Help with 'Speedy' valve gear and other issues...

[suctionhose]

Originally I thought just CNC "machining". As Julian says there's plenty other work besides machining.

But now I see CNC punched / folded cabs and tenders, wire cut valve gear, diecastings, lostwax castings, polyurethane castings, laser cutting, water jetting, 3D printing - haven't seen any instant boilers yet (although it does seem quite common to outsource them over there..)

My vision of the future, which isn't all that reliable I must admit, is models becoming collections of ready made parts made possible by low cost, short run production methods. Just order on line and superglue it together! (which is rather unimaginative)

There was an article in a local paper the other week about 3D printed titanium horse shoes. Just scan, pay and wait by the mailbox! My god, old blacksmiths and farriers would be turning in their graves!!!

Julian, let's hope they don't replace cereal packets with something you can't write on...

[Roger]

You're so right suctionhose, if there's a way of bolting servos and a computer onto it, someone will. I once saw a CNC exhaust pipe forming machine and that was amazing in action. The pick and place machines used in PCB manufacture take your breath away, as do embroidary and knitting machines. It's a long way from Jacquard looms with punched cards, but the programmable nature of these things is what everyone has striven for.

I can certainly see the 3D printing of lost wax casting moulds being a boon to any model engineer, particularly for our friends who build the model gas turbines. Googele GTBA to see what they get up to.

[Deleted]

The Glasgow Thai Boxing Association --- and "Help with Speedy valve gear" make odd bedfellows, methinks ??....(Chuckle, LoL....A deliberate GOOGLE mis-quote for the sake of humour )......ie}---We appear to be straying from the core thread, even though it is yours STRIPLAR and does include "other issues"....So, how far have you now got with the research etc. and could you post some images to view ------- EDIT}---Never mind the Jacquard loom }---- www.youtube.com/watch?v=f1Zzj9ZBYmQ ---- was not the Abbacus }--- en.wikipedia.org/wiki/Abacus ----- a form of computing ??

[Roger]

Nov 11, 2013 12:55:59 GMT @4930hagley said:

The Glasgow Thai Boxing Association --- and "Help with Speedy valve gear" make odd bedfellows, methinks ??....(Chuckle, LoL....A deliberate GOOGLE mis-quote for the sake of humour )......ie}---We appear to be straying from the core thread, even though it is yours STRIPLAR and does include "other issues"....So, how far have you now got with the research etc. and could you post some images to view ------- EDIT}---Never mind the Jacquard loom }---- www.youtube.com/watch?v=f1Zzj9ZBYmQ ---- was not the Abbacus }--- en.wikipedia.org/wiki/Abacus ----- a form of computing ??

Nice one, I'll have to check them out. And yes, most definitely, an abacus, punched cards and changing screw cutting gears are all computing and programming of sorts.

But back to the thread if I can remember what it was about...  ah yes, valve gear. Thanks to Don Ashton and others for much valuable technical information and opinion, I now have a fully working 3D solid model of the valve gear to Don Ashtons line drawing dimensions. It's been quite a challenge to tweak the various components that get very close indeed at half and full gear. The expansion link superficially looks unchanged, but it's narrower front to back, has the channel set closer to the back and the pin in the die block to the rear too. Each small change along with shaving the end off the lifting arm has allowed it all to clear (just) while looking innocently as if it was no problem at all.
I've added 1/8" to the frames between the cylinder and the front axle to make room for the brake hanger and that all looks spot on now.
Incidentally, the full valve gear can now be moved to any position by moving the lever with the mouse, and the wheels turned showing the whole gear in motion. It can be stopped in any position and drawings automatically created at that position so clearances can be measured. It's snug, but it all clears fine and the dimensions are true to Don's drawing. Two weeks of headache but worth every minute.
By all means let me have your email address and I'll send you a 3D model that you can manipulate and crawl all over. I've attached it here so you may be able to grab it and do that anyway. Open it with Adobe Reader X and click where it says 'click to activate'. It will take a moment to get ready, but then you can move it all around.

So the quest for the best valve gear is over and I'm currently agonising over the boiler. Watch this space....

Attachments:

Chassis.pdf (2.02 MB)

[Deleted]

Nice model Roger

Like you, I build in 3D first to find the errors before cutting metal. Must have half a dozen designs on the go at the moment. It's almost as interesting as building the real thing !

John

[Roger]

Thanks John, it looks like there's no chance of you becoming bored either then. What 3D modelling package do you have? I use what was originally called Alibre Design but is not rebranded Geomagic Design. It takes a bit of getting used to, especially when trying to get the best out of the CAM module. I've spent a fortune on it, way more than most hobbyists would pay even if they could afford it. I justify it for the business since it's allowed me to take on work that wouldn't have otherwise been economic or even possible. It's paid for itself time many times over.

[Deleted]

I use Alibre PE (now rebranded as Geomagic Cubify). I downloaded the trial version which was Alibre Professional some time ago and gave that a go but couldn't afford to buy the licensed version so went with the basic version. As you say, a steep learning curve as I had only ever used Autocad 2D but well worth the effort.

The first loco I 'built' was a 2½" gauge version of Keith Wilson's Ariel (rebuilt Merchant Navy) and that was a real babtism of fire! A very complex loco to start with but I learnt a lot doing it. Using the package is now almost second nature and I can build parts and assemblies very quickly now.



One of the projects is a much modified Simplex with a completely new boiler with much larger grate area to test Jim Ewins's theories. At the moment it's just a rolling chassis but will be getting back to it as soon as one or two 'foreigners' are out of the way.



Sorry if this is getting a bit off topic.

John

[suctionhose]

What interests me here is you can "build" a perfectly accurate machine in 3D but can you "build" the whole locomotive to that accuracy?

Sure you can machine parts accurately, although kit builders will tell you nothing actually fits and lots of hand work is necessary. Either that or fits are quite loose... Have seen simple ready-made locos, made by the methods we are discussing and as finished assemblies, they are quite poor.

In industry, repeatable accuracy requires a great deal more than a wish for it to be so.

Not trying to be provocative here, but I am interested to know how you cope with the realities of distortion due heat or stress, boilers that aren't precise to a thou, cumulative errors, etc. I.e. making the next piece fit the last.

Two lessons in life I have had:
1. Detailed long range planning based upon completing first step perfectly is unrealistic.
2. Each next step is defined by the actual outcome of the one before it and must be adjusted accordingly.

In other words, if the building the entire model is contingent upon zero errors throughout then how is this possible to achieve outside of your computer?

Intended for discussion not argument...

[ettingtonliam]

I know very little about CAD,(just struggling to learn the basics with Draftsight), but when we had mass manufacturing industries such as car manufacture, I understand that tolerancing (setting the tolerances for individual parts and ensuring that each part at its tolerance limits still fitted to its neighbour at the limits of its own tolerance) was an art in itself, and quite time consuming. Can any of these programmes do that automatically?

richard

[Deleted]

Nov 12, 2013 3:35:00 GMT suctionhose said:

What interests me here is you can "build" a perfectly accurate machine in 3D but can you "build" the whole locomotive to that accuracy?

From my point of view I'm using the 3D CAD just to prove the design and check the original drawings for errors etc. making corrections where necessary. I'm not using it to have parts made for me (can't afford that!. I much prefer to do as much as I can myself anyway) so I'm still going to use the finished drawings to build in the 'traditional' way. At least I know that the parts will fit together then and I won't be wasting precious time and money making parts that won't. Life's getting too short!

I see no reason why you could not have a complete set of parts made and just bolt them together. With modern machining methods such as CNC, laser cutting, etc. this should be perfectly possible. Polly seem to manage it ok. The problem with such as Winson/Modelworks kits is that the designs were never proven and the parts poorly produced.

Richard - I don't think there is the facility to introduce tolerances into the parts, certainly not in Alibre/Geomagic. Maybe the more expensive packages such as Solidworks have this?

John

[Roger]

Nov 12, 2013 3:35:00 GMT suctionhose said:

What interests me here is you can "build" a perfectly accurate machine in 3D but can you "build" the whole locomotive to that accuracy?

Sure you can machine parts accurately, although kit builders will tell you nothing actually fits and lots of hand work is necessary. Either that or fits are quite loose... Have seen simple ready-made locos, made by the methods we are discussing and as finished assemblies, they are quite poor.

In industry, repeatable accuracy requires a great deal more than a wish for it to be so.

Not trying to be provocative here, but I am interested to know how you cope with the realities of distortion due heat or stress, boilers that aren't precise to a thou, cumulative errors, etc. I.e. making the next piece fit the last.

Two lessons in life I have had:
1. Detailed long range planning based upon completing first step perfectly is unrealistic.
2. Each next step is defined by the actual outcome of the one before it and must be adjusted accordingly.

In other words, if the building the entire model is contingent upon zero errors throughout then how is this possible to achieve outside of your computer?

Intended for discussion not argument...

I take this entirely in the spirit it was written in and I fully understand your question.

Firstly, it's a mistake to look at this as a lesson in accuracy and precision, it's not about that. We all use calculators that give us umpteen places of decimals in the answer and we're happy to round them to the nearest centimeter if we're building a wall, and to the nearest micron if we're grinding an Air Bearing. That's what makes us Engineers, it's about designing and laying out our creations that just happen to be near perfect in their geometry and dimensional accuracy. Kitchen designers use CAD but you won't find them with a micrometer in their hand.

Secondly, the hand drawn engineering drawings of Speedy are to the same precision as my model. If a size says 2" then it's 2" and not anything else. Numbers in their own right are absolutely precise. There is nothing on the drawing to state that there's an overall tolerance of +/- 0.01" so the pedant would say that it's all to be made to perfection, just like you might suggest of CAD. Clearly both viewpoints are nonsense.

The only difference between two sets of otherwise identical drawings of Speedy, the ones I bought and the ones I could output from a 3D model is that the dimensional errors wouldn't be there in the CAD ones, and there wouldn't be missing lines or misleading radii shown for the flanges.

So that brings us to the question of accuracy and tolerance. The issue with the drawings of 'Speedy', and probably all of those that you'll ever see for model locomotives, is that they don't conform to rigorous Engineering standards, and are therefore inherently vague. There's no indication that the length of the boiler is to be any more tightly controlled than the fit of the bearings. Some efforts have been made to specify a press fit or things to be a 1/4" bare, but these won't cut it when you're trying to define the fit of a piston in a modern car engine.

It doesn't matter how you produce your drawings, the issues are the same. If you're to be rigorous and define things that will fit in all circumstances regardless of where they were made, you need to define three things. 1) The dimensional tolerance of the parts 2) The geometric tolerance of the part 3) The surface finish. You won't find any of these things on your drawings, and you won't find them on my CAD ones either. LBSC may have known about the first one but chose not to mention it, and the others weren't even formally laid down in his lifetime. I don't specify them on my own drawings for my own purposes because I know what fit or dimension or any of the other things are that I'm aiming at, just as most of you do.

On a proper Engineering Drawing to a modern standard, everything is defined so that the acceptable tolerance of every part is there. Most are defined by global tolerances with the more critical parts specifically detailed. The drawings we're buying fall far short of modern drawing standards and would never make it to the shop floor in today's factories.

I'm sure you all have a Zeus book, and I would recommend that you look at the last couple of pages and see what this is telling us. If I was providing proper drawings to a machine shop, I would show the shaft being a perfect geometric fit to a hole of the same size. That's the easiest way to represent it. When drawing the disassembled parts I'd again show them at the precise nominal size on the drawing for convenience, and then add the tolerance for the part from the selected fit if that's appropriate. If the hole is 10mm H7 and the shaft is g6 then the table tells you the complete range of sizes those parts can be and still obtain the required fit. That's the same fit, but different tolerance if you specified a 60mm nominal size. This is the power of the system.

I've said more than enough already, but the message is this. CAD is a visually perfect representation of an imperfect world, and it isn't intended to imply perfection. I could draw a wall brick with CAD, but you wouldn't assume it had to be made to micron accuracy. See my point?

[bhk]

I tried to post on this last night from my phone but i could not get it to go,

But I was trying to post the very same thing striplar has said, almost word for word in some cases.

There is so much misconception about modern techniques by those people who have not used them and mainly that misconception centers around the skill leveled required. you can not become a successful programmer until you understand fully how to machine, similarly with CAD you can not become successful at it unless you understand the fundamentals of draftsmanship.

As striplar has said the ISO tolerance system for limits and fits is in most pocket engineering books, I apply these to components that I'm currently building for my sweat William as I go, if I built it as drawn it would never go together never mind run.    

As for introducing tolerances into drawings with 3D packages, my understanding is that Autocad Inventor and solidworks both have the ability to do this to some degree. But this is not something I have experience with as we still used an engineering with a brain to do this process, just like the normal model engineering chap in his garage would.

[Roger]

I ought to say also that I don't accept suctionhose's accumulative errors assertion. If you make things accurately with machine tools then for most of the things you're ever going to make for a locomotive it's going to fit as machined. It's a completely different mindset to making something by hand. A rough and ready job would be machining parts to
+/-0.1mm I wouldn't expect any of the hole positions on a locomotive frame to me more than +/-0.025mm out of position however sloppy I was. This is a completely different world to marking it out with a steel rule centre popping the holes and using a pillar drill.
The reality is that most of the parts will be an order of magnitude more accurate than they need to be. The errors are certainly not zero, and things like the boiler will inevitably have greater errors than solid parts. The fact is that CNC machined steel flanging plates and CNC machined flue holes are going to result in a boiler that's a great deal closer to the drawing than if you make it all by hand.

Each step doesn't rely on the last with ever increasing errors of any consequence if the accuracy is high. If you drill holes half a millimeter out of position and then compound the error on another part that you add to it and end up with a millimeter error then that's possibly going to be an issue. If those errors accumulate to a tenth of that, it probably won't matter, and if it is, it's a small amount to adjust to get it right.

How would you build any mass produced commodity if your assertion was correct?

[Roger]

Nov 12, 2013 1:06:41 GMT John Baguley said:

I use Alibre PE (now rebranded as Geomagic Cubify). I downloaded the trial version which was Alibre Professional some time ago and gave that a go but couldn't afford to buy the licensed version so went with the basic version. As you say, a steep learning curve as I had only ever used Autocad 2D but well worth the effort.

The first loco I 'built' was a 2½" gauge version of Keith Wilson's Ariel (rebuilt Merchant Navy) and that was a real babtism of fire! A very complex loco to start with but I learnt a lot doing it. Using the package is now almost second nature and I can build parts and assemblies very quickly now.



One of the projects is a much modified Simplex with a completely new boiler with much larger grate area to test Jim Ewins's theories. At the moment it's just a rolling chassis but will be getting back to it as soon as one or two 'foreigners' are out of the way.



Sorry if this is getting a bit off topic.

John

That's a fabulous Alibre model you're created there, very nice indeed. I use it even if I knock something up by hand or on the lathe. For me it's as much about documenting what I've done and the convenience of having any dimension I need just by asking for it. I create a neat folder tree for the whole item, then sub assembly, ending in the individual part. Each part has it's own folder with any CAM paths and a Notes document if there's something I need to record about the making of that item. If I create a drawing from the model of that part, it goes into that folder too. All neat and tidy and instantly available for future reference without having to find that scrap of paper or envelope. I would never go back, it would be too painful.

[Rex Hanman]

Nov 12, 2013 21:23:49 GMT bhk said:

There is so much misconception about modern techniques by those people who have not used them and mainly that misconception centers around the skill leveled required. you can not become a successful programmer until you understand fully how to machine, similarly with CAD you can not become successful at it unless you understand the fundamentals of draftsmanship.

Thank you so much for that statement. :)Please tell the British education system! It is largely because of the opposite view that I gave up teaching technology. I seemed to be out on a limb trying to start with the basics instead of expecting kids to use high tech equipment to perform tasks they didn't understand!

Sorry if I've gone off topic!

[jma1009]

so far as the boiler is concerned i think myself and roger have made a bit of progress by way of considering the GWR standard No.2 boiler which was shortened in barrel and firebox length to produce the No. 10 boiler fitted to the 15XX class. same flanging blocks employed.

LBSC completely ignored the fullsize boiler dimensions and arrived at a boiler grossly overscale in diameter of the barrel fore and aft and in the taper on the sides of the outer firebox.

hence the lagging problems and much else besides. a shortened version of keith wilson's BULLDOG boiler provides a scale and very efficient steamer for those wanting a scale version of the SPEEDY boiler (see ME 18th May 1980). the LBSC PANSY boiler has similar faults so far as overall dimensions are concerned which doug hewson has recently attempted to correct by, so far as i can tell, copying keith wilson's DUKEDOG boiler without acknowledgement.

i have a table of all the GWR standard boiler dimensions compiled by K.J. Cook which is invaluable for this sort of work.

if anyone has any queries on GWR standard boilers i should be able to help, though in this respect i bow to TAFF's superior knowledge of same!

cheers,
julian

[suctionhose]

Nov 12, 2013 23:59:56 GMT Rex Hanman said:

Nov 12, 2013 21:23:49 GMT bhk said:

There is so much misconception about modern techniques by those people who have not used them and mainly that misconception centers around the skill leveled required. you can not become a successful programmer until you understand fully how to machine, similarly with CAD you can not become successful at it unless you understand the fundamentals of draftsmanship.

Thank you so much for that statement. :)Please tell the British education system! It is largely because of the opposite view that I gave up teaching technology. I seemed to be out on a limb trying to start with the basics instead of expecting kids to use high tech equipment to perform tasks they didn't understand!

Sorry if I've gone off topic!

Not off topic. This is at the heart of the topic!

Thank you all for the lengthy and passionate explanations! Ok I confess - it was everything I'd hoped for...

If one is fluent and familiar in these tools (3D CAD & CNC) it makes total sense to them for Model Engineering.

Obviously, they have more features than a cereal box but is the end product all that different?

"That's what makes us Engineers." - being able to render our thinking via different mediums.

Those tapered everywhere, nothing vertical or square GWR boilers where laid out with chalk and trammels - knowing what you're doing is what's important. And as Rex Hanman confirms, that, going forward is an endangered species!

Thank you Stripler for allowing the discussion to run it's course. Very thought provoking. I shan't be 3D Modelling engines any time soon but much obliged for showing us what can be done!

Ross

[pault]

Hi
I fully agree with the bulk of the pro CNC etc statements made, I would however ad the following things.
Most laser/water/plasma cutting companies will normally quote a general accuracy of + or – 0.3mm generally they will achieve much better, but they don’t have to unless you reach agreement before hand. This could be a problem if you are having batches done at different times by different suppliers.

It is a common misconception that CNC machines produce spot on accurate parts each and every time the green button is pressed. If you just write a program load it onto the machine, set the tools and press go you will get something that looks like the part should but will probably not be 100% within tolerance. CNC machines suffer from all the same problems as manual machines, tool/cutter deflection, tool wear, job deflection, heating/cooling of the job/spindle. Heat is a big enemy of close tolerance CNC work, to the extent that top end machines have heating/cooling systems built in to control the temperature of the spindles/screws/slides/coolant etc. When we machine manually we continually measure the job and then machine accordingly. Most CNC machines do not have the benefit of this feedback being done automatically, it has to be done for them by the operator.

Something else to consider is work holding/set up for second+ ops, this needs to be spot on as if it is not the first op and second op machining, whilst being correct may not have the correct relationship this becomes more critical as the number of ops increase. Again the choice of work holding and the set up of it are critical. In addition the diligence of the operator loading the second+ ops in making sure that the parallels/locating button/pin or whatever are clean and the job is correctly located.

In a previous life I sometimes used to run a couple of CNC lathes on tight tolerance production work where the average cycle time was between 40 seconds and 2 minutes. Monday morning was always busy, the machines had been shutdown over the weekend with only background heating on. On starting them up you would always be checking the job and tweaking the tool offsets for the first couple of hours until they had become thermally stable, once that had happened they would run continuously for 24/36 hours with little or no intervention.

All of these things are obvious to a lot of people, however if you are on minimum wage and just told to put that there do this up and press this button would you know or care as long as no one complains.


High Accuracy always has a cost and there are two main ways to gain it. The first is tight tolerance machining with rigorous QC checking to insure the work is in tolerance. The other is to machine to relatively wide tolerances and then very accurately measure and grade the parts. Fords used to do this with engine components which were colour coded, so you would used all blue or red or green etc components to build an engine. Ball and roller bearings are made this way, the factories don’t set out to produce a batch of CN grade bearing balls, cages, and races, they just manufacture and grade the parts. Clearly this approach only works with relatively high volume production.

Something which is often overlooked when people talk about precision machining is that it is impossible unless you can measure the job accurately. It’s no good having a machine CNC or manual) which can hold a dimension to within .0002” and claiming that you can work to this tolerance if you can only measure to within 0.001”. Imagine a two equally matched lathes and machinists, one with a mic that is spot on and one that has a mic with a 2 thou error. Unless the problem is picked up elsewhere the output from the man with the error on his mic could end up at a customers who decides it is scrap or needs to be reworked.The most important part of accurate machining is accurate measurement.

In some ways there is no difference between manual and CNC machining the end result is down to the skill and diligence of the person doing the job.

In short yes it is possible to build and engine from a kit of parts just by assembling it as long as the design and toleranceing are sound, and the manufacturing is carried out to the required standards. The moment any of these things are compromised for whatever reason then the fitting begins.

When considering a loco kit there has to be a fair degree of skill involved at some point in the design/manufacturing/assembly process. The designer has the choice where the skill will be placed be it the design planning, toolmaking (yes there is still toolmaking with CNC machining) manufacturing or assembly stage. Where the problem starts is when an unskilled (business) person thinks that there is no need for skills at any of the stages, or that something designed and manufactured to a poor standard can be simply assembled into a quality product.

As to whether CNC should be used for our models, I do model engineering for fun relaxation and because I enjoy it. How I get my results is up to me, as long as I am happy with it that’s all that matters. It also works the other way round as well. If someone tells me they have marked out, chain drilled, hacksawed and filed a set of frames recently I will take the p*^s. They are probably either young, skint, single, or a masochist I did it when I was young, but now theres not enough life left to waste time doing that
Regards
Paul

[Roger]

All in all I think that's been a very healthy and constructive exploration all round, and I don't think any of us are very far apart in our views. We just use different tools with their respective strengths and weaknesses, that's all. There are no compelling reasons for anyone to use CAD if it doesn't suit them. It's not cheap, and I personally think it's value is only partially realised if it's not followed through with the creation of tool paths to drive a CNC machine.
CAD may be many things, but it it's not a substitute for Engineering know how, experience and common sense. At the end of the day, there's no magic wand that allows us to circumvent a genuine feel for what's going on at the cutting tips and how much load and deflection is going to take place. Taking a cut on my mill with Servos instead of winding the handles is no different except you can't feel the forces coming back through the handle. You most certainly can hear, see and feel the vibrations though, and if that doesn't tell you anything, you have no hope of making something successfully.
In terms of the end product, the result is of a similar accuracy to any other manually machined parts but without the pain of tedious repetition and the likelihood of mistakes happening quite as often.

[uuu]

Nov 13, 2013 11:45:54 GMT pault said:

CNC machines suffer from all the same problems as manual machines, tool/cutter deflection, tool wear, job deflection, heating/cooling of the job/spindle.
Regards
Paul

Yes - and if you CNC mill one side off a strip of steel it can go banana shaped, just like a manual mill. Now all I need is a CNC hammer to welt the thing straight again.