passive drain of their inserters, and I've never had more than about 50 without a break. (Hmm. 59 kW is probably an overestimate, since the fast inserter won't be running nonstop, but not by more than a factor of 2.)

foamy wrote: Mon Jun 22, 2020 6:47 pm If necessary you back break things into multiple individually smaller loops, too.

Sure -- been there, done that . But whether you use a few big loops or many small ones, you still have to cover the entire frontage in both directions. You can only avoid that if you confine yourself to a single huge loop.

I was under the impression that huge circuit networks of hundreds of belts was a ups drain.

5thHorseman wrote: Mon Jun 22, 2020 9:24 pm I was under the impression that huge circuit networks of hundreds of belts was a ups drain.

All I can say is that with the continually moving loop I _did_ notice a UPS slowdown, but with the circuit activated belt, on a considerably larger scale, I don't.