Power feed for a VM32L mill

This is a bit of a catch up on stuff that's been going on for the last couple of weeks.

Following making of a Spindle lock for the mill, I cut the 4mm keyway for the 12mm shaft of the leadscrew in one pulley.

And a 3mm one in in the steel dog clutch half.

I thought I was doing well, but then I put it all together and found that the bracket is miss-sized (or the size I used for the pulley in the design was wrong) which meant it wouldn't fit. A bit annoyed by that.

After a few minutes expressive language, I realised that I could remove the bottom plate and replace it with two bars, one at each end where the bolt holes are. I believe I'managed reasonable precision (within a thou) of the length, so it may even be better.

It looks like that works (sadly no photo, but I'll take one when all the bits are together). The dog clutch parts engage with the various turning bits mounted by hand.

With that, I Loctited (638) the bearing into it's housing and left it to dry. Next is to loctite the shaft into the bearing and make sure everything moves freely. This is the bit which scares me the most as the tolerances are pretty tight.

Whilst waiting for the loctite to set, I started on the replacement handwheel which will get bolted to the X axis pulley.

I have a piece of 4 inch aluminium bar sourced from some show somewhere as a, 'That will be useful someday'. Well it will be, but I hadn't considered how I would cut it if it was wider than needed.

In the end I drilled and tapped the hole which the handle will go in, bolted it to an angle plate and clamped that in the jaws of the bandsaw vice. This lets just enough poke out.

Whilst this makes the job possible (and lets start by saying that cutting it by hand isn't work considering), it's going VERY slowly. I've got about a quarter of the way through in about 15 minutes.

I expect this would go better with a blade with coarser teeth (as it says on the machine), but I only have the one. Hmm. It might actually be worth investing in a coarser blade. The time it would take to swap the blade would hopefully beat the time saved in the cut.

I'm also thinking about automating the thing. Very brutal engineering. Basically just hang a weight on the end of the saw! I've already got a clip to keep the trigger down for when using the saw table.

Hopefully, next time I post, the mechanics will be done....

Iain

Iain

Cutting a thick lump using a blade with fine teeth goes better if you can drill some vertical holes down through the stock on the cut line. This gives the swarf somewhere to go effectively making the big lump a stack of thinner ones with spaces in between.

Still a slow job as there is a lot of metal to get through but the saw blade spends less time with filled gullets riding uselessly on the material.

Wastes slightly more than the drill diameter's worth of metal of course. Which may be an issue.

If the drill won't go right through the idea works fine drilling from both sides. No need for the holes to accurately line up. Just close enough for a clear path down.

I used an extreme version of the technique when needing to cut a slice off a 7" diameter lump by hand 'cos some dinkle-brain forgot that his Rapidor saw capacity was only 6" when ordering it.

Silly boy.

Not only did I drill vertically I also went horizontally. Long series 5 mm drill I think. Plenty of lubrication and careful feeding on the drill got the holes done. When all is said and done drilling is by far the fastest metal removing method accessible to any ordinary workshop.

Finished the job with a hard-point builders saw.

Radial drilling might well work too. Easier to get a clean start for the drill as you are always drilling from top centre position. Just rotating the stock a little between holes.

I shall try that next time.

Probably drill lots more holes too. This will involve going round a couple or four times to ensure the drill doesn't get caught up where the holes meet due to asymmetrical loading when there is a hole one side but metal the other. No need to have all the holes going right through. I imagine 6 or 8 going right to the centre and a similar number of intermediate holes going only half way would make like usefully easier and the job far faster.

Clive

Just going back to the first part of this post, has anyone used a magnetic clutch to engage / dis-engage the drive motor? look like it could be a simpler and cleaner option?

Bit of a faff that. I just have the timing belt pulley with a larger hole bored through it running on a shaft and have a pin through the disc of the handwheel engaging a hole in the pulley.

I built mine with a simple friction disk clutch. Works both as an engage dis-engage device and also will slip in case of a jam somwhere.

At last! It's done! Well... nearly

I've run this up with my test program and can get about 8mm/sec with 8x microstepping. I'm planning to try lower levels of microstepping and check how that affects quality and performance.

Now, I say it's done, but I'm lying. My control program is still being worked on. I decided after some red herrings to put the UI for the control on an Arduino, so I'm working on that at the moment.

Also I want to have end stops working before I use it in anger and that needs a bit more metalwork and electronics - and software.

In the meantime it's back to my vertical engine (at LONG last).

Iain

When that's done, I will do som

I'm getting very frustrated with my 'controller' for the power drive. I'm writing this hoping for inspiration from the collective brains, or perhaps just to clear my thoughts.

To recap. I've got the hardware and electronics built for the power feed. Using an Arduino nano, I have a program which can drive the power drive, with commands from a PC (other serial terminals are available).

Last I wrote, I'd been working out how to get a nice user interface. What I want to do is to a) have some presents which will cut at various speeds and (optionally) return to the original place. b) allow me to manually control speed, c) generate aforementioned presets and d) have it respect some limit switches in case I muck up my settings.

My other constraints are that I don't want to spend much, I want the user interface to be easy to use. and it shouldn't be too big.

My software skills should be decent (after 40 years man and boy), but are basically limited to microsoft tech.

I was expecting to be able to use one of several android tablets/phones I have, but that proved challenging as none of them can talk to attached devices (SB OTG apparently).. When I found something which would (e.g. my current phone or an el-cheapo), it wasn't obvious how to power the phone whilst it was talking elsewhere.

MY Pis are all early generation and have similar restrictions.I do have an old laptop and an old windows tablet which would do the job, the the laptop is far too big (and XP) and the tablet is too valuable - it's my only portable 'business' device.

What I'm trying at the moment is to use an Arduino mega with an LCD shield.. The software support is very low level and I've spent a bit of time starting to build a windowing system, but I find the touch screen access not terribly reliable - and I don't trust it much.

I've rousted out some rotary encoders I bought a while back and got those working (so I can have input with a rotary switch and output on the LCD). These (really) need to run off some of the interrupt pins on the mega.

Now, the LCD module sits on top of and covers nearly the whole of the top of the mega, using most of the interrupt pins, and leaving very little space to connect to the 3 unused interrupt pins (19, 20, 21). I would have to solder to the board. Which, of course is doable.

So the thing is, every stop forward I take results in two steps back - or so it seems. I can't help but feel that I'm going about this the wrong way - that there's an easier or more obvious way to do it. If pressed, I would confess I'm a little tempted to spend a little more to make my life a little easier - the bits and bobs for prototyping are adding up as a start and discard various approaches.

So I throw myself on your mercy! Any ideas for how to make this a little easier would be most welcome!

Iain

Iain

Have you considered an Android tablet using Bluetooth to communicate with the Arduino?

An app on the tablet could have the user interface and send the preset settings to be stored on the Arduino, so that it would not need to continually communicate.

I have built Arduino interfaces for the DRO for my lathe and mill to use with the TouchDRO app and these use a small HC05 Bluetooth module connected to the Arduino, so it should be possible.

Stuart

I use an Arduino Uno with cnc shield, a Bluetooth connection to my Android phone, running GRBL o the Uno and GRBL controller on the phone. It does what you are asking for.

Edited By John Haine on 28/09/2020 22:04:12

Just to enlarge on that post, GRBL runs on the Arduino and needs streams of g-code commands (and control) on a serial port. Apart from configuration data (e.g. steps/mm) it doesn't have persistent data - this sits on the controller. GRBL Controller has a number useful features, including the ability to stream g-code programmes to GRBL, but at the moment the only thing I use is jogging which does most of the things you want. In particular it has 4 predefined "soft keys" that can have g-code macros assigned to them.

More details:

**LINK**

**LINK**

Let me know by PM if this is interesting for your application.

Edited By John Haine on 29/09/2020 07:06:02

I am in the process of putting stepper feeds onto my Myford VMB mill. Priority is the Z axis which is a pain to manually wind up and down. The following is my KISS approach. I have no doubt it will invoke a tirade from the cognoscenti.

I didn't want the hassle of clutches. I have bought in a standard 5mm pitch gearwheel from BearingBoys and machined this to fit on the Z axis manual handwheel in place of the graduated scale ring. (I have fitted DROs on all three axis so I never use the scale rings). The gearwheel is belt driven from a 2Nm stepper on a 2:1 ratio. The stepper is fitted onto the Z axis assembly on a standoff plate. The stepper has a BearingBoys gearwheel to match and the drive belt also came from them.

The X and Y axis will follow the same method in due course. I have no intention of modifying for CNC operation as I have a Tormach/Fusion 360 combination for my CNC work. The mods to the VMB are simply convenience mods to make life easier and reduce my consumption of Shredded Wheat. It is very useful to have the combination of CNC and manual milling side by side.

The 'electronics' is an Arduino Pro Mini with minimal code content. This drives a 'bus' of three quad tristate buffers, one for each axis. The buffers for each axis when selected via the 'bus' drives the Pulse, Direction and Enable pins on the stepper driver (DM542Y). The axis selection line also enables a relay to switch on the 48V to the stepper driver. Only one axis can be enabled at any one time. An axis when not enabled allows full manual movement of the axis via the original handwheel. Also when not enabled the stepper driver consumes only a few mA so there is no heat dissipation to consider. Manual movement is not inhibited by the stepper braking as there is no enable or supply voltage to the driver. I can't wind the handwheel fast enough to generate any back emf issues. Limit switches could be added into the logic to drive the Enable function but that's a later refinement.

The control panel is simple. A 3 position rotary switch to enable the axis of choice, a centre off double throw switch to select backwards/off/forwards motion and a potentiometer to set the speed of movement. There is no reason why the pot could not be replaced with (say) a 12 position rotary switch with fixed resistors to represent fixed speed settings. The centre off switch disables control of all axis regardless of the position of the X,Y,Z selector switch. The control box is a lash up at the moment but could be a fixed panel on the apron of the mill or could be a 'pendant' type box. The connection is via an 8 way cable (to allow use of CAT5 stock).

I am in the process of engineering a PCB for the control electronics and cleaning up the prototype wiring. I have the stepper mounted on the mill ready to go. The 2Nm stepper drives the axis very nicely.

It's not rocket science but satisfied my minimal effort to get motorised control of the VMB, create some nice smooth cuts and save my arm muscles.

If you would like a more detailed description, pictures etc let me know and if the PCB, (when completed) is an attractive item I can run an extra one for you (it will be CNC milled). The PCB design will be my first dive into Fusion's Electrical module so progress is a little slow as I get my head round the software. I have to say that so far it is very impressive in its integration with the CAD/CAM functions. It is not quite like it used to be with graph paper, scalpel and crepe tapes ....

Alan

Mine is permanently coupled, just energise the 'enable' pin on the stepper controller and the motor winds freely. It is ever so slightly notchy, but with my 0.1" pitch leadscrew that is 0.0005" increments

Edited By duncan webster on 29/09/2020 14:43:24

Wouldn't that feed back voltage into the circuitry when turning the handwheel tho Duncan? Maybe not enough to damage anything...I don't know. But with the powerfeed circuit off, turning the handwheel the control panel was like a Christmas tree! **LINK**

Hi, All.

Stuart. An Android tablet with bluetooth was one of my earlier options. This fell by the wayside for two reasons. One I blew the bluetooth module (but have since restocked). But two, I do NOT want to write native Android code, nor learn how to do it (old dog, new tricks!). I've used Xamarin to build an android app (for an entirely different purpose) and Xamarin doesn't support bluetooth natively. Not with the Forms version anyone. My searching at the time didn't give me any useful results, though, inspired by you're suggestion, I've restarted the search (though at the moment I seem to have hung my development environment).

John. I'm particularly keen on having my own user interface working my own way - plus, of course, despite my bleating I want to build the thing myself rather than using stock components.

Alan. That sounds very interesting, but I think you've chosen a different route to me. What is frustrating is that I assumed (yes I know what that means) that the software part would be the easiest since I kind of do that for a living, but it's been a nightmare!

Thanks all.

Iain

So I thought I'd see if I could make some progress with bluetooth.

First I would get the new HC-05 modules. Oh. Where are they? Well that was a good 20 minutes, only to find the just over arms-length away.

Next breadboard it up and copy paste a test program from the web (that's how we developer's work these days, don't know you).

Then try and find a terminal emulator which works with the HC-05. The first 3 don't (or I can't work them). Hoorah! I have communications of a sort.

Now I'm going through trying to build the bluetooth test app. Which, frankly, is not going well. If I had a clue what I was doing it would probably be easy, but it's random walk.

I reckon it's close to enough for tonight...

Iain

Well. I spoke too soon. I've managed to build and deploy to the android simulator - which didn't seem to work. Then I attached my phone and downloaded to that. The download was successful and it does something, though I really don't know what. HOwever, it's a good sign and I am shutting down now before I find out something depressing.

Iain

Is it essential to have a GUI/Touchscreen? They use a lot of pins and are hard to program, specially on a small microcontroller. If the display can be squeezed into a character lcd instead, there will be no shortage of pins and the programming is easier. On the downside, the interface is less slick than a gui.

It's possible to read rotary switches by polling them, but I prefer to detect changes with an interrupt as you intend. The hardware interrupts are easiest to use but almost any Arduino pin can be used to detect interrupts. They're a pain to setup at low-level but the pin change interrupt library makes it easy - have a look at this example.

Even the early pi's should communicate with an arduino via serial usb, but they're quite limited and poorly supported compared with a 3b or 4, both of which can plug in a small hdmi screen (£32) If you're familiar with Visual Basic, Gambas is available for later Raspbians: not a clone, but similar in most respects.

I tried to use a Raspberry 2 recently and it was so slow and out-of-date I decided not to bother!

The User Interface is often the hardest part of a program...

Dave

I had a fit of insomnia last night so spent some time investigating options. Finally, at 2am I decided to embrace poverty in preference to frustration and bought a (relatively) cheap second hand Windows 10 tablet.

I don't need to worry about anything technical there, I can just write code. It also means that I have a reasonable chance of getting the thing (finally) working in place in the next week. Should take me a few hours to write something which will be enough to be usable and I can improve from there.

Iain

You could have tried MIT App Inventor

It's been a while since my last post - nearly 3 years in fact!

I got stuck at the UI phase for a long time - partly out of frustration and mainly stuck with an over-engineer head on (one of my weaknesses, I'm afraid).

I took up the challenge once again more recently as I had a piece steel 20x200x660 mm to square up. The idea of turning handles to do this gave me the shivers, so I took up the torch again.

I have sort of succeed - at least well enough for this project - but I've learned a lot of lessons. Here's what's happened.

Firstly, I abandoned my motor driver code. It worked to a point, but I was struggling with the maths to make it speed up or slow down (from movement). My maths is greying with my hair.

Instead, I swallowed my pride and used the FastAccelStepper library which did the trick perfectly well.

Secondly, I abandoned my clever UI in favour of some buttons and a potentiometer. I can now move left and right and adjust the speed. My idea of cleverness here is to have a 2 way center off toggle switch so I can jog or set the movement to continuous.

It works. Sort of.

What have I learned (apart from humility)?

POWER!

The motor really isn't powerful enough. It's an act-motor 23HS2442J3 capable of 1.8 Nm at low speed, dropping off with speed.

Even with no load the motor struggles to move the bed at all at the very end of travel. In the middle it will do 12 ish mm / sec (720mm/ min) which is fine, but I have to manually wind the last couple of inches and it won't cut at all there.

I can (fairly) easily take a 1mm DOC with a side mill in 20mm steel by hand, but am limited to about 0.25mm with the power feed.

I'm only running at 24V with a TB6600 controller which can take 40 and increasing the voltage would probably help, but I still think it's too weak.

I don't think a stepper was a good choice.

CLUTCH

The dog clutch I made mainly works, but it's awkward to engage and disengage. I have to undo a grub screw and force the moving dog along the motor spindle. The keyway seems to want to choke up so I have to lever the dog back and fore. Not a good design.

I will almost certainly be reworking this. In fact I will consider buying a commercial drive. It's fun to muck around, but I've probably spent a comparable amount to a commercial drive - ignoring the many hours of time I've put in!

Thanks for all earlier comments!

Iain