I don't think any questions about signals are old hat, John, I've been on this forum since the beginning, and when I finally got around to doing signals (you'll notice there aren't any in PO&N4, altho I started to TRY to add them in version 3!) there were no "experts" around to answer questions!
There still aren't any experts, most people that have gotten signals to work learned by trial and error and aren't 100% sure exactly what they did to make them work (unless they actually DO know and they're just holding out on us!), and a lot of the routes with signals on them only work "the way god intended" if the activity is written a certain way. That's been my biggest problem all along, trying to understand the incestuous relationship between the signal setup on the route and the activity editor. Long haired wizard set up some activities for PO&N4 with AI traffic with no signals at all, so it is possible.
I haven't seen MP3 yet, Bob would probably be able to answer that question better, but my experience has been;
1. The SignalNumClearAhead value in the sigcfg.dat file affects how many interlocking signals ahead of each train are activated, or "tickled awake" as I call it, so they show something other than red.
2. Interval signals might be affected by SignalNumClearAhead, but I'm not really sure how, and the above behavior still applies to intervals. I think I changed that value to 1 once just to see what it would do, and got green lights that should have been red. So I really don't know, I'll have to experiment more with it. There is a "flashing red" entry for "BNSF3LightA", which is supposedly the script for the interval, but I don't know how that works either.
3. HAL9000 (AI dispatcher) sets the switches for the paths X number of signals ahead of each train, the value of X IS affected by the SignalNumClearAhead value.
4. If the "player drivable path" box is not checked for the AI train, #3 is irrelevant, HAL just sets up the entire path start-to-finish for an AI train at the moment of spawning. My AI trains are set in the activity editor for "expected performance" (for the AI service, not the player service) of 25 to 30. This is APPARENTLY (again, nobody seems to know otherwise) the only way to control AI train speed, these values give me a more realistic 15 to 20 mph for the various 2dash937oil, gp3817mixed, 2dash940mixed, etc., consists. At those speeds the signals obviously need to be closer together, or the player will be waiting half an hour at a signal waiting for a train that's many miles, but only a few signals, away. Which leads to;
5. AI trains don't stop for any red interval signals. Interval signals are supposed to be 3 different types, but with or without the number plates and gradient tags they're permissive, "stop and proceed" is meaningless, just blow right thru them, and since the speed restriction for a red inteval is 20 mph and my AI trains are already below that they don't even slow down for a red interval.
I couldn't tell you about anything but BNSF signals since that's all I've played with (I live next to the former C&NW Galena line on the West Side of Chicago, and that's the signal type they use), and I'm not using them the way they're supposed to be used anyway. So the above diagram don't apply to the PO&N, since the AI trains always take the curve, the player always takes the straight (the default "at rest" position of the automatic switches), and the player path is only long enough to get started, and don't actually go onto the mainline at all. Once on the main, he's on his own, has to manually set the manual switches for different branches, and yield right-of-way to AI traffic, which has priority. Here's a section of the town of Sundermeyer;
AI path is purple, player path is red. the player path starts in the engine house at left, and an end point is placed on the yard lead to the right. The player main line is to the north, and in order to get there he has to cross over the AI path. The two auto switches are back-to-back left turnouts (seperated by a 10 meter straight), so altho it looks like the AI path is going straight, it actually comes in thru the curve and leaves thru the curve. After he passes, both switches return to the at-rest straight position. "HAL" won't set any switches for the player to go up there, since the player path don't go up there. So the player has to drive to manual switch 2, set it for the curve, and reverse thru it. Check the signal, if either auto switch is set curved instead of straight one of the signals will be red. If neither is red, he drives thru.
Here's the setup for the crossover, the "dwarf" signals are standard interval signals with the poles buried up to the heads. This avoids confusion about which red signals are OK to run, and also relieves the visual effect of having so many signals so close together that it looks like a forest of signal poles. On the right is a standard BNSF interlocking with the second head enabled and linked to the curved, or "diverge path", of the far auto switch. Just ahead of the engine (sitting on manual switch #2) is a kitbashed abomination made from a BNSF gantry-mount interval placed on the pole of the distant searchlight signal from Hank's NA switchstand set. The distant signal is linked to the straight track on the far auto switch, so it will show red if that's curved. The interval signal will show red when the near switch is set against it, it shows yellow if manual switch #1 is set against it, and green if manual switch #1 is set for the curve. That's the purpose of the two "dwarf" signals between the two auto switches, to prevent this signal and the one on the opposite end from indicating red if the far manual switch is set against it.
Here's the problem with answering "how do I set up signals" questions. The distant signal was made to return green for the exit it's linked to, and yellow for the switch thrown to the unlinked exit, since it's intended as an advisory signal. I hacked Hank's file to make it return green or red instead of green or yellow, since with my particular peculiar setup, any auto switch thrown for the curve is a stop order for the player this is showing that the far auto switch is set curved, therefore has to be considered an absolute stop by the player. The upper signal hasn't changed yet, so this is an opposing meet, the switch window shows the near auto switch is still straight.
The advancing path of the AI train now sets the near switch curved (switch window), so the upper interval goes red, since it can no longer trace a path to another signal. If the spacing of the signals beyond these are set up just right, the AI train should arrive within a few minutes or less.
After the AI train passes, both auto switches will reset to straight, which makes a clear path thru the auto switches for the player until the next AI train comes along. By the S-curve of the grain train, you can see that the AI train is not really going straight, but taking a double-curved path.
None of this works perfectly, but it does work. Unfortunately you're asking the wrong guy about the traditional way to set up signals, since my setup don't even come close to traditional. Normally you would use a double head interlocking to guard the entrance to the doubletrack sections, place intervals on the single track sections, and use all the "action event triggers" and "begin passing path" stuff in the activity editor to make the AI trains and the player follow a rigid schedule, the player also has to follow the path exactly the way it's layed out to make the interlockings work for him. I'm not sure if anyone has found a foolproof way to set up which train gets priority at a meet. In mine it's simple, the AI traffic always has priority, if the player is blocking an AI train and it has to stop, he'd better get out of the way and let it pass so they don't start stacking up behind each other and make a parade! The way mine are set up, I can't even place intervals on a single track section, or the player gets a green light to a cornfield meet. So no matter how long it is, every single track section between passing sidings is a single block. Not exactly a class 1 railroad!
Printer-friendly page | Top
In figure 1, I place a signal. The switch to the right of the signal set against it is an obstacle, so since it can't trace a clear path to another signal to the right, it indicates red.In figure 2, I place another signal to the right of the first one, which shows red since it can't trace a clear path to another signal to the right. The first signal on the far left now sees another signal, checks it's state which is red, and returns steady yellow.In figure 3 I place another signal between the first two and the offending switch-set-against-me, all the above applies, third signal red cuz no clear path to a signal to the right, second signal goes from red to yellow, first signal now flashing yellow.Figure 4 fixes it, at the cost of having a ridiculous number of interval signals. The fourth and last signal on the far right is red, you already know why. The third signal is now steady yellow cuz it sees another signal to it's right, the second sees the third is yellow and shows flashing yellow, the first one on the far left now sees a flashing yellow, and since that isn't steady yellow, red, or absent, that fulfills all the qualifications to return green. Halleleuiah. Somebody say I BELIEVE!I'm still experimenting with them, but it appears that one way around using enough signals to make it look like a runway for Close Encounters spaceships is to use the North American switchstand set, since the switchstands act like "advisory" signals and change the sequencing of the intervals, so it don't automatically return red for a switch set against it when that switch is linked to a switchstand signal.