Page 1 of 1
[1.0.0] Combinator behavior inconsistent below 50% power
Posted: Mon Nov 09, 2020 11:13 pm
by Theikkru
It took me forever to diagnose this, but it appears that combinators intermittently fail to respond to pulse signals when running below 50% power satisfaction. This is inconsistent with other devices like inserters, which operate slower under low power, but still generate single-tick pulse signals reliably all the way down until 0% power. The result is that a factory that uses pulse signal logic can suffer from all sorts of strange, unpredictable, and very hard to diagnose problems if even an intermittent dip in power occurs.
I could understand if all such logic became equally more sluggish as the power grinds down, but having combinators alone react like this is like having inserters drop items on the ground halfway through their swing.
A demo save is attached. To start, drop a filter inserter in front of one of the 4 test inserters. You can play with the ratio of solar panels to radars to see how the behavior changes at different power levels. The bottom-leftmost combinator tracks the running total of the difference between the full-power combinator and the low-power combinator.
Re: [1.0.0] Combinator behavior inconsistent below 50% power
Posted: Tue Nov 10, 2020 12:55 am
by valneq
I am not sure how combinators are supposed to work during a brownout. Other people were confused by this as well, see
viewtopic.php?f=23&t=90631
So this could be "not a bug", but I am not sure.
Re: [1.0.0] Combinator behavior inconsistent below 50% power
Posted: Tue Nov 10, 2020 12:56 am
by Rseding91
Combinators specifically just re-send their previous signals when low on power. That's by design.
Re: [1.0.0] Combinator behavior inconsistent below 50% power
Posted: Tue Nov 10, 2020 1:02 am
by valneq
… could be worthwhile to explicitly mention this on the wiki. I was just searching and did not see anything in that regard.
Re: [1.0.0] Combinator behavior inconsistent below 50% power
Posted: Tue Nov 10, 2020 3:18 am
by Rseding91
valneq wrote: ↑Tue Nov 10, 2020 1:02 am
… could be worthwhile to explicitly mention this on the wiki. I was just searching and did not see anything in that regard.
The wiki can be edited by anyone. You just have to request an account. Or you could post here:
viewforum.php?f=50
Re: [1.0.0] Combinator behavior inconsistent below 50% power
Posted: Tue Nov 10, 2020 4:29 am
by boskid
To be more specific about why it behaves like that:
Combinators (arithmetic+decider) are defined as consuming 1kW (wiki to fix as it states 2kW) which means they require 16.667 energy to update in a given tick. If the combinator has less energy than that it will not consume energy from its energy buffer and will not update leaving output signals at previous values.
Internal energy buffer inside of combinators has capacity of 34.3334 energy (2.06 * 16.667). If electric network is at 48.54% or above, electric network will deliver enough energy to keep combinators running every tick: If combinator has 16.667 energy in buffer combinator will update leaving buffer with 0 energy and then buffer will get (34.3334 - 0)*0.4854 = 16.66 energy back.
If combinator gets 8.3335 energy every tick, at one tick it may have not enough energy so it will only charge, and at the next tick it may have enough energy, will update and consume 16.667 energy going back to a not enough energy. It will appear as updating once every 2 ticks.
This behavior is consistent however it could happen different combinators are not updating at different ticks. In general they may sync up because combinator with less energy will get more energy.
If well discussed i could increase combinator internal energy buffer to be 4.06 of max energy consumption per tick (67.66802 energy) so they would keep running down to 24.63% (they would still consume 16.667 energy every tick, but they would be taking more energy due to fair share logic preventing other entities from taking that energy), however i am not considering this to be important change as it reduces problem solving on the player side.
Re: [1.0.0] Combinator behavior inconsistent below 50% power
Posted: Wed Nov 18, 2020 3:47 am
by Impatient
@boskid: Sounds about right.