Sunday, February 06, 2011

Train detection woes

I spent Friday evening hooking up power feeders in Emsingen, as well as  connecting the drops from the train detection sections to a Uhlenbrock 63350 detector. The 63350 sounds great: Loconet, 16 inputs, flexible programming. But, ... it just won't send a sensor signal to Loconet when cars or a locomotive enter the monitored section. However, when I send a Loconet interrogate command, the unit happily reports all occupied sections correctly. So it seems I have everything connected correctly on the electrical side. Since Maerklin is a 3-rail system, I can just insulate one rail, and whenever the axles of a car bridge the two rails track voltage can be detected against ground on the sensor input. Ground gets to the unit via Loconet, so this should work. Looks like it's time asking on the appropriate mailing lists, and contacting Uhlenbrock Support.

Programming Loconet Configuration Variables (LNCVs) via the Intellibox menu is a tad cumbersome. I need to figure out how to do this for a unit that is not directly supported by JMRI.

What's more annoying though, is that I found out while researching the detection problem, that the 63350 doesn't have opto-insulated inputs, so shorts can show up on LocoNet and possible cause havoc (in fact, I had one nearby switch, which is controlled by a LocoNet connected decoder, throw itself when I created a short voltage spike while fiddling with cables. Not good). Not being opto-insulated can cause issues with Digitrax equipment later on, too.

For now the plan of attack is to remove the 63350 from track detection, finish up wiring in Talheim which is using a Viessmann 5217 (S88 straight into the Intellibox), to verify the general approach, and use the 63350 for the control panel in Emsingen complementing a 63400 unit. That way the control panel will have only one power plug for the 16V accessory bus,  as well as the Loconet plug(s).

No comments: