I have thanked Philip for his kind offer but was lucky to be offered a pair from a local source. The waiting time at Ultrascale is only two months for this item I'm glad to say.
I have jumped ahead of sequence on this thread with all the talk about valve gear and drain cocks. Nearly three weeks ago now, the loco was tested on two layouts. First, on Calderside Exchange Sidings, the part of the WS4G layout Calderside that is currently erected in our new clubroom. (There isn't space for all 40 feet of it!). The loco derailed at just one place, a turnout set for the straight route. The tender derailed at a couple of spots running tender first.
Perth2018 - DAS - IMG_2817-1-1-1-1.jpg
As my 0-6-0 Barclay Tank runs through all the layout without derailing, as do other locos and tenders, I wasn't satisfied with this, even if the derailments could be said to be caused by track faults. (We found the turnout straight route was undergauge through the switch - a very typical problem.) I really want to prove to myself it is possible to get a 2-6-0 to hold the rails as reliably as an 0-6-0.
20190312_201846.jpg
20190312_201826.jpg
The second layout visited was John Stocks' layout Kettlewell. Firstly this was for structure gauging tests. Despite the large cylinders there were no issues.
John constantly looks out for track faults. Where any derailment takes place he makes checks and replaces the track if there is any fault. The Crab ran almost perfectly here. The only problem was that the front wheel sanding gear was so close to the rails it caused a short where inevitably the track levels are not absolutely perfectly matched on the entrance to the turntable/fiddleyard.
That's an illustration to me of the sort of track imperfection a model has to be able to cope with and where faithful realism has to be compromised. No other stock of mine or his has the same problem. So the sanding gear was modified to run above the brake cross rod rather than below it as per the prototype. It's completely unnoticeable to all intents and purposes.
From the Calderside tests I concluded the tender suspension would be more reliable with a fairly simple modification, to make it exactly conform to the 3 legged stool principle. So the springs on the front wheels were removed and a central resting place made for the axle (which runs in a tube). Getting this the correct height was the only difficulty. The middle axle spring remains.
20190302_102628.jpg
The loco derailment was caused by the pony wheel derailing. Only by running dead slow would it stay on the rails at this (undergauge) spot. People have varying opinions about such derailments and many accept them as just a normal fact of P4 life. Roger Sanders recent articles would suggest the problem is the Back to Back setting. My hypothesis remained that there was, despite the modifications I had already made, still insufficient weight on the front pony wheel, and that this is a typical problem with a front pony or bogie.
This loco, as I've made it, has a central roller above the pony bearing a proportion of the loco weight, and as per Comet design the front driving wheels are sprung with individual coil springs. These are modified to exert less force than the Comet design. (See page 4 31st December '17) With still not quite perfect trackholding it was just a matter of "tuning" the front driving wheel spring suspension to take still less loco weight, which results in more weight resting on the pony.
So the front wheel suspension was again modified by yet further raising the top of the spring fixing place in the frames by 1mm, thus reducing the spring thrust and (therefore) increasing the weight born by the pony.
20190228_144014.jpg
This resulted (on my bendy bumpy test track) in the front driving wheel not staying fully down on the rails at all times. So this was easily "retuned" by placing a 14BA washer above the spring, roughly halving the reduction of spring force I had just made.
With the front wheels out of the frames, it was opportune to reconsider the matter of the hornblocks I visited earlier - does the shape of the slot influence how well they work? With a reduction of spring force it is that much more important they work properly and freely. What I observed previously was that the front wheel downward movement was better one side than the other, and I'd found that it wasn't caused by any difference of spring force.
(This refers to my post on May 4 2018 on Page 6)
To briefly recap, the original hornblock design (as per Mike Sharman) had slots that were slightly V shaped. This enables the wheelset to not only move up and down as a pair but also to tilt where a single wheel needs to move up or down more than its opposite partner. MJT hornblocks I bought years ago are made to this design. But recent acquisitions including those with this kit are not. (Mike Sharman's Flexichas book starts with a whole spiel about the importance of the axle being able to tilt, so I think it must be fundamental to his concept.)
I could now see that, as I had guessed, I had modified one hornblock but not the other. So that fettling was done. Very carefully so as not to deepen the grooves. I don't have an ideal file which would probably be a very small ended curved triangular one with a safe edge. I put these fettled wheelsets in the frame and with Brasso tried to replicate the type of movement I was seeking to further wear the slots to shape.
20190228_204022.jpg
The result was what I'd hoped, the front wheels now actively descend into the track dip of my test track with similar vigour, in contrast to the video I posted earlier.
The video posted earlier this week (which was chiefly to show the working valve gear, set to about half way between full and midgear), shows that the loco and tender can run at full speed with no problems. The test track is quite challenging as I showed earlier in the thread but testing on the layouts demonstrates that - for me - there are always further hurdles to be overcome and modifications made before the model can be said to work properly. I prefer not to paint it before reaching that point because of all the handling needed in making modifications. But to go back to Philip's point of painting it as you go, if I was making another Crab from the same kit I would probably copy exactly what I've eventually found does work, so would be able to paint it at each stage.
Using 24hr Araldite footsteps have at last been added, using those supplied with the Bachmann model and located as per the locating holes and pegs. The pony wheel is more or less up against them on my test track 4ft curve.
20190303_162444-1.jpg
If I were to add guide springs there might be 1mm clearance enabling a sharper minimum radius - if the guide springs worked! - but I'm content with an already probably sharper than prototypical 4ft minimum. (I wonder how to find out what the actual minimum radius figure was?) So for Gavin's interest here is a picture of the tightest curve on Calderside. I don't know the radius but it's pretty tight. The Crab easily went round it but that was before footsteps were added.
20190312_213540.jpg
I might add that I could probably couple the tender a bit closer to the loco for a 4ft minimum. But the 243 Loctite seems to have locked up the EMGS adjustment screw coupling very firmly indeed and I might do damage to its fixing in the loco chassis if I try to use force here. Unless anyone knows how to unlock Loctite 243...?! Not sure it's worth the bother for the very minor adjustment that could be made.
The loco was tested a few days ago on both layouts again. The shorting problem on Kettlewell was solved, and the loco and tender no longer derailed at any place on the Calderside exchange sidings even at full speed which I estimate a scale 45 - 50mph, in either direction. Sadly I could not take a video to prove it here but will do sometime. Interestingly the sharpest curve was still no problem even with the footsteps now added so it may be less sharp than I surmised.
So basically I'm fairly happy to say it works and it's now ready for strip down clean up and painting with just the return crank issue to complete and cylinder stuff to add first.
My conclusion, with due apologies to Bill (page 3) and Will (page 6), is that the pony wheel does need to have substantial weight on it, much more than its own weight, to reliably stay on the track, however that weight is applied. (Note the hefty looking springs just visible on the picture of David Franks' Crab (Page 3 October 14th).) And probably much more as a proportion of the front driving wheel's weight, or total loco weight, is needed on the pony than the prototype, as the issues are completely different. And (with due apologies to Roger) that Back to Back exact accuracy of 17.75 is irrelevant. However obviously the downside is a loss of adhesive weight. To test how much loss, the next step must be to apply for a haulage capacity trial on Grayrigg to see how it compares...Allan!?