Fun with JMRI II and September 2011 Status
I’ve been playing around with JMRI some more, and trying to debug my transponding problem with the first of the electronics boards. This is really baffling. I checked the wiring, and it was fed through the RX sensor properly. I replaced BOTH the PM42 and the BDL168 circuit boards (I’ve got a stack of them waiting for more electronics boards once I get this one working) and I tried using other blocks. And I had more transponding sensor failures. On both sets of RX sensors. One defective set I might accept, but two?
So I tried a variety of things, and noticed that the non-functional detectors would, every once in a while, work. In fact, I discovered that with the train motionless, one of them would periodically cycle from detection to non-detection, emitting a LocoNet message reporting the change in status each time. I tried moving the wires. I pulled a fresh RX1 set out of a bag, and set it up atop a trash can (see above) with every wire fed through it fully separated from every other wire in mid-air (about the middle of this I was holding things in both hands and wishing I had a third arm). And that failed too, reliably as it were.
I move the wires even further apart, adding longer feeders to the track so I could route them well away from the BDL168 itself (which isn’t supposed to matter, but by this point I was trying to eliminate all possibilities, no matter how unlikely).
I tried running wires through the sensor front to back, and back to front (and came to the conclusion that both worked, or didn’t, equally well). I plugged in the LT5 tester (four of them, actually) and watched the LEDs indicate occupancy and transponding function at the same time I watched the LocoNet message log on the computer. Oddly, I occasionally saw LocoNet occupancy messages even when the LED said transponding wasn’t working in that block.
I thought the old, slow, computer I was using might be a problem, and swapped it temporarily for a recent Intel laptop with the same software; no change.
Then I pulled out my multimeter and checked the new sensors, but all four reported almost exactly the same resistance (2.3 - 2.8 ohms), indicating that it wasn’t a simple wiring problem.
It’s still unsolved. I’m tempted to believe that there’s some kind of noise issue here, but I can’t really figure out what it might be. I may move the computer and power supplies further from the board to see if that has an effect (it seems unlikely). There’s an interesting note in the BDL168 manual about adding a resistor across the input of the BDL168 to reduce sensitivity for outdoor layouts. I’ll probably try that, although my gut feeling is I have too little, not too much, sensitivity.
One final thought is that I’m using a DE10 equipped with a Digitrax DN163K1D decoder. There’s a note in the RX4 manual about wire-in decoders that observes that resistance on the F0 (white) lead is important to proper functioning of transponding, and that adding a 270 ohm to 470 ohm resistor across the white (F0) and blue (common) leads may be required to make transponding work. They also note, rather ominously, that with board decoders the resistor “may” already be installed. With LEDs on the DN163K1D I suspect there’s already adequate resistance there, but that’s another thing to look into. And I’m a bit worried that I may need to do something special to make my “DCC Ready” trains with the drop-in EM13/FL11 decoders work properly. However the FL11 driving LED headlights should have adequate resistance just like the DN163K1D, and I don’t need to detect the motor car if I can detect one or both cab cars. Still, another thing to worry about if I get this working at all.
A bit of Googling around suggests that other people do transponding with N-scale trains, and I can’t find anyone asking questions about similar problems, which leads me to think I’m missing something blindingly obvious, and will really feel stupid once I figure out what the problem is. Ah well, it’s a frequent feeling for me with electronics; I’m used to it.
Site Updates
I’m going to start noting significant changes in these postings, for those who don’t want to pull up the General RSS feed (found on the sidebar of the Home page) and see every change, as well as to create a record for myself of what changed when, since the RSS feed has a limited number of entries and deletes the old ones.
- Updated the collection page with new trains: JR West 125 Series, Toden Arakawa 7000 tram and Tōkyō Metro Ginza 01 Subway train.
- Added a Subway Trains subordinate page to the Collection section; I used to have a bunch of these, but wasn’t happy with the format, so instead of converting them I’m going to rewrite from scratch. This is the second (and the E231 one was done in Stacks and is going to get re-done).
- I’ve added a page collecting all of my info about the BDL168 in one place. This will be gaining more content in the future, and is somewhat redundant with other pages (although I likely won’t remove the info from the others, as the bits fit in where they are).
Administrivia and Monthly Status
Well, September was an eventful month. I finally got the site converted from iWeb to RapidWeaver (culmination of a two-month project to re-enter every page in the new system), only to discover the size of it was stressing RW to the point it crashed. That was traced to a problem with a plug-in called Stacks, which has a memory leak and issues with storing large amounts of photographic data (and I have a lot of that). So now I’m busy re-converting pages from the Stacks format to the basic Styled Text format. Fortunately that’s faster and simpler than the original iWeb conversion was. At the same time, I have other things to work on, so it’s not going as quickly as I’d like. Probably only a quarter of the pages have been converted so far, but they were the most image-heavy ones, so RW is running faster and much further from its memory limits now, but it will still crash reliably if I try to export the site and then save (I have to do: “save”, “quit and restart”, “export” to get it to work, and as a result, every page is always marked “changed”, which makes incremental publishing impossible at present).
The other thing for the website was to decide how I wanted to do the blog comments. I chose to create my own for lack of anything suitable out there. This entails writing a comment system from scratch. I’ve made a start on that, lost the better part of two days trying to get the CSS right because I was using the wrong kind of comment delimiter and it silently ignores lines with bad text on them (I wondered why rules seemed to work for a while and then stop; they stopped after I got them working and added comments so I’d know in the future why I did it that way, duh!). There’s still the actual programming yet to do, but I think I’ve got a good handle on that. It’ll be a few more weeks before I have working comments, but they’re coming.
In the past week I’ve gone through many of the site pages looking for links that got corrupted during my page conversions. In some cases the link is just broken, but in others it ends up pointing to a page added after I deleted the old one, which is never the right page. I expect there’s still many of these out there (and I’m still creating more with my page conversions) so I’ll be doing more of these scans in the future.
I noticed that the search box on the website top bar wasn’t working. The reason was I had the old URL in it (the new one omits the redundant “SumidaCrossing” after the domain name). I’ve fixed that, so you ought to be able to see what Google has indexed on the site now.
And that’s about it. I’ve run a few trains on the main layout’s DC tracks and on the tram layout, but other than the electronics board and JMRI work discussed above, there’s not been time for much else on the layout this month. I now have the second controller for the tram layout, and I’m thinking I may do some fixing down of track and adding of scenery there
