Sunday, August 20, 2017

CNC Z Axis with 150mm or more of travel

Many of the hobby priced CNC machines have limited Z Axis movement. This coupled with limited clearance on the gantry force a limited number of options for work fixtures. For example, it is very unlikely that there will be clearance for a vice on the cutting bed of a cheap machine.

I started tinkering around with a Z Axis assembly which offers around 150mm of travel. The assembly also uses bearing blocks that should help overcome the tensions that drilling and cutting can offer.


The assembly is designed to be as thin as possible. The spindle mount is a little wider which allows easy bolting onto the spindle mount plate which attaches to these bearings and drive nut. The width of the assembly is important because it will limit the travel in the Y axis if it can interact with the gantry in any way.

Construction is mainly done in 1/4 and 1/2 inch 6061 alloy. The black bracket at the bottom is steel. This seemed like a reasonable choice since that bracket was going to be key to holding the weight and attachment to the gantry.

The Z axis shown above needs to be combined with a gantry height extension when attaching to a hobby CNC to be really effective. Using a longer travel Z axis like this would allow higher gantries which combined allow for easier fixturing and also pave the way for a 4/5th axis to fit under the cutter.

Thursday, August 10, 2017

Larger format CNC

Having access to a wood cutting CNC machine that can do a full sheet of plywood at once has led me to an initial project for a large sconce stand. The sconce is 210mm square at the base and the DAR ash I used was 140mm across. This lead to the four edge grain glue ups in the middle of the stand.


The design was created in Fusion 360 by just seeing what might look good. Unfortunately the sketch export as DXF presented some issues on the import side. This was part of why a littler project like this was a good first choice rather than a more complex whole sheet of ply.

To get around the DXF issue the tip was to select a face of a body and create a sketch from that face. Then export the created sketch as DXF which seemed to work much better. I don't know what I had in the original sketch that I created the body from that the DXF export/import didn't like. Maybe the dimensions, maybe the guide lines, hard to know without a bisect. The CNC was using the EnRoute software, so I had to work out how to bounce things from Fusion over to EnRoute and then get some help to reCAM things on that side and setup tabs et al.

One tip for others would be to use the DAR timber to form a glue up before arriving at a facility with a larger cut surface. Fewer pieces means less tabs/bridges and easier reCAM. A preformed blue panel would also have let me used more advanced designs such as n and u slots to connect two pieces instead of edge grains to connect four.

Overall it was a fun build and the owner of the sconce will love having it slightly off the table top so it can more easily be seen.

Sunday, June 4, 2017

Six is the magic number

I have talked about controlling robot arms with 4 or 5 motors and the maths involved in turning a desired x,y,z target into servo angles. Things get a little too interesting with 6 motors as you end up with a great deal of solutions to a positioning problem and need to work out a 'best' choice.


So I finally got MoveIt! to work to control a six motor arm using ROS. I now also know that using MoveIt on lower order arms isn't going to give you much love. Six is the magic number (plus claw motor) to get things working and patience is your best friend in getting the configuration and software setup going.

This was great as MoveIt was the last corner of the ROS stack that I hadn't managed to get to work for me. The great part is that the knowledge I gained playing with MoveIt will work on larger more accurate and expensive robot arms.

Wednesday, April 19, 2017

CNC Alloy Candelabra

While learning Fusion 360 I thought it would be fun to flex my new knowledge of cutting out curved shapes from alloy. Some donated LED fake candles were all the inspiration needed to design and cut out a candelabra. Yes, it is industrial looking. With vcarve and ball ends I could try to make it more baroque looking, but then that would require more artistic ability than a poor old progammer might have.


It is interesting working out how to fixture the cut for such creations. As of now, Fusion360 will allow you to put tabs on curved surfaces, but you don't get to manually place them in that case. So its a bit of fun getting things where you want them by adjusting other parameters.

Also I have noticed some issues with tabs on curves where exact multiples of layer depth align perfectly with the top of the tab height. Making sure that case doesn't happen makes sure the resulting undesired cuts don't happen. So as usual I managed to learn a bunch of stuff while making something that wasn't in my normal comfort zone.

The four candles are run of a small buck converter and wired in parallel at 3 volts to simulate the batteries they normall run of.

I can feel a gnarled brass candle base coming at some stage to help mitigate the floating candle look. Adding some melted real wax has also been suggested to give a more real look.

Sunday, March 5, 2017

Non self replicating reprap 3d printer

The reprap is designed to be able to "self replicate" to a degree. If a part on a reprap 3d printer breaks then a replacement part can be printed and attached. Parts can evolve as new ideas come along. Having parts crack or weaken on a 3d printer can be undesirable though.

A part on this printer was a mix of acrylic and PLA, both of which were cracked. Not quite what one would hope for as a foot of the y-axis. It is an interesting design with the two driving rods the same length as the alloy channel at the back of the printer.



A design I thought of called for 1/2 inch alloy in order to wrap the existing alloy extrusion with a 3mm cover. The dog bone on the slot is manually added in Fusion 360 so it is larger than needed. The whole thing being a learning exercise for me as to how to create 2.5D parts. The belt tensioning is on a 6mm subassembly that is mounted on the bracket in the right of the image below.


The bracket and subassembly are shown mounted below. Yes, using four M6 bolts to tension a belt is overkill. I would imagine you can stretch the belt to breaking point quite easily with these bolts. The two rods are locked into place using M3 tapped grub screws. The end brackets are bolted to the back extrusion using two M6 bolts.


The z-axis is now supported by a second 10mm alloy custom bracket. This combination makes it much, much harder to wobble the z-axis than the original design using plastic parts.




Sunday, February 12, 2017

Printer bracket fix

Similar to many 3d printer designs, many of the parts on this 3d printer are plastic. Where the Z-Axis meets the Y-Axis is held in place by two top brackets (near the gear on the stepper is a bolt to the z alloy extrusion) and the bottom bracket. One flaw here is that there are no bolts to the z-axis on the bottom bracket. It was also cracked in two places so the structural support was low and the x-axis would droop over time. Not so handy.


The plastic is about 12mm thick and smells like a 2.5D job done by a 3d printer 'just because'.  So a quick tinker in Fusion 360 and the 1/2 inch thick flatland part was born. After removing the hold down tabs and flapping the remains away 3 M6 bolt holds were hand drilled. Notice the subtle shift on the inside of the part where the extrusion and stepper motor differ in size.


It was quicker to just do that rather than try to remount and register on the cnc and it might not have even worked with the limited z range of the machine.


The below image only has two of the three bolts in place. With the addition of the new bolt heading into the z axis the rigidity of the machine went right up. The shaft that the z axis is mounted onto goes into the 12mm empty hole in the part.


This does open up the mental thoughts of how many other parts would be better served by not being made out of plastic.


Monday, January 23, 2017

OHC2017 zero to firmware in < 2 hours

I thought I'd make some modifications along the way in the build, so I really couldn't do a head to head with the build time I had heard about (a lowish number of minutes). The on/off switch being where it was didn't fit my plans so I made that an off boarder and also moved the battery to off board so that I might use the space below the screen for something, perhaps where the stylus lives in the case.


I did manage to go from opening the packet to firmware environment setup, built, and uploaded in less than 2 hours total. No bridges, no hassles, cable shrinks around the place and 90 degree headers across the bottom of the board for future tinkering.

This is going to look extremely stylish in a CNCed hardwood case. My current plan is to turn it into a smart remote control. Rotary encoder for volume, maybe modal so that the desired "program" can be selected quickly from a list without needing to flick or page through things.

Friday, January 6, 2017

Machine Control with MQTT

MQTT is an open standard for message passing in the IoT. If a device or program knows something interesting it can offer to publish that data through a named message. If things want to react to those messages they can subscribe to them and do interesting things. I took a look into the SmoothieBoard firmware trying to prize an MQTT client into it. Unfortunately I had to back away at that level for now. The main things that I would love to have as messages published by the smoothie itself are the head position, job processing metadata, etc.

So I fell back to polling for that info in a little nodejs server. That server publishes info to MQTT and also subscribes to messages, for example, to "move the spindle to X,Y" or the like. I thought it would be interesting to make a little web interface to all this. Initially I was tempted to throw over websockets myself, but then discovered that you can mqtt right over a ws to mosquitto. So a bootstrap web interface to the CNC was born.



As you can see I opted out of the pronterface style head control. For me, on a touch panel the move X by 1 and move X by 10 are just too close in that layout. So I select the dimension in a tab and then the direction with buttons. Far, far, less chance of an unintended move.

Things get interesting on the files page. Not only are the files listed but I can "head" a file and that becomes a stored message by mosquitto. As the files on the sdcard of the smoothieboard don't change (for me) the head only has to be performed once per file. It's handy because you can see the header comment that the CAM program added to the G-Code so you can work out what you were thinking at the time you made the gcode. Assuming you put the metadata in that is.

I know that GCode has provisions for layout out multiple coordinate spaces for a single job. So you can cut 8 of the same thing at a single time from one block of stock. I've been doing 2-4 up manually. So I added a "Saves" tab to be able to snapshot a location and restore to it again later. This way you can run a job, move home by 80mm in X and run the same job again to cut a second item. I have provision for a bunch of saves, but only 1 is shown in the web page in the below.




This is all backed by MQTT. So I can start jobs and move the spindle from the terminal, a phone, or through the web interface.


Sunday, January 1, 2017

Keeping an eye on it

The CNC enclosure now sports a few cameras so I can keep an eye on things from anywhere. The small "endocam" mounting worked out particularly well. The small bracket was created using 2mm alloy, jigsawed, flapped, drilled and mounted fairly quick. These copper coated saddle clamps also add a look good factor to the whole build.



A huge plus side is that I now also have a good base to bolt the mist unit onto. It is tempting to redesign the camera mounting bracket in Fusion and CNC a new one in 6mm alloy but there's no real need for this purpose. Shortest effective path to working solution and all that.