More Learning

I was running a production run of parts, drilling holes accurate to .00001" position, and suddenly I noticed that Z was not returning to 0.  Was something else broken?  It appears that trying to center drill at 0.5 ipm was confusing it about where Z was?  At 0.2 IPM, no problem.  Perhaps switching to carbide center drill bits would be good.

UPDATE: Even at 0.1 IPM, Z was getting shifted.  I have changed to the glacially slow 0.5 IPM, and increased drill depth to 0.5 to compensate for possible Z errors on start.

Every Bad Thing I Have Said About Millie; Forget It

Within its limitations on power and size of workpiece, it works well.  This evening I was cutting little parallelograms of acetal on the chop saw, and one came out a bit wrong because my jig for holding the material in place did not quite work.  There was a time I would have thrown it away or sanded it to square, but I would not have tried to cut it square with a bandsaw or chop saw; too dangerous.  I would not have tried to fix it on a manual mill; too slow and repetitive.  But I just change the parameters to my excavate program, put the deformed acetal wi the square side down and run the program to salvage this piece.

Happy Customers

I have now shipped three sets of the new Vixen HAL casters, two as warranty to existing customers, one to a new customer, who raved about them:

I received the casters yesterday Clayton, and I’m very impressed. I had them installed in to time at all, and it’s a breeze to maneuver now :-)

And this picture:

Today's Milling Experiment

I successfully milled a 30 degree diagonal to use as a base for milling this:

Fig. 1 

And why oh why do I want this?  In the manufacturing of one particular part for ScopeRoller, I have a part with this cross section:

 

Fig. 2 

 I need to drll and tap a hole in the top, and trying to get this part in the jaws of the drill press vise and level on top is a pain.  The thought is to drop this part in the first block, to get it level on top:

Fig. 3

 

You are probably wondering why I didn't just put a flat top block in a tilting table and mill vise at a 60 degree angle, and then run an end mill across the top face.  The mill vise that fits into the tilting table does not have enough clamping force to hold acetal in place while milling even a light cut across it.  The big ugly drill press vise that I have squared and maimed to fit on the table does hold acetal quite well, and with the 30 degree support under it, making the part in Fig. 1 was pretty easy.

Cutting Angles With Millie

I mentioned yesterday that I was rewriting my angle cutting program to do diagonals in gcode instead cutting a series of horizontal steps.  The result left something to be desired.

 

 

The stripes are not cosmetic.  I used a ball nose mill, and yet the other 45 degree angle was from a square end mill: actually smoother.

Using A Round Nose End Mill To Cut An Angle

While I don't have an immediate need to do this, I thought it would be good to write the code for milling a slope while I am not under any pressure.  So I did.  The first results reminded me of Djoser's step pyramid:

This was a consequence of telling the program to do cuts of .1" depth.  I just redid it with .05" depth, and it is much better.  Still, putting a workpiece on a tilting table is faster and better looking, unless I go to .0001" depth of cut.

It,strikes me that I should rewrite the program to do it not as a series of horizontal cuts, but as a series of diagonal cuts:

G90 x0 y9 z0 f5

G90 x0.866 z-1 f2

G90 y.1875 f2

G90 x0 y0 f2

More Uses for Millie

I was busy constructing parts for the Vixen HAL tripod legs, and I realize that some of the acetal parts to go in the ends of the legs were a little bit too long. So I put the part in Millie, and trimmed off .28" to make them exactly the right length.

Why It Has Been So Quiet Here

I spent yesterday getting the replacement parts made for two absurdly patient ScopeRoller customers.  Even though there was no machining of aluminum to make the parts, I did use Miilie to drill the holes in a total of 10 parts./  Here is one face of the first part.

Why do these pilot holes this way instead of by hand on the drill press?  Because these were placed accurate to .0001", instead of perhaps to .01" using a micrometer and drill press.  The slow speed on the drilling is because even 1 inch/min. caused a lot of vibration.  But 0.5 ipm was pretty peaceful.  The first two were pilot holes at 0.5" and 1.5" from the left end of the workpiece (or more accurately from the left end of the vise, which I was using to register the workpieces against)..  These went down .2"--just through the .125" wall.  The last hole was a through hole so .3" down from the top.

I made 10 of these in an hour or two, and much of that time I was doing other things while Millie worked away.

The Square Sleeve Version of the Product

This is what the square sleeve version of the ScopeRoller casters will look like for the Losmandy G-11 tripod:

Before you say anything about that last picture, I am quite sure DSM-IV excludes scientific instruments from the definition of hoarding.  And the big telescope that did not make it into the picture is only one more.

Square Tubes

I mentioned a few days ago my intention to change over from round tubes to rectangular tubes for the ScopeRoller casters , and all the advantages of doing so.  I built a square tube version for the Losmandy G-11 tripod, and it was a great success.    Much faster to make, more repeatable where holes end up, and four bolts holding it to the leg instead of three.  Less clamping force required to lock it in place, so less finish damage to the tripod legs. Unfortunately, because the local Interstate Plastics store doesn't carry 2 3/4" thick acetal, the insert was made from a 2" and 5/8" thick section, which looks bad but works fine.  This will be for my  personal scope, untuil the thicker stuff arrives from California.

 For the Vixen HAL tripods, which is what the CNC mill was for,  I will use eight bolts to secure the sleeve to the leg.  And this way I can ship these replacements a bit sooner. 

One Drop of Solder

You have probably heard the story of how John D. Rockefeller in the early days od Standard Oil asked the guy who sealed up the cans of oil how many drops of solder he used to seal the cans.  The answer was 43.  "What would happen if you used 42?"  It worked, and saved 6/10ths of acent per can, $43 million a year increased profit.

At ScopeRoller, I am always looking for a way to play Rockefeller.  Switching from boring out cylinders of acetal to a round aluminum sleeve that bolts on to the tripod leg both reduces materials costs, labr, and made a more durable product. 

We were making a set for the Losmandy 8 Lightweight tripod.  These are a rectangular sleeve instead of a round one, because the tripod leg is square.  The labor was substantially reduced.  The sleeves being square are easy to locate in the drill press vise, easier to drill and tap, and the acetal part that goes in the end in which the casters  sit is now a rectangle, easy to make exactly right sized with a planer, insteade of turning on a lathe.  So I wondered: why are the sleeves for most of these legs round?  Just because the tripod leg is round?  The sleeve holds to the leg with three bolts.  Going to a square sleeve means using four bolts which reduces the force required on each bolt, producing less damage to the tripod leg and less stress to the threads.  Square tubes weigh slightly more and cost a bit more than round, and 4 bolts cost more than 3 bolts, but labor saving is substantial and the holes are more precisely located.

This morning, as I started gathering materials for an experimental set for one of my telescopes that the tripods that sent me buying a CNC mill could benefit from this as well.  The Vixen tripod legs are rectangular:

 

I have produced a solution in the past that goes inside the leg, replacing the foot at the right side of the above picture.  It's a somewhat odd shape: hence the CNC mill.  But better to produce a rectangular sleeve like the others that slides over leg and foot and bolts on.  Trivial to make, lower labor costs and not dependent on the exact locations of the screws that hold the foot in the leg.  This way I can either reduce costs and increase profits, or reduce selling price and increase volume of sales.

Sherline Repair

I doubt most of you care about this, but I am putting it as a resource for others who might need help.

I followed instructions at http://sherline.com/5011uinst.htm for installing new Y axis backlash nut  Problem is that the nut does not go on the leadscrew easily.  It requires grabbing it with channel locks to turn on to lead screw, which chews up the brass barrel of the nut, preventing it from sliding into the recess where it goes.  Trying to turn the lead screw back into the X-axis carrier base, requires turning the nut at base of motor mount, but that rapidly unscrews from the base because it is threaded the opposite direction of the lead screw.   Solution, Loc-Tite to prevent turning.

I ordered two more backlash nuts and a new lead screw (which was $15.50--much cheaper than I expected).  Why the new lead screw?  Because I suspected the problem wasw damage to the lead screw, and Sherline warned me that they thread their lead screws about .100" oversize, and just using a LH 1/4"-20 die would cause backlash problems.

The new backlash nut turned onto the new lead screw without difficulty, but the nut would not turn into the  hole, so I assumed the problem was debris in the hole and I removed the lead screw.  But removing the nut was harder than going on, and after verifying the nut would slide into the hole just fine by itself, it would not go easily back onto the lead screw.  It looks like the place where the set screw through the table hits the lead screw is the problem.  (Don't screw it down so tightly next time.)  I used a file and sandpaper on that section without success.

Because the problem was apparently damaged threads on the Y lead screw(s), and Sherline makes them .100" oversize, I used an adjustable die open as far as possible.  As I had hoped, the die met no resistance except in the sections where backlash nuts had stuck. As a result, I can now turn the backlash nuts down the lead screw(s) by hand without difficulty.  So I reassembled, and even though the star washer is grabbing the projections on the backlash nut, the backlash nut still reverses out of the hole, failing to pull the table back on the Y axis.  I tried stacking two star washers to increase the range of spaces to grab the backlash nut, without success.  I am rapidly developing gobs of experience on repair of this mill, but I still don't have a working CNC mill.

Some pictures.

Sherline Repair

So I needed to replace the backlash nut on the Y axis lead screw.  The replacement nut did not go on easily at all.  I ended up grabbing it with channel locks to turn it on the lead screw.  The nut is brass, so the lead screw effectively acted as a tap.  But it still wasn't turning smoothly, so I went hunting for a left hand 1/4"-20 tap and die.  The tap came from Industrial Hardware in Garden City; they were surprised to find they had one.  No left hand die; but I found that at Idaho Machinery & Supply in Meridian.   This is a dangerous store for an aspiring machinist to enter.  Next door is even more dangerous: the Haas CNC factory outlet store.

Anyway, I was warned by Sherline that using my own 1/4"-20 tap on the nut and die on the lead screw would cause backlash problems.  Why?  Isn't 1/4"-20 a standard?  But I settled for tapping the replacement backlash nuts, which now turn onto the lead screw pretty easily (which I had also given a quick rubbing with sandpaper, which might have been what it needed to get the nuts on smoothly anyway)..  The end of the nut that slides into the X table base was a bit chewed up by the channel lock, so I put the nut on the tap, then put the tap into the lathe, and smoothed down the section that inserts into the table base.  Trying to hold the nut in the 3-jaw chuck by itself was pretty much impossible because of its size.

Okay, the nut now moves smoothly on the lead screw, but the remaining problem is that the backlash nut has a series of projections which lock into a star washer (also made of brass) immediately adjacent on the table base.  This prevents rotation of the backlash nut on the lead screw.  It also somehow forces rotation of the lead screw to advance the X table forward, although I don't see how.  In this case, the nut advances without dragging along the start nut and therefore the base, so I am still unable to work.  I think the projections on the nut were damaged while trying to put this on the lead screw the first time, so I ordered some replacement nuts ($2.75 each) and in case I need it, a replacement lead screw ($15.75).  I hope to be operational tomorrow.  In the meantime I am learning an enormous amount about repair and maintenance of the Sherline mill.  I really just want to get these six pieces made that I need for customers, and then consider moving up to a more serious CNC mill.

Why My Wi-Fi Did Not Work in Workshop

I had had enough problems with aluminum chips on the notebook that controls the mill, and he cardboard separator I was using to keep mill and notebook isolated kept falling over, so I moved the notebook to a slightly different position, and wi-fi again works.  I think the blade on the chop saw was blocking the signal.

Help

I have a Sherline 5400 CNC mill, and I have a problem. The problem started as: The manual that Sherline hands out with their CNC mill emphasizes that you should know exactly what each command will do before running the program.

I had wanted to improve the finish on the completed parts, because there is a series of stripes in the direction of milling. It is only appearance; there is no detectable edge, so .001" or less difference. But I thought I would smooth it perpendicular to the direction of the milling. So I modified the excavate program to mill in the X direction, then Y, instead of X then Y. The excavateyx program would be called with parameters to remove .0001" of material.

Fine. I wrote the program, and failed to notice that when computing the end of the X movement, I needed to subtract the mill radius from the destination. The right test strategy would have been to start the program, single step each line, then override the feed rate for the first line. Nope! The mill banged into the edge of the workpiece at 6 in/min, the rotation then yanked the Y axis towards the table hard enough to strip the teeth from two star gears that control the Y axis feed. Fortunately, these are $5 for both parts, but I ordered two, in case I screw up again.

The problem is that to replace the antibacklash nut on the Y axis lead screw, you have to remove the old nut, screw the new one on, and then reinstall the lead screw. Easier said than done. I moved the X axis far enough left to unscrew the set screw that prevents the lead screw from coming out, unscrewed the stepper motor mount from the base, and unscrewed the lead screw. (It is a left hand screw thread, just to humble me.) Okay, the new antibacklash screw did not want to go on the lead screw easily, but eventually, I managed to grip it with a channel lock, and get it on the lead screw. It turned freely for most of the screw's length, but it would not slide easily into hole for the backlash nut, probably because the channel lock deformed it. The same thing with the other backlash nut. I have a solution to this--sand the outside of the nut.

The bigger problem is that the lead screw does not want to screw back into the X table. If I use the 3/8" nut just in from the motor mount housing, turning CCW (left hand thread on lead screw), it just moves up the lead screw instead of turning the lead screw in. How to solve this?

UPDATE: I see evidence the nut was held by some LocTite-like adhesive in position in the past, so I applied LocTite to keep it from moving up  the lead screw.  I am also buying left hand 1/4"-20 die and tap tomorrow to clean up the lead screw and the brass backlash nut.

Struggling With My Brain

I sometimes wonderb how disabled I am, but the last few days of programming this CNC mill have persuaded me that it would not be wise to return to software engineering.  The manual that Sherline hands out with their CNC mill emphasizes that you should know exactly what each command will do before running the program. 

I had wanted to improve the finish on the completed parts, because there is a series of stripes in the direction of milling.  It is only appearance; there is no detectable edge, so .001" or less difference.  But I thought I would smooth it perpendicular to the direction of the milling.  So I modified the excavate program to mill in the X direction, then Y, instead of X then Y.  The excavateyx program would be called with parameters to remove .0001" of material.

Fine.  I wrote the program, and failed to notice that when computing the end of the X movement, I needed to subtract the mill radius from the destination.  The right test stratgey would have been to start the program, single step each line, then override the feed rate for the first line.  Nope!  The mill banged into the edge of the workpiece at 6 in/min, the rotation then yanked the Y axis towards the table hard enough to strip the teeth from two star gears that control the Y axis feed.  Fortunately, these are $5 for both parts, but In ordered two, in case I screw up again.

Not Everyone Uses CNC

I am building a part that replaces the bottom of the tripod on two Vixen telescope tripods: the HAL-110 and HAL-130.  They are the same shape internally, but the screw holes are slightly differently spaced.

I have historically drilled and tapped the holes in my part by using the legs as patterns.  For the CNC project, best to get the locations exactly and use the mill to drill pilot holes.  But it the spacing is different, and there is no symmetry at least at  the .001" scale.  My guess is that some badly paid Chinese worker has a pattern for the holes for each leg.  I just hope that he doesn't damage it, because then I will have to revise my program.

Generalizing The Solution

I have received a number of useful suggestions on machining from readers gfor which I am very grateful.  Centering the wotkpiece in the vise using parallels to lift it up into position means a bit firmer grasp on the workpiece.

I also decided to come up with a more general solution to excavating chunks of aluminum.  I had originally written C programs to produce gCode for each machining operation, and they were very specific to the face being machined.  This was bad because the definition of width varied depending on work[piece orientation.  I just wrote a more general solution.  Instead of parameters for length to cut, width, height,etc., it now takes six parameters: xs, xe, ys, ye, zs, ze for the X, Y, and Z starting and ending points.  Then it positions the cutting tool appropriately and does the excavation using the feed rate and cutting depth specified on the rest of the command line.  Several different programs have now been replaced with one using different parameters for different faces.

Starting A Production Run

I have commented before about how the Sherline CNC mill is not powerful enough to do large scale production work, but it still amazes me that I can afford a tool like this! 

I did have an odd failure last night.  The Y axis was suddenly not going where it should.  The coupler that locks the stepper motor to leadscrew had disengaged.  This was easily fixed, but one more thing to check before starting the mill.