The View Up Here

Random scribblings about kites, photography, machining, and anything else

Archive for December, 2009


Posted by Tom Benedict on 04/12/2009

I’ve owned a benchtop CNC mill since the middle of 2000.  The first real machine tool in my shop was my Taig lathe, so when it came time to find a CNC mill to go with it, the Taig mill was a natural choice.  The two tools have identical spindles and take much of the same tooling, so this was a money-saving measure at the time.  But in terms of transferring work from one tool to the other it has added benefits as well since the same chucks will fit the lathe spindle, the mill spindle, and the rotary table on the mill.  All in all, it’s a nice combination to have in the home, even if it is a little small by industry standards.

My shop has seen several moves over the years, the most recent being a move from Texas to Hawaii.  It wasn’t the kindest thing to do to my tools, but a set of good stout shipping crates kept things from going too horribly wrong.  Even so, the performance of the mill began to degrade once it arrived, and a little over a year ago it finally gave up the ghost for good.

Luckily none of this had anything to do with mechanical damage of any sort.  I’ve tested the tramming on my mill over the years, and unless I do something really stupid, like dig in a tool and keep driving one of the axes, it’s never had any problems of that sort.  I replaced the spindle motor on both the lathe and the mill with 1HP variable speed DC motors at one point, which was a good change for the lathe and a vast improvement over the mill’s stock 1/10HP AC motor.  Other upgrades came along as well, such as a relay box for the mill’s spindle and coolant, a quick change tool post system for the lathe, and a number of other things.

But once the mill was dead, none of that really mattered.  So it sat.  And I fumed.  And eventually I more or less walked away from it.

If it had died a violent death with smoke and loud noises and all the trimmings, I probably would’ve had an easier time of things.  Instead it died a slow death of having random position loss in various axes.  Parts started to come out wrong, and toward the end I couldn’t make a “there and back again” pair of moves in any of the axes and have it come back to the same place.  The thing still moved, but in essence it became useless as a machine tool.

Digging into the electronics indicated that there was a problem in the high voltage power supply.  But as with most problems, that was just the symptom rather than the cause.  Digging deeper, it became apparent that the mill’s electronics weren’t designed well in terms of heat extraction.  Over the years they had run a little too hot a little too often, and things were starting to fall apart.

Despite what people may believe about computers, heat doesn’t instantly kill electronics.  But it does reduce its expected lifetime.  Every electronic assembly carries with it an expected mean time between failures, or MTBF.  It’s been a while since I’ve done an MTBF search, but once upon a time most hard drives had an MTBF in the several tens of thousands of hours.  Running a hard drive hot wouldn’t instantly kill it, but it would reduce the expected time before failure for that individual device.  A well-treated drive might have an MTBF of 20,000 hours.  One that was in a computer with a busted cooling fan might have its TBF reduced by a factor of ten or more.

Which is essentially what happened with my mill controller.  The big electrolytic capacitors in the high voltage power supply had cooked.  And chances are the FETs in the motor drivers were cooked as well.  Everything was suspect.  So I either faced a whole string of test-and-replace operations, with the certainty that it would fail again unless I re-designed the case and cooling fan layout, or I could skip all that and replace the whole mess.

I chose the latter option.  But not having the money readily available, I couldn’t actually make good on the plan.  So the mill sat.  And sat.  And sat.

Then the miraculous happened and I got a bonus at work.  Ever since I first bought my lathe and mill, my wife and I have had an agreement:  Bonuses are bonuses.  Half goes to the family accounts, but the other half goes to the individual who earned the bonus so they can enjoy the fact that they were rewarded for good hard work.  This is how my mill was purchased in the first place, and this was exactly what I needed to get it back up and running again.

I knew the controller I wanted: a Gecko G540.  It’s a beautifully engineered piece of electronics capable of driving four stepper motors at up to 48V at 3.5A.  My motors weren’t a great match for it, but I knew I could replace those later.  The G540 is just the driver electronics, though, so in addition I knew I’d need a power supply.  And a case.  And a cooling fan.  And an E-stop switch.  And a power switch.  And a whole host of other doodads that seem simple but add up fast.  In the end I skipped all that and picked up a G540-based controller from Keling.

But Keling is in China, so while the unit was shipping I had some time to think through everything else I wanted to do.  Since this constituted something of a “do over” from the standpoint of my mill, I made a list of annoyances, both major and minor, and started work.  The list is by no means finished, but here’s where it stands now:

  • Cleaner – At one point I switched to a coolant/lube that did something I’d never run into before.  Even better, I found there is a name for it: “varnishing”.  It had a tendency to cover everything in a nice layer of oil (a good thing for a machine tool!) and then dry out to a hard, tacky film.  This is other-worldly bad for tools.  I took the whole mill apart, cleaned everything to bare metal, and re-assembled it with good quality way oil on every moving part.  The goobery coolant/lube is now gone.
  • Storage – My mill work area has been horrid for storage.  So I ran an eight foot shelf across the top of the mill and lathe bench, and moved a number of tools that had previously lived underneath the mill bench to the shelf.  This keeps them clear of swarf, coolant, and other crud that seems to happen to everything below benchtop level.
  • Lighting – The lighting around the mill, and to a lesser extent around the lathe, has been gawdawful.  I mostly work in the evenings, but I can’t turn on the full lights in the shop once my kids are in bed because they can shine in their bedroom window.  I picked up some under-shelf kitchen LED lighting, and got a nice spotlight set up that points at the mill spindle.  These are switched from the same switch unit that powers on and off the mill controller.
  • Cables – The cables on the original mill motors were very short, and of uniform length.  Which makes sense from one standpoint, but was painful when trying to locate the mill electronics.  The one place that really made sense to mount the electronics was on the mill’s work surface.  Which is a terrible idea when it comes to keeping flying metal out of the cooling fan.  (Hm!  Now I know why there was such poor ventilation on the original controller!)  I installed longer cables that let me put the electronics on the shelf over the mill.
  • Computer – I’m still not where I want to be with this, which would be to run my mill using Mach3 controller software.  But I got a new installation of EMC2 on a faster computer, so my pulse stream to the mill’s motors should be cleaner.
  • Way covers – The mill moves things around using big stepper motors attached to lead screws.  If metal chips or coolant gets on the screws, it can seriously mess them up.  The way covers I had were basically plastic sheeting.  I replaced them with rubber accordion way covers from Little Machine Shop.  It’s a vast improvement that’s hard to appreciate unless you’d spent almost ten years using plastic sheeting.
  • New controller – The Keling G540 controller finally came.  I plugged everything in, set up EMC2 to use it, and away it sang!

One of the big drawbacks with the previous controller was that it did nothing to address something that happens with all stepper motors: midband resonance.  Every stepper motor on the planet acts like a wheel with a detent spring holding it in place.  To move the motor, you move where it wants to detent to by changing the configuration of the magnetic fields in the motor’s coils.  But it’s still acting like a shaft on a spring.  If you make the motor move to the next step by changing the configuration of the magnetic fields being generated by its coils, the motor will very quickly accelerate toward the new position, and decelerate once it hits it.  And it’ll bobble back and forth (very quickly and with very very small moves) kind of like a car with bad shock absorbers will bounce around if you jump on its bumper.  The real catch with this is that at a particular speed, you wind up hitting a harmonic of that bobble frequency, and the motor will resonate.  Typically this results in lost steps and lost position.  And the really unfortunate thing is that this happens at a low speed that you really do want to get past so it can run at high speed.

The G540, in addition to being a 10 microstep driver, also compensates for midband resonance.  So when I said I plugged everything in, set up EMC2 to use it, and away it sang, it really did sing.  The mill is quieter now, doesn’t hit that “growly” sound as it accelerates up to speed, and has no speeds at which it can’t really operate at all.  This is a massive contrast to the way things were.

My mill still isn’t 100% ready for prime time.  There is still some work to be done like adding home switches, installing relays and outlets in the Keling box for spindle control, getting the VFD output from the G540 to drive the Digispeed I’ve got on my mill’s spindle motor so I have closed-loop spindle speed control, and a whole host of others.  But much of that is in the category of “want” rather than “need”.  It can wait.  For now, I have a mill that’s almost ready to go.  I expect the first of the new parts will be coming off the mill this weekend.  Quite frankly, it’s overdue.

– Tom

Posted in Electronics, Engineering, Machining | Tagged: , , , , , , , , , , , | Leave a Comment »