All change - a small OO roundy

[fourtytwo]

As described here viewtopic.php?t=58281 I have decided to scrap my first OO attempt that was a terminus to fiddle yard in favor of a round layout to be better able to watch the trains go by! It is small being only 6ft x 8ft and has a single track mainline.

First task has been to strip and clean the existing baseboard modules (mostly 6ft x 2ft) together with recovering everything reusable. Sadly most of the points are medium & large radius that will have to be sold as the new layout only uses small radius.

P1180291.JPG (55.37 KiB) Viewed 3003 times

Presently before track laying I am planning the control system as I am determined it will be at least semi-automatic leaving hands and eyes available for other things than just controlling the trains. This is something I used to do but gave it up some 10 years ago to become fully manual and I now think that was a mistake. Strangely digging out old data has coincided with my local model railway club looking for a degree of automation and we just happen to use exactly the same analogue controller design so maybe two birds with one stone.

So at least the baseboards are up & ready but a long way to go..... I have 3 out of 13 points so far!

Please excuse the jumbled attachments, the software and I seem to disagree about how it works!

[aleopardstail]

looks good, there is a lot to be said for automation and DC, a lot comes down to relay switching logic really and a few sensors, then whatever braking/acceleration system you want - with the massive benefit that with DC you are driving the track and not caring about the actual train..

pity on the points but at that size something has to give, trains will be shorter anyway and to be honest the small points don't look that bad and work just fine

[fourtytwo]

Something else I discovered just in time that has to give is track spacing with tight radii....so an update to the plan to increase this from 2" (50mm) to 2.6" (66mm) in the loops. Unfortunately this has further squashed up the plan and the reverse loop firmly appears part of the station ho-hum.

[Phred]

fourtytwo wrote:
Something else I discovered just in time that has to give is track spacing with tight radii....

Wish I had learned that before a couple of hair raising collisions taught me better!

Your track plan looks really interesting BTW.

[fourtytwo]

Thanks Phred
Track laying has begun but not before altering the plan a smidgen to try decouple the loop from the station a bit more .......

The biggest stupid gotcha so far in track laying is I failed to drill the hole for the servo's to operate the points. Instead I have drilled two small holes though the tiebar hole in it's two positions and hope to open this up with a burr from underneath. This is a horrible job as your eyes fill with sawdust and if you are unlucky enough for the burr to contact the tiebar ......................

To much haste but it means I have been able to assemble a loco and 4 coaches to test stopping distances, for anybody interested unloaded & loaded is 2 & 3.5" from a fairly high speed.

P1180292.JPG (64.02 KiB) Viewed 2890 times

So sawdust here I come, it's 12mm marine ply

[fourtytwo]

Today we got the 1st fully powered circuit and.................derailments!!

It's those settrackish curved points some older stock just doesn't like them there being lots of lurid tales of woe on the net that I discovered after the event!!
However I am fairly certain they are just sorting out the wheel standards I should have long ago sorted out but have been to lazy having just gotten away with it. I have unpacked one rake of coaches and loco that run faultlessly over them so it was just sods law I picked the wrong set 1st.

I tried re-planning the layout to eliminate them but no go so it's wheel standards or bust!

So apart from the choosy points I also found the room floor was not as level as it looked so some baseboard legs got trimmed, the old drill that has done me at least 15years with a new set of bearings some time ago proves to have a dead unobtainium variable switch that's now sitting on the workbench looking for a way to open it, had to use a hand drill (remember those) for the first time in years!

So tomorrow lot's of speed trials as I have a weird control mechanism in mind where it's latency could be life and death on such a compressed layout.

Happy modeling.

[aleopardstail]

had some of them on a layout a decade or so ago, found it was partly crap wheels (and dirty wheels) causing problems performance improved by taking a file to the back of the outer switch rail, specifically the nose of it, get that sharp and problems largely went away

most of the problem in the facing direction was stock being pushed flange to the rail, hitting the nose of the switch rail and climbing it to escape to the wilderness, getting that narrower made a huge difference. other bits were the usual "check rails that may as well not be there" which a thin plastic strip can help with to stop wheels climbing over the frog.

though a lot can be done by accepting that bit of track is slow running.

good to get stuff going and with a complete loop you have gotten further than me

[fourtytwo]

Progress is being made but mostly in the S&T dept. Parts are being assembled literally a 64pin TQFP (0.5mm pitch) chip destined to be the block signalling processor successfully soldered without a hot air gun! Other parts shown are the track circuits and controllers refurbished from last use two layouts ago so ~2010, and there is a newly constructed track circuit bias supply (large blue capacitors) that generates ~28V. All this lot together with an area processor and a dozen DPCO relays (not shown) is enough to control the mainline and loops. A lot of additional complexity comes from the entire layout being bi-directional, a first for me in automation. I am also hoping this new system will deal with dmu's as well as traditional loco hauled stock, the problem being position of the leading end of the train, not just the motor bogie

[aleopardstail]

Electrikery!

We have found a witch!

[fourtytwo]

Greetings all, more & more electrickery!

I lost about 2 weeks trying to find a decent set of libraries for the STM32 processor and in short they do not exist as ST try to force you to use there own horrible Hardware Abstraction Layer (HAL) that is so bug-ridden no bare metal programmer in there right mind uses it!

So instead back to good old Microchip PIC & try and avoid spending £100+ on new debugging tools that try to force you to use by making newer versions of there debug software incompatible with the old ones........ BUT finally I found a solution (mplabx5.35) that supports all the chips & tools I am using, this can only be found by repetitively downloading different versions, installing & trialing them a slow process but at least the libraries for both C & assembler are sane.

So I got to build some hardware! The 6 section signal box that also controls 6 points and some signals is complete and slowly being connected to the layout (power only so far) this has it's own processor along with 23 relays and 18 track circuits that does what the central controller tells it via a simple serial protocol. Meanwhile the central controller using a more powerful processor is also complete from a hardware perspective but much software remains to be written, it also has a USB serial port for status & debug message reporting.

There is also a smaller signal box to build for the rest of the layout that presently remains as the mainline loop (along with all it's points) only.

I have not yet determined what will be on the control panel (that will be connected to the central controller) but the kiss principle will definitely apply.

[aleopardstail]

looks very good, I did consider PIC stuff myself, lovely range of hardware but the software to use it all and programmers £££ mean I stuck with Arduino and ESP32..

presumably you are going with interlocking logic etc using the relays?

[fourtytwo]

I can understand the problems with programmer prices they are well out of order, I am just lucky to have bought a PK3 years ago when they were a lot cheaper. I like Arduino too mainly because of the freedom from expensive tools but in my honest opinion it's to hard to do bare metal & realtime. A particular problem I have is the delay between an occupied signal and opening the stop relay and how far the train will have traveled

The relays do traction switching of one sort or another as that requires dry contacts, each section has a reversing & stop relay, that's 12, then another 6 for point frogs, a couple for reversing the loop rail polarity and two spares for luck!

Interlocking will be entirely in the central controller software, especially hair raising as everything is bi-directional (head on collision potential)

[aleopardstail]

I can understand the problems with programmer prices they are well out of order, I am just lucky to have bought a PK3 years ago when they were a lot cheaper. I like Arduino too mainly because of the freedom from expensive tools but in my honest opinion it's to hard to do bare metal & realtime. A particular problem I have is the delay between an occupied signal and opening the stop relay and how far the train will have traveled

The relays do traction switching of one sort or another as that requires dry contacts, each section has a reversing & stop relay, that's 12, then another 6 for point frogs, a couple for reversing the loop rail polarity and two spares for luck!

Interlocking will be entirely in the central controller software, especially hair raising as everything is bi-directional (head on collision potential)

PIC and similar are way better for "I said do this" levels of control.

there is a lot of switching stuff where relays are the simplest solution, yes solid state is possible but a relay works, and best of all when it changes you can hear it.

best way to handle that sort of delay is really to do what the 1:1 scale version does, look at likely line speed and allow the blocks to be longer than otherwise with overrun space.

have relays handling frog polarity here, pondering interlock with the block system so a wrong way movement, via a second relay, can depower the frog to make it an insufrog to prevent the short circuit.

[fourtytwo]

SUCCESS!! After what seems like weeks of hardware & software de-bugging/improvements I have two trains running around the single line circuit (in the same direction) controlled by the block signalling and pausing as required to avoid collisions, it's been running an hour without errors so I call that a pass A signal block uses this much code for one direction & no junctions

EDIT copied code replaced by screenshot so that indentation is visible
Each block is individually coded as they are all different insofar as junctions etc. The state machine may seem over the top but things were getting in a right hair brained mess trying to manage without one, I think it makes the code a bit more readable rather than clever tricks and I have no need to save code space.

Now back to some track laying beginning with the middle loop to make life more interesting, still no control panel mostly because I cannot think of the totality of what will be on it and how big to make it, geographic or not etc zzzzz

[fourtytwo]

Talking of birds nests here is one of mine under the central controller wiring the processor chip to some of it's peripherals (an 8x4 mux and 16 relay drivers).
This week has almost entirely taken up with dasterdly software tool chain problems, you know what you want to do but the tools are either broken or defective and you waste your time trying to find workarounds. Here I am dealing with two generations of processor, two generations of software tools and an obsolete hardware debugger! Not only do you have to find the right combination for each processor but everytime you need to move to the other one everything is upsticks again! In desperation I have had to resurrect a 32 bit laptop capable of running winXP for one of the combinations.

One fine day I will find utopia in this department being a processor suitable for bare metal realtime programming with an inexpensive WORKING tool chain (dream on)

In the meantime I see lots of interesting progress on other layouts whilst mine languishes in limbo