Workshop working tolerances

[Dave CParticipant @davec87625]

Hi guys

I am currently attempting to build a 3" scale Mclaren traction engine and was wondering what sort of tolerances would be acceptable. Obviously spot on would be fantastic but this is not always possible.

I am currently 80% completed on the machining of the crankshaft which has been a trial but enjoyable and a massive learning curve to boot.

On the two off big end throw diameters I have had to finish 10 thousandths under due to the fact I was struggling with chatter and therefore had to keep going. Sods law but the finish was pretty good until the final passes. The big end bearings are not yet made and so they can be made to suit however I have got all other dimensions to within 2 thousandths after a very long week and am extremely happy.

I had the gears professionally machined and find the tolerances are ( well not what I expected ) Am I being too exact as I find myself getting pretty downhearted when I miss my final exact dimension or am I just aiming too high. It is a hobby after all and I am not a time served toolmaker.

The traction engine is 3" scale and not a pocket watch so what sort of working tolerances would others find acceptable.

Just a very quick side note I am machining on a Harrison M300 and one of the cam locs falls through into the spindle and has since I owned the machine. Is this purely held in situ by a spring and detent / ball bearing ?? any help at all is appreciated.

Dave C

[Dave CParticipant @davec87625]

[MuzzerParticipant @muzzer]

There should be a pin at right angles to the cam to hold it in place. On my Bantam D1-3 spindle there are 3 cams and 3 slotted screws. Remove these to have a closer look. Presumably one of them has been broken.

This thread shows some drawings from the standard that defines the dimensions etc.

Murray

[Mick B1Participant @mickb1]

There can't be a hard-and-fast rule because, as you note, you have control over the actual value you reach for dimensions on each part and full interchangeability with other examples of your model isn't a requirement. What is required is to reach an acceptable fit with interacting components.

A very rough rule of thumb is a thousandth of the diameter – clearance for a close running fit and interference for a solid drive fit. But there are other dependencies; for example if you've got 2 bearing bushes about 3/8 diameter and 4 inches apart, a couple of thou clearance is more realistic, because you have to take care of alignment as well.

You're unlikely (AFAIK) to meet them in Model Engineering, but there is also a labyrinthine system of Geometric Tolerancing for defining the limits of dimensions with respect to datum features elsewhere on the drawing. I can remember tackling these as a Tool Designer nearly 40 years ago, but have had little to do with them since.

[JasonBModerator @jasonb]

It really comes down to what "fit" is required for each part.

For example your gears that are fixed to the shaft want to be a closer fit than the ones that will slide along the splines, a solid bearing will need to be made more exact than a split bearing that can be adjusted for fit. Your flywheel wants to be bored so it is a light press fit on the shaft so that when the gib head key is fitted it is not tilted on a loose fit resulting in a wobbly flywheel, and so on.

Add that that everything is bolted to a big bit of steel tube that will move when heated so you don't want fits so tight that it all binds up the first time you put a fire under it. On the otherhand you don't want any noticable slop where it is not wanted.

Edited By JasonB on 02/09/2018 13:28:53

[Clive FosterParticipant @clivefoster55965]

Back when hand work was commen ± 1/64 " / 15 thou / 0.5 mm was considered realistic for ordinary craftsmen doing careful work. Maybe twice that ± 1/32" for ordinary, but not ruff, work.

For machined parts a fairly common seems to be that ± 5 thou / 0.1 mm is a sensible tolerance for non critical dimensions. Should always be able to do better on any half decent machine unless there are serious material or tool issues. Portass "S" , Sooper Adept and other inexpensive, low end machines will probably struggle tho'.

For parts that have to be to size but not to a fit ± 1 thou / 0.02 mm should only need a bit of extra care on any decent machine. Ought to be pretty much routine on industrial style machines if the job warrants that bit of extra effort to properly account for backlash et al. That's about what I normally expect to work to. But my Smart & Brown 1024 is a high end toolroom machine which seriously helps and my normal work practices are mid way between the "get 'er done quick'n close enough" appropriate to ± 5 thou and "extra care". Mill is a Bridgeport with DRO's on all axes which makes it easier too.

For press fits, bearing clearances and the like you have to work to measurements to get paired parts right. Tolerances don't really apply there for home shop folk. If I really concentrate I can get into the ± tents thou range but, frankly dear I'd rather not.

In practice its more important to know what your personal "quick'n close", "careful work" and "really concentrate" capabilities are. The range of skills, machines and simple practice for home workers are so great that its impossible to be dogmatic. The important thing is that you don't waste valuable workshop time trying to hard when routine would be fine or by do-overs when routine wasn't enough.

Worth remembering that in practice a thou is about as close as can be reliably measured using normally casual methods. Comparative measurements between parts of very similar sizes can be much closer, this sort of thing is what the teeth's thou vernier on micrometer is for. Generally helps if you can use the same measuring stick for both but realistically a thou is it.

Geometric Tolerancing is a weird and wonderful world of its own. Strictly its for CNC boys. Even the pro's have problems applying it to manual machines. Common practice is often to convert to normal rectilinear (or polar if rotary table is involved) tolerances and forget all the fancy multidirectional gains.

Clive.

[Neil WyattModerator @neilwyatt]

Just make sure everything fits everything else.

You aren't mass-producing them, and no one is going to creep up behind you with a micrometer to check.

Neil

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Neil Wyatt on 02/09/2018 17:05:48:

Just make sure everything fits everything else.

You aren't mass-producing them, and no one is going to creep up behind you with a micrometer to check.

.

I believe that the creepers are more interested in colour-matching and rivet-counting

MichaelG.

[MWParticipant @mw27036]

I remember "This Old Tony" on youtube saying, "The devil's greatest trick was convincing you to buy cheap tools"

Well, guilty as charged in my case I'm not sure how true it is but it did make me laugh.

Michael W

Edited By Michael-w on 02/09/2018 19:15:15

[RevStewParticipant @revstew]

Posted by Neil Wyatt on 02/09/2018 17:05:48:

Just make sure everything fits everything else.

You aren't mass-producing them, and no one is going to creep up behind you with a micrometer to check.

Neil

Wise words indeed.

[John RuddParticipant @johnrudd16576]

Posted by Neil Wyatt on 02/09/2018 17:05:48:

Just make sure everything fits everything else.

You aren't mass-producing them, and no one is going to creep up behind you with a micrometer to check.

Neil

So why am I always looking over my shoulder? Guilty complex?…

You are quite right tho'…. make everything to fit… and all will be good…

I try to be accurate, within a thou or so… (no wonder the scrap bin is always full!!!) but its a trade off between time and accuracy and materials on hand…

[HOWARDTParticipant @howardt]

How accurately can you measure. A shaft can be fairly easy to measure whatever the diameter, but a hole is entirely different. Unless you have inspection room gauging equipment just make it fit.

[Anonymous]

Posted by Dave C on 02/09/2018 12:19:39:

On the two off big end throw diameters I have had to finish 10 thousandths under due to the fact I was struggling with chatter and therefore had to keep going.

I feel your pain! Been there, done that; I hate SG iron. More chatter than an Islington "liberal".

On the other hand I'm not the best person to comment on tolerances; too pernickety. I'm building a couple of 4" scale Burrell SCC engines; I've been told several times on the TT forum that I'm building traction engines, not a clock.

For general shaft and bush (no relative rotation) I aim for about half a thou. For parts with relative rotation I aim for about a thou. For other parts it depends. I generally aim for 2 to 3 thou give or take. I tend to make parts as the fancy takes me, so parts that fit together may be made weeks, or months, apart. I expect them to fit together, and ideally be interchangeable between engines. Of course some things don't matter, or close tolerance isn't possible, like the smokebox and other sheet metal parts and some casting features.

I've made all my own gears. Generally the spur gears and the steering worm and wheel have 5-10 thou backlash. The differential bevel gears have no backlash, which will need to be sorted with shims as they need at least a thou or two to run smoothly.

Really it's a case of knowing where tolerances are important and where they are not. I like things to be just so. That also means some considerable redesign effort as the drawings are error ridden, are often not true to prototype, and some items will never fit or work as drawn,

Andrew

[Anonymous]

Posted by HOWARDT on 02/09/2018 20:29:29:

How accurately can you measure. A shaft can be fairly easy to measure whatever the diameter, but a hole is entirely different. Unless you have inspection room gauging equipment just make it fit.

I grind my own gauges, compared against gauge blocks, and use those to set the fit of internal holes.

Andrew

[Paul LousickParticipant @paullousick59116]

"I had the gears professionally machined and find the tolerances are ( well not what I expected )". Do not accept the gears if they are outside of the tolerances specified on the drawings. Send them back and have them re-made.

Machined parts to an exact dimension is extremely difficult which is the reason to specify a tolerance on drawings, it is a range of sizes which are acceptable. Drawings should specify a general tolerance for all dimensions and specific tolerances for critical parts. If not shown, you have no come-back from suppliers.

Do not make tolerances smaller than needed because as the tolerance goes down, the cost of machining goes up.

Paul.

[Bill PudneyParticipant @billpudney37759]

Amongst many other things, I used to run an estimating section for a manufacturing division. Obviously we had all sorts of metrics available, but one that used to impress people was "…halve the tolerance, quadruple the time….". Sometimes it would be worse than that, when an "easy" tolerance could be achieved off the mill, whereas for the same feature but with a tolerance that demanded grinding, or even lapping, it could be off the scale……

To make things exciting some of our "draughtsmen" (Engineers on a CAD station) didn't have the faintest idea how any given part would be made. Good fun, I miss it like you miss the flu!!

cheers

Bill

[HopperParticipant @hopper]

As Bill says above, tolerances are a matter of horses for courses. If boring a cylinder and making matching piston you might want to work to half a thou or so, or at least a thou. Ditto your crankshaft. But many other dimensions can just as successfully made to the closest 10 thou or even more. The trick is deciding when to use which tolerance. Common sense and experience I suppose.

As for measuring holes, a set of good quality ball gauges for half inch and under, and telescopic T gauges for over half inch will give all the accuracy you need in the average home workshop when measured with a good micrometer.

[JasonBModerator @jasonb]

Posted by Paul Lousick on 02/09/2018 23:49:33:

"I had the gears professionally machined and find the tolerances are ( well not what I expected )". Do not accept the gears if they are outside of the tolerances specified on the drawings. Send them back and have them re-made.

Problem is the drawings were more than likely dimensioned in fractions and no other tolerances given on the drawing so the maker may have used the general rule that a fraction is +/- 0.015"

[Brian GParticipant @briang]

Judging from sound alone, I suspect the fit on many full size engines is far looser than would be obtained by scaling up the tolerances from a miniature.

Brian

[Paul LousickParticipant @paullousick59116]

If the drawings were only dimensioned in fractions and no other tolerance you cannot expect them to be machined to a fine tolerance. (+/- 0.015" sound good for general machining). The other general rule for tolerancing that we used to have is +/- half the value of the last digit. (eg 10.0". assumed tolerance = +/- 0.05", 10.00" tol. = +/- 0.005"

Paul.

Edited By Paul Lousick on 03/09/2018 08:26:42

[Dave CParticipant @davec87625]

Hi thanks for all the advice. The camloc is now sorted. It is held by a detent and spring secured by an Allen bolt as advised. A quick fix.
Sadly the gears are pretty random in their accuracy with the worst case being 1/16″ under size on thickness. The others in the thousandths. I have asked for a refund but as no tolerances are described on the drawing I am having no luck. Sadly I think a very expensive and hard lesson learnt. I machined the crank with the diameter half a though under only to discover the bore on the relevant gear is 2 thou over. Again my own fault for not checking properly. The gear is keyed but obviously not a good fit.
My option now is the expensive route of replacing some of the gears at my own cost though this really does hurt however I still need further castings from the supplier so I am possibly in a no win situation. I know the gears are wrong and I can adapt other dimensions to suit in the future but feel pretty disheartened when expensive drawings and castings are purchased which seem full of errors.
Rant over. Not happy but determined to make a success one way or another. Any recommendations for a good gear cutter anyone?
Oh and the relevant gear with the bore 2 thou oversize is also the one which is a 1/16 too thin.
Hope next week is a better one.

[JasonBModerator @jasonb]

I machined all the castings and barstock gear blanks for my 2" Fowler and just had the actual gear teeth cut, that way I could control the fits as much as possible. Might be the way to go if you need some doing again.

[Dave CParticipant @davec87625]

Yes on reflection I wish I had done the same. I work away an awful lot so workshop time is scarce and random. My thoughts were initially of saving time and also or as I thought ensuring an accurate and professional job. You live and learn. The hard way usually sadly. I intend to machine the replacement blanks myself next time.that way at least if they are wrong I can only blame myself.
Thank you all for your help and advice. It is appreciated. Recommendations for a gear cutter would still be good to hear about.
Dave

[Alan Waddington 2Participant @alanwaddington2]

Can you not alter the gears you have ? For example the crank gear you mentioned which is bored 2 thou over, maybe bore it out much larger, and turn up a new centre to size, which could be a press fit or welded/soldered in.

I doubt a gear being 1/16 undersize in width will make any difference to anything.

[Dave CParticipant @davec87625]

I had thought of that Alan but then with the additional work and the gear still being wrong I thought it better to just replace the gear with a correct one. I am at the early stages of build making parts as i can with available funds and stock etc and want to keep on track as close as possible with dimensions. My fear is that small errors on lots of parts with escalate and bite me later on. Probably not the ideal build sequence but I am a relative novice working from plans only. Like I say this whole thing is a massive, frustrating. Expensive. But enjoyable learning curve.
I will get there one day ha ha
Thank you for you’re advice.

[Author]

Posts