[cgodwin]

I have a lower power driver than you, but the wiring looks to be the same. I'm using a Keling KL-4030 and a Keling 36V power supply which seems to have no issues pushing my 425 oz-in steppers. The wiring diagrams I got from Keling worked fine. Here is how I'm wired up:

ENA- No Connection
ENA+ No Connection
DIR- Axis direction pin on breakout
DIR+ +5V
PUL- Axis pulse pin on breakout
PUL+ +5V

My breakout board is a C10 by CNC4PC, which I purchased from Keling as part of my full set of electronics I got from them. So my breakout board side is quite a bit different than yours.

Like you I also have a dead pin either in my breakout or my parallel port. My did work for a little bit before stopping, which has me worried since I don't know why it failed. I haven't traced it yet, but a nice feature of the C10 breakout is the isolation chips are socketed so they can easily be replaced if I burned up a channel there.

[evengravy]

Hey, cgodwin,

Thanks for the info there. I think there's some differences for sure. I found some info on mycncuk forum which suggests my solution is working for others with the same breakout board and drivers here M542 wiring - MYCNCUK

I'm a little more confident that I have it right now, although some users are suggesting to disconnect the ena- pls- and dir- lines all together, presumably since the drivers are grounded to the same psu as the pcb they have a reference to ground anyway. I think il keep mine as it is for now and be carefully with my grounding. The isolation is a concern as you mention, maybe il put some opto isolators in between for peace of mind,

Do you know if any company makes a stand alone isolation pcb for between the parallel port and the pcb?

[evengravy]

So, its been a long time since I have updated the build, things have been hectic, I have been flat out building 3D printers and have recently became a father for the second time, not much free time.

Since the last post the machine has been painted, I chose standard external door paint by duluxe, seems to be very hard wearing so far. Oxford blue for no particular reason apart from it was the most attractive in the small selection in store locally.

I sourced some low viscosity epoxy resin, 1.5kg of it (calculated at almost 3mm depth for the volumetric of my extended base) this epoxy was a two part 2:1 ratio epoxy to hardener (I chose the slow hardener) and was provided by a crowd called Reactive Resins in the UK. They have an ebay store if anyone is interested and prices were pretty reasonable.

So a few days back I levelled up the base in preparation for epoxy pouring, this was done with standard builders levels by stabila, obviously I would have loved to have access to a machine level but maybe that's just overkill here. Rather than level the machine in reference to the uncoated base as suggested in the build documentation I chose to level with respect to the Y axis rails, it turns out my base was not very flat so I thought it best to level from the rails on which the Y axis will run. I feel this worked out well so far.

The epoxy was applied two days ago and has taken until today to cure (maybe the slow hardener was a bad choice) due to the temps in my "workshop" being quite low (11 degrees C max). I didn't degass it or anything just mixed it in one container, transferred it to another clean one and mixed it again to ensure thorough mixing.

The resulting base is exceptionally flat, to my eyes at least, and has given me a nice surface to start to measure upward to the rails in order to get them as accurately levelled as possible.

Today I began to level the rails with respect to the hardened epoxy base and began by creating a jig out of a flat piece of stock and a bent piece of 8mm threaded rod with a sharpened end. The build doc outlines the procedure of using feeler gauges to determine the clearance between the jig point and the rail underside. I quickly found this to be rather a painful process (i have very little patients) and was quickly abandoned in favour of another method.

I had a look through my stock to see what else I could come up with and came across a Dial Test Indicator that I had previously bought to set up the delta 3D printer I constructed over christmas. I spent half an hour or so making a hillbilly jig for the DTI so that it could be used to measure the top of the rail at each bolt location during setup. The jig itself consists of a long piece of 20mmx40mm aluminium extrusion for a base with a piece of 20mm aluminium angle to provide vertical mounting of the DTI, the extrusion base was slightly "rocky" so I also added a small length of 20mm extrusion as an outrigger to stablize it further.

The jig was positioned so that the DTI point touched the centre of the rail above a bolt location and the rail was adjusted in reference. This was applied for every bolt location on both sides. I was really surprised by just how much measurable difference there was from one bolt location to the next and just as surprised at how much I could bend the rail by hand to rectify. The process took a little while but I will sleep better knowing I used a measuring tool rather that relying on my (poor) ability to determine feel of a series of feeler gauges. As far as I can measure I have set up the rails within 0.01mm from front to back on both sides, I understand there will likely be some error in my jig but I am guessing this will be close enough? Comments and corrections welcome, this is my first metalwork project and I'm making it up as I go to some degree.

Anyway, here's some pics, next onto levelling the X and Z axis then onto tramming I guess... some other things to think about.

Cheers, John

[evengravy]

some pics

[evengravy]

Following the adjustment of the rails I went ahead and fixed the Y axis and performed the same levelling procedure, again I made a jig for my dial indicator and set it up as best I could.

When adjusting the X axis to be at 90 degree to the Y rails I noticed that a relatively small amount of force was required to mis-align it again. The mounting points attaching the bearing carriers to the gantry tube ends would shift, I could not produce enough tension on the threaded studs to eliminate this movement.

After reading a while on the forum I came across a thread entitled "make your gantry rock solid" whereby a moving knot of wire rope and pulleys allows racking to be minimised. With this in mind I decided to see if I could incorporate this arrangement into my momus.

I fixed some 8mm eye bolts to the front of the unit as a solid fixture for the wire rope and at the back of the unit fixed two horizontal 8mm eye bolts that can be tensioned to apply tension to the wire rope system.

This arrangement has improved my issue no end at the expense of some added resistance in Y movement.

I'm hoping my choice of 380 oz inch motor will be strong enough, comments welcome.

Hopefully the images are self explanatory.

[evengravy]

When I ordered the metal from the UK I omitted the cable carrier parts, the reason being that I have extended the dimensions of the unit and therefore wasn't at that stage sure of the dimensions and arrangement.

At this stage I am pretty strapped for cash so instead of ordering more aluminium from the UK I decided to see what I could come up with using materials I had to hand.

I am currently having my new house wired and there is lots of plastic tubing laying around, these are roughly 1" diameter and are used to carry the house mains wiring in the walls.

I also had a few different types of bearings laying around and a 3d printer that I built over xmas so I decided to combine these to create a cable carrier. The main lengths are made up using plastic conduit and I designed and 3d printed bearing carriers to connect to the gantry, to the back of the wooden frame and a dual bearing hinge for the bending section.

The bearings were pressed into the plastic parts whilst they were still hot on the 3d printer heated bed and the shrinkage of the plastic as it cooled held them in nice and firmly.

Hopefully the images are self explanatory,

The results seem ok and it should be strong enough to carry the wiring, time will tell I guess.

[ldanut]

I can't any attached image anymore. Do you know why?

[evengravy]

Not sure, they're showing up fine from here

[evengravy]

So, the project cnc has been on hold for a while as I built the space to house it, over the past few days I have been back at the unit in hope of getting it cutting this week or so.

I built magnetic hall switches for the homing of each axis and implemented those, allegro A3144 if I remember correctly, although I don't have a great picture. I have the unit completely wired bar the wiring of the router itself. I chose to keep the 48v psu external to the enclosure for fears of overheating, my chosen drivers use quite a bit of room and although I probably could have squeezed them in I didn't. I have been very careful to try to minimise the potential for noise issues, as such you will see a lot of copper tape and twisted pair wiring, in addition I have used clip on magnetic ferules on motor and limit switch wiring.

[evengravy]

The cable carriers that I mentioned previously with 3d printed bearing carriers of my own design are so far working out fine, I may replace them in future but for now they will certainly do the job required.

The past few days have been spent in attempt to get the unit as accurate as I possibly can. I have attached a dial test indicator to the router mount as close to the tool position as possible and have been measuring accuracy from that point.

The side to side accuracy at tool position is within 0.1mm. I would ideally have liked to improve that figure but after many tries to rectify it appears that there is a bowing of the cold rolled steel X axis transport which I can not overcome, looking a specs (briefly) for this material it appears this is within spec. 0.04 inch and is also within specification outlined by the manual so further trials to rectify this are unlikely.

The accuracy front to back (575mm travel approximately) is much better and is within 0.03mm (0.0011") and the left to right inaccuracy of 0.1mm is consistant from front to back so it is likely that I will surface mill a sacrificial board when up and cutting. I also intend to use the Autoleveller cnc software (Autoleveller | CNC workpiece levelling) in conjunction with a DIY probe (more later) to allow accurate engraving over the entire volume of the cutting area, both for engraving front panels and for milling pcb stock.

I have purchased some dibond aluminium composite that I intend to use for the electronic enclosure doors and since I intend to use this material extensively in future it will likely be one of the first materials that I mill on my machine, after the machinable wax that I made some time back.

I did have concerns about my choice of drivers and motors (2m542 and 381oz inch) and whether they would be able to move the Y axis considering the addition resistance added by my moving knot arrangement, as far as I can tell I am not going to have issues in that regard. The machine has been tested at rapids of 30,000mm/min (1181 inch/min) without any stalling, this could potentially be improved with the addition of some resonance dampers on the motor shafts, which I may include as some point, although it is unlikely that I will ever run the machine to that speed.

Once the machine is cutting I intend to add some additional functionality:

Firstly a coolant misting system for aluminium work

Secondly a 2W 445nm laser diode for engraving on wood and laser etching dibond (which I've already purchased), custom driver is currently under construction and Picengrave (PicEngrave CNC Image Engraving Software) will be used to create the gcode for rasters.

For now though, i'm just looking forward to getting the momus cutting, it's been a long road.

[Mac.CNC]

Thanks for posting, it's been a little quiet around here lately, hmm...

Frankly, from what you wrote, you got something close to a watch-makers machine there. Getting it to that point of accuracy at this stage is a feat, congratulations :-)
Don't loose your mind over the pcb thing and the bent cold rolled, you can make up for that later, if you're not too far off at this stage - it's easy.
Just surface a spoilboard with the machine itself and carpet tape the pcb to that before cutting it. All the off stuff in the rails will be wiped out that way.

I'm afraid you will not get any good answers on the choice of motors around here, it really depends on what you're cutting and at which parameters.
You'll learn that soon, once chips are flying and you can listen to the noise (if that sounds right or wrong), and also if your motors get crazy hot or not.
My z-axis motor for example, gets so hot during complex carvings you could fry an egg on it, touching it is possible, but not for long without risking a blister.
Yet, I've been waiting for it to fail for the past six months and here it is, still carving along :-)

As far as speed goes, don't worry so much about it - the fastest cutting speed you'll ever be dealing with is about 150ipm from my experience, anything higher will cause the gantry tube to twist and cause shatter in the cut (in wood).
That speed in itself with a 1/4 mill at full diameter depth is quite impressive to watch on a small machine like this, but the Momus can do it. I usually dial that back to 90 or 120, still - quite nice to be able to do that.

I've come across the autoleveler thing too before I started on the pcb thing, what kind of kills that process is >>>time<<<.
Just to get an accurate point cloud the machine is going to scan points in your surface for like several hours, depending on the size of the thing.
Afterwards you have to crunch the numbers, overlay it all and hope it worked...
It's much easier to surface a new scrap mounting board instead and cut the PCB on top of that - MDF reacts to climate changes in thickness by the way, in the range of +-0.03 I've heard.

The laser thing you're planning intrigues me greatly!!!
I've also been thinking about integrating a laser next to the spindle motor to be able to burn lettering or a logo into my projects without changing anything in the machine.
Please, please pretty please keep us posted on how that worked out for you, I'm super interested in that :-)
--
Mac

[evengravy]

Hey mac,

Thanks for the kind comments, I'm assuming a watchmakers machine is pretty accurate so I'm happy with that so the last thing I'm going to do is try to tram the spindle and that's it for alignment. In hindsight the modifications for easier tramming I should have included as per the update stickies here but I guessing everyone is like this with their cnc, it's never finished 100%

I know what you mean on the drivers and motors, time will tell although my plan to remedy heat issues is to tame the voltage down if I have any issues, as you say cutting speed is only a fraction of maximum speed so I'm. More interested in cutting performance over maximum rapid speed (although I'm far from an expert)

Autoleveler isn't something I've used per say, It just seemed like a good concept, my plan was to mill myself a touch probe pretty soon and create a point of the entire bed without stock so that this single point cloud could be used on all stock in future rather than creating a point for all jobs individually but from what you say it seems this is not way to work with it. As you say I think a milled spoil board and carpet tape should work, il try that first.

On the laser engraving there are some result with the diode on aluminium composite here metalbot
For the audio equipment panels it looks ideal. For aluminium I'm considering painting stock flat black, laser engraving to remove paint then using a saline sulphate etch like detailed here:
nontoxicprint | Nontoxic Printmaking, Safe Painting & Printed Art

This PDF from picengrave gives a good idea of the setup of laser diode with cnc http://www.picengrave.com/Laser%20Setups.pdf
It's likely to be slow for raster compared to a CO2 unit, but it doesn't cost too much to implement and can be hacked to work from a spare axis step and direction signals from my controller which I have. I may actually code my own controller using the arduino and a digital to analog converter for analog modulation signal to laser driver depending on time.

By the way I've been admiring you build for some time mac really beautiful, hope mine cuts half as well as yours

[evengravy]

Cutting!

So i'm very glad to report that the momus is cutting, rather than waste time I jumped straight in today and milled a spoilboard and attached some 3mm aluminium plate and performed some test cuts.

The edges are very nice with no noticeable stair stepping. Feedrate was 300mm/min with 0.3mm Depth of cut using a two flute 3mm carbide upcut bit.

The accuracy is a little off in the Y axis and I'm more than a little bit puzzled as to why...I have checked the steps per mm multiple times and confirmed with micrometer under no load condition, it holds perfectly as far as I can measure.

I cut a 12.5mm square pocket straight through the plate, X dimensions are within a fraction of a millimetre of the input values but the Y is almost half a millimetre to tight. At first I thought it was the CAM package at fault so I hand coded a GCode file to test, same issue. During the milling of the spoil board it seemed to be accurate in the X and Y so this leads me to believe it is the extra force required for the aluminium?

At this point i'm a bit stuck to be honest, the only thing I can think of is that the Z axis is deflecting during cuts on the Y axis, there is a noticeable amount of play in the Z axis assembly in that direction due to my redesign (not as per the plan)

The could possibly be missed steps in that direction due to the moving knot arrangement, but unsure how to best test for this.

I would appreciate any comments that might help me get to the bottom of the error. Thanks guys,

[Mac.CNC]

I don't think you're loosing steps, you would be off consistently in either direction.
This sounds more like what you already said: Deflection. Although you should see considerable stair stepping then.
I would start there, tighten that up somehow and see if it goes away. That could have something to do with the y belt also, in theory - if you over tighten that belt, your cut parts become bigger as opposed to if it's not tight enough you might see what you're seeing. Then again, it would be doing that with and without load.
--
mk

[evengravy]

Hi, thanks mac for the advice, after much trial and error today I was able to track the error, definitely deflection of the z axis and bit in the y direction.

I performed test cuts in wood using the same bit (3.16mm 2flute) and got consistent accurate cuts, down to the .00mm

When testing with alu I still had errors in the region of .4mm, my thought then was that the bit and z were bending slightly on first pass (too deep) which begins in the y direction and any successive passes were falling into the mis aligned groove and an error results.

I ran a test cut where I began a shallower first cut in the x direction, 0.1mm doc, then move z up a few mm and plunge before the first y travel of the square. The process repeats during each pass. The resulting inside cut is within 0.02mm in the Y on inside cut

The z travel is quite low during these cuts so I may have to raise the bed up further and cam my work to take into account the flexing.

On strange thing however is that the leftover square piece from the middle of this test cut, which should measure 10mm x 10mm, I fact measures 10.00mm (X) x 9.68mm (Y) , strange since the outside cut is accurate. Seems I'm getting 0.21mm deflection here.

One other thing I may consider is cutting direction, would this make any appreciable difference? Thanks mac.

[Mac.CNC]

I'm no expert on this since I haven't tried cutting metals on mine yet, what I do know if that topic is called "climb vs. conventional milling" and there is a really good article about the differences available on the cnc-cookbook:
Climb Milling vs Conventional Milling

The way I understand it, climb cutting (milling the part out clockwise) tends to deflect the bit away from the cut and conventional (anti-clockwise) towards it.
I suppose to minimize deflection, you should mill conventional and use no more than 30% of the bit diameter for depth of cut.
Putting a riser into the machine will also help, I did that too at first on mine. This could be as easy as gluing/bolting a few plywood rectangles on top of each other.
Avoid MDF in this case, that stuff "breathes" and changes in thickness by a few thousands of an inch depending on the weather. If you stack a several on top of each other to make the riser, you would be compounding that effect. Then make sure you resurface the top of the stack to make it true to the machine again.
--
mk

[evengravy]

Thanks mac, my thought train is alongside yours, il be ordering myself some marine grade plywood to build up the build area for aluminium work, il likely make a frame similar in structure to the momus base to make up the height that can be bolted on/off quickly. most of the work I intend to do will be within 1" travel.

I took a quick measurement of the z and the cutting bed that I used today is circa 180mm from the nearest bearing on the Z axis, some flexing is inevitable I suppose. My router is farther from the z axis plate than I'd like too. By rotating my router 180 degrees and a few quick mods I can get my router upwards another 20mm which should also help. At the moment the top of the router interferes with the z axis motor mount even though I did raise that part.

I'm considering a complete re-design of my z axis to use v bearings and get the kress router a bit closer to the x axis in order to reduce the leverage but at that point I may just start a newer, bigger cnc, not a priority just now however.

I'm a bit gutted I have this problem, the x axis cuts have exceeded my expectations by a long way, I can not measure any inaccuracy in that direction with 0.00mm digital callipers.

Thanks for the links on climb vs conventional, I will certainly be looking at those

[Mac.CNC]

You know, I wouldn't bother with making a mini torsion box for the riser. I sort of did that and now I regret that I spend so much time on it after taking it out of the machine (see my thread).
Don't worry about it so much, just stack some scrap plywood - doesn't have to be marine grade ;-) and do tests - see what happens - you may find something totally different going on.
For all I know, it could be a loose pulley, belt problem, perhaps your belt clamp isn't tight, it could be the bolts into the gantry tube or something on the x-trucks, you never know - until you narrow it down.

For example my original problem was different, more x related that anything else but I blamed it initially on deflection also (or even the belt).
In the end it turned out to be loose x-rail bolts in the first place, making it a little better, then after I added the second x-axis motor and getting everything tightened up, pinned and threadlockered down - that wasn't it. I still had problems I couldn't put my finger on. I still don't know why, but a combination of a y-motor with a pulley that resulted in a mach setting that wasn't dividable by 10 caused the whole mess.

I just measured mine and my closest bearing is 240mm off the bed (with no riser), I'm getting accurate cuts and even dragging a 4" dust hose around the machine doesn't affect anything.
Again, I'm only cutting wood, but my general feeling is that you might discover other problem areas if you look closer.
Have you tried to put a dial indicator on the bottom of your z while fully extended, poking at it with a finger from different directions to see if there's anything loose?
That's kind of what ticked me off at the problems I had in the first place.
--
Mac

[evengravy]

Hi mac, il try as you suggest, thanks for that. I did check the y pulley and it is fine although I'm not keen on those as they are aluminium and appear to be cast, they certainly are lightweight so il likely replace them with steel. I locked up the gantry and tried to see if it would move but no noticeable movement in the y under force by hand. One thing I did notice is that the Y belt is pretty tight, much tighter than the X, essentially I cut it one tooth too short and struggled to get it on but I presumed that it would stretch a bit, I will try to refit it looser if i can to see if that could be an issue. There isn't much info on tensioning that I could find.

So plan of action is

Raise bed with scrap plywood surface mill and test
Test with different end mills
Fix dial test indicator and measure deflection.
Try looser and tighter belt tensions on Y
Disassemble and reassemble Z

I'll give that a shot on Friday and see how I get on, thanks mac

[evengravy]

So, as mac predicted I think the source of my issues lay elsewhere.

Today I spent around nine hours stripping rebuilding and checking every aspect of the machine, I managed to find a problematic bearing in the x truck and replaced it, breaking a stud from the x bearing carrier in the process, what a headache. The machine is back together after much blood sweat and cursing and after a quick test the problem I had in the Y was transferred to the X, the Y is now exactly correct and half a millimetre is missing off the Y in my 12.5mm test.

I noticed a pause on the X travel a couple of times in this case, certainly that shouldn't happen, I think it may be my computer and something kicking in to disrupt the steps transmission causing the inaccuracy. It is a bit of a relic by all accounts and is running onboard graphic which could be an issue. Driver test came out system excellent so I presumed it would be fine, maybe not. My brother has a dell that I'm going to swipe to test it as soon as I can.

The only other likely cause as far as I can see is improper gcode formatting causing pause and lastly noise somewhere,

Plan of action is to test the other machine first,
Check gcode via another cam application
Lastly check step and enable lines for noise with my scope.

Computer issues are well in my usual realm so I'm more comfortable now I am reasonable sure that mechanically things are as good as they can be.