Signal bug when placed too close from rail junction

[JasonC]

In this train station, trains arrive and depart on the tracks at the top left. I dropped a cargo car blocking the departure track as a test and it left the signals in the following state:



Signal X is red from the dropped car. Signals A1/S1 and A2/S2 are red because of the trains in the station. Signal E is a chain signal and is blue. This is as expected.

Dropping the car turned signals D1 and D5 red, but left D2-4 green. This is what I am confused about.

Why are D2, D3, and D4 not also red? They are outbound chain signals, shouldn't they be synchronized with X?

I cannot figure out what makes lanes 1 and 5 special. It's not the presence of trains (every combination of red/green and train/no train is shown here, there is no correlation). I can't figure out what is going on. The stations are all identical except maybe 5 (since it's at the end), but signals D1 and D5 are behaving differently than D2, D3, D4 when I expected them to behave identically.

The train in lane 2 is waiting at the signal. The trains are behaving correctly, but the signal S2 still displays green and I'm trying to understand why.

Here is the mouse over for both trains showing their paths:

[sillyfly]

This looks like a bug to me...

[Zeblote]

They might be too close to the curve? If you mouse over the chain signals, do they cover both rails?

[vortex1967]

I have just done a test of this setup but only 3 stations wide. I had the opposite happen, I had D1 green and D2/D3 red.
All relevant signals covered both rails. It didn't matter what order I placed the signals, the result was the same.
I removed S1 and D1 turned red. I moved the signals right to the next rail piece and they indicated correctly as expected removing and replacing cargo wagon.
I have noticed before as I've played Factorio that signals don't always operate as expected when placed so close to the junction of a curved rail and straight rail.
You could do a quick test by moving D3/D4 and S3/S4 to the next rail piece to the right.

[Kelderek]

They might be too close to the curve? If you mouse over the chain signals, do they cover both rails?

I suspect this is the issue. Signals D and S should probably be shifted one square to the right so that they are only touching straight rail, likewise signals E and X might need to be moved 1 square north also.

You should also backtrack and double check to make sure there are no signals on the right side of your entrance rail line (where E is) going back beyond what we see in this image. If there was a rail signal somewhere back there, then that would explain why the signal at D2 shows green, the train there might think it can go back the way it came. But is more likely to be the first case - your signals probably need to be one square to the right to avoid touching the curved rail.

[vortex1967]

This might help if you want to keep the same design for your stations.
On the left you can see that the 2nd track is now free for a train to enter.
Despite the position of the cargo wagon a train can still enter the 4th track.

Chain signals are used extensively to prevent trains entering the intersection if the exit isn't free.
3 track spacing is used on the horizontal tracks to allow for placement of some of the signals.
On the right can be use in a blueprint to extend as many stations as required.
It just needs the top two signals removed.

[vanatteveldt]

@vortex: Isn't it problematic that the entry and exit signals are not on the same location? I thought that if you want 2-way track you need to have both signals directly opposite of each other (why) doesn't your placement make the station track inaccessible?

[JasonC]

This might help if you want to keep the same design for your stations.
On the left you can see that the 2nd track is now free for a train to enter.
Despite the position of the cargo wagon a train can still enter the 4th track.

THANKS. This finally makes sense. You have no idea how long I struggled with exactly that issue. Today is going to be spent rebuilding that station yet again .

The only other addition I'd make to that design is another outbound chain signal right exactly where the train is, or keep the stops at least two train lengths away, so that if a train stops at one of those signals it doesn't stop halfway out of a station and confuse the inserters there. I spent a good amount of time flushing solid fuel off the iron ore belts of my steel mill because of that .

@vortex: Isn't it problematic that the entry and exit signals are not on the same location? I thought that if you want 2-way track you need to have both signals directly opposite of each other (why) doesn't your placement make the station track inaccessible?

It's only problematic on two-way track. Those little curved sections I think you are referring to, where the signals are, are one-way. E.g. a train won't (and can't because of the one-sided signal) travel inbound on the upper curve (with the chain signal), and won't/can't travel outbound on the lower part (with the regular signal).

I'll try some experiments with the station from the original post. I haven't been responding here because I redesigned that station (like 8000 times) and forgot to save the old one. I'll recreate it and continue these tests. Thanks for all your replies so far.

[vortex1967]

It's not problematic in that setup because there are 2 tracks leading to a single track. One track as an entry and one as an exit.

Looking at setup 1 & 2 in the image below. (1 is left track traveling and 2 is right track traveling)
Referring to both, the train on the left wont proceed onto the middle track because it's exit is blocked by the cargo wagon.(a plus of chain signals)
However the train on the right can proceed because it's exit is free.

Setup 3 is a double header setup that requires signals to be placed opposite each other.
The train at bottom left is waiting at signal because I have it set to travel to train stop at top right which is currently occupied.
The train at top right is set to travel to train stop at top left.
When train at top right has left and arrived at it's destination at top left, train at bottom left can now proceed.
They are set to travel back and forth. Only double header trains can use this setup.

Hope I haven't confused things.

[vanatteveldt]

Thanks for your explanations!

I understand the part about the segments, that makes sense.

What I still don't understand is (e.g.) D2 is green in the original post?

[vortex1967]

The cause of the unexpected indication is that the signals are placed too close to the junction of where the curved rail meets the horizontal rail.
Visually the signals might indicate that they are in the correct position by showing the yellow arrows just before placement but when placed
in world they are in an ever so slightly different position that the signals are in fact only reading the horizontal track and unable to see
along the curved rail piece to the signal behind the cargo wagon.

I experimented with the same setup, I also had false indications. But I had D1 green and D2-3 red
Moving D1-5 and S1-5 more to the right solved the unexpected and confusing indications.

Hope that explanation has helped.

[vanatteveldt]

Sorry if I'm being daft, but you're saying that the signal is essentially placed behind the junction, and only controls the straight part, right?

In that case, why doesn't the second train drive until the signal just before the cargo wagon? or is it simply still (un)loading and could it drive there if it wanted to?

[vortex1967]

After a lot more testing it appears to be a graphical glitch when signals are placed so close to the curved rail straight rail junction and there is a normal signal opposite as indicated in the image below.
To a train the signals are red and will stop. The issue seems to only manifest itself when signals are so close to a junction.
Most times I tried, I could change red indicating signals to green by removing and replacing the curved rail next to the chain signal.
Remove and replace cargo wagon would reset some not all chain signals but not all the time.
Remove signal opposite and the chain signal indicates correctly.

In one of my previous posts I mentioned "I removed S1 and D1 turned red" and I didn't pickup on the significance of that comment or that action at the time.
Signals are a mine field of head scratching when they don't work as expected.

Hope this has sorted confusion all round.

[vanatteveldt]

Should this be posted as a bug report?

[Loewchen]

Simpler depiction:

[JasonC]

Should this be posted as a bug report?

I think we're all too lazy to actually do it.

[sillyfly]

It seems to me it's related to this bug, perhaps even exactly the same, so presumably it is waiting for 0.13. I'll leave a link there to this thread.

[Koub]

I'll say duplicate of 21464, and resolved for the next major release (0.13)
[Koub] Moved to Duplicates

[sillyfly]

I'll say duplicate of 21464, and resolved for the next major release (0.13)
[Koub] Moved to Duplicates

As I understand it the resolution is pending 0.13, meaning it is not yet resolved, but they will work on it for 0.13. Not that it matters all that much