CNC'd Loco Wheel & Earning Brownie Points

[philf]

Having finished my CNC miller I wanted to try to cut a wheel for the model "Lion" loco I'm building.

I modelled the wheel in ViaCad:



and today I had a go at cutting one:



The code was generated for me in Vectric Aspire. I used a 3.5mm ball ended cutter for profiling, 3d roughing and 3d finishing. There were about 18,000 lines of code and it took only 19 minutes to cut.

I was really mystified by the order the wheel was cut - it seemed totally random and I was somewhat disappointed that the finished item didn't look like the model. Chamfers on the model weren't completed on the part (only the chamfers adjacent to the spokes were cut!) and the spokes weren't radiused like on the model - it made some attempt at rads but they weren't complete. The rim certainly isn't good enough to run on a rail.

I'm wondering if using a much smaller cutter might improve matters (I can't think why it should) - I'll try get the code redone for me using a 2mm b/e cutter for the 3d finishing to see how it ends up.

The 19 minute cutting time was so quick because it's cut in MDF (painted gold afterwards to make it look like brass) at 1,000mm/min feed rate with a 6mm depth of cut. If cut in mild steel or cast iron I'd probably be using about 100mm/min and max 4mm depth of cut so the cutting time would be more than 3 hours (and with a 2mm cutter probably more than 8 hours!) It will be an interesting exercise but in the end I think I'll have to shell out for the castings.

I've just finished painting the front door and thought I'd finish it off with an engraved number courtesy of the CNC. I've earned a few more Brownie points (which equates to more time in the workshop!)  :



Cheers.

Phil

[raynerd]

Hi Phil, your door plate looks great, nice one!

I think your wheel looks amazing too but I`m taking from your message that your not that impressed. Your finish is good for MDF - I know you gave use the feed but what spindle speed did you use?

Chris

[philf]

Hi Phil, your door plate looks great, nice one!

I think your wheel looks amazing too but I`m taking from your message that your not that impressed. Your finish is good for MDF - I know you gave use the feed but what spindle speed did you use?

Chris

Hi Chris,

I was quite happy with the door plate and , more importantly, so was my wife. To earn some more points I need to get some plastering done on the hall, stairs and landing - she can't see the character in bare brick walls!

The wheel shows some promise and that the miller works OK. Another friend, who used to work for Cosworth, has offered to produce the code for me on a high-end CAM package and so I might take him up on the offer. I think it will still take too long to machine out of a useable material. I used my maximum spindle speed of 10,000rpm and stood holding my vacuum cleaner over it for the 19 minutes - MDF dust isn't very healthy.

I suppose if I were to get into casting I'd at least be able to make the patterns ................

Cheers.

Phil.

[75Plus]

What's wrong with this picture?

With such a beautifully finished door plate the attaching screws cry out to have their slots in perfect alignment!!!!

Joe

[philf]

What's wrong with this picture?

With such a beautifully finished door plate the attaching screws cry out to have their slots in perfect alignment!!!!

Joe

Thanks Joe - I'll go out and line them up now!

(They're now all vertical - is that OK - or do you think they'd be better horizontal?)

 

Phil.

[75Plus]

Phil, I tend to think vertical is best. I believe horizontal would tend to make the plate appear shorter.

Joe

[Stilldrillin]

I think Posidrive screws would have looked much better.........   

               

David D

P.S.
Nice looking wheel. I thought it was a bronze casting.......... 

[HS93]

can you straighten the barrel on the lock please, and please don't put pizza screws in, they are for amateurs, real men use slots.

anyway really nice job lets see some more and the mill doing it 

      Peter 

[philf]

can you straighten the barrel on the lock please, and please don't put pizza screws in, they are for amateurs, real men use slots.

anyway really nice job lets see some more and the mill doing it

Peter, David D

My wife spotted the barrel yesterday - it will be sorted as soon as it's fit to open the front door - it's snowing heavily at the moment and it's very windy.

I agree with Peter (sorry David) - pozidrives are rather industrial and would look out of place on my 116 year old Victorian front door (3' 9" wide and no chance of it ever being replaced with a white UPVC one where pozidrives might be acceptable).

I'm going to have another go at an MDF wheel. I think some of the problem might be that I exported the model from ViaCad at too low a resolution. I need to get the code redone and I may try videoing some of it being machined again.

Cheers.

 

Phil.

[Stilldrillin]

Peter, Phil.
Yes, I have to agree, on an aged door. 

With a "Victorian", stainless plate and Yale lock......   

David D

[philf]

Peter, Phil.
Yes, I have to agree, on an aged door. 

With a "Victorian", stainless plate and Yale lock......   

David D

Hi David,

To quote the Yale website: "Linus Yale Sr.'s son joined his father in the business in 1850 and perfected and patented his father's pin tumbler cylinder lock and became the considered locking expert of his time." (I'm sure mine doesn't date back to any time earlier than the 1960s or 70s - a couple of years ago I had to turn new pins for it out of bronze welding rod because it had become unreliable - it should be OK for another 30 years or so.  )

The door plate is actually brass! I satin finished it with fine wet & dry before applying a coat of lacquer. It was going dark when I took the photo so the colour doesn't show up well.

Cheers.

 

Phil.

[Stilldrillin]

Brass!!       Ah...... That's more like it! 


I'm wrong on all counts, again!   


Incidentally, we've got the same weather....... And a garden crew stripping the turf off my front lawn......... 

David D

[Bluechip]

And a garden crew stripping the turf off my front lawn......... 

David D

  

Wossat then ??? AFAIK if the turf is gone, you don't have a lawn anyway ..

Dave BC

[philf]

I had another go at my loco wheel today.

I downloaded a free 30 day trial of MeshCam last night and was quite impressed of how easy it was to get some output from it.

I used the same stl model which was used to generate the code from Aspire.

MeshCam generated two files for me - roughing & finishing.

The Aspire generated code took 19 minutes to finish the wheel. The MeshCam files a total of 80 minutes! (44 for roughing and 36 for finishing.)

The finished item looks much better!  The rim would be good enough to run on (if it wasn't made of MDF) and the roots and rads on all the spokes are finished properly. (There is a mystery pyramid shaped lump on one of the spokes that wasn't cleaned up (on the spoke at about 7:30 near the hub).  )

First picture is of the wheel after roughing:



Next is after finishing:



Finished Meshcam wheel:



Finished Aspire wheel - note the lack of full radii on the section of the spokes and odd terminations of the fillets on both ends of the spokes:



I tried to video some of the process (turn the volume down - it's very noisy with the CNC and vacuum cleaner running and I don't know how to edit it!):

http://www.flickr.com/photos/16648905@N04/7048228979/in/photostream

I'm sure I could dramatically reduce the cutting time by roughing the spokes with Cut2d and finishing with MeshCam after adjusting some of the settings. I think it would still take too long to cut out of steel. 

Next thing to try is one without the radii on the spokes - I think it's these that take most of the time and the wheel may look OK without them.

Phil.

[Brass_Machine]

Hey Phil...

What are you using for a CNC? Have you answered that already? I am losing it...

Eric

[philf]

Hey Phil...

What are you using for a CNC? Have you answered that already? I am losing it...

Eric

Hi Eric,

I did a post about it a couple of weeks ago:

http://madmodder.net/index.php?topic=6964.msg74694#msg74694

Cheers.

 

Phil.

[Rob.Wilson]

Hi Phil

The MeshCam wheel looks allot better than the wheel cut with Aspire ,,,,,,,,  wonder why    it also did not cut into the rim inner as Aspire did .

Whats the cutting time for one in steel/cast iron ?

Rob

[raynerd]

I`m sure an obvious question, but how are the estimated times being generated - i.e does it have a input for the material (so it "knows" what it is cutting) or are you choosing the feed rates and obviously a slower feed with the metal?

Both look good BTW!

Chris

[philf]

Hi Phil

The MeshCam wheel looks allot better than the wheel cut with Aspire ,,,,,,,,  wonder why    it also did not cut into the rim inner as Aspire did .

Whats the cutting time for one in steel/cast iron ?

Rob

Hi Rob,

The Aspire created file didn't cut into the rim too much - it just didn't bother to cut the chamfer in between the spokes at either the rim or the hub ends!

Both wheels were cut at 10,000rpm and 1,000mm/min feed with a 6mm max depth of cut. In steel I would only use perhaps 100mm/min and 3mm (maybe 4mm) depth. I'm guessing that the MeshCAM created code without any tweaking would take about 12 hours in steel! It's an interesting exercise if nothing else and I'm learning all the time.

I've just given the new wheel a coat of varnish to keep it looking nice.

Chris,

Both programs need you to input what spindle speed, feed rate and depth of cut you require. I guess the high end CAM packages work it all out for you based on the material you are using.

Mach3 can tell you what time your GCode will take to run.

Cheers.

Phil.

[Brass_Machine]

Hey Phil...

What are you using for a CNC? Have you answered that already? I am losing it...

Eric

Hi Eric,

I did a post about it a couple of weeks ago:

http://madmodder.net/index.php?topic=6964.msg74694#msg74694

Cheers.

 

Phil.

Yes you did!

Told you I was losing it!

[David Jupp]

I have no direct experience of either package that Phil used - I do know from my limited exposure to CAM (mainly Alibre CAM) that there may well be 'non-obvious' setting options that can make a big difference to the outcome.

It would be interesting to understand if the 'better' looking results with MeshCam are due to inherently better gcode generation, or just default settings (for something in the program) that suit this job better than those in Aspire.  I.E. with 'equivalent settings' (not necessarily easy if the terminology doesn't match) would the results be similar in both packages?

[Stilldrillin]

Don't care what prog's been used. I left all that far behind, when I finished working for a living.   

The first wheel's very nice. But, the second one's superb!   

Well done Phil!   

David D

[philf]

I have no direct experience of either package that Phil used - I do know from my limited exposure to CAM (mainly Alibre CAM) that there may well be 'non-obvious' setting options that can make a big difference to the outcome.

It would be interesting to understand if the 'better' looking results with MeshCam are due to inherently better gcode generation, or just default settings (for something in the program) that suit this job better than those in Aspire.  I.E. with 'equivalent settings' (not necessarily easy if the terminology doesn't match) would the results be similar in both packages?

Hi David,

I don't have Aspire - the code was generated for me - so I can't really comment on why it didn't work out. I didn't mess about with the MeshCAM settings - I just installed it and it worked!

I'll experiment with the MeshCAM settings to see if I can speed everything up without cocking up the result too much.

I will probably end up buying Vectric Cut3d and then I can at least discuss the problems with Vectric. (I bought Cut2d and am very pleased with it.) Although MeshCAM seems such a bargain at the moment. Maybe I need both!

Alibre was recommended to me but it's cost rules it out.

Cheers.

Phil.

[David Jupp]

Phil - I certainly wasn't recommending Alibre CAM (it's based on Mecsoft VisualMill).  I don't have anything like enough experience to pass any kind of judgement on it.  It may not be the most expensive, but certainly isn't bargain basement either.

Actually getting sufficient information to understand if a particular CAM package might be good for you or not seems to be quite tricky.  The fact that your were easily able to get such a good result (apart from the minor bleb you highlighted) is a plus for MeshCam.  Ease of use is probably more important than absolute capability for many hobbyists.

[John Stevenson]

Phil,
send me the stl file and what tooling you wnat to use, speeds and feeds and I'll bung it thri Cut 3D and post you the code.

john [ at ] stevenson-engineers.co.uk

[philf]

Phil,
send me the stl file and what tooling you wnat to use, speeds and feeds and I'll bung it thri Cut 3D and post you the code.

john [ at ] stevenson-engineers.co.uk

John,

That's very kind of you. 

Email sent.

 

Phil.

[John Stevenson]

Phil, Can you resend it please, not showing up here.

[philf]

Phil, Can you resend it please, not showing up here.

Hi John,

The first file was nearly 20MB - maybe that was why it didn't get to you. I've sent a much smaller one.

Cheers.

Phil.

[John Stevenson]

Smaller one wouldn't open.
Found the first file in junk on the server.

This is what Cut3D reckons it's going to do.



Five minutes for the roughing pass and 24 minutes for finishing path but I told it to do a second finishing path at 90 degrees to the first so total of 51 minutes.

Email sent with code.

John S.

[philf]

Smaller one wouldn't open.
Found the first file in junk on the server.

This is what Cut3D reckons it's going to do.

Five minutes for the roughing pass and 24 minutes for finishing path but I told it to do a second finishing path at 90 degrees to the first so total of 51 minutes.

Email sent with code.

John S.

Thanks very much John, 

I'll be down in the cellar shortly trying it out.

Cheers.

Phil.

[philf]

John S very kindly send me a Cut3d generated file which I have just tried to machine.

The wheel should have been finished in 2 raster passes at 90 degrees to each other.

Unfortunately, 3/4 of the way through the first finishing cut, my Z axis lost about 5mm worth of steps and I couldn't rescue the part. 

My machine still has an acme leadscrew on the z axis and can't run very fast without stalling. I have a ballscrew to fit but the machine will need a lot of modification to accomodate it. A gas strut might help to take a lot of the weight of the knee.

The first picture shows how it cut and how the fillets on the spokes look. They look very good and don't show the problem that the Aspire wheel showed! (Strange that Cut3d and Aspire are both from Vectric!)



The next picture shows the steps which are produced by rastering. The second finish cut (at 90 degrees to the first) would have cleaned these up.



John estimated 51 minutes - Mach3 told me 1hr 14minutes which is probably down to my slow z axis.

There is one feature in MeshCAM which ensured a good finish to the rim and the inside of the spokes - it has a 'waterline' finishing feature which runs round following the profiles in steps . It also has raster finishing in either one or two passes.
 


I don't know why MeshCAM needs to do the inside of the spokes in many passes as the sides are vertical and would machine just the same in one pass. I've asked the question of the software developer as I may be missing something in the settings.

I'm doing no more in the workshop today.

Phil.

[fatal-exception]

The raster looks good for an ACME z-screw. The Waterline finishing would improve the overall finish. \

If I were to do that part, I would probably start by pocketing everything that can be pocketed. Raster with a ballnose in probably a couple of directions, then do the final finish with the waterline type of program. If you want an extremely fine finish, decrease your stepover and ramp up your feeds, and use the largest tool possible.

Anyways, looks good! It's always fun to finish 3d parts!

Paul

[philf]

The raster looks good for an ACME z-screw. The Waterline finishing would improve the overall finish. \

If I were to do that part, I would probably start by pocketing everything that can be pocketed. Raster with a ballnose in probably a couple of directions, then do the final finish with the waterline type of program. If you want an extremely fine finish, decrease your stepover and ramp up your feeds, and use the largest tool possible.

Anyways, looks good! It's always fun to finish 3d parts!

Paul

Hi Paul,

Did you see the photo of the wheel I did with code generated by MeshCam? It did include waterline and pencil finishing - if it had been in metal, the rim would have been good enough to run on.



The rads on the ends of the spoke are 2mm and that's why I used a 3.5 mm b/e cutter with a 0.35mm stepover.

I have looked at some old loco wheels and think that I could get away with having rectangular section spokes without the corner radii. I could then machine a blank in 3d and cut out the spokes in 2.5d. MeshCam doesn't have any facility for 2.5d machining so I'd have to use Cut2d as well.

The ACME screw will have to go but it can wait until I'm in the mood for stripping the machine down. (I have a job for the Acme screw on a diy surface grinder project). I acquired a freebie 2505 ballscrew to fit but it has a double nut which is 150mm long - I don't want to lose that much travel and so I am thinking of using just one of the nuts. I'm not sure how much backlash there'd be. There is another option and that would be to use one nut above the mount and one below. I'd have to raise the base of the mill because the screw would try to bury itself in the concrete floor!

Cheers.

Phil.

[fatal-exception]

Yeah, that wheel looks great!

I've never really used MeshCAM, probably had a look at it at one time, but I settled on ArtCAM Pro, which seems to do a nice job of any 3d I throw at it. I use SheetCAM for all the 2.5D stuff, been using it for years and it's never failed me.

I know about the ball screw retrofitting. Making sure everything will align and fit takes considerable time. I have a largish mill that is next on the list for ball screw CNC conversion. (ZAY7045) Not really looking forward to the project, more towards the finished product.

 When I designed my router, I used double nuts on each axis. That was about 5 years ago, and there is still no noticeable backlash, and I've only adjusted them once. I guess it all depends on how accurate you need to be. Try to use both if possible. It sounds like you will be driving the nut(s)?