I should soon be pulling finished structural components of my CNC build from the maw of this machine (if we want to adhere to technical accuracy, replace "pulling" with "handed" in the preceding sentence).

Every minute spent on the internet is a minute that I am not pushing this into being. This is the sort of under breath muttering the passer by will hear from the apparently distracted bearded individual as he bicycles to and from work today.. tomorrow... etc.

Tore apart the Ratelco 24 volt battery charger in order to assemble a very primitive unregulated power supply for the stepper motors of the CNC project.

Essentially this entailed ripping the mains transformer, rectifier block and caps out of the supply, leaving all the current sensing & regulation for the parts bin. The bay in the frame is 5U tall, and the Ratelco was 6U tall, felt good to tear it apart.

I recall reading that a rusty transformer has one foot in the grave and another on a banana peel. That's just the sort of amusing imagery it takes to stick something in my mind for years. Not that I'm going to adhere to such a sentiment, mind you. I fully agree that stopping further advance of rust is a good idea, in so far as stopping the advance of rust is reasonably possible at any rate. After embarking on a read of Practical Transformer Winding I was reminded that the oxide layer of rust performs much the same electrical function as the laminate varnish insulation, and while infinite resistance is highly unlikely, the resistance should be high enough to effectively isolate one laminate in the stack from the next - so long as moisture is removed from the equation. So, next time I photograph this thing it will have a fresh coat of paint on it.

Here's a rough layout of the PSU, I will bolt down the base plate before assembly. The 1N1188 bolt on DO-5 diodes purport to be good for 35 amps, I don't see the four motors pulling that even if each is under full driven load at the same moment, so I may leave them even though the heat sink will be hot (electrically) which is something I'm not too excited about. Still on the drawing board there.

There are a few windings here that I don't have a direct application for, time will tell if they'll serve a purpose. The original circuit was designated 24 volts, it provided roughly 28 volts unloaded, if I can boost/buck to bring it up to 30 or so I'll still be well under the 40 volt cap of the 64000 uF capacitors.

I think this served as current sensing in the battery charger, I'm still contemplating applications. There's a distinct lack of standard objects for scale reference in this post, sorry.

Mayday! Time to get some progress and self serving industriousness laid out, and resume paving the path to the future.

Forging a bit of progress on the CNC project. Smoothing out nicks and surface irregularities of the structural material through a gentle file hone and fitting the bearings onto the screws as outlined following.


In preparation I stuck the screws in the freezer for the afternoon, and the bearings in an oven (initially at 250 f) for ten minutes. I forget the actual room temperature dimensions, as they were compared at ODD a while back, but it did amount to an interference fit. The hope here was that between thermal expansion and contraction a fit could be made without the necessity of an actual press.


Note the frosty condensation at left offset by the heated metal at right, the metal cylinder was the bearing race "anvil" to alleviate hammer shock from being transmitted through the bearing itself. The first two involved hammering at the coupling end. This wasn't a huge concern as the coupling end is not a super critical precision surface as it will be floating to allow drift from minor misalignments to be taken up at the coupler as opposed to premature wear on the bearing surfaces (no mushrooming occurred).

The second bearing was a bit more resistant to installation, probably due to the cooling of the bearings during the first installation; so the third bearing was heated at 350 for five additional minutes. Additional heat was the key, as the third bearing required no hammering at all.


Installed, ready to move on to structural plotting, based on actual measurements of installed screws.

Made another metal run yesterday, now I have the fundamental material for the gantry on hand.


Here's a cross section shot of the X axis gantry beam, it's a touch more involved than the SketchUp drawing I drafted in October. One point of concern I have here with the current loose mock-up is the linear bearing rail situated over the blue faced material will require a countersink channel on the inner angled face of the C channel wall to accept mounting hardware (unless I opt for threading the holes in the C channel, which is a design tactic I'm avoiding at the moment).


My main objective is to confirm the clearances I came up with on paper. The dimension of that nut pushed me into an adaptive frame of mind.


The bearings / Z axis car has a clearance of 0.2865 on paper, which adheres to my desire to keep roughly 0.25 clearance between moving parts. However the nut, suspended inside the channel as pictured, has 0.0335 inches of clearance. Initially my thought was to modify the nut to reduce the chance of interference, but there's also the matter of the bracket that ties the nut to the Z axis car. Not to mention that the nut itself is a point of wear, and will eventually need replacement.


I want to keep the spatial displacement of the axis' as small as possible, as off axis distance equates to magnification of slop. My current plan is to "elevate" the screw in the channel to position the nut so the collar rests on the inside face of the Z car, essentially putting 0.3730 of the mounting flange into the baseplate of the Z car. Since placing the flange on the outside edge of the car will reduce usable travel by roughly 0.3785 (widening the car), I plan to cut a slot in the baseplate of the car to accept the nut. This will remove one screw, the remaining three will tie into an L bracket milled into a horseshoe shape on one side.


By repositioning the screw the motor mount hardware will clear the 3 inch C channel, which will come in very handy should I need to R&R the NEMA23 motor. Not pictured here is the 0.5" thick "wall" that will cap off the X axis channels, provide the plane to mount the motor & bearing at far end of screw and tie into the legs that are not pictured, but are rendered in the drawing below.


So, visualize a 6x14" plate of 0.5" thick aluminum bracketing the ends (and obstructing the view of the cross section) and you've got it. I may have to remove some material from the car side of the upper branch of the 8 inch leg channel, but that will wait until I've got the gantry assembled and am mocking up the car.

It took me a week to execute a simple task. Referring back to the interference fit for the Z axis screw, my planned fix involved cleaning the threads on both parts and filing down the folded thread on the screw.


My initial mock up here, in which I've placed a knife in the channel between threads to act as a protective barrier between my filing work and good threads, proved cumbersome. In the end I simply held the file as I would a pencil and exerted more care than slop.


Success! Sometimes the fastest fixes are the most satisfying.

Next phase of the CNC project entails getting metal milled for the gantry. Once the gantry is solidly defined I'll build the frame on which it connects to the table, most likely out of wood since my budget is wearing thin. I reason that the framework is a bolt in layer, that can be upgraded if wood proves inconsistent in dimension.

It also looks like my initially just hoped for feature of a spindle RPM control for the Porter Cable 75192 fixed speed 3 1/4 HP router motor is achievable with the Super-PID closed loop controller. I think I would have easily burned up more than twice the cost of that trying to engineer something myself.

Have run into a bit of a back and forth in terms of getting my transport sorted.


These parts are so close as to not actually work as designed. This brass nut was sent as a replacement for a mis-picked Delrin plastic nut (originally shipped in the larger size of the other screws). I'd like to employ the clearly higher quality metal part, but need to make them work together.


My hopes are that the out of whack thread starting at the point it is turned down are what is prohibiting me from threading the nut. I'll file those bent ridges down without side loading and hope that allows freedom of movement.


It's either that or tapping out the nut.

Inching forward with the CNC project, my screws and some hardware have arrived.


Here's a quick mock-up of my X axis, well, hardly a mock-up, but I digress. Framework and data handling aside, these are the important parts.

X axis is to ride on THK linear bearings, driven by a Pacific Scientific Powermax II stacked stepper turning a screw, powered through a GeckoDrive G201X which will receive commands ala dedicated desktop PC running Ubuntu/EMC2. Having all these components laid out in front of me means I really need to address the power supply sooner than later (now).


Front and center sits 112 pounds of power supply, waiting for me to lug it to the bench.


It's only after executing that little juggling operation that I remember this is the unit with a faulty breaker on the DC side and what I'm really after is the 107 pound power supply that was sitting next to it.


After a brief inspection confirming that the power transformer is strapped for 120 volts and nothing is visibly leaking, bulging or dead inside the case, I fire it up. The hum of operation borders on a buzz, but no smoke or sparks thus far... good.


There's a standard cassette tape at upper right for scale.

There's a point of concern with this PSU, the DC breaker. I've gathered that it's inadvisable to to switch the DC side of the power at the drivers, as they will face back EMF from the motors which can burn them up, setting me back replacement costs on up to four G201X, which isn't something I'm interested in.

The supplies are also tailored to battery charging, which has a different set of operational parameters than driving motors. My Y axis will have two motors working in hard sync to move each leg of the gantry, the two stacks of each motor are going to be wired in parallel, offering a path rated at 7 amps per motor. So the last thing I want is some battery configured current sniffer to crowbar output voltage when I most need it.

In short, I came to the realization last night that my best tactic is to gut the logic/regulator portion of the power supply, reconfiguring it into a primitive rectified DC beast, which, as I understand it, is best for this brute force application to begin with. I guess I could have stayed with the heavier supply after all.

It would appear (at least in my convoluted install/current rung on the learning curve) that the precision capabilities of SketchUp are somewhat limited, though being able to confirm my visualizations and having drawings upon which to base further visualization is well worth the quirks. I've executed reasonably correct renderings of the actual metal and finished dimensions of the gantry framework. Omitted (in addition to all linear movement components) are the end caps that will assist in rigidity and provide for motor and screw mounting.


Homage to TRON:


I doubt I'm going to wind up rendering the entire machine in 3D, since I can't use the files for anything other than eye candy. So expect some really boring 2D CAD stuff in the future.

Speaking of eye candy, here's a gutshot of the oscillator bank in a Baldwin 45 organ. This picture amounts to being essentially gratuitous since the Baldwin 45 is really, really low on my to-do list. I do really like their approach at the voltage buss (seen at bottom) though, one of the grey area aspects to a tubed modular.


I've sort of committed myself to guts or gear pics with each post, so while I'm busy on the data entry side of things I'll continue to dig up more or less random shots from the archives. I expect some actual current electronics shots soon.

After assigning permissions and editing the virtual registry of my simulated windows machine (in wine, so I'll name it Drunk) I've successfully launched Google SketchUp 8 in Ubuntu. The following is a screenshot of a rough (and partial) render of the plotted gantry:


I'll spend the next few coffee soaked mornings drafting dimensionally correct components and fitting them together.

I had the good fortune to play with an early 70s PAiA modular last night, it fortified some of my ideas about future builds. Mainly, the elusive aspect of personality. I aim to harness a reproducible approach that emulates aged and failing parts without the sense of impending doom.

On the guts front, here's a couple shots of my 1973 Fender Princeton (queued up for a recap).



When I bought it there was some serious anaemic distortion going on. As it turned out, the ground reference to the phase inverter had lifted. I'm hoping for some robust distortion after recap and a little tube rolling.

In terms of being a documentarian, I seem to be having one of those weeks in which all of my forward push is transferred directly to a wheel I'm standing within. Bit pieces for the CNC project have been trickling in, I dug through the aluminium scrap heap at Coyote Steel last Friday and procured some C channel that will comprise my gantry. Before posting about that, however, I'd like to have some sort of 3D rendering/CAD cut-sheet of my gantry to accompany what would otherwise be a dull picture of a pile of metal.

This is where I'm reminded that my passions do in fact not lie in the realm of twisting software around to my needs and flexing my digital muscles. Shame that, since I'm morally locked into the Linux philosophy, which requires that I think for myself (ie: fit together digital puzzles) once in a while. So this morning I've installed wine so I can have another go at Google Sketchup, which as I recall has a nice and soft learning curve. We'll see if my mild apprehension in getting nested software to run is warranted. Then I'll be able to bang out a proper CNC update with eye candy.

Also, on the machinist front (remind me to self apply that label only AFTER I am pulling finished work out of the mill), I picked up a Mitutoyo digital caliper over the weekend, which will help me immensely in committing the correct dimensions to the drawings I'm about to draft.


For the record, these knobs are in fact different sizes.

And, so as not to rob ourselves of the pleasure of looking at guts, here's an internal shot of a Leader Electronics Corp. LCG-388 Color Bar Pattern Generator. Hailing from the days in which a television was repairable (I'll let you do the math on that). Given the compact nature of the unit I fully expected to see some ICs in there but no.. THIS:


Discrete transistors, germanium diodes and a lot of the letter "3".

CNC Build, part one.

So, in order to be able to make two sided PCBs, cut face panels/enclosures and basically be able to reproduce multiple units of various things, it has been decided I need a CNC mill. I've been buying bits and pieces for going on a year now, and today I hauled home this old metal cabinet that will perform the duties of foundation to the machine:

I've laid my linear bearings out in a sort of imaginary mock-up, the gantry pair (horizontal in the pic) are pictured laying back and will actually be situated so as to support a vertical Z axis (the shortest of the lot). I'm planning on housing one of my huge 24 volt power supplies inside the case, greatly reducing tip over factor. Here's the fast and dirty low down of the planned build:

Phase 1) Work area 18x18x6 inches, moving gantry design. Linux EMC2 controlled.
Phase 2) Fabricate superior framework.
Phase 3) Vacuum table
Phase 4) Computer controlled variable speed control for the AC motor

We'll see how far down that list I get, I'm just happy to finally be able to photograph something besides to contents of boxes for this!