Factorio Forums

I went into a deep dive into modules recently, and was surprised to find them more complicated than I initially thought. For example, this topic on how adding a speed3 module to an assembler with 3x prod3 modules in it is actually more energy efficient than adding a eff3 module. Wild stuff.

One thing I came across that was repeated again and again was that quality and speed are "antagonistic" due to speed modules reducing quality%, and quality modules reducing speed% (though at a flat rate compared to their level).

But I wondered; what if the speed bonuses which increase production throughput could outperform the significant quality loss, netting more quality items over a given time period than just quality modules alone.

So I whipped up a calculator in Excel and turns out it is better to have one speed module for every four quality modules!
At level 2 modules having 4 quality nets you 6.4% "net" quality = 6.4 quality items per unit of time that the assembler makes 100 units of output per unit of time.
Compare this to 4 quality and 1 speed (still level2) and your output is 7.15% "net" quality = 7.15 quality items per 100 units (per unit of time)!
The math:

• total output = production * speed%. if we assume 1 units produced per unit of time, for example a 30% speed increase is 1*30% = 1*1.3 =1.30 total output

• total output of quality items = total output * quality%

• for level 2 modules: speed2 = 30% speed, -1.5% quality. quality2 = -5% speed, +2% quality

• if we have 4x quality2 modules = -20% speed, +8% quality which is also the totals without a speed module.

• total output = 1 * (1-0.20) = 0.8 units

• total output of quality items = 0.8 * 0.08 = 0.064

• totals with 1x speed module = 30-20=+10% and total quality = 8-1.5 = 6.5%

• total output = 1 * 1.1 = 1.1 units

• total output of quality items = 1.1 * 0.065 = 0.0715

So you get a whole 0.0075 more quality items per unit just by adding one extra speed module!

Now the smart ones among you are going to say "hey wait a minute! If you use a beacon to get that single extra speed module, the beacon will multiply it by 1.5 due to its "distribution %" so its actually going to be 1.5 speed modules for every 4 quality modules!" and you would be right, but the beacon distribution% is only based on how many beacons are touching the assembler. As in, you can have two beacons touching your one assembler with only a single module in one of them you can get 1.5*sqrt(2) = 1.06066 "total" modules added!

In actuality, running 1.5 extra speed modules (aka one beacon with one speed module in it) is usually the better choice than the math I showed above. But I wanted to make that tangent to show off that weird niche thing I learned about beacons

If the above math is right (which seems to be the case given my testing in-game) then here are even more calculations, with the added bonus of all the levels of quality of modules (boy was that a confusing sentence). These numbers are assuming a base production of 100 units per unit of time, and the values are how many quality units per unit of time are created (they also work out to be the "net quality%" and I think they are easier to read than numbers with more zeros in front but if the crowd disagrees I can upload the "real" numbers not the "actual numbers *100") As you can see, at certain module levels (assuming all modules are the same levels as each other because if I added *that* then the matrix would be massive), it is usually better to run one speed module in a beacon next to a 4x quality assembler.

Finding this this got me thinking; what is the actual best ratio between speed and quality modules?
Turns out its not straightforward as one would think. Adding quality modules changes the whole game of course, though the 1.5speed module seems to be the right choice if you aren't using level 1 modules (which lets be honest here, you probably won't be if you are even considering adding beacons to quality builds haha)

You might also be wondering why I included a "level 3*" column: well that's because while doing some calculations I found that because:

• speed1 reduces by 1%, and speed2 by 1.5%, and speed3 by a whopping 2.5%.

• if the quality level matches, a 1x speed1 + 1x speed2 = a 1x speed3 in both speed and quality % changes.

This also means that in terms of the ratio between speed and quality loss should be speed1=speed2, but speed3 should be worse by about 33%. In other words, if you are to use speed only use speed1 or speed2.

The actual effect is that using a speed2 gets you roughly the same, if not equal quality items per unit of time than a speed3 as can be seen above. but that is if you are using a single speed module......
yes I re-checked my math at this point because man does it not make sense. Its most obvious with the "total" 3 speed modules in the center where at default quality it is better to use speed2, but after rare quality it flips! As far as I can tell, this is due to how quality2 modules scale with quality, and why quality2 modules are very powerful (and don't take superconductors!).

I played around with production modules with quality, but it looks like a single speed is the way to go. Adding prod modules means removing a quality module so it really hurts the %. This is of course only if you are trying to maximize quality items/sec. if you are trying to increase throughput or reduce how much input items get consumed, then use the calculator as I've included production modules as well.
I also ignored power usage, efficiency modules, and pollution because, well, I just didn't want to do the math. If you want to do it, be my guest!



Here is the excel sheet in case anyone wants to both test my math and play "download random thing from the internet roulette" I might convert it to a google doc eventually so the wider audience only has to do the former.

TLDR: A single speed module in a beacon is actually *not* antagonistic to quality, and in fact help your quality output per unit of time! I will be updating the wiki as soon as I have an account there
Also beacons don't work like you think they do when you don't use them as intended

Aaaand that's my first post on here! Hopefully my math isn't wildly wrong and my username is forever tainted as "the un-calculable one". Hope you enjoyed reading, now get back to making your factory bigger


P.S.:
I used the following shortcut types during the above text in case you got lost:
prod = production
qual = quality
1x speed3 = one (1) level 3 speed module
4x eff1 = four (4) level 1 efficiency modules

TauPlays wrote: Fri Apr 17, 2026 12:53 am But I wondered; what if the speed bonuses which increase production throughput could outperform the significant quality loss, netting more quality items over a given time period than just quality modules alone.

Intuitively, yes, i agree and understand, one can picture it as tossing a coin for tails. If you toss faster, you have more tails per hour, but unfortunatly, tossing faster also causes some misses, when the coin is still rolling it doesn't count as tail. Less than 50% of tosses will result in a tails. And if you throw super super fast, then the coin never has enough time to land, you're juggling at this point, there is no more tails.

A bit the same with "quality result", if you increase speed , you have a diminishing yield on your input, but more output per module, until you increase "too much" and you don't get any result at all.

TauPlays wrote: Fri Apr 17, 2026 12:53 am This also means that in terms of the ratio between speed and quality loss should be speed1=speed2, but speed3 should be worse by about 33%. In other words, if you are to use speed only use speed1 or speed2.

Consider the following setup :
Crafting speed x % quality = quality result per unit of time

Left : 1.6875 * 5 = 8.4375
Center : 1.4375 * 6 = 8.625
Right : 1.3125 * 6.5 = 8.53125

Here the result for what is generating the most quality output per unit of time is Speed 2 > Speed 1 > Speed 3

1 and 2 do not have the the same result, even if they "should" ?

It sound like the intermediate conclusion i quoted may be valid in the particular case where the ratio of module is the one you calculated as best, "4 quality and 1 speed " , which requires at least a beacon. But if you consider what players may do in their mall, then the conclusion is a little different. If players wants to have "more quality fluid tanks" from their single assembly and they are considering modules options, they shouldn't upgrade their speed module too a tier 3, but a tier 2 is preferable over a tier 1.

Or am i missing something there ?

mmmPI wrote: Fri Apr 17, 2026 5:17 am If you toss faster, you have more tails per hour, but unfortunatly, tossing faster also causes some misses, when the coin is still rolling it doesn't count as tail.

That is a great way of articulating it!

mmmPI wrote: Fri Apr 17, 2026 5:17 am Or am i missing something there ?

You aren't missing anything, I just did not articulate my thought process very well.

What I was trying to communicate was that without a calculator to actually do the somewhat complicated math or by testing it in-game, it appears that speed1=speed2>>speed3 but in actuality that is not always the case like you have shown. In other words, its not intuitive on what is the "best" answer to maximize quality items per second

my calculator does match up with your finding that in the "3xqual3 1xspeed#" case, a speed2 is the way to go and Speed2 > Speed1 > Speed3 Given that, I would like to re-make my calculator to include all the variables for module levels (maybe just speed modules to start off).

Thanks for the response!

TauPlays wrote: Fri Apr 17, 2026 12:53 am At level 2 modules having 4 quality nets you 6.4% "net" quality = 6.4 quality items per unit of time that the assembler makes 100 units of output.
Compare this to 4 quality and 1 speed (still level2) and your output is 7.15% "net" quality = 7.15 quality items per 100 units!

You might want to clarify that you mean "per 100 units of time" here and are not talking about resource efficiency.

mmmPI wrote: Fri Apr 17, 2026 5:17 am [Speed module] 1 and 2 do not have the the same result, even if they "should" ?

If they had the same result, they would essentially be the same item. What OP meant is that the ratio of speed gain vs. quality loss is the same for both modules; preserving that ratio might be important, if you're targeting something specific. Since beacons scale all effects of modules (not just the good ones), swapping speed 1 and 2 in beacons can help reaching the desired effect.

Nidan wrote: Fri Apr 17, 2026 12:19 pm You might want to clarify that you mean "per 100 units of time" here and are not talking about resource efficiency.

Great point, I will update that.

The calculations I did were all in the effort to maximize how many quality units are created per unit of time.
In other words, if you only care about maximizing how many quality items you get, what is the best module combination to get that outcome? This is most relevant when your inputs are near infinite and you don't care about total item throughput (or maximum output per unit of time) or productivity. The side thing I should have mentioned in the main topic is that adding speed modules to a quality module build not only increases your quality items per unit of time, it also increases your normal items per unit of time (at module levels 2-3 at least since speed1 will usually break even on the +-speed%).

TauPlays wrote: Fri Apr 17, 2026 12:08 pm my calculator does match up with your finding that in the "3xqual3 1xspeed#" case, a speed2 is the way to go and Speed2 > Speed1 > Speed3

That's a good sign i think

I tried to follow your explanation, i only made the setup to verify mid way something because it appeared counter intuitive to me what you said and i wanted to be sure i was following the reasonning. I was surprised by the findings x).

Adding the beacon is an extra layer of dfficulty, i think i understand better the charts and your approach to narrow down the math to a useful result and the choice of illustration now.

I try to avoid using calculator when playing , i like to toy with them outside a game to learn rules of thumb i can then apply when playing, that's a new one for me so thank you for the calculator, made me learn a thing

Your charts also shows the total item produced, not only the quality ones, to me in game from the information available from the tooltips i would math it this way :

Crafting speed * % quality = number of quality output per unit of time // item produced in 100 unit of time

Left : 1.6875 * 5 = 8.4375 // 168.875
Center : 1.4375 * 6 = 8.625 // 143.375
Right : 1.3125 * 6.5 = 8.53125 // 131.125

That's just 100 times the crating speed, and you can subtract the number of quality item to have the number of non quality item , "in the same amount of time".

I think then comes the recyclers, the setup with a speed module 3 produce more "non quality item" per unit of time, if you recycle them until the ingredients are reused and made into quality output or voided, the ratio may be different, it may require another calculator to find out how in game it would behave if you automate a setup that include those recyclers, "for the sake of having run without clogging", in early game you would maybe keep the quality things like solar pannel and accumulator for a space platform, and use the non-quality ones. But later in game or for other items where you maybe tempted to achieve the "most quality output per hour" , it may matter that you have the possibility to recycle "more failed attempt".

For the coin toss analogy, it would be looking at pictures when the coin was still rolling, and sometimes the picture shows a tail, so even if it missed the first time because the coin was still rolling, some of them initial throw are still couting as a tail in the end if you add that second layer of verification.

It makes "over tossing" sounds more appealing, maybe speed 3 modules aren't that bad ? I'm way too lazy too make a calculator myself though x)

Nidan wrote: Fri Apr 17, 2026 12:19 pm If they had the same result, they would essentially be the same item. What OP meant is that the ratio of speed gain vs. quality loss is the same for both modules; preserving that ratio might be important, if you're targeting something specific. Since beacons scale all effects of modules (not just the good ones), swapping speed 1 and 2 in beacons can help reaching the desired effect.

I disagree with the first part. You can imagine a infinity of modules stats that would yield the same amount of "quality output per hour", or the "same result" for that context. Consider the assembly machine speed is 1 and the quality chances are 10 %. If you make a module that put the speed to 2 and the quality to 5 %, you have in practice the same result as if you make a module that put the speed to 4 and the quality to 2.5% chance. Since speed modules have 2 parameter that matter here, speed increase and quality reduction. But those module would be different items, different tier of speed module, yet in practice, if you are not interested in recyclers, and you only look for the quality output per hour at the end, you wouldn't see any difference if you swap them in your mall.

I think the calculators helps correcting those intuitions, i don't understand why only speed 1 and 2 in your reformulation , what's the big deal in preserving the ratio, it's like 1 speed 2 has same impact as 2 speed 1, but that's different with speed 3 ? To me it appears different for 1 and 2 in some cases the same way as it's different from 2 and 3 so it's hard to make a rule of thumb there.

mmmPI wrote: Fri Apr 17, 2026 5:14 pm ...it's hard to make a rule of thumb there.

That's why I made the calculator, to find that rule of thumb. Outside of setting up all the different options in a test world and running them for a long while, then doing calculations on the results, it felt like a better use of my time to just calculate from the raw numbers.

Knowing now that the "rule of thumb" is "best is a beacon with only a single speed module" I might have to change my builds. I was honestly surprised that this was the output of the calculator

TauPlays wrote: Fri Apr 17, 2026 12:08 pm my calculator does match up with your finding that in the "3xqual3 1xspeed#" case, a speed2 is the way to go and Speed2 > Speed1 > Speed3

Though to be clear, it is worse than running just all 4 quality modules (at least for quality items per unit of time):

mmmPI wrote: Fri Apr 17, 2026 5:14 pm Your charts also shows the total item produced, not only the quality ones

Ive got two calculators in the excel sheet. One outputs both total items and quality items per unit time, the other just the quality per unit time but has all 5 possibilities of module quality and is color coded

mmmPI wrote: Fri Apr 17, 2026 5:14 pm I think then comes the recyclers...

Recyclers are certainly something where the calculations can get complex. In early game for use off of Fulgaris, the input to them might be limited, but like you said, the thing to maximize might change depending on what you are trying to accomplish.

TauPlays wrote: Fri Apr 17, 2026 6:18 pm Knowing now that the "rule of thumb" is "best is a beacon with only a single speed module" I might have to change my builds. I was honestly surprised that this was the output of the calculator

It's not always the case, consider this setup :
crafting speed * % quality chance

a) 3.375 * 6.2 = 20.925
b) 2.1875 * 8.1= 17.71875
c) 4.5625 * 4.3 = 19.61875
d 2.9125 * 5.4= 15.7275