Most intriguing, and it would appear you are on to something really big here. At this very early point in my investigation I can merely suggest a change in your product terminology for the American market: in English (and in typical American railroading terminology) trackside is one word, not two. This is a compound word in English, not two separate words. I look forward to further installments. You present yourself well and your concepts very clearly in these videos. Thank you.
I'm glad that you are diving into this topic. I've just been thinking about train detectors for my layout and was looking at the various options. I was wondering why I was not hearing that people were using some of the cheap current detectors for Arduinos that are available on EBay (5A Range Single Phase AC Active Output Micro Current Transformer Module ZMCT103C). Although I never got an answer, I have decided to experiment with one later but in the meantime I'm going with the MERGDTC2-T DCC Dual Train Detector. It appealed to my desire to build and it worked out cheaper than any other commercially available current detector. I'm looking forward to finding out if you had considered one of these and also what else you have up your sleeve for train detectors. Keep them coming Hans!
Hi. Like you I have been interested in detection for a very long time. The trouble is that amounts the current offering of detection types there is nothing that is (a) universal an (b) fit for purpose for all needs every time. This then translates into having one set of detectors to deliver (for example) block occupancy but doesn’t cover engine/wagon identification, direction of travel, loads carried or specific track location. Because of this if one is to detect the full gambit of conditions one would like to have we are then require to multilayer, say current detectors and separately specific location detector and finally say technology like rfid/rail com for specifics on rolling stock. In my view the worst part of this is not that there are three layers of detection at all, rather that their technology to use and pass this information back to some central repository is totally fragmented. I dream of the day we have a can type producer consumer bus that treats all sensors as “an event” an pars that to a central repository to be integrated into an intelligent place to deal with the information from everything.
Hi Hans, I find your concepts very interesting. However, I model the 1960s and block detection on the real thing reigned supreme up until the 2000s. Since then axle counting in my country (New Zealand) has been installed in some areas, mainly our two major suburban networks, track block detection is still the major method of detecting prototype trains. This is also easy to implement in the model world and train current detection has been around since the 1960s. Diode and CT current detectors are the most common. I built my first TwinT DC detectors in 1970. My current layout is equipped a variant of Bruce Chubb's Optomised DCC Detector, and has proved reliable for over 20 years of operation. I use JMRI and find it easy to detect train detection by using the concept of. A train travelling in Block Detector A and enters Block B it is travelling North. If travelling from B to A it is travelling South. I do have some some automation on my layout and rely on slowing a loco entering a block and then stopping on the next block, usually using the Acceleration and Deceleration CVs of the decoders. To overcome differing loco characteristics and to stop a train in front of a signal or at a platform, I need to install location sensors. See ua-cam.com/video/0rUtKtrvSa4/v-deo.html I have modified, designed and built quite a lot of circuits over the years and detection circuits include with my comments. For small scale trains. the two most common ones are. Block detection. The very common Diode Forward Voltage Drop block detectors must be opto-isolated and trains slow when passing through from unequipped blocks, unless undetected blocks are equipped with some "dummy Diodes". I have designed and built boards for installation at a local MR club. CT Current transformer, self isolating, require high voltage protection on the transformer output and amplification for axle resistor detection. These are my preferred option. I have looked at and experimented with the modern current detectors, but they have some short comings such as their current noise threshold. Location Detection Photo-transistors. Easy to implement and connect direct to a micro. These are excellent for a day time lit layout. but no good on layouts with night lighting. Infra-Red. Easy to implement, but current Ebay and some commercial offerings use 5 mm IR LEDs and IR transistors which are hard to hide and fit into existing track work. I have replaced these with 3mm IR devices with success and it makes them easier to install. The main aspect of the above is they don't require any modification or installation of "special" circuits a to the locomotive or rolling stock, and this makes it easy for the average modeller to install, that's why manufactures have generally stuck to the above options. Magnetic. I am currently working on a signalling system for a live steam and battery loco garden railway. In this case Hall effect sensors above the sleepers, and magnets installed on the locos by simply placing them on the cow-catcher (Pilot), where the tend to stay. Visitors are supplied with a magnet. Light detection was no good due to varying light conditions, cloud cover and evening running. In small scale magnet detection is usable unless you have visitors. Not easy to install magnets though. One thing I have discovered over the years is that the KISS principle is very important especially when supplying others equipment to install. I am following your ideas and experiments with interest and I know the amount of time you spend is huge. Keep it up and look forward to your upcoming videos. Graham
Hi Graham, thank you for your comment. Very interesting to read how you do layout automation and watch your video. I look at the problem from two different view angles: model railroading and industrial automation technology. From the model railroader's perspective, particular when modeling past epochs as you do, current sensor block detection is the way to go. It is prototype-like and very reliable. But in order to ensure reliable automated operation, you need either locomotives with very similar characteristics, or location sensors (at point sensor), as you say in your comment. If your model more recent epochs, you probably want to have more modern concepts as well. Nowadays trains use moving blocks, dynamic braking distances and so on, all in the interest of reducing safety distance between trains and increase traffic density while maintaining safety levels. I have no intention to prescribe to anybody how to do layout control, but I think a more diverse set of technologies should be available and this is a shortcoming of the mainstream manufacturers. From the industrial process automation perspective I see a lot of development in the areas of sensor technology and control systems, that simply did not make its way into the model railroader's world. And since no-one would try to control a numerical axis with technologies from twenty years ago, I don't see why model railroaders should not make use of these new technologies. That is my motivation to experiment with various concepts.
For your “GPS dream” take a look at GamesOnTrack.com, I am sure you have found them by now and I would be very interested in your technical opinion of this system. Great work with the videos.
Thanks for the link. No, I was not aware of this system. From what I understand they use ultrasound triangulation (time of flight) to calculate the location and send trigger pulses to do so via radio. I like this concept. Time of flight is definitively the right approach to do triangulation for localization, but I would like to see that with radio. Some of the new Wifi routers can do it, but the accuracy right now is only about 10m. I am sure in a few years down the road radio localization will be the way to go. And by then the devices will be small enough to fit in rolling stock ;-)
Yes, this is how most control software does it (integration of speed over time to calculate the distance). This works reasonably well as long as there are no power interruptions and the software knows the speed step to speed ratio for each address. And of course, also the length of each block.
Yes, that would be a simple form of a trainside sensor. The other possibility would be using an incremental motor and communicate increments instead of speed steps. None of those solutions is part of any standard, as far as I know.
You make very good videos, but I wish you would get the mic closer to you. Too much echo in the room and makes it harder to hear you. Thanks for all your work in these.
Thanks for the feedback. Unfortunately, my cell phone seems not to support an external microphone. But I consider to use something else as recording device.
Most intriguing, and it would appear you are on to something really big here. At this very early point in my investigation I can merely suggest a change in your product terminology for the American market: in English (and in typical American railroading terminology) trackside is one word, not two. This is a compound word in English, not two separate words.
I look forward to further installments. You present yourself well and your concepts very clearly in these videos. Thank you.
I'm glad that you are diving into this topic. I've just been thinking about train detectors for my layout and was looking at the various options. I was wondering why I was not hearing that people were using some of the cheap current detectors for Arduinos that are available on EBay (5A Range Single Phase AC Active Output Micro Current Transformer Module ZMCT103C). Although I never got an answer, I have decided to experiment with one later but in the meantime I'm going with the MERGDTC2-T DCC Dual Train Detector. It appealed to my desire to build and it worked out cheaper than any other commercially available current detector. I'm looking forward to finding out if you had considered one of these and also what else you have up your sleeve for train detectors. Keep them coming Hans!
Hi. Like you I have been interested in detection for a very long time. The trouble is that amounts the current offering of detection types there is nothing that is (a) universal an (b) fit for purpose for all needs every time.
This then translates into having one set of detectors to deliver (for example) block occupancy but doesn’t cover engine/wagon identification, direction of travel, loads carried or specific track location.
Because of this if one is to detect the full gambit of conditions one would like to have we are then require to multilayer, say current detectors and separately specific location detector and finally say technology like rfid/rail com for specifics on rolling stock.
In my view the worst part of this is not that there are three layers of detection at all, rather that their technology to use and pass this information back to some central repository is totally fragmented. I dream of the day we have a can type producer consumer bus that treats all sensors as “an event” an pars that to a central repository to be integrated into an intelligent place to deal with the information from everything.
Hi Hans, I find your concepts very interesting. However, I model the 1960s and block detection on the real thing reigned supreme up until the 2000s.
Since then axle counting in my country (New Zealand) has been installed in some areas, mainly our two major suburban networks, track block detection is still the major method of detecting prototype trains. This is also easy to implement in the model world and train current detection has been around since the 1960s. Diode and CT current detectors are the most common.
I built my first TwinT DC detectors in 1970. My current layout is equipped a variant of Bruce Chubb's Optomised DCC Detector, and has proved reliable for over 20 years of operation.
I use JMRI and find it easy to detect train detection by using the concept of. A train travelling in Block Detector A and enters Block B it is travelling North. If travelling from B to A it is travelling South.
I do have some some automation on my layout and rely on slowing a loco entering a block and then stopping on the next block, usually using the Acceleration and Deceleration CVs of the decoders. To overcome differing loco characteristics and to stop a train in front of a signal or at a platform, I need to install location sensors.
See ua-cam.com/video/0rUtKtrvSa4/v-deo.html
I have modified, designed and built quite a lot of circuits over the years and detection circuits include with my comments.
For small scale trains. the two most common ones are.
Block detection.
The very common Diode Forward Voltage Drop block detectors must be opto-isolated and trains slow when passing through from unequipped blocks, unless undetected blocks are equipped with some "dummy Diodes". I have designed and built boards for installation at a local MR club.
CT Current transformer, self isolating, require high voltage protection on the transformer output and amplification for axle resistor detection. These are my preferred option.
I have looked at and experimented with the modern current detectors, but they have some short comings such as their current noise threshold.
Location Detection
Photo-transistors. Easy to implement and connect direct to a micro. These are excellent for a day time lit layout. but no good on layouts with night lighting.
Infra-Red. Easy to implement, but current Ebay and some commercial offerings use 5 mm IR LEDs and IR transistors which are hard to hide and fit into existing track work. I have replaced these with 3mm IR devices with success and it makes them easier to install.
The main aspect of the above is they don't require any modification or installation of "special" circuits a to the locomotive or rolling stock, and this makes it easy for the average modeller to install, that's why manufactures have generally stuck to the above options.
Magnetic. I am currently working on a signalling system for a live steam and battery loco garden railway. In this case Hall effect sensors above the sleepers, and magnets installed on the locos by simply placing them on the cow-catcher (Pilot), where the tend to stay. Visitors are supplied with a magnet. Light detection was no good due to varying light conditions, cloud cover and evening running.
In small scale magnet detection is usable unless you have visitors. Not easy to install magnets though.
One thing I have discovered over the years is that the KISS principle is very important especially when supplying others equipment to install.
I am following your ideas and experiments with interest and I know the amount of time you spend is huge.
Keep it up and look forward to your upcoming videos.
Graham
Hi Graham, thank you for your comment. Very interesting to read how you do layout automation and watch your video.
I look at the problem from two different view angles: model railroading and industrial automation technology.
From the model railroader's perspective, particular when modeling past epochs as you do, current sensor block detection is the way to go. It is prototype-like and very reliable. But in order to ensure reliable automated operation, you need either locomotives with very similar characteristics, or location sensors (at point sensor), as you say in your comment. If your model more recent epochs, you probably want to have more modern concepts as well. Nowadays trains use moving blocks, dynamic braking distances and so on, all in the interest of reducing safety distance between trains and increase traffic density while maintaining safety levels. I have no intention to prescribe to anybody how to do layout control, but I think a more diverse set of technologies should be available and this is a shortcoming of the mainstream manufacturers.
From the industrial process automation perspective I see a lot of development in the areas of sensor technology and control systems, that simply did not make its way into the model railroader's world. And since no-one would try to control a numerical axis with technologies from twenty years ago, I don't see why model railroaders should not make use of these new technologies. That is my motivation to experiment with various concepts.
Looking forward to going on this journey. :-)
Great info, thanks for sharing.
Thanks for watching!
For your “GPS dream” take a look at GamesOnTrack.com, I am sure you have found them by now and I would be very interested in your technical opinion of this system. Great work with the videos.
Thanks for the link. No, I was not aware of this system. From what I understand they use ultrasound triangulation (time of flight) to calculate the location and send trigger pulses to do so via radio. I like this concept. Time of flight is definitively the right approach to do triangulation for localization, but I would like to see that with radio. Some of the new Wifi routers can do it, but the accuracy right now is only about 10m. I am sure in a few years down the road radio localization will be the way to go. And by then the devices will be small enough to fit in rolling stock ;-)
Train side could also be a combination of the decoder telling how far it has driven and last known location.
Yes, this is how most control software does it (integration of speed over time to calculate the distance). This works reasonably well as long as there are no power interruptions and the software knows the speed step to speed ratio for each address. And of course, also the length of each block.
@@IoTT If the decoder counts actual revolutions -calculates the distance and feeds it back, it will be very precise.
Yes, that would be a simple form of a trainside sensor. The other possibility would be using an incremental motor and communicate increments instead of speed steps. None of those solutions is part of any standard, as far as I know.
Thank you
Welcome!
You make very good videos, but I wish you would get the mic closer to you. Too much echo in the room and makes it harder to hear you. Thanks for all your work in these.
Thanks for the feedback. Unfortunately, my cell phone seems not to support an external microphone. But I consider to use something else as recording device.
@@IoTT Dose you phone support a headset? If so you have a microphone option there.