Saturday, June 23, 2018

Trama Lama Lama Ding Dong


No, that has nothing to do with this post.  This has to do with a procedure we need to do to a milling machine called tramming (which had me singing that title to myself all afternoon while wandering around the house). 

Maybe it'll make more sense if I back up and start at the beginning. 

This week, I moved onto the next part in my flame eater engine: a support for the flywheel.  For a couple of reasons, I didn't like the original author's approach and this seemed like a good part to redesign.  Why didn't I like it?  Two main reasons.  First: he lets the flywheel shaft, 1/4" steel rod, just rotate on aluminum and one thing I've learned about these engines is it's worth it to remove as much friction as you can.  I wanted to use ball bearings, which causes the uprights of the part to be bigger than Duclos used.  Second: he makes this part from a solid 2-1/2x1x3/4 chuck of aluminum and turns most of it (about 65-70% by my calculations) into waste chips.  So I decided to make it out of three pieces of 1/4" aluminum (all you're left with in his design).  Two uprights and a bottom piece to hold them in place.  Some time with CAD resulted in this drawing for the uprights:
I then created a drawing with two of these side by side, spaced far enough apart to put a 3/8" cutter between them.  A little time spent with a couple of pieces of software and porting it all to the mill ended up with this:
The problem concerns those holes.  Here it gets a little tricky to understand so I'm going to slow down. 

After cutting out most of the shapes, I marked the center of all six holes with the center drill I created last weekend.  I have the exact X/Y coordinates for the center of all the holes from the drawing, and since everything in the file is made using the same reference point, they should be perfectly positioned.  Then, because neither of the holes fit one of collets I have (like the center drill does), it was time to put the drill chuck in the mill's spindle. 

When I put the 9/32" bit in the chuck, it was visibly not centered over the hole it was supposed to enlarge.  I've seen this problem before and thought it was due to a prior mistake in finding the edge of the part, and I just re-assigned the X-coordinates.  I can't change them, though, because everything needs to use the same coordinates.  Troubleshooting, I put the drill bit (too small for the small holes) back in a collet in the spindle and it was exactly over the center of the mark.  Put a drill bit in the drill chuck and it moves a visually noticeable amount.  Since the big (9/32) hole was kind of an arbitrarily chosen size, and I have a 5/16"collet so I figured that instead I'll drill that with a 5/16 drill bit in the collet. 

When I went to drill the small holes, I had to use the small drill bit that wouldn't fit any collet, so back to using the chuck.  I ended up centering the holes visually by moving the drill bit around .020" to the left. 

So why should the coordinates for the part differ if I'm using the drill chuck instead of a collet?  Both of them are supposed to centered in the spindle, and since whatever is in the spindle is rotating, if they're far off center, it's very noticeable.   Time to ask the collected wisdom of the internet to which I was told "check your tram". 

Tramming is a process that's done to verify that the Z-axis motion is perpendicular to the table.  A common way of doing it is with a dial indicator and the way the angle is verified is that the indicator is moved from one side of the center (spindle) to the other side and the head adjusted until the indicator reads the same (ideally).  Fixtures like this are used, but the farther apart the dial indicators are the more sensitive it is.  My fixture puts the two spots I'm checking over 10" apart.  Here's the way I checked mine.
This is an optical spherometer I made for grinding telescope mirrors back around 2005.  The indicator's dial shows .0001 per minor division and each of the large print numbers represent .001".  I measure one side, rotate the fixture 180 degrees and check the other side.  I need that mirror on the right edge of the picture for that. 

I checked the tram with this tool and found that the two readings were within about .0005 - call it perfect.  Now what? 

What this is really checking is that the center of spindle is perpendicular to the table, "pointing straight down" if you will, but what if the Z-column is not really vertical?  The head will drift to whichever side the column is tilted but the center of the spindle will still point vertically at the table.  The drift in inches will be (height * sin(error))  Wait... I have an angle indicator, what if I stuck that on the column?
That says my Z column is 0.4 degrees off.  That's a lot!  What would that look like?  Let's say I used the centering drill and then raised the head 3" to fit the chuck and a drill bit in there?  3* sin(0.4) = .021" - eerily close to how far I had to move the drill to put the drill bit in the center of the hole (and saying the vertical difference was 3" is approximate).  Sounds like we could be in the right ball park.

The Z-column is held in place by four large socket headed cap screws (M12).  Getting them out was an ordeal when I took it apart for the CNC conversion, but easier now that I have the right tools.  Once all four bolts were loosened, it was trivially easy to get that indicator to read 90.00 degrees - it actually almost fell into place. 

But doesn't that knock the spindle away from pointing straight down?  Yes!  When the column was inclined 0.4 degree to the right, I adjusted the headstock to compensate for that; now that the column was vertical, I had to loosen the headstock and readjust that.  For a final check, I rigged up a way to hold a dial indicator (.001 minor division) and ran it vertically up the edge of a 5" machinist's square. 
The needle didn't budge.  Is it perfect?  I can't say.  If I assumed the needle did move and the error was .0005 (since I'm using a .001" reading indicator, I'm sure I would have seen that) and solve for the angle needed to give that, I find it would have to be .0057 degrees. I'm pretty confident the error is less than that.  

A check with the center drill and then using the chuck to recreate the parts used to drill the small holes in the parts shown at the top resulted in the drill bit being visually centered and the drilling the hole without the bit appearing to wander or bend at all.   I think everything I can do agrees that it's fixed.


  1. I get such a kick out of watching you find these problems, and then solve them, that it always brightens my day.

    Thanks, SiG!

    1. When the geometry was explained to me, it was easy to visualize in my head. Plus, it's especially easy to seize onto the concept of the Z-axis being off because it's something I did when I did the CNC conversion and had the column off. I'm sure it was good when it came from the factory.

  2. Good work! I would still be out in the shop looking for a non existent chip in the taper or something....

    Question- the angle gauge- that references to level or plumb?

    So the table measured zero? (level)

    Or can it be pre-set to any surface as a zero reference, then used to determine angular difference?

    This would be a useful application for a laser centerfinder- just run the table or head up and down and see if the location changes.

    1. The angle gauge can be zeroed on any surface, and I zeroed it on the table to set a reference and measure the angle between them. iGaging claims accuracy to 0.2 degrees repeatable to 0.1 and displays to 0.05 degree resolution. We used to use things like this (only a little bigger) to zero the antennas on airborne weather radars.

      The laser center finder is a good idea. It would be useful for quickly verifying the setting hasn't been knocked off by vibration or other use.

    2. I bought a "Digital Protractor" to use to zero my satellite antennas.

      Turns out it's an order of magnitude (or two!) "better" than I need. It's also a bit 'jumpy' when resting on an antenna boom. The $10 "angle finder" from Home Depot works just fine.

      We used one at Boeing to zero our telemetry antennas, but those mounts were so stiff that they didn't bounce around like my puny Ham Radio stuff!