Do you make time for experiments?

[BobWarfield]

Sometimes, we get so busy finishing projects, we have no time for anything else. We fall behind, and we stick to tried and true "fastest most familiar" methods. That's a shame. Do you make time for experiments? I find that's where I learn the most. I try not to be so busy that I can't stop and go try something out.

What kinds of experiments?

Well, like trying a bunch of different ways of measuring precise Z-axis location on my mill. What works better:

- Touch off by feel: Move the cutter down until you feel it touch the workpiece without the spindle running? I get an accuracy of 0.012" doing that. Lousy! And repeatably so!

- Touch off by sound: Fire up the spindle and lower until you hear it cut. 0.0085" accuracy, and repeatable. Better, but no great shakes.

- Touch off with paper: The traditional old school method. Fire up the spindle, insert a slip of paper atop the workpiece, move down until the cutter grabs the paper. Most paper is around 0.001" thick, so allow for that too. Alas, I got a very repeatable 0.010" inaccuracy on this. BTW, repeatable inaccuracy can be factored in! And be sure you use a long enough slip of paper so your fingers holding it are in no danger!

- Use a fancy Z-axis presetter gizmo:

By now you guys know I'm a total gadget junkey. Those presetters are available from 800watt and others on eBay. They have a self-calibrating standard inside. You run the cutter down until you detect any motion at all on the needle. At that point you are exactly 2" above the workpiece. Use your quill DRO or other aid to close the gap exactly 2". This bad boy put me exactly onto my target. I can't say how exactly other than that it was well under a thousandth. At that time I had not the ability to measure to a tenth, but it was dead on to a thousandth.

Recently, I did another experiment. I wanted to measure runouts on my mill spindle and some ER32 collet chucks. Have you ever checked your runouts?

I ran an 0.0005" Interapid DTI on the inside of the R8 taper using a magnetic indicator stand sitting on the mill table:

I flicked the spindle on at lowest revs and then hand rotated it to max runout to measure. It was 0.0005". That's very tolerable for this kind of machine.

Then I stuck a collet chuck in and measured the outside barrel:

About the same, even on an El Cheapo 800 watt chuck versus my 'spensive Bison chuck. Hmmmm.

Now I pulled out a pin gage from my set. I verified it was straight and true on my surface plate, stuck it in a collet, and checked the runout there:

Oops! Bad collet, bad!

And that's how I knew I could've stayed with my cheap chuck, but should've bought nicer collets. Sigh....

 

[Brass_Machine]

Bob

Put the gadget down and step away. We have a 12 step program for you to undertake... :big:

Eric

 

[Loose nut]

When I first put digital scales on my mill/drill I did a little experiment. I put a wiggler in a collet and tried running it up against a parallel held in a vise just to see if the scales were repeatable in their accuracy. After five tries I had three with no visible error and two that were out .0005" so either I am very good at getting exactly the same error most of the time (quite possible too) or the scales have good repeatability or maybe I was just lucky?. Experimenting can give you a better idea of the capability of your equipment.

 

[BobWarfield]

Loosenut, it sounds like the error in your scales is something less than 0.0005", but not too much less. Your experiment, 5 trials with 3 of one results and 2 of the other can be used to extend the accuracy of what you're measuring. Average the trials together. You get 0.0002". More trials would extend the accuracy further. It's too tedious to use this method very frequently, but sometimes you gots to know! I wonder what 10 trials would get you? If you still see 0.0002" after 10 trials, you could be pretty sure.

Frequently I will select a project because it lets me try something new as much as because I want whatever it is to be built. I built my disc sander before the belt sander because I wanted to try to make a round balanced disc much larger than my lathe could handle. I learned a lot doing it and got a great new machine for the shop in the end.

Cheers,

BW

 

[Airhead]

- Touch off with paper: The traditional old school method. Fire up the spindle, insert a slip of paper atop the workpiece, move down until the cutter grabs the paper. Most paper is around 0.001" thick, so allow for that too. Alas, I got a very repeatable 0.010" inaccuracy on this. BTW, repeatable inaccuracy can be factored in! And be sure you use a long enough slip of paper so your fingers holding it are in no danger!

Try taking a small piece of paper and sticking it on the work with a little oil. Kind of like putting a piece of toilet paper on your chin when you nick yourself shaving. Make sure to smooth it down so that there are no edges sticking up mess you up. Works good for me.


Rick

 

[Philjoe5]

My machining hobby is one big experiment. I've built three "identical" steam engines. Parts were machined using various materials and techniques. Being at this for only a little over a year I wanted to experiment and see what works best for me and the equipment I use. A few conclusions I've drawn:
1. I can make a better cylinder bore (less taper) using the 4 jaw chuck on the lathe compared to using a boring head on the mill even though I use the same brazed tip boring bars
2. One experiment leads to more experiments, so I've started engine #4
3. I don't know how to stop and get off this merry-go-round :big:

Cheers,
Phil

A later post from me is going to address how to best compare the relative efficiencies of these "identical" engines ???

 

[Loose nut]

Loosenut, it sounds like the error in your scales is something less than 0.0005", but not too much less. Your experiment, 5 trials with 3 of one results and 2 of the other can be used to extend the accuracy of what you're measuring. Average the trials together. You get 0.0002". More trials would extend the accuracy further. It's too tedious to use this method very frequently, but sometimes you gots to know! I wonder what 10 trials would get you? If you still see 0.0002" after 10 trials, you could be pretty sure.

BW

Bob, the scales only measure to .0005" so anything less than that is a guess and I would rarely need that kind of accuracy for what I make, mostly I just wanted to see if the scales were repeatable and as you said the average is less than a half thou. variance which is pretty good for no name Chinese scales, unless my ability too reproduce the same error repeatedly skews the data. I think I just confused myself.

 

[ianjkirby]

Hi guys,
Bob, in mentioning a few ways of setting the Z height, left out one that I have used for years with great success. A 6"/150mm ruler lives in my shirt pocket, and I use it for lots of things, even some not intended by the manufacturer! By design, it is .5mm thick (the 12"/300mm ones are 1mm thick), and I use it in the manner of a feeler gauge under the cutter, without the spindle running of course. I believe I can get within .001" every time. This must be a common practice, surely?
Regards, Ian Kirby.