Converting my SX3 mill to CNC

[MMan]

This is my conversion of an SX3 mill to CNC but before we get into that I think it is worthwhile setting out what I am aiming to achieve:

I want to be able to make one-offs and small numbers of parts that would be too complex for me to make manually. I know some people on this forum create masterpieces using handles only but this is way beyond my abilities.

Some people seem to get hung up on rapid speeds on CNC. This is not really a concern for me as precision is more important than speed. I plan to set the rapid speed a bit above any cutting speed just to make it obvious when it is rapidding but squeezing the last bit of performance out is not a necessity.

Why an SX3?:

I thought about all sorts of ways to get what I wanted; building a gantry type machine that would have very low depth of cut but might nibble away for hours to converting my existing SX2. I rejected the gantry machine because I did not think I could get the accuracy I wanted. The SX2 was more tempting but the challenge of trying to make parts for the conversion on the machine I was converting was quite daunting. Plus there is not much room under the table to fit  a ball nut for the X axis

I was coming down to maybe having to get another SX2 from somewhere when a second hand SX3 came up on ebay. So meet Dusty an SX3 (named because he got here covered in dust and anyone called Miller gets called Dusty).

First pic is of Dusty, second is of Dusty chomping into a 50mm cube of steel, making a new QCTP for the lathe.

I am going to start off with the stepper drivers and then work through the rest of the conversion.

[vcutajar]

Interesting. Will be following along.

Do you plan to get a commercial kit for the cnc conversion?

Vince

[MMan]

Hi Vince,

i have bought the CNC Fusion kit which gives the ball screws, nuts, mounting blocks, etc. It looks well made but I am already seeing it will need a few mods to fit the mill. Some are tolerancing and product variability issues and the other thing I have spotted, so far, is a need to mod to get lubrication to the ball nuts.

Other than the above the other parts of the project are the electrics, limit switch mounting and lubrication.

All the best,

Martin.

[Barneydog]

Hi. Several years ago I converted my X1 fully to CNC. I did not use a kit but bought all the part from many suppliers. All three axis have ballscrews driven directly through Oldhams by decent stepper motors. Control is through parallel breakout and four individual step controllers all built into the controlling computers case. Fourth one is for fourth rotary axis I have built. I have used this setup for quite a while with no constant ball nut lube. Now and then I spray light oil on the ballscrews when I service and clean the machine. The only benefit I can see from a kit is the parts all come together and have instructions with them. My accuracy is down to 0.01mm on all axis.
You say you don't need speed but would say build for the maximum speed, torque and accuracy. You can slow it down when you want. Mine cuts just as we'll at high speed as at slow speed. I have no limit switches fitted as the ones I did fit we're just a nuisance.
I have gone on to design and build a large CNc router table and to CNc a small lathe.
The learning curve to do this is steep but quick and short as logic counts for a large part to work it out

Julian

[mikemill]

MMan

I have had a CNC Syil X3 mill for around five years now and for my money it’s the best CNC combination on the market. Using CNC machine tools is a sharp learning curve. You mention rapid speeds the G0 (rapid command ) is set in your controller, please set very low initially and set your Z clearance height above work clamps to avoid crashing tool into clamps.
Have you thought of which controller to use, I have had Mach from the outset; it is user friendly and has very good support (you will need it). CNC machines are extremely useful tools and great fun, good luck with the conversion.

Mike

[MMan]

Apologies for my late replies, life has been getting in the way. Good things, but in the way never the less.

Julian: I did think about the X1 but, once the SX3 appeared on the horizon, that was the way I was going. I agree with you on balls screws  - I know some people have done great work with ordinary screws (there was a great write up by someone making an orrery on a cnced Sherline mill and lathe) but it is one challenge I will happily skip.

I have found that the kit has needed some work to fit it so certainly not just a bolt on (to be expected across the variability of the machines). I have found it be good components and well made though. I did think a lot about building from screws up but the availability of the kit was another pile of unknowns out of the way. Maybe if I do this again I might choose your path - I am curious to see if I can turn the hardened outer off of the ball screws.

As for the lube,  the ball nuts have an open hole for lube so I either have to make a plug or rig up lube. While I have everything pulled apart it is not a lot of additional complexity (I hope).

Mike: I plan to take it very gently to start with. As I built up the interface box the Estop was the first thing tested (I am behind in writing that up too).

I have gone for Mach3 to drive it. I know there is an open-source alternative but I have heard both good and bad things about it.

I am looking forward to making my first real cuts with it

Thanks All,

Mman.

[MMan]

Electronics (feel free to skip if this is not your cup of tea)

There are a number of ways to buy stepper drivers; Really cheap Chinese imports with a bad reputation for reliability (but very cheap), all-in-one boards like the Gecko G540 and separate driver modules.

The Gecko Drive G540 is probably powerful enough for this job and gives 4 drivers and a built in breakout board. The hassle is that to buy this in the UK is almost $600 rather than $300 in the US. I did a bunch of soul searching on whether I should import a pile of G540s, keep what I needed and sell the rest on. In the end I went with discrete drivers and an interface card, marginally more expensive than importing a single card but if one driver dies it can be swapped individually. The real cost is increased complexity.

I wanted to fit all of the interface electronics into a rack case that will live above the mill with the PC that drives it. I had a good strong, spot welded, rack case that I built an amplifier in many years ago and which had not been used for a decade or so. When I pulled it from the shed I found that it had been occupied by a family of mice so the first job was to clear all that up and strip it down.

The basic plan is to have Estop and No Volt Relay (NVR) feeding a toroidal transformer the output of which will then be rectified and smoothed to power the drivers at around 42V. There is an auxiliary power supply the provides 12V for the interface board and also drives the cooling fan and some relay logic that looks after limits, interlocks, etc. The 12V supply is ahead of the NVR and Estop so that it can still do its job if the rest is tripped.

The box is split neatly into three section by the Driver assembly: Mains to the left, drivers in the middle and interface and logic to the right. The driver assembly (first pic) is a piece of folded-up ali on which the drivers are mounted and which screws to the bottom of the box. It forms a tunnel for air to flow from the fan across the driver heat sinks. There is a small plenum visible to allow the air to calm down and spread out before hitting the heatsinks.

This view also shows the Breakout Board (BoB) mounted on the assembly. One of the tricks to building electronics is knowing where to split things apart so that they can be dismantled. A variation of sods law says 'if it is hard to dismantle then it will have to dismantled more times before it works'. There is more wiring between the BoB and the drivers than going into it, so it makes sense for it to be mechanically attached to the assembly. I have also mounted the bridge rectifiers on the assembly - it is in the right place and a large piece of sheet metal with plenty of air flying over it is ideal for getting rid of heat.

The second pic shows the front panel being made. I decided to fit ammeters to each driver so that I could see what they were up to and spent some time wondering how to make square holes for them. Having had a press at work many years ago didn't help, it just made me think I had to punch them. Then it dawned on me that the mill would do the job no problem. Amazing how the obvious isn't obvious. Pic 2 shows the set up for drilling round holes, clamps were then moved to mill the square ones. All holes were done by DRO from a sketch in DraftSight. I chose brass for the front panel because ali was too soft, steel hard work and I liked the look.

Pic 3 shows the start of front panel wiring and pic 4 the completed panel waiting on a shelf for the rest of the box.

Pic 5 shows the box basically complete. On the left you can see the large relays for switching power to this box plus mill spindle and auxiliary out, 12V supply is at the back. In the middle is the driver assembly now plugged into the stepper cables. On the right you can see the small relay board that looks after limits and logic.

There is a lot of discussion on home built CNC forums about noise and earthing/grounding particularly on limits and Estop. A lot of the earthing/grounding discussion is a bit mixed up so I have tried to simplify it in my own mind and we will see if I have something acceptable. The other thing I have done is use small fast relays to isolate and clean up noise. So my limit signals come in as a switched 12V signal that only goes to a relay. Anything noise sensitive, such as the BoB, only connects to isolated contacts on the relay with its own, isolated, supply. Similarly for other signals.Hopefully this will do the trick.

Pic 6 shows the unit being tested, stood on the mill table. Up and to the left Mach3 is visible on the computer monitor. The unit is only connected to Estop (left hand cable), X motor (cable next to it) currently. The sockets under the meters go to the motors while those at the right go to limits and other sensors (TBD).  The limits sockets have shorting test plugs inserted into X and Y. This is the normal case when the limits have not been triggered (NC switches). The Z limit switch is missing its shorting plug and you can see that the indicator under the third meter is illuminated showing it has been recognised.

After a tussle with some bad crimps it now all works. All four drivers have been tested and the logic works as intended, Mach3 responding correctly as well. Of course I do not have screws and motors fitted yet so cannot do anything actually useful yet.

On to ballscrews and limits next.

Mman

[Dave Otto]

Very nice work Mman!

I'm in the middle of a CNC project my self and I can appreciate all the work that went into the control box.

Dave

[MMan]

Thanks Dave,

It took me back to skills I haven't used for a long, long time. Enjoyable.

Mman.

[MMan]

Hi All,

Apologies for the long break - partly life getting in the way. Partly I do not think I have the hang of writing build logs yet - I keep on saying to myself that I will 'just get X done before I post', of course there is always something else and on and on. Hence I have some catching up to do.

This post covers off the bulk of the mechanical conversion. The kit from CNCfusion is well made but I found I had to make the odd alteration. The general form of X and Y is as pic 1. A stepper drives a coupling and then directly a ball screw. This gives around 1 thou per step which is then scaled by 10 micro-steps in the driver, principally for smoothing. I found that my Y axis did not fit the mill. Close examination showed that one of the mounting holes on my mill's saddle was mis-drilled. Fortunately there was room to mill down the mounting block to fix it pic 2. You can just see M5 grub screws in the mounting holes - making it easier to clear the swarf.

The other mod I had to make is for the ball screw lube holes. All the ball screws have 6mm lube ports and it felt wise to connect them up . The problem is that there is not enough room to get a lube connector in. Hence a second cut on each of the X and Y ball screw mounting blocks.

The Z axis works somewhat differently. The motor is a larger Nema34 and drives the ball screw via toothed belt. The motor mounting plate needs a slot milled in the underside to clear locating pins on the mill column Pic 3. A slightly different mod is needed to get the lube to the Z ball nut pic 4.

Getting the X and Y axes fitted needs a little thought to get the ball nuts parallel and the bearings in the right place to give free and smooth movement. Essentially it is a question of moving the table back ward and forward and nipping up the right thing at the right moment. Pic 5 shows the table and saddle assembled.

Incidentally, I initially thought I would only need a shaft on one end of the stepper motors. Some thought lead me to the realization that they have to be double ended. This is to make it easier to drive the screw manually during fitting but also to allow the gibs to be adjusted from time to time. I have not made handles for mine because I will not need to do this often. However a collet block does well enough when needed.

Fitting the Z axis is easier the only tricky bit is getting the locating pins out of the mill's bearing mounting blocks. Once free they go in and out easily so there may have been some retainer in there originally. Pulling those pins out was a real 'why did I start this' moment but all went smoothly after they were free. Pic 6 shows the Z axis fitted.

Pic 7 shoes the electronics box that fits on the back of the mill. The bottom of the panel that divides the electronics from the mechanical side of the mill has to be removed to give room for the Z stepper motor. Because this would open up the electronics to some risk of swarf blown back round the column I added a plate above the motor. This would then block air flow over the electronics so I put a small fan in the plate, blowing downwards. Airflow is now in at the top, back of the mill and out round the base of the column Pic 8 shows the modified electronics box.

That is it for now, next up is the limit switches.

All the best,

Mman.

[MMan]

Hi All,

Now to the limit switches to prevent the machine crashing into the end stops.

First the Z axis, I couldn't think of a sensible place to put a limit for the lower end of Z. Either it was going to be so high as to be in the way of useful work or so low only be useful in the case where no tools were mounted and nothing was on the table. So I have skipped a lower limit. The upper one is a waterproof micro switch which is mounted up near the top of the metalwork. It is nicely out of the way of most swarf and its wires go through a gland into the electronics box at the back. See Pic1.

X and Y limits are in a nastier location and so I am using real limit switches. The nice thing about these ones is the wheel is off to one side which makes mounting easier. The X limit switch is mounted to a piece of 50x25 box section mounted behind the table. The section is just high enough to lift the wheel up to the table. a piece of chamfered steel is attached at each end of the table to trigger the switch. See pic 2

The Y switch is mounted under the table where a DRO was previously mounted. A piece of bar holds it up so that the wheel rests on a piece of milled 25mm angle iron attached to the saddle. In this case the limit is triggered when the wheel falls off the end of the angle. See Pic 3.

I mentioned in my last post that I was a bit behind with posting and the more observant will have noticed the swarf in the last two pics - it works. And surprisingly well (odds and sods to sort out but basically there). So I have videos to post: The first is the first moves the mill made - with a marker writing "Hello World" on paper. The odd looking shape to the end of the marker is due to a table crash as I was trying to understand offsets (that is why I started with a marker anyway).

<a href="https://www.youtube.com/watch?v=Wde81QwWm5w" target="_blank">http://www.youtube.com/watch?v=Wde81QwWm5w</a>

The second is the first metal cuts - a mounting plate for fitting a stepper to my rotary table (I have a lot of gears to cut).

<a href="https://www.youtube.com/watch?v=te1FDGr16WA" target="_blank">http://www.youtube.com/watch?v=te1FDGr16WA</a>

I did not use CAD to create this but was trying out the Mill Wizard software from the same people who make Mach3. It is between what you might write with Gcode by hand and a full CAM application. It makes it simple to do simple things and gives you a graphical interface to see what is happening. I think it will see most use for tooling and simple work. Pic 4 shows the job that created the plate. As a piece of software it feels very new and a bit buggy - in part I may not be using it quite as the writer intended but that is what we users do.

Pic 5 shows the motor and plate and spacer assembled before mounting on the rotary table and Pic 6 shows it all together.

All the best,

Mman

[MMan]

Hi All,

I posted this thread a long time ago. This is a quick update on a mod I did last year to increase spindle speed a bit (posting triggered in part by the current debate on CNC in home shops).

Last year the PC that runs this CNC mill blew up. Replacing it lead to a pile of 'if you are replacing that then...' mostly necessary as I have learnt more about CNC. One that has been on my mind for a long time is that the spindle speed is low for my typical CNC uses. Cutting times can be long for some parts and the contoured large clock frames I have been making could run for an afternoon.

I have considered everything from speeders to bolting on a secondary spindle but realized there was a different approach (not necessarily better, time will tell). The SX3 is intended for typical manual milling so tops out at 2800rpm. That is a good compromise for a mill that might be using larger sized cutters but I find with CNC that I tend to use smaller cutters that do not need the torque. Poking around I realized there is room in the top cover to replace the motor drive pulley and double the  rpm. Here is my ingredient list: 455mm HTD5 15mm wide, 44 tooth HTD5 pulley and appropriate taper lock bush bored 14mm. Total cost £25 including shipping. Fitting took around 30 mins and is reversible if I should change my mind.

5600rpm is not a lot in comparison to some/real CNCs but should be an improvement for not much outlay. The spindle end pulley could be changed too but that would mean remaking the quill splines in some way I do not want to tackle for now. I do not know if this may impact the spindle bearings, even though the forces are likely lower but they are on a longer term list for replacement anyway.

All the usual caveats YMMV, etc.

Martin

[Admiral_dk]

I will say that for that small amount of time and money - a fine improvement 

I'm guessing that you are happy with your CNC and use it, since you did the upgrade of PC etc.

Per

[MMan]

Hi Per,

Yes, I do use it. Not for everything but for complex shapes and contours it is great. For things without complex shapes the manual mill is usually quicker.

All the best,

Martin.

[Jasonb]

That is very similar to what Sieg are now doing with the SX3.5 to make it more suited to CNC conversion of simply if you want to run small cutters. The have a duel pully system with the slower ration being toothed belt (no slipping) and the high ratio being via a polyflex belt which is nice and quiet and gives 5000rpm top speed

As you say the smaller cutters don't need as much power to drive them so the loss of mechanical advantage due to the higher ratio is not an issue.

The KX-3 does have a lot higher spec matched pair of bearings with a suitable high price to match

[Vixen]

Hello Jason,

Do you have any photos of the SX3.5 duel pully drive to give the 5000 RPM spindle? I went a different route (as ever) and fitted a VFD to speed up the motor to provide the 5000 RPM spindle speed. It makes a bit more noise at 5000 RPM compared with the standard 3000 RPM spindle speed.

Mike

[MMan]

Hi Jason, Mike,

I do have an SX3.5 as my manual mill. It could be where the idea came from, just an idea swimming in the chunky veg soup in my brain. I will post pics of the 3.5 pulleys when I am in the workshop next.

All the best,

Martin.

[Jasonb]

Not the best of photos but you can see the timing type belt in position for slow speed and the multi vee grooved pullies below that take the poly-V below for high speed

[Vixen]

Thanks for the photos Jason. They make is quite clear how Seig did the two speeds on the SX3.5

On my F1 mill the spindle pulley is part of the tool-changer mechanism, so I'm limited there. The smaller drive pulley under the brushed DC drive motor could have a larger diameter pulley fitted together with different toothed belt. That would work, but it's tight inside the spindle head casting which could limit my options.







Sounds like this is may be how Martin did his speed change

Mike

[MMan]

Hi Mike,

Yes, just so.

All the best,

Martin.

[Vixen]

Hi Martin,

I think I can see the image of a 44 tooth HTD5 pulley in my tea leaves.   

Mike

[MMan]

Hi Mike,

Glad it has been of use.

All the best,

Martin.