As long as your outputs are all open collectoor, you can do a simple wired-OR. Otherwise you need to add an additional logic layer.
As valiant an effort this is to build a system that works with no modifactions to the rolling stock, it is FAR easier to do it the prototype way and just detect that there is something in the block, ANYWHERE in the block, instead of counting objects entering and comparing it to the count of objects leaving. There is a place for this, but it's just not going to be suitable for more complex track arrangements. A combination system may enable complete trains of unmodified rolling stock to otherwise be detected , but for mainline trains, doing what Sheldon does and just have the engine and caboose detected is sufficient. For switching moves this may not work in the current sense system because there are plenty of photos to prove that there are many times when you might have the caboose ahead of the engine and 3 cars trailing behind, or the order being engine-caboose-rest of train.
--Randy
Modeling the Reading Railroad in the 1950's
Visit my web site at www.readingeastpenn.com for construction updates, DCC Info, and more.
Just a little insight on how prototype signaling fits into this. Generally, manual turnouts in CTC controlled blocks put the block into apsolute stop at both ends when thrown to the diverging route.
So a local freight switching a siding on the main locks out other traffic simply by throwing the turnout, even if he leaves the mainline completely. Or, in the case of our models, leaves undetected cars on the main.
That is how my few mainline industries are set up. So the block stays red no matter where the caboose and loco are. They are either on the main, making it red, or the turnout is not set to the main, making it red. Or, the whole train is in the siding and the dispatcher requests they clear the main for other traffic.
Sheldon
It seems to me that the key to making this work is to have a "total" for each block and all the detectors on the inbound side add to that total and all the detectors on the outbound side subtract from the total. No matter how many inputs there are.
On block 4 there are 4 detectors on entrances to the blocks so the total in the block is the sum of all the input detectors, regardless of which one was the input, less the outbound on which ever detector measured outbound flow. If you did that then there would be no need to split the block. If I had the end of a 6 track staging yard that fed into a single main that fed into a yard with a main, siding and switch lead, I would have 6 detectors on one end and 3 detectors on the other. The combinations of which route inbound and which routes outbound are immaterial. If the sum of any of the inbound routes less the sum of all the outbound routes is not zero, then the block is occupied.
The problem that it has opportunities for failure issues if the train car count changes (whether its and increase or decrease) is still a concern.
If the system would accomodate an "infinite" number of exit and entrance points on a single block, then the way to fix the set out/pick up problem is that you put a detector at every switch into the block, then the totals will always be right (assuming a high degree of detection reliability).
With that many detectors, how to hide them may be an issue.
Dave H. Painted side goes up. My website : wnbranch.com
dehusman It seems to me that the key to making this work is to have a "total" for each block and all the detectors on the inbound side add to that total and all the detectors on the outbound side subtract from the total. No matter how many inputs there are. On block 4 there are 4 detectors on entrances to the blocks so the total in the block is the sum of all the input detectors, regardless of which one was the input, less the outbound on which ever detector measured outbound flow. If you did that then there would be no need to split the block. If I had the end of a 6 track staging yard that fed into a single main that fed into a yard with a main, siding and switch lead, I would have 6 detectors on one end and 3 detectors on the other. The combinations of which route inbound and which routes outbound are immaterial. If the sum of any of the inbound routes less the sum of all the outbound routes is not zero, then the block is occupied. The problem that it has opportunities for failure issues if the train car count changes (whether its and increase or decrease) is still a concern. If the system would accomodate an "infinite" number of exit and entrance points on a single block, then the way to fix the set out/pick up problem is that you put a detector at every switch into the block, then the totals will always be right (assuming a high degree of detection reliability). With that many detectors, how to hide them may be an issue.
This system works that way, but for only 3 inputs, limited by the number of pins available, and the requirement for absolute communications reliability.....that's the reason for the 3 pins for each detector communications link. Anything less would not be reliable.
My system is designed the way it is specifically to remove the computer from the installation....the fact that the Arduino is used to implement the internal logic is immaterial. Logically, the computer is invisible.
Your proposal sound wonderful, but when you come back down to reality, no system is capable of handling an infinite number of entry points.
I know what you want, but the design goals were for a system were simplicity and low cost. The only real way to allow for what you want would be a complex system that transferred packets using some protocol with error correction and verifications along a common buss of some sort used to link the various entry points, which also would require some semi-complex configuration to define how it all goes together.
Configuring and troubleshooting such an installation would be completely beyond the capabilities of just about all modelers, in both capability and equipment required.
But I'm sure that you're up to it....please keep us updated on you progress.
crusader27529Your proposal sound wonderful, but when you come back down to reality, no system is capable of handling an infinite number of entry points.
Actually, existing current-detection systems do that just fine, no computer. Don't they? A few cents per car for a resistor-equipped wheelset.
Edit: Plus, no visible detectors sticking up all over the layout -- which would be the biggest non-starter for many with your proposal.
Layout Design GalleryLayout Design Special Interest Group
crusader27529This system works that way, but for only 3 inputs
What if you added a multiplexing device to your system that could combine the signals from multiple detectors and forward them to one pin on the counter? Design it so you could build trees of them as required.
I have the right to remain silent. By posting here I have given up that right and accept that anything I say can and will be used as evidence to critique me.
I know what you want, but the design goals were for a system were simplicity and low cost. The only real way to allow for what you want would be a complex system that transferred packets using some protocol with error correction and verifications along a common buss of some sort used to link the various entry points, which also would require some semi-complex configuration to define how it all goes together. Configuring and troubleshooting such an installation would be completely beyond the capabilities of just about all modelers, in both capability and equipment required.
Infinite just means many. The point is the existing systems allow for the situation I have described, a single block with many routes in or out. Since they have been in existence for decades in many different versions, and used by hundreds of people with signal systems, I can't help but think that that can't be beyond the capabilites of all modelers since virtually all modelers who have signal systems have blocks with multiple routes in or out of the block. I would think for your proposal to be useful to a broader audience it would have to replicate that type of capability.
I know I can make a block with multiple routes in to the block work, its really simple if you use a system other than yours.
The primary issue in these discussions are whether current detection is adequate with only head & tail being detected, or making all rolling stock detected. Both have issues, specifiically loss of tail detection or having SOME rolling stock not detectable (somebody bringing rolling stock that isn't detectable), and my system, which counts things as they enter or leave a block. ALL these approaches have some problems, and ALL solutions have tradeoffs.
Since I'm the 'new kid on the block' related to detection, people who already have current detection defend their systems, because they work, and the advantages of a new system don't justify replacement of a functional system. That makes perfect sense.
BUT, you've got to separate a detection system from a complete, implemented signaling system. Detection is the building block for signaling, just like any detection technique.
In the fantasy world of having perfect detection, detection with no power in the rails using metal wheels and axles would be almost perfect, but we don't have mineature batteries that last long enough to be really usable, or charge fast enough to be practical, or batteries that have charge/discharge cycles long enough to have really long life.
IMHO, the perfect detection system would be a scanner to read a barcode off every piece of rolling stock as it enters or leaves a block. This is doable, but scale sizes would be an issue, not to mention the cost. A commercial barcode reader has a rotating mirror to control the scan of the barcode eliminating the issue of the speed an item passes the scanner.....that's large and expensive. A laser diode source (IR or visible) would be requied at each block.
An RFID reader is an alternative, but the coils needed are somewhat large, and HO is the likely current minimum size that would work. Most RFID readers installed are used to keep track of rolling stock, and NOT for detection, because it would require readers at the beginning and end of EVERY block, which means MANY readers. And, it still would only read instrumented rolling stock.
Overall, the barcode scanner would be best, because adding a barcode is as simple as pasting it to the car.....since none of these fantasies are currently practical, we need something new and/or different, IMHO.
Come up with some ideas, and maybe we can make something practical that's affordable and not too complex.
In all these discussions, I've been remiss in pointing out one other major advantage with my system:
IT DOES NOT REQUIRE ANY ACTUAL ELECTRICAL BLOCKS. IT REQUIRES NO GAPS BE CUT IN ANY TRACK.
So, with a DCC system, the only actual electrical blocks required would be for reversing loop segments.
crusader27529Come up with some ideas, and maybe we can make something practical that's affordable and not too complex.
It already exists, as a number of people have pointed out. And works fine on hundreds (probably thousands) of layouts.
crusader27529IT DOES NOT REQUIRE ANY ACTUAL ELECTRICAL BLOCKS. IT REQUIRES NO GAPS BE CUT IN ANY TRACK. So, with a DCC system, the only actual electrical blocks required would be for reversing loop segments.
That would be an extremely unwise way to wire a DCC layout (or any layout). Good practice calls for gaps for troubleshooting as well as to allow for expansion and contraction of the benchwork with humidity changes.
crusader,
Maybe I'm just lucky, but I don't have all these derailment related losses of cabooses you are so worried about.
It's called good trackwork and good equipment standards (coupler heights, etc).
The proposal is basically for an "axle-counter" approach.
Axle-counters are certainly used on the prototype but only for specific applications where a simple in/out count is useful and adequate for safe operation, not as a general purpose track occupancy detection approach. Robust safe-working systems are simply not designed on the basis of a single person being able to reset anything in the system.If you need a reliable axle-counter (and understand the limitations of the approach) then the proposed detctecor may be worthwhile considering, but its nowhere near a complete track occupancy detection solution.