Shop Made Tools

[Lew Hartswick]

Been doing that for years. Use the carriage as a stop for the tailstock . That way just back the ram a quarter turn and when the chips have been cleared, slide the tailstock back till it hits the carriage and lock. Start to crank and the bit starts a quarter of a turn later. That way you don't risk hitting the bottom of the hole with the bit too hard or it grabbing.
...lew...

[mattthemuppet]

For me that would still be a hassle as it takes a couple of turns with the spanner to lock and unlock the tail stock.

[Toolguy]

That method only works well with the camlock lever type tailstocks. The screw clamp ones are a little more cumbersome.

[wern]

Nice work Mat, the end of the rod is begging for a ball of some kind, black plastic commercial, or a brass ball turned using a ball turning attachment.

Werner

[mattthemuppet]

thanks Werner, a ball turner is on the list although some way down it. At least I'll be able to thread any handles for it with this tool

[Dan_the_Chemist]

Just a round chunk of 4140 with square ends

So, having made the level and used it to align the lathe, the next project was a cylinder square.



I had a nice 6" length of 3" OD 4140 for the project. The first step was high pucker factor - I mounted it in my 8" four jaw, indicated it in within a thousand'th, and then faced it and then cut a recess without a steady rest. Lots of light cuts and made sure I wasn't in a possible trajectory. Spot drilled it for a live center, flipped it end for end, and faced the other end. Honestly I wasn't very comfortable so I didn't worry about the worlds best facing cuts in the recess. Once I had both ends recessed and drilled for centers, I tapped in a 1/2-13 hole on each end to insert a threaded drive pin and turned it between centers after that. MUCH HAPPIER.

I tried using a carbide insert for the turning but I wasn't happy with the result. I ended up using a HSS tool that I reground and stoned before every cut. It still looks a bit "frosted" but it feels pretty nice. Even though I had trued up the lathe using a 14" test bar, I found that I was out almost a full .001" on this 6" cylinder. So I spent some time truing up the tailstock back and forth until I couldn't measure any diameter difference end to end with my cheap 2-3" micrometer. This took a couple of passes to get the tailstock adjusted correctly. Let's just say the cylinder went on a diet and ended up closer to 2 3/4" than 3". Once I had the outside surface trued up I took one final pass with the HSS taking off the least I could and with a fairly slow feed rate. I touched up the .2 wide contact surfaces on the ends and called it enough. I decided against any sand paper or scotchbrite even though it was a little frosty looking.

I painted the faces of the recesses (mostly to cover up the less than perfect facing off cuts), inserted set screws into the drive holes (no mechanical reason - I just think it looks more finished). A dial indicator shows that it's about 2/10'th out of equal diameter from end to end.

I then used my machinist squares as a comparison. I used my 6" square and found that there is a little tilt - far less than .001 Of course the tilt is away from the square in one orientation, towards the square when it's rotated 180 degrees, and they meet nicely at a middle rotation. I wanted to see at what rotation the machinist square and the cylinder square most closely paralleled each other, so I set up a laser single slit diffraction test.



I attached a laser LED to my height gage and shined that on the slit created by almost touching the machinist square to the cylinder square. The actual size of the spacing doesn't matter... I just got them close enough to get a good diffraction pattern.



Then by cranking the LED up and down, I could watch how the diffraction pattern changed along the height of the slit. If the slit gets narrower the spacing of the diffraction pattern will get wider (and dimmer). So, I just kept rotating the cylinder square until I got an even spacing from top to bottom. I marked that position, and then repeated it with a second square. I got nearly the same position on the cylinder square. I then checked the two squares against each other and found that they are not perfectly parallel.

So, at least one or both of the two squares aren't perfectly square. However, since the two positions on the cylinder square were within 30 degrees of each other, I assume they must all be close. I'm going to have to figure out how to tell which one is the culprit, where perfectly square is on the cylinder square, and how to measure the change in the diffraction pattern to get absolute numbers instead of using it to detect constant comparisons. That's gonna take a while. Any suggestions are welcome.

[10KPete]

I am sooooo pleased to see that someone out there knows how to check things using light diffraction!!! Excellent!

Pete

[Dan_the_Chemist]

I am sooooo pleased to see that someone out there knows how to check things using light diffraction!!! Excellent!

Pete

First time I did a laser diffraction experiment I had to set up the optical bench, clean the Brewster windows, spend some time with the gas mix, and then futz and fiddle with the mirrors. It took over a day. Then again it was 1968.

Today I just went to the electronics spare parts, found a laser LED, two resistors, borrowed a LiPo, a little duct tape and was running in 30 minutes. Progress is amazing.

[10KPete]

Ain't it though!!

Pete

[Andre3127]

For things like that you could use a fine oilstone and some light solvent (kerosene) and massage the high spots. With a day of careful measuring and fine-tuning, you could be within a tenth. You'll also be left with a very nice surface.

[TGTool]

There are cylindrical squares that are not made perfectly square. If you picture it, a tilted square will have a high and low side by indicator measurement. At 90 degrees from those points it will be exactly square. Further, if you know the angular tilt at the maximum, you can also mark the calibrated tilt from max to zero around the cylinder. So in fact, you don't technically need a perfect square if you can measure and mark where square is on your cylinder.

Also, squares, both cylinders and right angle squares are considered self checking from a surface plate. If you know the blade is perfectly straight (which you can check with an indicator to the surface plate) on both sides, then an indicator can give you the double error of any out-of-square with the right setup.

[rcaffin]

I'm going to have to figure out how to tell which one is the culprit,
I think you will need a third good square. Whether you could use the cylinder ... not sure. Harder.

You are not quite at the NIST level yet, but close enough that the rule of thumb will apply: 'It gets an order of magnitude harder every time you try to halve the error.'

To go much further, you may also need to bring the gear into a temperture-stabilised room and to start wearing gloves, to avoid contaminating the surfaces and to avoid differential warming. Fun stuff!

Cheers
Roger

[TGTool]

I'm going to have to figure out how to tell which one is the culprit,
I think you will need a third good square. Whether you could use the cylinder ... not sure. Harder.
Roger

You don't need another square. You can't prove squareness by a vote among the references. There's a possibility that several may err in the same direction. With a surface gauge you have everything you need to prove the accuracy (or lack thereof) to the limits of your measuring equipment.

Here's a demonstration at Moore Special Tools book Foundations of Mechanical Accuracy.



First check that the blade is straight, neither bowed nor tapered.



Then, using the indicator and surface gauge touching the blade at two points, bottom and top, compare both sides as this guy is doing. The indicator reading is the double error. With the square shown at Moore, you can adjust it. If yours isn't adjustable you may have to lap or otherwise adjust, or at least note how far off and in which direction.

[mattthemuppet]

made another little widget, a holder for 3/4" shank annular cutters (got one in an eBay lot a while back)

[rcaffin]

What's he doing?

You don't need another square. You can't prove squareness by a vote among the references. There's a possibility that several may err in the same direction.

That's how we make optical flats: lapping three units against each other in a cycle. They routinely get down to lamba/10 that way.
You can also make right angles that way, by lapping three units against each other. I never said it was easy though.

Here's a demonstration at Moore Special Tools book Foundations of Mechanical Accuracy.

Nice old photos, but I have to confess, I cannot see what the guy is actually trying to achieve there. Perhaps you could explain it in some detail to me?

Cheers
Roger