I've been on the road for the last 8 hours, but have kept up through email notifications throughout the day and gotten several good ideas. I'll try and run through all the comments and respond to each now. There seem to be 3 camps of things to cover; functions such a device should have(software), the form it should take(hardware), and how to select or upload different functions and adjust them(programing).
I'll start with the miscellaneous ones that don't fit neatly anywhere else. RTR12, I've been following along with GRJ's insulated rail module but just haven't had anything useful to add to the conversation up through now. It's likely to be a useful device for many folks. The discussion has also served to really drive home a point to me, though. Nothing against the thing because it is some good solid engineering that fills a need, but it serves to show that folks are willing to spend some dollars to avoid working with even the most basic electronics. That has set the tone for my project here, as I think one of the most important aspects will be making it dead simple for people to use, with no learning curve whatsoever. It needs to be available ready to use out of the box, and easy to set up for folks with no inclination to learn any new skills.
As a side note, as of right now I've been unable to find a way to duplicate the functionality of pre-built relay modules from china for the same or lower cost than those modules. I agree with the problem of insuring a consistent design to the modules over time that was pointed out in GRJ's thread, and need to consider that further for a design here.
Functions:
PLCPROF, "I would like a detector system that works in pairs, so it can detect a short train in a long block without insulated rails. One sensor at each end of the block."
Do you mean, if using something like an IR sensor of some sort on each end of a block, the output would activate when a sensor on either end is triggered, then would remain on until the second sensor is triggered and 'un-triggered'? This is, of course, possible but may require some fiddling to get it to work smoothly, if the input sensors give intermittent feedback, such as from gaps between cars and such. I think I can make it work in any case. I'm mostly basing things on using insulated rails at the moment, but other sensors could be incorporated.
John Graser, Do you mean having TMCC Commands injected into the base when the device is triggered by a passing train? This is possible, but substantially more complex to accomplish than what the scope of this device would cover. I do have another project in the works that might handle that, however.
I lost track of where it was mentioned, but the idea for toggling a switch each time a train passes is certainly feasible.
Hardware:
Consolidated Leo, I tend to agree that special circumstances are best handled on a case by case basis. However I think that simple timers will work well for most cases. I don't think I want to get into modifying accessories at this point. I think that is beyond what the average person would want to do. I'm not opposed to allowing for limit switches to be connected to the device, but will have to see how much I/O room is left when all is said and done. I have to put a lot of thought into anything that will add board space, as I need to keep the real estate down to make a cost effective PCB. On the last point, I'm in a toss up right now between using a ProMini board or just going with a straight ATmel328 chip on the board. I like the nano for general tinkering, but the ProMini offers the same functionality for an embedded device at a lower cost and in a smaller package. The nano would only be better if a user-accessible USB is needed…and I have other ideas on how to allow users to adjust the program. I'm also kicking around with using an entirely different uP.
Stan, At this point I really am looking for the kitchen sink, since I don't have any idea what folks would find useful. I'm sure that I'll have to trash some ideas out eventually. On the idea of a modular design, that's feasible, but I think it would drive up the cost in the long run. You would need three modules instead of one in anything that existing products can't do already. I like the idea of piggybacking on GRJ's module, but it would require a complete redesign to be useful here. and would needlessly drive up the cost of that product to add support for features that wouldn't be utilized by most people. On the other hand a 99 cent relay module from china and 20 cents of components would provide an optically isolated input and output for a uP.
Programing:
I've tossed around everyone's comments on this all day. At this point I'm down to three options I like. The first is the SD card. In theory a user could pop the card into their computer and have some simple piece of software to click a check box for which function they want and set any simple parameters for that function. The second is currently beyond my ability to do, but I think I know a couple folks that could get me started. That would be to drop the arduino, and instead use a BLE module as the brains. Here a smart device App would be used to program the thing. I think this would actually provide the best, easiest to use, functionality, but it would require my learning to program the BLE module, and likely finding someone to write the app. I really like this option, but it is presently not something I know how to accomplish. The third option to have a 'programer device' This would be a separate unit that a module would plug in to to be programed. The programer would have easy to use controls, perhaps using an LCD display of some sort to show the various settings. While writing another option occurred to me of using a Wifi connection for programing. A simple 'webpage' could be hosted on each device allowing settings to be adjusted from any wifi enabled device. This is something I do know how to do, even if it wouldn't be very pretty.
Mostly, I think it is important here to make things easy to use. I feel that most people would be turned off by any complex loading process, and some even by having to do anything on a computer at all. To that end I think that having all of the 'simple' functions pre-programed into the device would be useful, only requiring outside setup for more complex things. I think folks would be more willing to have a slightly more complex install if they are getting a function that either doesn't exist from another product, or that is prohibitively expensive in other options on the market.
So, after the feedback thus far, I think I'm looking at hardware that is capable of detecting at least two inputs and triggering at least two outputs, and having various options of what to do with those. I'm thinking that jumpers or dip switches for selecting various simple functions could work and one of the options there would be to set the unit to 'advanced' mode for outside programing. I'd likely also like to have some multi-turn potentiometers for timer adjustments on the board.
JGL
