go_away

Author Topic: My homebrew CNC  (Read 9056 times)

0 Members and 1 Guest are viewing this topic.

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
My homebrew CNC
« on: February 27, 2011, 03:31:38 PM »
I just finished up a CNC machine I've been building over the past couple months. Today I just had to assemble it making sure things were square and parallel, etc.
I intend to use it for isolation milling for prototype PCBs. That should be nice for proving out a design before sending files to a board house to have some professionally made. I'm a machinist and this design is intensive in that area, so wouldn't recommend it specifically to anyone not skilled in the field. There are designs out there which are much easier to make. Now in electronics I'm a green hack at best, so time will tell what kinds of things the machine will end up making, but fortunately I have a buddy who is my electronics guru. This was actually a product originating from our paired desire to create some cool electronics widgets, so it actually stands a good chance of being a very useful machine.

It could also machine some small parts in soft materials too, and of course do some engraving as well.

It's not actually "complete" as there are a few things which will be improved in the future (such as a better spindle and various other things), but as of today it's a working machine and capable of doing some work.

I built it on the cheap for the most part. Much was pulled from the scavenge and scrap pile. All together the bought stuff totaled around $100 (driver board, power supply, hardware, etc.).

Couple for overall view.





Here are the driver board and power supply.


Offline BANE

  • Supreme Robot
  • *****
  • Posts: 639
  • Helpful? 4
  • E=roboticsC^2
Re: My homebrew CNC
« Reply #1 on: February 27, 2011, 04:24:15 PM »
nice job and looks quite rigid,
I've been considering making something like this for pbc too since the printer toner method seems not to be working well for some of the boards ive made
Quote
$100 (driver board, power supply, hardware, etc.).
can you provide a bill of material for the electronics?

Also, are those linear bearings you're using?

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #2 on: February 27, 2011, 09:05:00 PM »
I've heard mixed reviews on isolation milling PCBs, but having been in machining for a while I've learned that many of the things you think should be simple end up needing a bunch of head-scratching and tweaking to get the process ironed out and dependable. I think that must be where the negative reviews come from because there are several sites I've seen where the authors are getting some very good results. It's a matter of working the problem and learning a process that works well for you, which will happen if you keep at it. By all this I mean to say that I expect to have a bunch of problems at first, but also expect to get it figured out.
Here's a link showing some pretty simple successful methods.
http://millpcbs.com/

Like I mentioned in the original, several things came from the salvage pile, which include the stepper motors and linear bearings. I think if you bought motors and linear bearings off eBay you could add another $75 or so on an average day, less if you shop around. And of course you could get creative and skip the expensive linear bearings, which is what I would have done if I didn't have some in surplus.

The stepper driver was about $33 + shipping from eBay:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=270706351680
(They've upped shipping $10 since I bought mine. Uugh)
That little stepper driver is popular and inexpensive from several sources on eBay, and it seems to work pretty well.

The power supply (24V 6.5A) was about $15 + shipping from MPJA:
http://www.mpja.com/prodinfo.asp?number=16854+PS
(Now out of stock. Uugh)
Of course any clear-thinking tinkerer pads the order a bunch here to spread the shipping charges out. You can always use some more heat-shrink tubing or something.

I already had wire  ;D That's about it for electronics. Didn't really build a thing there, just plug in and play. Since I can now (or soon anyway) make my own boards, that's likely to be a different story on the next project. But for the time and money involved I think I'd just buy a stepper driver unless I had special power requirements for big motors or something. They seem to pack some punch for the money these days.

A surprise to me was the dang hardware. Screws, nuts and washers totaled up after four trips to the hardware store to about $50 and can nickel and dime you to death. Granted that I probably only used 2/3 of what I bought because I had to change the design after it was "done" the first time but not performing acceptably. That stuff is just a lot, lot more expensive than I remember it to be last time I did something like this. They must be making them out of crude oil or something. I'm going to shop around for less expensive sources in the future.

Mach3 would cost about $150 more, but I purchased that already for a previous project. EMC2 would work just as well for free (some say better).


« Last Edit: February 27, 2011, 09:12:32 PM by vinito »

Offline Anoroc

  • Full Member
  • ***
  • Posts: 55
  • Helpful? 2
Re: My homebrew CNC
« Reply #3 on: February 28, 2011, 12:53:34 PM »
I would LOVE to have a small table-top CNC router table for my house LOL. Looks like you did one hell of a job! I have a 15" wide KNK, which is only 2.5 axis, (Only up or down for the head). Since I have no real z-axis control, I can't mount a rotary head to it. Not for a lack of trying! I could probably get away with doing some plastic engraving, but that would be the limit of its ability.

Offline garrettg84

  • Robot Overlord
  • ****
  • Posts: 187
  • Helpful? 8
  • Armchair Roboticist Extraordinaire
Re: My homebrew CNC
« Reply #4 on: February 28, 2011, 01:12:04 PM »
I'd like to see a video in action if at all possible =)
-garrett

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #5 on: February 28, 2011, 05:33:42 PM »
Yea I'd oblige if I could. Maybe I can borrow and pull that off sometime. However, in operation it looks and sounds (noisy!) just like the few dozen videos of similar machines on youtube.

Here's a picture of a fairly successful test I did last night though. Traces are too narrow but that's a simple matter of adjusting a parameter in PGB-GCODE.



Here are a few particulars:
Softwares:
Eagle for PCB design. It's free for small boards, pay for unlimited (?) size files. Seems to be the industry standard for most folks I run into.
http://www.cadsoftusa.com/

PCB-GCODE for generating G-code from the Eagle file for both isolation milling and hole drilling. It seems to work pretty well. And it's free !
http://www.pcbgcode.org/
This runs "inside" Eagle and it's easy to deal with.

Gsuite for optimizing the code generated from PCB-GCODE. Basically eliminates a ton of rapid moves from what PCB-GCODE generates for whatever reason and does a very good job of it. For a slow stepper motor machine, it's a godsend. It has a couple other handy utilities as well. Pretty neat. Also free.
Gsuite v1.3

Mach 3 for running the machine. About $150 but all I'm doing is running the G-code to the machine and for that EMC2 would work just about as well (some say better) for free. I think Mach 3 is a little overboard for this, but I bought a copy a while back so I'm using it.

Steppers are NEMA 23 and 1.5A, so about as small as this frame gets I think. For the force required for an engraving bit and those little bitty drills, these are more than adequate.

Lead screws are 1:1. Just 5/16-18 threaded rod. I was finicky about picking a pretty straight one and it wasn't too bad. Seems to be plenty accurate enough for the purpose, and dirt cheap. 200 Steps per revolution on the motor (which is typical and easy to find) and 18 tpi gives a nominal positional accuracy of .00028 per step, so more than adequate for me.
I machined the nuts from delrin (yer basic drill & tap) and that makes for a free-turning but close fit. Of course wear & tear will probably loosen things up over time, but they are easy to make and at the moment it indicates zero backlash.

Downloaded the board file in Eagle from here in case you are curious what it is:
http://arduino.cc/en/Main/ArduinoBoardSerialSingleSided3
« Last Edit: March 04, 2011, 07:44:01 PM by vinito »

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: My homebrew CNC
« Reply #6 on: March 03, 2011, 04:48:27 PM »
Hi,

Nice job with the PCB  ;D

How long did it take to route it?
Are you planning to route/drill the holes as well?
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline photomark

  • Jr. Member
  • **
  • Posts: 44
  • Helpful? 0
Re: My homebrew CNC
« Reply #7 on: March 04, 2011, 02:16:46 AM »
nice job , even more so considering it is a budget home built machine , I have also started on CNC and bought my self a cheap machine from China , I fried the electronics on the first day and even though it was coverd by warranty it just was not worth the cost of me sending it back to them , the company was good about it though and gave me a token $50 refund , I got a new power supply for $50 and a controller board identical to yours for around $60 .

I am running this with Kcam and using Turbocad 17pro and SolidWorks 2010 , for PCBs I found a great program called FreePCB , I also have Ventric Cut3D for generating 3D tool paths .

You got me thinking on just using all thread for the lead screws , I am intending on building a much heavier machine , not so much larger but I have a need for a very heavy duty machine and good lead screws are very expensive to buy .

The best thing about building a CNC anything is that you learn a lot about stepper motors, I had a box full of them ,all sorts and types ,mostly from old record turntables , I have had a lot of fun with them and intend on using them in a lot of future projects .

Well that is if I get some time

MT       

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #8 on: March 04, 2011, 07:00:21 PM »
Hi, Nice job with the PCB  ;D
How long did it take to route it?


Thanks for the kudos.
1) My motors were tuned to turn as fast as they could and still repeat reliably, which will move the table at 13 inches per minute. That's pretty slow, but not for steppers (a little over 200rpm).
2) With the parameters in PCB-GCODE set the way I did it for this test, it made three paths around each trace (to achieve the width I "instructed" it to make). That's 6 paths between close traces and with the runout I'm getting out of the crappy Dremel performance which makes the cuts wider than it should, that's way more than it needed. Actually I think it could have been done with a single pass around each trace, which would cut the time to 1/3 the time this one took.

So anyway, it took a bit less than two hours to trace that board out, so a single pass path would have cut it down to maybe 30-40 minutes.

Quote
Are you planning to route/drill the holes as well?

Yea that's in the cards, but this was just some double-sided material I had from Radio Hack from years back and I was just testing the trace routing, so no sense doing that to this board. It wouldn't have functioned anyway if I had actually finished it since the other side is solid copper. Anyway, I have run a second test of a smaller board and added drilling. The process worked OK, but the Dremel is in the way again, making a bit of ratty hole at the entrance to each, plus some ripping and tearing. Plus, I couldn't find my selection of drill bits and fudged it with what I could find, which were too large and didn't help any. Obviously a better spindle is also in my future, and was from the beginning. But a Dremel got things kind of going quickly and easily. Some parts are on the way and some I have started making already. It is a little bit "on-the-cheap" too, but will still be tons better than a Dremel and still better than several I've seen out there on the web. Of course having less runout will mean I probably won't get by with single-pass trace routing, but I'll be happy to trade that out.

I'm an old hand at machining, but this particular thing is pretty new to me. I'm doing it for several reasons. First, I want to be able to make my own boards of course. Next, I also consider this a prototype concept in which I build my way into a pretty nice servo-driven machine with more bells & whistles eventually - in other words, making my biggest mistakes and fixes now. Sure it's been a few dozen times before me, but I kind of like the process anyway.

And finally (which actually is more immediately applicable to #1 and similar to #2) I want to develop a reliable process for making boards. There are several variables and possibilities to work out. I need to add some stuff to this machine so I can mechanically line up the board on the flip and insure things are in register (alignment) for double-sided boards. I'm still exploring engraving cutters (angles, material, style) to find out which ones work the best and last the longest. Drilling the holes seems simple, but there are at least three ways I can think of to do it and need to experiment to find which turns out the best. Plus each solution tends to illuminate more questions to answer and I'll have to solve those current unknowns as I discover them. It's a good thing I enjoy this process.   :D

Interesting that photomark is using a whole different set of software to accomplish this same thing. Just goes to show there's a whole rainbow to choose from. As for all-thread for lead screws, I'm pretty happy with how mine works. That's one of the few things on the machine I haven't had to fiddle with at all - it just worked OK out of the gate and is still working fine. If it was a milling machine and needed to make precision parts, it probably wouldn't be adequate. But for routing PCBs it works plenty well. You just have to check them out when you buy them to make sure they are pretty straight. Just roll them on a table or floor and look for wobble. All-thread from the hardware store is notorious for being bent, so be picky. Usually threaded rod is used to simply hold stuff together so being bent is a non-issue. But for a lead screw you want to minimize that of course. Also, you might look here (click) and maybe it would be a good, inexpensive source for an even better solution. At 4tpi, you'd have to use belts and pulleys to change the motor/screw ratio to get acceptable resolution for accuracy, but that's not a big deal. They'd be beefy and smooth anyway.

By the way, if you want a decent spindle and don't feel like making one yourself, this one looks like a pretty darn good deal:
http://www.angelfire.com/az2/proff/
« Last Edit: March 04, 2011, 07:28:17 PM by vinito »

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: My homebrew CNC
« Reply #9 on: March 04, 2011, 08:28:33 PM »
Hi,

So anyway, it took a bit less than two hours to trace that board out, so a single pass path would have cut it down to maybe 30-40 minutes.
That's not half bad - I talked to a science teacher who started a shop with robotics parts and he used a ready bought machine at the faculty to make a similar size PCB in 7..8 hours.


[...] It wouldn't have functioned anyway if I had actually finished it since the other side is solid copper.
A lot of HF PCB's are made with one side copper, where a small pad of copper is removed from where the through holes are - makes a great ground plane.


[...] Of course having less runout will mean I probably won't get by with single-pass trace routing, but I'll be happy to trade that out.
As long as you have adequate isolation (and not doing anything extremely high impedance), there's no (electrical) reason to remove the in-between islands.


[...] I'm still exploring engraving cutters (angles, material, style) to find out which ones work the best and last the longest.
A rod with a splinter of wolfram (tungsten) brazed into a slit in the end and ground to a "slanted V" the shape of most engravers I've seen (I'm sure you know what I mean) is easy to grind into the size you need and is very long lasting.


Drilling the holes seems simple, but there are at least three ways I can think of to do it and need to experiment to find which turns out the best. Plus each solution tends to illuminate more questions to answer and I'll have to solve those current unknowns as I discover them. It's a good thing I enjoy this process.   :D
Yes, you really have to - to get it done  :)


Interesting that photomark is using a whole different set of software to accomplish this same thing. Just goes to show there's a whole rainbow to choose from.
Way more than in the early eighties when I first considered making one - perhaps I should get around to actually do it some time soon.
But it would be more for engraving, as I prefer photo transfer when it comes to PCB's.
Wonder if I could find a program to take a bitmap and trace that down on a piece of something, or are they all vector controlled.


As for all-thread for lead screws, [...]
Suitable screws can be found in car jacks and in several other places.
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #10 on: March 04, 2011, 10:03:55 PM »
Quote
That's not half bad - I talked to a science teacher who started a shop with robotics parts and he used a ready bought machine at the faculty to make a similar size PCB in 7..8 hours.
That was no doubt stepper driven as well. I've seen videos of servo driven machines and they whip stuff out tons faster. I wonder if he was going full-bore - I've seen some routed boards which have large bare sections of board, which as you pointed out, is not necessary for function. I'm sure your parentheses meant that other reasons may be helpful, like reducing the opportunity for solder bridges in shaky hands, etc.

Quote
A lot of HF PCB's are made with one side copper, where a small pad of copper is removed from where the through holes are - makes a great ground plane.
Yea and PCB-GCODE generates that side too in case anybody wanted to know. I'm not yet set up for flipping over for side 2 yet so that's moot for me so far. Coming soon though - no self-respecting hobbyist is thoroughly satisfied with single-sided boards. 8)

Quote
A rod with a splinter of wolfram (tungsten) brazed into a slit in the end and ground to a "slanted V" the shape of most engravers I've seen (I'm sure you know what I mean) is easy to grind into the size you need and is very long lasting.
I've found some pretty good looking solid carbide engraving bits which come in different angles and cost around $5. I'm using something like that now too. That's a pretty good deal and I hope they work out. I have a nice little grinder which could sharpen them or make from scratch, but it's not fully together as yet and all I need is another project. I think the style you describe is probably the most common style. The single-lip cutter is very stable and lasts a long time. The only problem might be with very high-speed spindles which would amplify the half-grind into some undesirable vibration. But I'm not spinning that fast yet, nor will the replacement spindle I'm building (I don't think).

Quote
Way more than in the early eighties when I first considered making one - perhaps I should get around to actually do it some time soon.
But it would be more for engraving, as I prefer photo transfer when it comes to PCB's.
Wonder if I could find a program to take a bitmap and trace that down on a piece of something, or are they all vector controlled.
Besides the software for generating various code, the whole DIY CNC arena has become much easier just within the past 5 years. With Mach3 and EMC to run it, and the cost and variety of driver boards or the knowledge base to pull from to build your own, resources for finding parts, and on and on. Not too long ago it would have taken a seriously knowledgeable hacker to have his own CNC anything for less than $2K minimum. Now an experienced hacker (i.e. a bunch of junk in boxes) without a lot of knowledge could essentially have one without spending a cent. And starting from scratch would still only cost between $100 to maybe $400 depending on what you wanted to spend money on.

As for bitmap to G-code, there are free packages out there to do that. Normally they just do a grid or spiral pattern over the work envelope and Z will vary depending on the gray value on the image, so the result will show the path pattern. Converting to vector and generating code to directly draw lines is trickier, so that stuff is generally for cost, but I haven't looked either and there might be some good stuff available these days. A chain might be easily workable, i.e. converting bitmap to vector, then vector to some kind of CAD standard (DXF, etc.) or something could probably be done pretty easily by someone with your knowledge. Then converting CAD to G-code would be cake. Text (TT fonts for example) are already vector and generating G-code from TTF is easy. I'd give some links but I haven't looked into this for quite a while so have nothing at hand.

Quote
Suitable screws can be found in car jacks and in several other places.

And Acme threaded rod can be bought new for fairly cheap too. It is sold with standard and precision grades. More expensive than standard V-threaded rod, but tons cheaper than ball screws (new). The nuts tend to be more expensive than they oughtta be though, but all any of this takes is a little shopping.
« Last Edit: March 04, 2011, 10:07:52 PM by vinito »

Offline Ro-Bot-X

  • Contest Winner
  • Supreme Robot
  • ****
  • Posts: 1,431
  • Helpful? 25
  • Store: RoBotXDesigns.ca
Re: My homebrew CNC
« Reply #11 on: March 05, 2011, 07:43:39 AM »
Nice job vinito, I love it. I've been following in the shadows for a while, but I have to ask for something:
Quote
I've found some pretty good looking solid carbide engraving bits which come in different angles and cost around $5.

Can you post a link? I need to get some for myself.

I built a home made CNC a while ago but did not try to make PCBs because I did not have a proper bit. My construction is way less precise than yours, perhaps I'll give in and spend some bucks to buy a ready made solution. But I'm waiting for a strong motivation, right now I don't have it. Actually, I would like to have a laser cutter (for plywood, plastic and perhaps 2mm aluminum) and a 3D printer like a Reprap. But I need to be able to make things out of them to recover for the costs. Perhaps an All-In-One machine would be the best thing, just have 3 tools on the tool holder.

Can't wait to see your progress!
Check out the uBotino robot controller!

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #12 on: March 07, 2011, 07:14:31 PM »
Sorry this took so long X.

The cutters I saw are $6.50 each unless bought in a lot of 10, which brings them down to $5 per (plus shipping in either case).
Here's the link:
http://www.drewtronics.net/

In my case, shipping was around $5 no matter how many were coming, which makes sense because they are pretty bitty things.

This site has some pretty good guidelines which have proven to pretty much work out true for me as well:
http://millpcbs.com/

and he has some well-explained info regarding use of the engraving bits here: (though I caught a couple errors which you will also catch pretty easily)
http://millpcbs.com/index.php?option=com_content&view=article&id=23&Itemid=47
He gives his personal preference for what angle of cutter to use. I'd offer my own opinion but haven't done enough experimenting to give an informed one yet.

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #13 on: March 14, 2011, 09:03:35 PM »
Just figgered I'd post an update, if that's what you can call it. Not much has changed lately, mostly because I'm waiting on a bit coming on a slow boat from China. It's the collet chuck. The shank is long enough to pass through a couple bearings in a block I'll mount on the machine to replace the Dremel. It takes a collet type called an ER8.

Picture of the collet:


In retrospect I should have just made it myself. Probably all I'll hold in this collet will be 1/8" shank tools, so I should have just machined my own spindle with a simple dedicated 1/8" collet for it. It would have been relatively simple and possibly more accurate anyway. I don't need the gripping range which ER style collets provide. Oh well. We all have brain farts now & then.

I still have to deal with a pulley arrangement and motor. Not a huge deal, but by the time I'm done I probably could have spent about the same $$ and just bought one of those ready-to-go spindles I posted about above and just made the mounting block to fit my machine.

I did drill and ream a series of 1/16" holes in the table so I can use pins in any two of the array for origin and alignment. This will enable easy setup for double-sided boards and "production" of multiple boards whether single or double sided.

Also have experimented a little and have improved a bit on the process. I actually made a couple usable boards (a single-sided "Really Bare Bones Board" Arduino thing I downloaded off the net - the official one can be seen here). I handed them off to my guru buddy to analyze. They are good enough to function, but I'd like to be able to do better and the Dremel is proving to be a dead end. I hope to take the quality up a few notches with a good spindle.

OK that's all for now. Things are busy for me right now but I hope to have the spindle up and running in a couple weeks. So new pictures, discoveries and revelations are forthcoming.
« Last Edit: March 14, 2011, 09:27:25 PM by vinito »

Offline PurdueRobotics

  • Beginner
  • *
  • Posts: 2
  • Helpful? 1
Re: My homebrew CNC
« Reply #14 on: March 21, 2011, 06:30:52 PM »
Very nice work.  Do you find that it is adequate as far as finished quality?  Have you tried it on alum.?  It appears that the z axis may be the weak link, meaning that the spindle of the Dremel tool is not as rigid as it could be.  I also am in the process of making one of these machines.  During the design process I gave the spindle considerable thought having had extensive experience with CNC machines on a large scale (such as MAZAK and Cincinnati Micron) and the spindles are nearly always where the accuracy is compromised after some hours are logged.  What I choose to use on my micro CNC is a DeWalt "dremel" style grinder.  It is about 50% larger in diameter, 2" longer and heavier, but the DeWalt uses larger bearings in the spindle and has a higher power motor with equal RPM capability. Between these two features it can handle larger cutters (up to 1/4" which is way more than is necessary) and will last considerably longer while maintaining accuracy and minimizing deflection on the spindle.
Again, great job in fabrication and an excellent posting on the details! 

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #15 on: March 21, 2011, 07:36:12 PM »
On the spindle, as mentioned a couple times above, the Dremel is not adequate for much but it was an easy way to get it up and running quickly. I'm absolutely sure that putting side pressure with an endmill toward anything, even plastic, in a Dremel will end up with not much but a bunch of howling and chatter. The Dremel I used was even one of the "nicer" varieties - kind of a pricier step-up one. But it's still only what it is with a bunch of vibration and not much precision to brag about. Honestly, Dremels barely hold themselves together just spinning air at the top setting. They're good for what they were made to do, and this is not what they were made to do.

I have similar background to you in CNC I guess in that I've been around the big boys a lot, ranging in value from around $50K to $800K. Oh heck yea, they work great. My budget doesn't even comprehend that kind of outlay though. So it's Dremel to start, then a spindle out of my own hands to actually do decent work. The piece arrived from China today, so it's R&D time again. I have the motor and bearings (angluar-contact high-speed ball bearings), so have to make the rest of the parts and mount them. Plus it's pretty busy at work right now so probably be at least a couple weeks before it's built. I designed the front of the Z-axis to be a simple plate with four drilled & tapped holes so it will be a very simple matter to slip the new spindle on and hit the ground running.

Thanks for the kudos. Once I have some interesting developments I'll report back with more pictures and such.
« Last Edit: March 21, 2011, 07:39:42 PM by vinito »

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC - revived
« Reply #16 on: May 22, 2012, 12:49:47 PM »
Well after about a year of pause, I figured it was time to revive my CNC project and get the thing to finally being "done", i.e. actually able to make things and do a decent job of it.

First, I did finally determine that my cheap eBay TB6550 driver board I bought off eBay turned out to be the source of the low performance of the steppers, so I bit the bullet and ordered a Gecko G540 and it should be here tomorrow. I won't really have time to play with it much until this weekend, but I'm looking forward to seeing a big improvement.
Those TB6550 driver boards do have a lot of negative reviews and I've read through a long post where some folks have looked into them and corrected errors or upgraded things to make them work much better. But I figured I'd just skip over the uncertainty and purchase a driver which has proven over and over to perform well, at least for now. I'll try fixing that cheap old thing later on I'm sure. That will be fun to try, but for now I want a quick fix and reliable machine.

I also considered purchasing a smoothstepper, but then I read that several people have had problems with them as well so figured if I want to add an interface like that I'll just look into it later and make sure I get something proven to work reliably on that as well. I hear that the ethernet smoothstepper is supposed to be much more reliable, but at the price (quite a jump from the USB for whatever reason) there might be a better solution anyway.
Edit to add:
Turns out the ethernet smoothstepper is about $30 more than the USB smoothstepper - for some reason I thought it was much more difference. Since it reportedly is much more reliable and less buggy than the USB, I may yet get an ethernet smoothstepper at some point. For now I've got an extra parallel port card I can use for more IO if I need it (I'd like to make a nice control panel too eventually). We'll see how things wring out with what I've got and watch the continuing development of smoothstepper and decide later if that would be beneficial or not.


So by the end of this weekend I should have some results on the drive system to report about.

I ended up making a spindle and tried a motor on it which didn't work, then made a coupler to drive it from a dremel which exposed a bad rattling (maybe just dirt in the bearings) so that didn't work out so well either. It turned out decent and almost worked and I might look into making it work correctly later too, but again, quick fix and reliable spindle is wanted at the moment.

A pic of the unfortunate spindle experiment:


So I'm planning on trying out one of those high-speed air-cooled spindles from eBay. A query on a different board helped me determine that I should get the spindle separately and purchase a separate Hitachi VFD. So that's an upcoming upgrade as well. Any advice from experienced members here is welcome too. I won't be pulling the trigger on that for a little while.

So the machine initially cost roughly $100 to build and now I'm biting off another $600 or so which makes no sense, right? Well there is a more complex master plan of course.

A while back I purchased a stepper driven CNC device for scrap/parts price. It is a little larger, much heavier and more precise. It appears to have been some kind of video inspection device of some kind maybe. I'll snap a photo of it and post it later. But the lowdown is that I should be able to use the new drive and spindle on the old machine to make some PCBs for decent home-brew stepper drivers which will ultimately be installed on itself so I can move the Gecko to the new, bigger machine. At that time I will move the VFD spindle to the new machine as well and work out a different spindle fix or solution for the smaller machine.

Redundant? A little. But the bigger machine will be rigid enough to do some milling and of course the larger work envelope won't hurt. Also, the bigger machine is too heavy to be portable while the smaller one is very easy to carry and transport. Sometimes I do demonstrations or the like so a portable CNC machine will be handy now and then. I could even loan it out to the local robotics or MAKE clubs or something like that now and then.

OK, that's it for now. More updates soon.
This stuff sure is fun.

Edit to add:
Since I'm going to try to make my own stepper driver once the machine is running, any suggestions for a good choice of the one to build would be appreciated.
I was most impressed by the one this guy was able to make and could copy it since he provides what you need to roll your own:
(link)
This one looks to be smooth and fast. Of course that doesn't necessarily mean it is tuned with a bias toward torque at all, but maybe it would be a good one to try?
That's the one I would make if I was going to do it today. If anybody knows of another good (better?) one for a CNC machine, please share and discuss it here. My head is pretty open on that project currently. Thanks.
« Last Edit: May 22, 2012, 09:43:12 PM by vinito »

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #17 on: May 23, 2012, 08:38:51 PM »
Woohoo!
Ups delivered the G540 today already. Sure seems a bitty thing. I'm currently loading XP into an old desktop so I can start from clean scratch. I even found a spare little 1024x768 LCD monitor for it - perfect.

I have a couple steppers in the parts bin so I can fiddle with it a little bit. I actually loaned the little machine out a few months ago to my electronics guru so he could mess with it and try to learn Mach a little, but he didn't find any time anyway. So we're getting together this weekend so he can bring the machine back and we can start configuring things.

Trudging forward...

Offline jkerns

  • Robot Overlord
  • ****
  • Posts: 269
  • Helpful? 12
Re: My homebrew CNC
« Reply #18 on: May 24, 2012, 09:47:55 AM »
A lot of variations on this sort of thing here: http://mtm.cba.mit.edu/
I get paid to play with robots - can't beat that with a stick.

http://www.ltu.edu/engineering/mechanical/bachelor-science-robotics-engineering.asp

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #19 on: May 26, 2012, 03:55:45 PM »
Kind of working slowly on the resumed project.
First reading documentation on the G540 to make sure I don't burn something up.
Soldered up the connectors with the correct current resistor tucked up in the plug and made up a terminal strip to put between the connector ends and the motor. The terminal strip is so I can both anchor the wires coming from the motors securely and switch the wiring around sans soldering if needed.

I may have the motors turning again later tonight if I don't have too many interruptions. Should be a half-hour job really, but I have a few other irons in the fire today too.

I looked in the PC boneyard trying to find an adequate box to run this machine. Everything is either too far broken or too weak to do a decent job :'(, so I've thrown more money at this project. I think it will be worth it though. What I ended up getting is a mini ITX, nice little case, 2GB RAM, 250GB Hard drive  and etc. to just build a new little system tailored to operate the machine. Mach3 requires a 32-bit OS so I'll be installing a bare-bones XP on the hard drive. It has a built-in parallel port (getting harder to find that these days) and I got another PP card to fit the one slot on the MB in case I want to add more easy IO in the future. For less than a couple hundred bucks, this little PC ought to work out pretty darned well. So interjecting the new brains into the system (and probably re-tuning the motors to suit) will be next weekend's project.

This new control stuff will eventually be migrated to the "new" CNC project. Here's a pic of it in its current state:


It's kind of a mess, but trust me, it's a pretty nice little machine and should do a great job. In the meantime, the one I built from scratch should be able to make some of its own replacement parts. If you look closely, you might see a vacuum pump and vacuum table. ;D

Why have just one when you can have two at four times the price?

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #20 on: May 27, 2012, 12:41:00 PM »
I set everything up and tuned the motors the best I can. It's just not performing very well.
The videos I've seen of Mach3 running steppers through a G540 all show pretty smooth and solid motion. Some are even pretty fast. Mine are pretty awful, though better than with the cheap Chinese driver.

We'll see if there is any improvement with the new PC. Better performance is likely but I suspect my motors, which are the smallest NEMA23 frame I've ever found, are simply not well-suited for this kind of task.

If anyone has experience with this kind of stuff and can recommend a good motor to work with Mach3 and the Gecko, please let me know. There has to be a way to get this thing running right. I'm shooting for reliable, smooth, fast motion (in that order) and it seems like I should be able to get it with the Gecko.

Thanks for any advice.


Edit to add:
I set up a desktop box and got no better performance, then I discovered that the bad-performing motors both had one wire come loose from the terminal block I installed under the table. Uugh. I fretted about it all day, got confused and frustrated, and the fix was a simple snugging up of two lost connections. So the good news is that all three axes are now fast, smooth and torquey. Tons better than before. I think this machine might be worth having around after all.
« Last Edit: May 28, 2012, 02:44:55 AM by vinito »

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #21 on: May 28, 2012, 10:48:44 AM »
Wow the Gecko is working great. With the Chinese driver I could get max feedrates of 13 IPM on the table and 4 IPM on Z. Now I am repeatably getting 40 IPM on all axes, which maxes out the Mach3 kernel so may get even faster if I can jack the kernel higher with the new system.

Here's a short video:
CNC machine Mach3 and G540


Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #22 on: May 31, 2012, 08:49:23 PM »
Made a bit more progress.
If you watch the video, you will see that above the spindle there is a bracket holding nothing. I made that to hold a dremel and at first it was mounted at the bottom of the Z-axis plate thing. After I made the new spindle and found that the motor I intended to use on it was waaay too wimpy to work on it, I re-thought and decided to modify the dremel bracket and move it to the top so maybe it could drive the spindle. So I hooked the dremel back up and connected it to the spindle with a coupler and tried it out last night. There is some vibration, which is worse at some speeds than others so I just have to fiddle and get between the harmonic nodes. But it does work OK. I was afraid that the bearings would just burn up at those kind of speeds, but it ran for 20 minutes or so last night at about 20-25K rpm and it hardly got warm. A pleasant surprise.

I went a little further and milled flat a piece of plastic to true the table up to the axes and that worked like a champ. I can tape copper clad to the now flat plastic and etch PCBs. I'll be giving that a try this weekend.

Also got the new system built and installed. It's a sweet-running little thing and I've been able to ramp the rapids up to 50 inches per minute. Since I'm only driving my motors at 25 - any faster than that and it loses torque and skips. I'm happy with that. 50ipm is so much faster than I had before that I have no plans to increase the power supply voltage just to get a bit faster rapids out of it. And of course the new system is just generally tons smoother and faster than that relic I had it running on before.

Once I get it etching some PCBs I'll make another small video and link to it.

Getting pretty close now.

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #23 on: June 16, 2012, 07:31:59 PM »
I finally milled my first usable PCB with my little machine.
It's not perfect, but it would be functional. You'd just have to be a little more careful with the soldering I think.

The method I used to flip the board to mill side 2 was fairly simple. I just drilled holes of a proper diameter to fit a couple small pins I had at 0,0 and 3,0. On this board, the way it was laid out on Eagle had the origin (0,0) slightly off the board outline, so the registration holes will be gone once the board is cut off the material. You could use any two holes, but using the origin for one of them is handy and quick way to get things set up for the flip.

I'm very pleased with the motion of the machine, but the spindle still leaves something to be desired. So I'll be doing something different eventually.

I'm not sure how this occurred, but you can see that the registration kind of got spread out from one side of the board to the other along the X-axis. The motors didn't skip steps, it's just off in the code for some reason. I'll probably try some other method of generating the code and see if I can fix that. The PCB-GCODE thing is a bit awkward to use anyway, so maybe the next thing I try will be better anyway.

Side 1:


Side 2:


Here's a short video of the machine making the board this morning. There is no sound on purpose - it's just an annoying whining sound from the noisy spindle, and the sound and video got progressively out of sync as the video progressed for some reason, so I just left it out:



By the way...
The machine is successfully rapid positioning at 60 inches per minute but I have it ramped to keep things solid, so it takes about an inch of motion to get up to full speed. I can probably reduce the ramp time, but the long side of milling on this board was only 8-1/2 minutes. Not bad.
« Last Edit: June 16, 2012, 07:35:03 PM by vinito »

Offline vinitoTopic starter

  • Full Member
  • ***
  • Posts: 101
  • Helpful? 6
Re: My homebrew CNC
« Reply #24 on: October 14, 2012, 11:18:22 AM »
Well OK. I finally got my little machine working pretty well. Yaaaaay!!!

I bought a fairly inexpensive spindle off eBay. As far as I know, they are kind of new on the scene. I've seen them before, but not common and not on eBay, though it could be that I just wasn't using the correct search string. At any rate, the spindle spins about 1K to 30K rpm as per the specs and the runout is very small (less than .0005"). It is very quiet as well - MUCH quieter than a router motor and even quieter than a Dremel by probably 3 or 4 times. You just hear a whoosh and little hum. It's pretty torquey too even at low RPM. It takes an ER collet like the one I built, but it's the next size larger (ER11 rather than ER8) and will accept up to 5/16" diameter shanks if I ever have a need for it. I won't on this machine, but that's the capacity anyway.

I wired a relay in to switch it on and off from Mach3 and plan to eventually have Mach control the RPM as well. It's pretty nice just to be able to let Mach turn it on and off - seems more professional or something. For now a simple 5K potentiometer controls the RPM. Another nice thing is that it comes up to speed in probably a quarter of a second, so no delay/wait necessary in the programming.

I did a little testing then ran the same board I did before. The results are dramatically better. I haven't figured out how to take a good picture of a PC board yet, but following is a picture of the spindle and the results of the routing. Hard to see (can't see it?) but the result is very, very clean and crisp. It's just tons better than before and I'm happy the long-awaited good performance is finally here. Looking forward to being able to actually do some good work on it.



Yesterday's new routing on top. Old routing (from before) on bottom.
Same exact code, cutter and depth as before. The old routing cut so much wider because of the runout. That was with my home-made spindle too, meaning a Dremel performs even worse.


That new little spindle is a nice setup for a reasonable price and I highly recommend it if you have a similar project. Search "300 watt spindle" under Business and Industrial on eBay and you'll find them there (with various options, i.e. with mounting bracket, with PWM speed control, with power supply, etc.).

Or here's a source in the US who sells them for a little higher price for the same thing: link. This source also carries a good selection of other pretty nice stuff for decent prices. Interesting source. Their home page
« Last Edit: October 14, 2012, 01:34:35 PM by vinito »

 


Get Your Ad Here

data_list