Sieg SX2.7 (Long Bed) CNC Conversion

Okay, it's been a few weekends of work but finally got my SX2.7 under power...

https://youtu.be/9T3QncRvuR4

Bench was the original frame for a router I'd intended to build, but chickened out (it was going to be too large for the shed... so became a pretty handy bench). As part of that original design I'd planned around supporting 2 19" racks - using one of these to fit a HP server, the control box (PSU steppers, motion controller) and a server keyboard/monitor - for which the backlight on the monitor finally failed today...

Attachment 25777

The other 19" bay (out of shot) now simply holds my compressor (a silent 25l ex-dentist job). The central bay just holds stock material.

Attachment 25778Attachment 25779

Z Axis - it's all been done before, I'm just copying others who've converted their 2.7s. Z is by far the easiest axis and the original ball nut holder takes a standard 1605 ball nut. 2:1 belt drive as a general approach to all axis.

Attachment 25780Attachment 25781

Some detail on X - ignore the hack job on the ball nut - I was trying to expose the nut holder's bolt hole in situ to unbind the screw that was slightly misaligned after I'd tightened all bolts. That was a difficult few minutes with a Dremel and none too pretty.

Attachment 25782

Y Axis - very much like the others. I'm not sure that I like the limit switch so exposed, so might re-do this once I regain my enthusiasm.

Control box is a 800VA 50V dual secondary toroidal transformer; rectifiers and 10,000uF/160VDC caps. With bleed resistors - despite arguments on here to the contrary. Stepper drivers are 2DM860H (4 of to support a 4th axis), Steppers are 3.1Nm 4-wire - just cheap eBay jobs. Motion controller is an UC300ETH with a standard parallel BOB. I finally found a use for the 5/24V PSU that I'd stripped out of a laser printer.

Both Y and Z axis the screws are unsupported, X has a phosphor-bronze bush to support the end of the screw (butchered with a band saw) mounted in the end plate. General approach was limited to hand tools or what I could fit on a Denford Starmill (150x80 table area - somewhat limits options). Oh, and everything with an option to recover the machine back to the original manual control if necessary.


I've tested the machine at 1000mm/min and it's faultless (I shouldn't say that so soon), though my original wet-finger - I didn't realise that UCCNC defaulted to 20% jog speed, and at one point had it running at 4000mm/min - sounded sweet but that's crazy fast for a mill this size, so it's now got a more sensible limit,.

Next is to sort the limit switches properly then consider belt guards for each axis. I've also removed the neutral switching on the e-stop and used the now spare way of the switch to provide the e-stop signal back to the control box. This, and the probe and limit sensors to be routed through a connector mounted into the 8x4" plate on the top/rear of the column.

What I don't have is a solution for the spindle - I've eyeballed the brushless motor controller, think I understand the basic operation but don't fancy reverse engineering that one. I might instead scope the control board in the pendant and see if I can create a closed-loop emulation of the control buttons to get a form of speed control. For now, manual operation of the spindle. I'd be interested if anyone else has a solution for this.

Quote:

---

Originally Posted by Doddy

I'd be interested if anyone else has a solution for this.

---

Well!, that might be easier than I thought... there looks to be a PWM at 10kHz between the front panel control and the spindle motor driver on the main power board. I'll get probing when I can be arsed and it looks like an interface could be easy to interface with a standard 0-10V supply (microcontroller, ADC and PWM driver). Details to follow.

Hello.

Nice!

I am attempting this conversion on a new SX2.7. Do / did you find any problems created by the belts creating a small amount of backlash as the motors change direction? How critical is belt tension and does the additional side loading of the motor shaft bearings created by tensioning the belts concern you?

Thanks

Dennis

The belts each have tensioners (just a M8 or M10 in a machined slot to allow any slack to be taken up, packed with bearings to present a rolling surface to the belt). There's no discernible backlash (yes, of course there's theoretically "some" but not enough to worry me) - choice of HTD belts obviously helps to some extent.

How critical is belt tension?, don't know - took out the slack and I can measure a step on the stepper translating to linear motion.

Side loading on the motor shaft bearing?, if it's a problem I'll replace the motor, it's worked well for 18+ months with light use. The bearings are designed for radial loads so I'm not losing sleep on that one.

You have to consider the scope of the mill - is a mini bench top mill - nothing more, nothing less.

I should also add... half the reason for a belted solution was to turn the motors back in on themselves to reduce the external footprint of the machine. It needs to sit on a 1300x700 mm bench top (originally intended as a router frame), Even with the motors turned inwards the longbed version of the 2.7 challenges that footprint.

Thanks for your quick reply. It is encouraging that you have 18 months + use. I am currently machining the ball screw housings on my Arboga mill, modelled in Fusion and test printed on my 3D printer for fit. Using cast iron for housings (because I have some). If you were to start again, what would you do differently?

Thank you for your post which motivated me to buy the SX2.7 and modify it. I now have a neat lockdown project!

Thanks again and stay safe

Dennis

Yes, I do like the idea of reducing the footprint. Not many of us has the luxury of space in our home workshops.

D

What would I do differently?

I’d start from the outset with an enclosure! I probably wouldn’t have gone for the long bed.... but I don’t think that was a major cost component.

I’d put the home sensors outside the envelope of each axis to avoid crashes. I was taking a puritanical approach that the supplier recommended end-on sensing, but my experience has taught me that the performance difference doesn’t worry me, not as much as inadvertent crashing the axis. That wasn’t helped by me using one control box for two mills, resulting in weird configurations that could result in the sensors being unpowered and subject to crashing.

I’d roll the x axis stepper towards the rear of the bed rather than the front. I believe there’s space there if drawn close to the bed. Y I’m happy with. Z I don’t really worry about so not much in the way of introspection there. Definitely retain the quill lever

I have belt covers on each axis now... 3D printed was easy.

Unspoken of here is the replacement of the bldc motor with a servo... a result of me blowing up the spindle controller trying to adapt it to support 0-10v operation. That is now commercially available from sieg in any case. My solution, not necessarily recommended, requires me to think of adapting the spindle controller pendant. (Another roundtuit job)

Part of this evolution is the chassis wiring isn’t quite what I’d intended, I’d plan that as bit more carefully.

There’s a sieg user group on Facebook which has a couple of cnc’d 2.7s, if you’ve not seen that already. Useful to see what others get up to.

Quote:

---

Originally Posted by Doddy

Unspoken of here is the replacement of the bldc motor with a servo... a result of me blowing up the spindle controller trying to adapt it to support 0-10v operation. That is now commercially available from sieg in any case. My solution, not necessarily recommended, requires me to think of adapting the spindle controller pendant. (Another roundtuit job)

There’s a sieg user group on Facebook which has a couple of cnc’d 2.7s, if you’ve not seen that already. Useful to see what others get up to.

---

Have you got a wiring diagram of your inputs / outputs from your breakout board to the servo drive and vice versa please?.
I recently got a 1.8kw one for the spindle on my amat25 conversion.

Daz.

Diagram?, you mean a design?, Paper? Nope, just configured the Servo for step/dir control and hooked up two outputs to drive those. As simple as that.

Further to last: The Chinese servo controllers (only ones I know) have 5V optical-isolated inputs for the step/dir control, with separately excited inputs for the other discrete inputs (so you can use 12/24 for e-stops, etc). But, and this is the big but - the spindle speed inputs are 5V and can be directly coupled to a BoB output.

Edit: I liked the performance so much I replaced the phase/vfd on my ML7 lathe with a 2kW servo and drove that in exactly the same way.

Quote:

---

Originally Posted by Doddy

Further to last: The Chinese servo controllers (only ones I know) have 5V optical-isolated inputs for the step/dir control, with separately excited inputs for the other discrete inputs (so you can use 12/24 for e-stops, etc). But, and this is the big but - the spindle speed inputs are 5V and can be directly coupled to a BoB output.

Edit: I liked the performance so much I replaced the phase/vfd on my ML7 lathe with a 2kW servo and drove that in exactly the same way.

---

Mine has a DB44 socket, I have a cable and a DB44 screw terminal block. I was hoping to avoid using an axis control for it.
It says that in speed mode it has a direction control on one of the DI ports, as in relay on is one direction, relay off is the other (apparently). Can drive this from pin 17. Then has the usual on/off which I have a seperate rocker switch for, + various others I'll likely not need in speed mode. Was hoping to use it this way, with the 0-10v servo input driven straight off the bob (board has it built in and I've tested it with a meter).
The outputs I'll just daisy chain to the estop via relays until I update my controller + boards. Then with more inputs I can go straight into the bobs optocouplers and ditch all the relays.

Sounds basically about right doesn't it?.

Ah, I should have been more clear - the inputs I'm using for speed control are as above. Of course it has the usual analogue inputs as well (from memory, at least!). What you've said makes perfect sense - and is a perfectly do-able solution. It sounds like you just need to export a direction (CW/CCW) the PWM output (0-10) and the servo-enable signal.