modelrail.otenko Trains, computers, vintage, retro, model railways… the lot…

Final PCB Board Complete!

Well, Silver Circuits manufactured and delivered my PCB much much quicker than expected and I'm not complaining!

And, last night, after a rush of construction... I couldn't get the board to work... The 'trip' sensors were permanently active! After a 30 minute stare at the circuit design today I'd realised I'd viewed an older circuit schematic when designing the PCB and had put the inputs to the LM339 the wrong way around!
After a quick test to correct this I had the booster operational.

Unfortunately this now means that all boards in this initial run are defective, but I have devised a way to correct the issue without too much hacking. A small veroboard piggy-backed on has allowed me to correct the inputs to the IC.

I was very much relieved once it was all operational. I now need to get the final components (you can see a dodgy-hack of resistors wired on) and the heatsinks in place. Then I'll need to find a neater box to mount it all in and all should be set to go.

PCB Design Complete

Well, after a long fight of finding appropriate software and then design I've finally sent off the final (if not entirely messy) design to the manufacturers...

Sure, it's not as small as I would have liked... but it's all set in stone now. I should see 8pcs on my doorstep in around a fortnight and then I can finally produce a finished product.
I can't wait actually... Last weekend I tested the final prototype on my own layout... which really only allows for around 3 trains running (one on a branch, one stuck in a siding and one on the main) and my battery ran out on the laptop before I had a chance to fully test it all out... either way, the quick test ran perfectly on all accounts:

I also purchased two DCC Accessory Decoders from eBay. I coded those in to trainControl (you'll see a few little squares above the console messages and below the train throttles) and they work great. Unfortunately they cannot be wired directly to the Kato UniTrack points (you'll need to wire a few relays in between) but I'm told work well on the Peco points. I didn't get to test them at the time.

The DCC Accessory Decoders did require me to do a bit of fine tuning on the Booster circuit though; it turns out they suck quite a bit of current on power-up to charge their little capacitors. This meant that the initial current draw on the booster was exceeding the trip circuit and the whole system would not power on.
After a little help the 1K2 resistor providing the op-amp comparison voltage was incremented to a 1K4 and everything worked fine.
This made me realise that the circuit will have different requirements for different layouts and so the final design will now incorporate a trimpot (initially set to 1K4) that can be user adjustable [but WILL void warranty] for 'expert users' to adjust.
Anyway... the work continues... everything is now set. I'm quite proud of the entire system and I can't wait to see the final circuit boards!
UPDATE: The boards are already done and on their way (priority) to me... thanks to Silver Circuits!

DCC Booster Prototype Mark II

Ok, hot off the development floor is the Booster Prototype 2.

This now includes short-circuit protection! The final design will have 2 outputs (for two separate segments of track) at 3 Amp each (depending on power supply capabilities.) There are also now 4 status LEDs (Supply Power, Data, Overload, Track Power/Data)... More to come as I finalise the circuit board design.

trainControl evolves…

After a few more days of work the trainControl application has evolved further:

Notable features:

• Loading and saving of trains/connection config
• Resizable
• Add/Delete trains via interface
• Connect/Disconnect on the fly

Still more to go... but very functional now... Download the source here. Of course, be warned! ... The source is constantly being updated by me and may DESTROY your computer. I've also considered the next challenge on this project... Handheld control via a GamePark GP2x.

trainControl v0.1 created!

After 3 days of coding/learning/coding-again I managed to whip together a fairly usable *nix console application to control the DCC System.

It is a client for the srcpd service mentioned previously and can control any number of trains.
There's still a fair bit of work to do on it... dynamically resizing windows, mouse input, etc... but I thought I'd post to show some progress.
If anyone is interested the source is here and can be compiled with gcc -o tc main.c -lncurses -lform -lpanel

...work continues...

Controlling the Booster from Linux

After losing the source code to my DCC Throttle application I decided it was time to get the whole system controlled from Linux (or, in my case, Debian.) I'd known that the DDW Server I'd been using was based off srcpd and that they were more-or-less compatible.

After a lot of tinkering (making sure libraries/compilers/etc... were installed) I finally got srcpd compiled and running.
There is a nice web-app to create the required configuration file which is then thrown in /usr/local/etc/ or can also sit in the same directory as long as you specify it.

It seems that you need to make sure that the 'NRMA/DCC Translation Table' is set to '1'. I have a copy of my configuration file here.
Right, now that the server was running I grabbed Telnet and connected to localhost:4303. I attempted to use the same set of commands as I had with DCCThrottle but to no avail... it turns out the version of srcpd I had installed was based on the 0.8.3 srcp specification... much had changed.
Instead of reading up on what I should be doing differently I connected up dtcltiny 0.8.2.

After a lot of tinkering with compilation/libraries/etc... I finally got it installed. On debian make sure you have qt3-apps-dev and libqt3-compat-headers installed... do this with apt-get. Anyway... I'd done all this on a spare Thinkpad 600E I had lying around (PII 266mhz 256mb RAM) and I'd had a few issues with lag once it was up. I'd turn the headlights off on the control panel and my Twilight Express locomotive would then go dark around 1.5 seconds later. After turning off nearly all debug output I had a relatively stable system... but it does need a lot more investigation.

The next trick will be to investigate building a client on the console... I don't need fancy Xorg to get a gui and an ncurses console app should do. I saw rcsh and rcman and the rest, but python isn't my friend... although it'd be a nice opportunity to learn.

Anyway... all of this was a great learning experience... and I was quite impressed that my old notebook handled Debian and got the locos moving.

Prototype Booster Complete!

Well... after a night of work (and using my breadboard version as basis) of soldering and drilling I have a prototype! I've used off-the-shelf components to house the board and everything fits nicely... secure... and works perfectly.

The next version will have an override switch (since the current version will only activate when there is data on the serial port) to allow power all the time.

Now is time to use this version as a test for how many amps it can push out... note I'm still using my computer power supply (which is still powering this machine I'm surfing on!) and all is well.
Finally... The going price for one of these will be AUD$65.00.

Anyone wanting one can leave a comment on this post or email me at stevenhoefel at hotmail dot com.

I've also written a new client app... very basic... controls the train that you specify.