November 20, 2015

Shiny Lathe Pictures

Here's a pile of tiny lathe pictures:

Since taking these, I've also re-cut the spindle taper in place to eliminate runout, mounted an encoder to the back of the motor for closed-loop spindle speed control, and leveled the tailstock.  

November 3, 2015

The Tiny Lathe is Working!

After great struggles with the lathe electronics, the lathe is finally working.

One of the main challenges was form factor.  I wanted all the electronics to fit into the base of the lathe, and I wanted the base to be thin (~25mm thick)  so I needed a 600 watt, 45-50 volt power supply that fit in a 25 mm thick package.  Bayley suggested using a Vicor DC-DC converter to generate 48 volts at 600 watts.  Vicor makes some extremely compact converters perfectly suited for this, and they can be had pretty cheaply on ebay.  Only downside is that they need around 300 volts in to turn on.

The basic electronics system would then be wall to doubler for 340 VDC, doubler to Vicor converter for 48 V, and 48 V into an AMC servo drive donated by Dane to drive the spindle motor.

This seemed like a good idea, so I got an appropriately sized Vicor brick on ebay and built up the rest of the system:

And so began the the second challenge: regen.

When the spindle rotates, kinetic energy is stored in the rotating inertia.  When the reference spindle speed is suddenly turned down quickly, the extra kinetic energy in the spindle is converted back into electrical energy and shoved onto the DC bus.  This is all well and good when your power system is supplied by a rechargeable battery, but is much less okay when your source is the wall.  This problem manifests itself as skyrocketing bus voltage when the spindle reference is set to zero from full speed.  The spikes were on the order of 20-30 volts despite the huge bus caps used, which was enough to cause the servo drive's over voltage protection to kick in and shut it down.

One solution is a shunt regulator.  The basic idea is to have a circuit which looks at the bus voltage, and if it sees the bus voltage rising too high, short the bus through a resistor to burn that energy.  Sounds easy enough, should just be  comparator,  FET, big resistor, and some passives, right?

Version 1.  The TO-247 is a big diode to make sure any voltage spike doesn't make it back to the power supply:

Version 1 was derp because the comparator sometimes couldn't pull down the mosfet gate hard enough by itself, or something like that.  The symptom was that occasionally the FET would latch on.

Versions 2/3 had real gate drivers, and actually worked:

At this point, my electronics (minus Vicor brick) looked like this.  Getting less attractive, as I had to shove in an extra 10 V supply to run the shunt regulator.

More problems.  The Vicor converter I got on ebay was DOA.  I returned it, and ordered another identical one from a different seller.  New one shows up and is also DOA.....

Time for something different.  Bayley pointed me to another Vicor product, Vichip BCM converters like this one.  Normally these cost ~$75 a pop on Digikey, but can be found on ebay for $10.  These are unregulated, fixed ratio DC-DC converters good for 300(!) watts each in about a square inch of chip.  They output 1/8 of their input voltage, so like the other Vicor converters I tried, they needed >300 volts in to get useful voltage out.  They also come in a funny surface mount package.

I ordered a few of them, and figured while I was at it I should lay out a PCB for the doubler, Vichips, and shunt regulator:

The shunt regulator is the top left corner.  The large D2-Pak devices are the fet, power resistors, and diode.  In the middle are 3 Vichips, and to the right the rectifier and caps for the doubler.  For gate drive and logic power there's a TI LM2594 12V buck IC.

Through the magic of 3pcb, boards appeared:


The board required a little fixing to get working.  I left the comparator ground pin unconnected accidentally (the net in the eagle schematic looked connected but actually wasn't).  More importantly, I didn't read the Vichip datasheet carefully enough, and missed two important details.

First, they really don't like output capacitance.  I had to remove all the output capacitors to get the thing to turn on.  Instead you should put the caps on the input.

Second, the Vichips can be expected to draw 6.5-13 W idle.  They got seriously hot just sitting there doing nothing.  A kill-a-watt confirmed that the board consumed 22W idle.

I didn't design the board for easy heat-sinking, but I was able to make it work by machining an aluminum spacer which gets sandwiched between the Vichips and lathe base with some thermal pads.  I forgot to take any pictures of the final electronics assembly, but here's the first chips generated under the lathe's own power (not a bench supply):

And a beauty shot:

I'lll do another post with more shots of the whole thing soon.  The lathe is currently functional, but there are a number of things I'd still like to do, such as:

  • Take a finishing pass of the inside of the ER-32 taper to guarantee its concentricity with the spindle.
  • Finish the new tailstock riser.  I've machined it, but mismeasured somewhere and it's a millimeter too tall.
  • Add closed-loop spindle speed control.  
  • Make chuck backplates so I can use real chucks, not just collets.
  • Add a leadscrew and motor for power feed.

October 17, 2015

Ball Turner

I recently stumbled across a neat design for a ball turner on youtube, and ebay-rage-ordered all the parts to build one.

It consists of:

- A CXA (fits the MITERSlathe) boring bar holder with a 3/4" split bushing
- The cheapest 2" boring head I could find
- A 3/4" shank for the boring head

Total damage was ~$100.

I turned a couple 3/4" I.D. thrust bushings:

A handle:

The little radial-keyway allows the handle to be rotated 180 degrees and still have positive interference with the back of the boring head shank (which I filed a matching keyway into).

Threw together a quick carbide insert holder.  Someone else was on the mill, so all the shaping was done with a file, belt sander, and bench grinder:

Almost done.  The two set screws are used to tighten the play out of the split bushing:

To test it out, I turned a ball for the end of the handle.  Success!

The ball was tapped, threaded onto the end of the handle, and locked with some red loctite:

Not bad for an afternoon.

September 7, 2015

Motor Mount and Bearing Shields

l spent way too long agonizing over how to mount the motor.  My basic requirements were:

  • Adjustable motor position for varying pulley sizes
  • Must only use one fastener to loosen the motor
  • Looks nice
  • Elegant, for some definition of elegant
To hold the motor, I made this piece which clamps around the motor housing (clamping bit not in this picture).  I didn't use the mounting holes on the face of the motor to leave more clearance around the pulley.  The ID was cut with a boring head, and the outside was roughed to an octagon, and hand-filed to round:

A steel square nut sits in a slot, so the motor position can be adjusted:

Here's the interesting part.  Rather than just bolting the motor through the slot, I cut a 45 degree counterbore in the plate that holds the motor, and a matching 45 degree brass washer.  When the bolt is tightened, the washer wedges the motor plate down into the aluminum plate below.  This way, the motor is firmly located against the metal both behind and below it.

I also made some simple bearing shields to hold the spindle oil in and stop chips from getting in the bearings. 

In the next episode I'll put the electronics together.

September 5, 2015

Tiny Tool Post

A tiny tool post for the tiny lathe.  The body of the tool post is 35 mm on each edge.  The standard tool holders  up to 3/8" lathe tools, and the boring bar holder is also for a 3/8" shank boring bar.

Brass wheels for setting the tool height were knurled on the MITERSlathe

A steel t-nut holds the tool post down to the compound slide.

Turning and boring tool holders:

August 27, 2015

Mini Lathe Pulleys

I'm back from the left coast, and for once I'm not busy building giant wooden contraptions for orientation.  Instead I'm trying to finish the tiny lathe before classes start.

I made some 2L sized v belt pulleys for the motor and spindle.  I considered using timing belts, but the spindle's going to be pretty high speed (>7k rpm max) and figured v belts would be a bit quieter.  Not to mention, v belt pulleys are much easier to machine.

The pulleys were cut turned using a grooving tool ground from a HSS blank:

For the motor pulley, I had a lot of trouble getting a clean 9mm bore using the MITERS small boring bars, so I tried a technique seen here for making reamers.  I turned some 9.5mm semi-hardened steel shaft down to just under 9mm, and ground three facets in the end.  The faces were cleaned up on a whetstone, to debur the corners and get a sharp edge.  I drilled a slightly undersized hole, and expanded it to final diameter with the reamer.  The results was a pulley with a nice 8.99mm bore, which I tapped onto the motor shaft with a hammer.

Both pulleys are retained by set screws.  Normally I greatly prefer split collar style clamps to retain round things, but in this case it was hard to beat the compactness of set screws.  And it's dealing with fairly minimal torque compared to most things I build.  On the spindle side, the shaft diameter is fairly huge, which makes the force on the set screw smaller, and on the motor side, the pulley's press fit should deal with most of the torque.

On both the motor shaft and spindle I milled a little recess for the set screw to sit in:

Also notice the pair of lock rings for preloading the spindle bearings

And here's the spindle and motor pair:

Next up: Building a tiny quick change tool post.

June 7, 2015

Spindle and Headstock Machining

Machining the spindle turned out to be kind of a disaster, but I was mostly able to salvage things.

I started out with a chunk of crufted 1.625" steel round of an unknown alloy.  For whatever reason, I found it pretty much impossible to get a good surface finish on the stock with the MITERS carbide insert turning tools.  The only way I could get a finish resembling anything reasonable was to crank up the lathe's speed to the maximum and take really heavy cuts.  Unfortunately, this meant it was impossible to take finishing passes, so I screwed up the surfaces that mate with the inner spindle bearing races, and made them a little too small.  I did the final surface finishing using sand paper pressed against a precision ground steel block.  This seemed to fix the surface pretty nicely, but I was a little overzealous with sanding, and took off at least a thousandth too much off the front bearing interface, and a tiny bit too much off the back interface.

Since I didn't have an extra piece of steel to start over with, I attempted to salvage the spindle by using the newly acquired MITERS knurling tool to expand the shaft at the bearing interfaces.  It did work, although I lost a little concentricity in the process.  Fortunately, I can virtually eliminate the spindle's runout by taking a final pass off the critical surfaces on the mini-lathe itself.

Some design notes about the spindle:
- The front of the spindle is tapered and threaded to accept ER-32 collets.  I did the M40 x 1.5mm threading on one of the fancy Prototrak CNC lathes in the new N51 CNC shop.  This was a huge pain to set up.  All I wanted was a menu to input thread pitch/feed per revolution and a lever to engage/disengage the autofeed, but instead I had to actually program a job to do all the passes for me.  It took many hours poking through menus, reading the manual, and asking for help to finally figure it out.  Conclusion:  I should find MITERS a set of change gears for the Clausing lathe to approximate metric threading, so I can do my metric threading there.
-The spindle is extremely disproportionately large in diameter for a lathe of this size, and for good reason.  The bore through the spindle is .875" in diameter.  A stock taig spindle has a bore of just over .375", so this will be able to hold much longer large pieces of stock.

To put the hex at the front of the spindle, I grabbed a hexagonal 5C collet holder and fastened some scrap stock in it.  I then put the matching ER-32 collet and holder on the spindle, and tightened the two collets down face to face.  The vice stop made this a quick mill-and-rotate operation:

The headstock was machined from the wonderful 4"x4"x12" billet I scavenged last year.  I bandsawed off a 4.5" chunk, roughed it to size, and fly cut all the faces:

The quickest and most repeatable way I found to measure it was using a granite block scavenged from and old atomic force microscope, and those measuring surfaces on calipers that everyone seems to forget about.  By tramming the Bridgeport (a relatively rare occurrence at MITERS) and tapping down the block with a hammer every time I tightened the vice was able to get a thousandth or less variation between the corners on each face.

The headstock grips onto the dovetail bed.  I cut the slit for the clamping mechanism with the Epic Slitting Saw.  It made the poor Bridgeport's spindle sound extremely unhappy, but did a nice job.  Two counterbored M6 screws tighten down the dovetail to lock the headstock in place.

Bearing bores completed.  The tapped M4 holes are for a bearing shield on the front, and a motor mount on the back:

And here it is assembled with bearings.  Both the inner and outer races of the bearing are tightly pressed at the front.  At the back of the spindle, the outer race is a sliding fit held with retaining compound, and the inner race is a light press.  It should possible to disassemble without trashing the bearings by heating to loosen the retaining compound.

I have yet to machine the nuts that will thread onto the back of the spindle and preload the bearings with a disc spring washer.

And here the lathe updates come to an end, as I'm currently ~3,400 miles away from MITERS.