Jump to content
 

Dongits Model Railway -- Three layers around the Study


Bloodnok
 Share

Recommended Posts

On 18/06/2020 at 18:24, Bloodnok said:

Well, unless I find the one that went 'ping'...


Unbelievably, three and a half months later, I did indeed find "the one that went 'ping'". It was hiding in the crease between the carpet and the wall. I'm not sure how it avoided all the vacuuming of sawdust and so on that's happened in the mean time. It must have been well stuck in that gap...

Edited by Bloodnok
Link to post
Share on other sites

  • 4 months later...

Well, that went quiet.

What did you all miss in the five months since I last posted?
Nothing really.

Last thing I did was solder up an ABC4 module, and a CANVOUT to control it. The ABC4 was complete, but never made it as far as the layout. The CANVOUT was ready for initial testing, then fitting the chips to the sockets. I ... just never went back the next day to test it and put the chips in. It just sat there. It's still sitting there waiting.

I did continue to pick up a few bits of rolling stock, and I built up a backlog of things that had been chipped or re-chipped but never configured.

Four ish months later and things changed. I got a new work PC, and it dawned on me that the "temporary table for three weeks to flatten the curve" had been in place for a year. So instead of perpetuating that, I should use my main desk. That triggered a full desk rebuild, which triggered both a study reorganisation (as everything had to move off that desk), and a cable audit (as nothing needed the same length cable as it did before). A week after the desk rebuild was complete, and I had the perfect old PC in the right place to dedicate as a layout controller, and the correct USB cable to avoid unreliable extensions, dangling cables and blocking the doorway, all of which the previous solution suffered from.

During last week I got the PC running right, with all the software installed necessary to run the layout, and the old layout configuration ported across.

This weekend, I've moved everything off the layout, cleared the backlog of DCC chips waiting attention, fixed a cable that got pulled out from where it should be, then played with by a cat, and replaced a servo that wasn't responding properly.

And ... it works again. I still need to work out how to tell JMRI to turn on and off the ABC blocks, and that CANVOUT still needs looking at, but the NX panel works on the new PC, and I ran some trains today. I didn't lose connection to the layout once, which is refreshing. The lights on my Bachmann 108 and Hornby DVT are finally correct and controllable. And the top goes back on my Bachmann 08 without me worried I'm stressing the plastic to make it all fit.

And still have one more day off tomorrow. I wonder what I can get done?

Edited by Bloodnok
  • Like 2
  • Craftsmanship/clever 1
Link to post
Share on other sites

I've just had first life out of the ABC controllers.

Like all good bugs, there were two issues sitting right next to each other. One was the prototype circuit board for the switchable ABC converters had a common 0v when the CANVOUT was expecting a common 12v. But they are only relays, so swapping the protection diodes around and connecting them backwards solved that. The second was that my CANVOUT node had forgotten everything I'd ever told it, including it's own hardware address.

Having fixed both of those issues, I can now switch on and off ABC braking using the configuration program.

Now I just need to teach JMRI when it's appropriate to switch it on and off.

Link to post
Share on other sites

On Tuesday, I got the first two switchable ABC sections live. Wednesday saw the first one programmed into JMRI as a "light". Once the hardware worked, this turned out to be relatively straightforward -- the biggest problem was figuring out exactly what type of device to call it.

Controlling it with JMRI Logix is a bit more of a mind-bender though.

JMRI Logix does not appear to allow a distinction between a "when" and an "if".

I had initially sketched out this logic:
When the berth track circuit becomes occupied, if the signal is at danger, turn on the ABC module.
When the signal clears, turn off the ABC module.

However, because "when" and "if" are the same operation to JMRI, I also got it turning the ABC back on when the signal returned to danger with the berth track circuit still occupied -- e.g. when the first wheel of the train goes past the signal into the next block.

Not really a problem if the loco is leading, as it will read normal DCC through the front axle with enough reliability it will keep going. But the moment you have a powered vehicle in any other position, it all goes very wrong and repeated re-stopping is common. 

An alternative solution was found - checking if any valid routes were set from the signal, rather than whether it was at danger. This required a lot more clicking, but should be a lot more reliable. I obviously can't use routes set from a signal when that signal is an automatic though -- that is a problem remaining to be solved.

But ... it works. I've only got one functioning section so far, but I can run a train into that section, have it switch on the ABC when the berth track circuit gets occupied, have the train stop (much tweaking of braking distances will be required to get them to reliably stop in time), then I give it a path into the storage yard and it starts to pull away again by itself once everything is ready.

This evening I modified the remaining ABC modules to have a common 12v, not a common 0v. And I finished testing the second CANVOUT, which has passed it's first power on test.

Tomorrow's job is to mount it in the concentration point and configure it. That would give two consecutive approach blocks and two yard tracks, plus the yard throat. That will make a much more thorough test setup...

Link to post
Share on other sites

It's ... alive.

I have successfully run two semi-autonomous trains. They followed each other down the approach road, queued up behind each other, and waited for a path. I pathed one into a storage road with the NX panel. It pulled away by itself, and parked at the correct location. The other shuffled itself up to the yard entrance to await another storage road. I assigned it a different storage road, and it drew forward and parked there too. I can set roads out of the yard too, and the correct train pulls away.

I say semi-autonomous, as with the notable exception that I cannot force a train to drive beyond it's authorised limit, I am still in full control.

Tomorrow I will attempt to beat the nearly 2GB of phone video into watchable shape and prove it.

  • Like 2
Link to post
Share on other sites

20 minutes ago, 5BarVT said:

Nice.

Are the short sections of wooden sleepers significant in respect of the ends of the blocks?  If so, very clever.

Paul.


Yes. Initially by accident, but I liked the effect so I (mostly) stuck with it.

Why by accident? I had short bits of reclaimed wooden sleeper track from a yard, long bits of reclaimed concrete sleeper track from a mainline, and a fresh box of concrete track (which was all that was available once lockdown had started).

I initially used the shorter pieces where I needed shorter pieces -- to cover overlaps and places where the front or rear of a train should be detected as blocking an adjacent line.

I also needed a fresh, uncut length of track as the berth track, so these all came from the box of new concrete track. That way all the braking distances are the same, and if a loco stops in one of them, it will stop in all of them.

So for both yard throats and the junction at the bottom of the ramp, there is a track section break where the sleepers change colour.
For the yard approach line (that runs around the rear of the yard) that is not the case -- I was running out of track, and used whatever I had available to make it fit. One of the stretches of wooden sleepered track does line up just after the end of a berth track, but the other doesn't -- it's in the middle of a block. And I'm not tearing that area up now just to line up a particular colour of sleepers.

  • Like 1
Link to post
Share on other sites

I dropped into the local wood yard, picked up a piece of 2" by 1" prepared, and just like that we're back in baseboard building mode.ramp_top.jpg.d98f0ef04460e56f5041e593dea8e79d.jpg

This marks the first place where there is something above a storage line. I now need to get all my AC electrics out and, errr, extensively test the clearance at this location...

Edited by Bloodnok
Picture Fixin'
  • Like 2
  • Friendly/supportive 1
Link to post
Share on other sites

I've started tuning the ABC stopping distance and scale top speed of my locos.

I very rapidly learned that you tune for speed *first*, and then tune for stopping distance afterwards. Because tuning the top speed will dramatically affect the braking distance, and you'll need to retune that from scratch afterwards.

I measured out a distance to time locos over (in my case it's a 3m straight run) and worked out some timings over this distance:
 

Real speed:	Scale speed:	Time to cover 3m:
155mph		0.909m/s	 3.30s
140mph		0.821m/s	 3.65s
125mph		0.733m/s	 4.09s
110mph		0.645m/s	 4.65s
100mph		0.587m/s	 5.11s
95mph		0.557m/s	 5.38s
90mph		0.528m/s	 5.68s
87mph		0.510m/s	 5.88s
80mph		0.469m/s	 6.39s
75mph		0.440m/s	 6.82s
70mph		0.411m/s	 7.31s
60mph		0.352m/s	 8.52s
40mph		0.235m/s	12.79s
27.5mph		0.161m/s	18.60s
20mph		0.117m/s	25.57s
15mph		0.088m/s	34.10s


How are these worked out?

* Convert mph to m/s -- multiply by 1609 and divide by 3600 (that's "metres per mile" and "seconds per hour").
* Divide that by the scale (76.2 for 4mm/ft)
* Divide the distance between your markers (in metres) by the scale speed you just worked out.

A human's ability to start and stop a stopwatch accurately at each end of a run is probably around +-0.3s. The resulting tuned speed error can be up to 20% at the fastest end of the above chart, around 10% in the middle where most of my locos should be, and about 2% at the bottom. This can be reduced by using longer distances and averaging multiple runs.

I'm not looking for perfect accuracy here so if I'm off by 10%, it doesn't matter. Untuned, I have several locos that will rocket off at a scale Mach 1 and happily tip themselves off at the next tight curve, and this needs fixing. But given they need tweaking anyway, I may as well try to get it reasonably "right".
I also want to show up the distinction between locos in freight and passenger service, where a freight loco is not capable of the same top speed as a passenger loco.

I'm also attempting to tune for linear speed -- to try to get 50% on the controller to be 50% of the max service speed. This will help getting mixed traffic locos running at about the right speed when hauling slower services.

Edited by Bloodnok
  • Like 1
  • Interesting/Thought-provoking 1
Link to post
Share on other sites

  • RMweb Gold

If you’re using JMRI, you should be able to get timings off that.  Using loconet monitor you can track occupancy to a much better accuracy.  It’s a couple of years since I have done it, but if it’s not obvious, squawk, and I’ll try and remind myself on my set up.

Paul.

 

  • Like 1
Link to post
Share on other sites

1 hour ago, 5BarVT said:

If you’re using JMRI, you should be able to get timings off that.  Using loconet monitor you can track occupancy to a much better accuracy.  It’s a couple of years since I have done it, but if it’s not obvious, squawk, and I’ll try and remind myself on my set up.


Yes, I'm using JMRI. I'm using MERG CBus rather than Loconet as the bus, but it still ends up in JMRI once it gets to the PC.

I will look into speed monitoring via the PC, it would be interesting to show the speed of each train. But what I've described above works, and will also work for anyone, even if they aren't using layout feedback.

  • Like 1
Link to post
Share on other sites

One step forwards, one step ... sideways.

There I was tuning loco speeds and ABC stopping distances.

A Bachmann 25 didn't seem to be able to get to it's scale 90mph at all -- it topped out at ~75mph ish. I'm not sure if there's a mechanical issue here or if that's all it is geared to do.

A re-motored Lima class 73 with a CD drive motor was previously capable of well north of 200mph, and has been tamed to run at a more appropriate 90mph.

A Bachmann 108 was previously capable of about 140mph, and will now barely crack 70mph.

One Heljan 47 was tamed from a top speed significantly in excess of 150mph to a shade below 100mph.

Then came the other Heljan 47. I'll confirm the same original top speed, then put the same values in -- hopefully it'll be close enough that's fine.

However, it didn't want to do that. It repeatedly had issues at one specific location on the track. There would be a grinding noise along with a slowdown in progress. Maybe wheelslip or flange squeal? When run slowly over that particular location, it would even stall. Run faster, it often popped a wheel off the track. Once when running rapidly against the normal flow of traffic, it jumped off the track entirely, left the baseboard, and dropped 6" onto a pile of stock boxes. (Thankfully no deadly 3' drop onto carpet though).

So ... off we go to troubleshoot. What's different between the two locos? The problem one has the old, narrower bogie sideframes (which I wish I'd noticed before I bought it, as now I need to figure out how to get hold of a replacement wider set -- the narrow ones really do make the loco look far too fat). I wondered if these were causing a lack of horizontal movement in the floating centre axle, so I pulled them off. No improvement.

However, crawling over that location really slowly, I did see a brief blue flash and heard a peep from the booster. Definitely a short.

The location was an insulating rail joint, and it was naturally right round the back of the yard behind six rows of stock. So I moved everything out of the way and had a closer look at it. The joint was proud -- standing a mm or two above the rest of the track where the thick insulating rail joiners were sitting on top of the sleepers. The bogies were tipping inwards slightly as the loco high-centred, and the short was happening where the innermost wheels of the bogie touched the underside of the chassis block. There was a few marks in the paint where the wheels had made contact, and if both wheels got to a bit that had no paint, that'd be why there was a short.

Figuring the sleepers plus the insulating rail joiner was pushing the rail up, I cut the two closest away and tried it again. No improvement.

Looking at the other 47, there were less marks, but they were still there in the same location.

Looking at how much angle the bogie needs to high-centre like that though made me realise that something bigger must be wrong. Not just the joint.

Then I noticed that the joint just so happened to be right over a where the baseboard frame ran under the board. Looking along the track, it did seem that perhaps there was more at play than just the joint. Out came the spirit level. Sure enough, the track was rising and falling some 3-5mm in a wave pattern, with this joint being a centre high-point. A clamp was used to lift the baseboard and make the approach side flat. And the loco ran fine, no longer having any issues with this joint.

Out came the 2x1, and a bit of new bracing was made for these two bays. Screwed in place, the board no longer sags by a few mm, and the locos run through again smoothly.

  • Like 1
  • Informative/Useful 1
Link to post
Share on other sites

I am now half-way through tuning the speed and ABC braking distance of my chipped fleet. The clearance at the first overhang point is about 20mm more than that below a Dapol catenary mast, and at maximum height all three of my AC electrics clear without issue. I'm concerned about the Hornby 87 though, as it's pantograph appears very fragile and difficult to pose at a consistent height.

BTW, does anyone have any experience with Bachmann Mk1 Pullmans? I have had one sitting on the layout while testing locos, and I noticed today it had got warm, which I wasn't expecting.

Link to post
Share on other sites

  • 2 weeks later...

All of the fleet are now tuned to have an appropriate maximum speed, a linear throttle response, and an accurate and reliable ABC stop within a 1 yard braking zone from a range of different speeds.

I really do like Zimo decoders. Their ABC braking looks very natural. From low speed, it runs at the entry speed until braking is necessary, then brakes to a stand. I'm not so happy with some other decoders, which will slow to a crawl and then crawl up to the stopping point. However, only my 08 actually has one of these in it (because a Zimo wouldn't fit), and I don't plan to run the 08 on the mainline all that much, so it shouldn't be an issue.

I turned down most of the momentum settings to make top speed tuning easier -- I'll re-tune these when I get more layout built and get a feel for how I want the trains to behave on scene.

  • Like 2
Link to post
Share on other sites

  • 3 weeks later...

588911009_lift-outbridge.jpg.594cd1de35032ea0696dccf0c0b5b283.jpg

The plan was always to dress up the lift-out section as a bridge. I wanted a truss for the visual spectacle of the train going through a bridge (rather than merely over a bridge).

I bought these laser cut bridge sides as they are the same length as the planned lift-out section.

However, now they have arrived, I'm a little concerned about how tall they are compared to trains. Even sat here level with the bottom of the lift-out board, the train is well below half the height of the truss. Even with clearance for catenary and assuming there will be a roof connecting the sides together, it still feels like the sides are significantly too tall. This span scales up to a 61m real-life distance -- not a massive span requiring such a tall truss, surely?

  • Interesting/Thought-provoking 1
Link to post
Share on other sites

The GCR bridge over the LNWR at Rugby was at least as long as your gap https://www.warwickshirerailways.com/lms/lnwrrm3047.htm and is not much taller than the train passing over.  Sadly, it looks like the bridge you've got is for 7mm or larger. Perhaps something like this would be more suitable. (No connection - just random Google search) https://lazzylozlasercutter.co.uk/product/large-model-railway-oo-gauge-bridge/ 

Link to post
Share on other sites

On 01/06/2021 at 06:38, Nick Holliday said:

The GCR bridge over the LNWR at Rugby was at least as long as your gap https://www.warwickshirerailways.com/lms/lnwrrm3047.htm and is not much taller than the train passing over.

 

That's very much the kind of appearance I wanted -- albeit I have four parallel tracks here rather than two.

 

Quote

Sadly, it looks like the bridge you've got is for 7mm or larger.

 

I suspect it's not designed for any scale -- the creator got a file and just uniform scaled it to a given length.
 

Quote

Perhaps something like this would be more suitable. (No connection - just random Google search) https://lazzylozlasercutter.co.uk/product/large-model-railway-oo-gauge-bridge/ 


Do they make one four tracks wide?

(I'm wondering if I should have got this one: https://www.ebay.co.uk/itm/122492762700 ... although I'd have to replace the laser cut two-part floor in that one to use it as an actual lift-out section).

Edited by Bloodnok
Link to post
Share on other sites

  • 2 weeks later...

Time to make some of this hidden storage ... hidden.

moar_baseboard.jpg.0dc7bf4f7628b5e9a42045d37ba69c1c.jpg

This is "plonked" at the moment, looking for where it doesn't line up.

Job one -- take the nice straight edge of the ply, deliberately lined up on the edge of the sheet because I wanted it nice and straight, and put a really subtle but significant curve in it, because even in a new-build house the wall isn't precisely dead straight.

Edited by Bloodnok
Picture Fixin'
  • Like 1
Link to post
Share on other sites

  • 2 months later...

This bridge is a ... vexing problem.

To progress with building the layout, I need to build the structure that will support the end of the bridge / lift-out section. This means it needs to be finalised.

However, I want the bridge to look right. I don't want to settle for something I'm not going to be happy with long term purely to hurry up construction now. And I'm zero for two in trying to get a bridge that looks right so far, so I clearly need to do something different.

So ... I've been trying to work out what I actually need to make it look right.

I started by trying to find some prototype photos of something similar to what I'm trying to achieve. This is proving *very* difficult, which is cluing me in that I'm asking for something wrong, and it probably won't ever look right unless I modify something on my side to fit  with how a real bridge works.

Hungerford bridge used to have four tracks (now reduced to three and the end of a platform) including junctions within it's northernmost section. But this is a fairly unique case, as it forms the station throat for Charing Cross station.

There are many four-track truss bridges of a similar span length to what I'm trying to achieve over the Mersey and over the Manchester Ship Canal. But none of them have all four tracks inside a single pair of trusses. All of them have at least one central truss. That  ... breaks my track plan.

Having concluded that I can't really make something that looks right without a centre truss, I started doodling track.

What would it take to move the junction north of the bridge? Answer: Get some Tillig curved points (tighter radius than Peco streamline). Build three custom curved diamonds. And then hide all of it in tunnels.

That doesn't seem reasonable to me. If I am going to the effort to build three custom curved diamonds, I ... kinda want to show them off. I think this one is a non-starter.

What about a single lead junction? That fits with the initial crossover on the bridge, the actual single lead just north of the bridge, and a second crossover either from Tillig curved points or using a single slip in the tunnel. The major disadvantage here is I lose the double junction, which was an intentional feature of the original plan, something I wanted to include. I'm not a fan of this option.

A third option is to replace the four track bridge with a two track bridge, and have a double junction at either end. I could pull this off with peco diamonds, but that requires a  really sharp (second radius) track with no transitions between both diamonds. Okay, the tight curve is in a tunnel, but ... it's still mainline, and I'd like to at least have transitions on such a tight track in mainline use. It would fit a lot better with a couple of curved diamonds. On the plus side though, one of those would be visible this time.

The fourth option is to move the double junction south of the bridge and around the corner. This demolishes one end of the engineers yard, and I lose my yard lead as that area is replaced with junction. However, it actually fits reasonably well both horizontally and vertically, with the gradients still sane. And there are no junctions on the bridge, leaving me free to have a much more conventional bridge.

Now I need to ruminate on the yard a bit to see if what I've drawn there is insane or workable, as I've been primarily concentrating on the mainlines, and not really on the yard...

T2F_Mk6h_Main.png.45908922d9e654e67f184c83e811fd62.png

Edited by Bloodnok
Picture Fixin'
Link to post
Share on other sites

  • 1 month later...

Shapes like this are complicated when nothing is flat or square. The trackbed against the back wall is rising at circa 1 in 50 here, and the trackbed towards the camera is rising too. This is a triangular junction, so the right hand tracks have to bottom out and start rising up again as they turn...

 

1671688679_2021-10-2520_50_18.jpg.6afe93bf5485ff298543acc8eca7c3b3.jpg

Edited by Bloodnok
Picture Fixin'
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...