AndyUK's Build - 1.2x1.0m Gantry

Hi All,

Hard at work as ever, just about finished welding the frame and will be painting soon... I've also got the main gantry structure together, and have starting cutting the aluminium components to form the Y and Z carriages. Photos soon, I promise!!

Quick question though - my linear rails on the gantry are collecting a bit of dust - whats the best way to clean and (oil?) them? Any recommendations for grease to flush the HGR20 carriages through with? And I need a grease gun... any advice appreciated.

Thanks in advance folks!

I'd clean and flush with oil then use some multipurpose lithium 2 grease as its quite cheap to buy, some links on it below :).

Spec me some lube!

Profiled rails - Installation guidelines and manufacturer comparison chart

Lithium Grease 400g - Toolstation

Legend! Thankyou - Your forum searching foo is far superior to mine ;)

Promised a couple of photos of the aluminium parts... so here we go - Don't judge my sloppyness too harshly ;)

I've been using the printer at work to get drawings printed out as accurately as possible. I'm then using contact adhesive to attach them to the aluminium, which makes the drilling process a lot easier. The drawings get a little busy and unprofessional, but it helps when I'm checking how accurate the printout is to be able to quickly know what each relative dimension should be.

Attachment 25537

Oh my life - the counter bore cutters are amazing. I love these things.

Attachment 25540

So, after some bandsaw action, some broken drill bits, a screwed up aluminium part (try and spot whats asymmetrical!), and some other parts made, I finally got around to some rough first-time assembly this evening. The rails aren't perfect yet - they're dialed in to about 0.02mm from the surface of the Ali extrusion whilst I'm awaiting my precision straight edge which is back ordered, but the carriage went together like a dream, and runs nice and smoothly along them.

Attachment 25536

Attachment 25538

Attachment 25539

Looking good Andy,

I and a few others I know print things out like this and it works quite well, if you set your printer to print at a 1:1 scale, it should print out "reasonably close" to true 1:1, then you can just measure what you get on the print out and adjust your printer settings accordingly, though I've never had to do this as its always been good enough.

Great build keep up the good work !

Hi All,

Small progress update; welding and grinding on the frame and bed is completed, I'm currently prettying it up, and will start painting soon. Once my ugly welds are covered up sufficiently I'll share an updated photo ;) I've got the Epoxy sorted, so that'll be the next job on the frame once the weather improves sufficiently.

I'm starting to ponder the electronics, and its all a little overwhelming. I don't think my budget stretches to servos, so steppers it is, and from the quick calculations I've done it appears I'll be wanting the standard ~3nm steppers with a ~70V homebuilt PSU. I'll be putting more thought into specifics and posting my thoughts before ordering them, but I'm utterly lost when it comes to narrowing down drivers.

Lets take Zapp Automations 3.1nm NEMA 23 stepper for example (it appears to me to be the standard) - it has a bipolar parallel (which appears to be the standard config) max rated current of 4.2A, so I'm thinking any stepper motor driver able to cope with >4.2A will be able to use the motor to it's full potential (more to the point, I'll have to limit the current on the driver). I'll probably want a motor driver capable of microstepping - and in this regime there appears to be named brand and off brand:

Gecko G2xx Drives - the difference appears to be the native stepping resolution and their microstepping capabilities, generally all in the £100+/ea region. I've seen a few US-based bloggers I trust using these and seem quite happy which is tempting.
Leadshine as used by a few people here... £50-100/ea from that I can see.
Random Brands, Same black box... £40-50/ea

Regards microstepping I have 2010 on X, 1610 on Y, and 1605 on Z (currently all planned for 1:1 but pulleys so room to gear) - lets take the 10mm pitch screws, 1.8steps/degree means half stepping nets 0.025mm/step - is that a reasonable target?

So can anyone shed some light on Drivers and point me in the right direction? Maybe point me to a resource which explains whats what, and what I need to look for to identify a reasonable product?


Second, I had originally thought of using a CSMIO/IP-M controller - I liked the ethernet functionality, 24V lines, and the fact that it was well documented, recommended, and provided a command buffer - BUT - I've been very aware that it has two very key draw backs for me: no good slave axis functionality for my dual x-axis screws, and four axis limit which would be a pain in the future if I wanted to add a 4th axis. I'm not keen on the cheaper parallel port breakout board style (rightly or wrongly, I'm kinda dumping the UCxxx controllers in here - please correct me if I'm wrong), but I have been thinking about the ESS ethernet smoothstepper. From what I can tell, it has the capability to handle up to 6 motors which allows future expansion, and can handle the dual screws and independent homing well - but it needs to be combined with a suitable BoB and only provides me with 5v IO which I'll need to put more effort into sorting.

Can someone tell me if I'm on the right lines, or if theres something drastic I'm missing or some better option? I'm really keen to have a robust and smooth machine.

Thanks for listening to the fevered rant this far - double thanks if you're able to put me straight on a question or two!

Microstepping - yes, half-stepping probably gives you as good a resolution as you need for a router, but the thing will run like marbles in a cement mixer! Go to x8 microstepping and it runs much more smoothly. Go much higher and you start to lose torque - x8 or so is a good compromise. That's my experience - steppers just don't run at all smoothly unless you microstep them.

Try some critical speed calculations (there are online calculators available) to see what max speed you can run your ballscrews at. Those X screws will probably go up to around 800RPM (exact figures depend on which calculator you use - my machine uses 1750mm 2005 and I can run up to about 900-1000RPM). Then look at what kind of rapid speed you would like, and work backwards to get pulley ratio. My 3Nm steppers run at around 1000RPM before starting to lose torque, so 1-1 with my ballscrews gives me about 4500-5000mm/min rapid. That's fine for me but your 2010 with, say, 2-3 pulley ratio, would give you more speed and the pulley ratio will increase available torque on the ballscrew to help. But play with the numbers - there are trade-offs here. I am happy with my machine because I mainly do small fiddly work which needs acceleration rather than high max speed but your needs might be different. There are also 4Nm steppers around (I think that CNC4YOU sell them, for example).

Drivers - I use EM806 and am very happy with them. Gecko are popular in the US but remember that they are a US product and there is an automatic bias towards "locally made". Understandably - and i believe that there are now tariffs which penalise Chinese imports! But in the UK we have a wider choice. Main points are enough current capacity (you have identified this), enough voltage headroom (EM806 is rated at 80V which works well with 68-70V PSU), and digital control - partly for smoother motion, partly for stall detect which I feel is vital if you are using dual-motors. Every so often I get a stall for some reason and the machine stops immediately without tearing itself apart. The 806 is a more recent Leadshine offering but there are clones of the previous model whose number I forget) widely available at lower cost. Any driver in this kind of range will support adequate microstepping, almost certainly. Plenty of people use the cheaper version although I chose the 806 as it looked as if the previous version was becoming obsolete. And so it was, but it seems only from Leadshine and the clone industry then took over!

I also use an IP/M. No, it does not properly support dual-axis homing, although assuming you have a way to home your machine, it does understand the concept of a slave axis and will drive it happily. For a long time, I homed my machine via the usual Mach3 option, then disabled the drivers to allow the slave stepper to be turned by hand, and tweaked it until the LED on the proximity switch tripped. Crude but effective. However, someone recently reported on the forum that you could go into the IP/M config tool in Mach3, disable slaving, then use a modified Mach3 homing routine to home X+A at the same time. This works very well. You then go back and re-enable slaving and carry on as before. My machine was built with adjustable targets for X and A, and by tweaking I can get good reliable squaring this way. It's a slight fiddle, but you normally only need to do it once at the start of a session.

The 4-axis limit is more problematic with the IP/M. If I were doing it again, I would probably go UC300ETH plus UB1 breakout board. As far as I can see, this is very similar to the IP/M with 24V signalling and plenty of I/O. And 6 axes! Can't remember now if it does differential signalling to the stepper driver, though, which is an CSMIO feature and, I feel, worth having if possible. Better noise rejection, in short, as with 24V signalling. 24V is useful if you use proximity switches as these typically run with 10V plus. When i tested mine, I found them a bit flaky at their nominal lowest voltage but they are fine at 24V, and I happily run them in pairs in series on X and Y so upper and lower limits go to the same input pin. Keeping XYZA on separate pins is useful on a larger machine as you can home multiple axes simultaneously that way in Mach3 (I home Z for safety, then X and Y at the same time, or XYA at the same time for initial homing as above). Using the UC300 also means that you could look at the UCCNC software in place of Mach3. I still have reservations about Mach4, and you have to be very careful about the exact version of Mach3 you run with the CSMIO plugin.

Anyway, just my thoughts, and I'm sure that others will disagree!

Thanks Neale, that was a super helpful post, and its going to take me a few evenings to reply in full!

Lets talk numbers first, because I've being doing these calcs behind the scenes. I've written my own version of irving2008's motor calc sheet. Its basicly the same, but I'm doing the calcs myself so I understand it, and I'm doing all three axis at once. I've also modified it to cope with dual screws on the X (I've doubled the screw inertia then doubled the Est Torque output - which in my head counters the additional drag of driving an extra screw against the additional power output.)

I've set the sheet up here for 5500mm/min, with a 50N (Ali) cutting force. Its currently not setup for gearing at all - and I have a microstepping factor of 1x because I haven't thought how to include these factors yet.

Attachment 25633

So, with 2x 4.0 Nm on the X we can just achieve this setup (I say just - as per the original spreadsheet there's a 3x Margin on the required torque), and rapids at 9000mm/min whilst staying below 900rpm (below corner and 80% of critical speeds). Note that although I've simulated 3.1 Nm on the Y and Z, they're massively overkill. Especially the Z.

Questions are - does this look realistic and okay? Is my dual screw fudge acceptable or are there any obvious holes?

Any tips on how to integrate the pulley factor and microstepping? How do I take these results and figure out what gearing to use?

Thanks!

Quick Update;

I've decided to go with a 1:1 ratio for now (which was always the intention) but by designing in the pulleys, I'll have the option for change in the future. This means that the dual 4.0Nm on the X should do okay, and despite the overkill, I've gone for 3.1Nm on the Y and Z (thinking being if I have issues with one motor, I'll have an identical one to swap in and out for troubleshooting).

I've also done a lot of investigation around motor drivers, and have ultimately followed Neale's recommendation with the EM806s - they seem to tick all the boxes for me and the stall detection is a really big draw. That, and Zapp's sale means they're very reasonably priced! (£62/ea rather than £105-£110/ea elsewhere).

Still quite undecided about breakout boards / control boards, although I'm moving further away from the CSMIO offerings. Watch this space!

Now that I know my motor and drive choices, my thoughts have rushed ahead to the PSU for the drives. This is all speculative, but I wanted to air my thoughts.

We have 2x 4.2A Max steppers, and 2x 4.0A Max steppers. Thats a load of 16.4A at full pelt, but the drivers only need 60-70% of that current. So worst case, I need between 9.9 and 11.4A. Lets call it 12A for the sake of overbuilding things.

The highest inductance of my motors is 3.2mH. So, 32*sqrt(3.2) = 57.2V Ideal voltage. Drivers can take up to 80V.

The secondary voltage of an unregulated PSU is increased by the capacitor bank by a factor of sqrt(2) - So I'm aiming for an ideal secondary voltage of 41V.

12 A * 41 V = 492VA. This is the minimum power of transformer I'll need. Standard size options are therefore a 500VA transformer, or a 625VA transformer would be the next size up. Standard secondary voltages in my range are 2x25V (in Series = 50V), 2x40V (in parallel) or 2x45V (in parallel). I'm therefore looking at a 500 or 625VA transformer with 2x45V secondaries, giving a voltage to the drives of 64V.

Capacitors wise, to achieve a 5V ripple at 50Hz under full load (which keeps me above the ideal supply voltage) I'm looking at C = 12A / (4*50Hz*5V) = 12,000uF. But more won't hurt. Also, while we're talking capacitors, I ideally want them to be able to cope with 2x V (so 120V), but 100V rated caps will probably do the job.