machining accuracy

[mklotz]

Bob's got the right idea. I always try to make the last two cuts the same depth so that the cut after the final measurement is almost identical to the cut that immediately preceded it.

A typical closing-in-on-final-size sequence looks like:

roughing to size cuts
0.010 DOC (that leaves approximately 0.020 to remove from diameter)
0.000 DOC (spring cut to establish where the tool is really cutting)
0.005 DOC (first finish cut)
measure with micrometer
0.00x DOC (final finish cut where 1<=x<=5)

(Note that my crosslide is calibrated directly - a 0.005 cut removes 0.010 from the diameter.)

Depending on the rigidity of your equipment, you may want to adjust those numbers slightly.

 

[vederstein]

My lathe is old (80 years) and sloppy so when I got to do Precision fits, where I can I use reamers. Then I use cold rolled stock for pistons. With a little bit of lapping usually everything is good.

...Ved

 

[KJE]

An interesting topic at any level. Being a machinist and machineshop supervisor for over fourty years this topic has come up many times. The answer I have always given is be fussy where it counts and use the tolerances given where it doesn't ....to machine everything to a tight tolerance is just a total waste of time. People who think they are working to a .0001" tolerance in any shop that doesn't have stringent quality control, which includes regular calibration of measuring tools and temperature control is just kidding them selves. However working to .0005" is perfectly obtainable on most home shop machine tools with some practice. A good understanding of fits and allowances is critical for any kind of success with model engineering and can be obtained by most home shop Hobbiests with time and experience. Enjoy your hobby!

 

[mayhugh1]

Since this is a hobby and not how I earn a living, I can afford to try to machine everything to within a thousandth. It's part of the satisfaction. When I don't achieve it, I try to figure out why so I have a better chance next time. I don't scrap parts unless they really won't work or they have a cosmetic defect that I can't live with. But, I'll be the first to admit I have a pretty low threshold with cosmetic defects. After doing this for a while, I found common sense things like keeping work-holding and tool-holding surfaces clean, periodically correcting tram, and cleaning collets with solvent tends to generally improve precision and repeatability. On a difficult or particularly important setup I also make it a point to not do any cutting until after the setup has been completed and allowed to sit a while - maybe hours or even overnight - so I can think about it. - Terry

 

[el gringo]

Allowing myself some time, especially over nite, to rethink a setup has always helped in many ways...all your comments are right on Terry.

 

[bazmak]

I have broached this question on a previous thread .The original has answered itself. Generally 1 thou is good enough
Most dims are acceptable to say +- 5 thou but a few need to be say 1/2 thou. Important dims can be sized to suit each other
If you make a cylinder to say 1" and it finishes at +2 thou then you make the piston to suit. A baseplt for mounting which
is unimportant can be say +- 10 thou or more.The general answer to your question is that the dim needs to be as accurate
as it needs to be. The advise i remember from my apprentice days are burned in my brain. Close tolerances are expensive
and the more tighter they are the more expensive and difficult to achieve. Choose the peg that best fits the hole. An example
i was given (dont know if its true),that Rolls Royce made the car bumper bars to 5 thou etc etc Why when its not required
The answer is because they could but it all came out in the price of the car

 

[TonyM]

I agree with Bazmak.

For one offs, which is mostly what the majority of us produce, fit and function is more important than individual tolerances.

 

[Hopper]

A quick touch with some fine grade emery paper at the end of the job is the absolver of many sins.

 

[bazmak]

If im looking for a fine fit then a fine finish is needed so i always use fine grade emery to finish

 

[Charles Lamont]

Very often, geometrical accuracy is far more important than dimensional.

It does not matter within a few thou how long a steam engine cylinder is, but if a flange is used to hold the cylinder, it is very important that the face is square to the bore.

It is also highly desirable that the bore is parallel, even if the size is a bit off.

With a twin cylinder marine engine with a one-piece cylinder block, it does not matter exactly how tall the columns are, but they need to be accurately the same length.

It is important to understand sources of geometrical error within our machines and in our set-ups, and to plan machining sequences to minimise such errors where it matters most.

 

[alchr]

Don’t forget to ask those sudo perfectionists what temperature they do their .0001 work at!

 

[Schnupfhuhn]

Take a slip of paper. Take a micrometer or a good calipher with 0,01mm resolution (Tesa, Swissmade, but still no match for a micrometer). Start with a part. Put down your aimed at dimension. Do the work, measure. Put the measure on paper. You´ll get a feeling what precison you can achieve without any worries pretty fast.

On average sized parts, I aim for 0,1mm on general dimensions and something around 0,02mm when it gets hairy. This equals an h7 tolerance field in ISO tolerances. If you and your lathe can´t hold that, something is wrong in your chain of tools, measuring equipment or your process...

 

[machinejack]

Not really answering the main question. When I am bidding and making questimations on cost the more 0.0000 on the right side of the decimal the more it cost, but that is for making money. Now as to the hobby side. I will make close to the final dimension but if I miss by a bit I adapt the mating part to fit. It all goes to ones skill level. Perfection is almost unattainable. 40 years tool and die experience and I still screw up. We are here to have fun.

 

[ICEpeter]

Schnupfhuhn,
Arent you typically posting on Peter's CNC Ecke in DE?

Peter J.

 

[Schnupfhuhn]

Main question: " What do you consider to be an acceptable level of accuracy for model making? I have always strived for .001" as a standard where it counts, some times I make and ......"

My answer: see what you can achieve. Usually h7 is a good tolerace field for a lathe. You can loosen up on general tolerances to 5 times (0,1mm) the value.

We never talked about bidding or contract work.

 

[bazmak]

I did a post on making a 1" dia test bar.First down to 1 thou and then down to 10 microns
its not difficult but time consuming.I did it for an excercise.but general rule of thumb is
1 thou is good enough for most things and half a thou when needed,for model engineers
If people have better lathes and more accurate measuring equipment then go for gold if its needed

 

[Hopper]

When I worked in machine shops -- when Noah was still a junior midshipman -- generally speaking jobs with tolerances of .001" and upwards were done on lathes and mills. Jobs to be done to tolerances of less than .001" were done on cylindrical grinders and surface grinders. In the home shop, with no precision grinders available, that translates to use a bit of emery paper below .001".

 

[Cogsy]

Personally I think you get used to your measuring equipment. 1 thou equates to just over 0.025mm and on close fitting parts I like to hit close the nearest 0.01mm. I mostly use metric measuring equipment so that hundredth is easily measurable. I know in steel I'm going to be there a while with emery paper to remove 2-1/2 hundredths so I'll try a spring pass of two and see what that does.
Another trick I picked up is to stop and measure just before I get to where I'm headed, take a spring cut or two, measure accurately, then run a couple of grades of paper over the part and remeasure so I can account for surface finishing on my final dimension. Then it's relatively easy to hit very close to desired dimension without having to 'sneak up' on it as much.

For those that wonder about hitting close tolerances versus temperature changes, it really makes little difference. For example, relative to other common materials, aluminium expands/contracts quite a lot by changes in temperature, yet for every degree Celsius it changes only around 2.2x10-6 metres, per metre. So for a 1 metre (over 3 foot) long piece of ali raised in temperature by 10 degrees Celsius, it will grow in length by 0.00002m, or 0.02mm over 3 feet - virtually immeasurable at the part sizes we deal with.

 

[alchr]

You might want to consider t3mperature with tolerances in .0001!!

 

[Cogsy]

You might want to consider t3mperature with tolerances in .0001!!

Not much though - my 3.3 foot chunk of ali detailed above will grow in length by almost exactly 0.0001" per degree Celsius increase in temp. but I don't work with 3.3 foot lengths, especially anywhere near that precisely. In the real world, the ali part I'm trying to measure might be 50 degrees hotter than ambient from cutting friction, (so about 0.005" longer over 3.3 foot) but if I'm measuring with a mic it's probably less than 2" in diameter, so the change is in the region of 1/20th of that. Total error from temperature ~0.00025", total error from surface finish + measuring error is far greater than that (at least for me). And if you really want to get finicky you just let the part cool off for a minute or two and most of the error goes away.