View Full Version : What were your lessons learned?

12-07-2004, 09:13 AM
Hi folks, I have a question to those who have gone before me. You might have build a dozen CNC routers, or just finished your first after being inspired by this website.

So my question is this: Could you list the things you would advise those of us still in the planning stage? Like, the things to look out for, or the things you wish you would have done differently.

Your sharing will greatly help those of us reading, thinking, and planning.


12-07-2004, 02:22 PM
I like that idea.

12-07-2004, 04:51 PM
The biggest things I can think of are....

"Sort of Rigid" is not good enough.
"Sort of accurate" is not good enough.
What seems "fast enough" rarely is.

12-07-2004, 07:54 PM
Not to make this into a shrink session but it depends on your personality.
When those who know you hear that you are going to build a machine do they think "Yeah sure, yet another half-fast unfinished project." Or do they think "That guy is always making the neatest stuff."
If you are an "up like a rocket and down like a rock" sort of person when it comes to projects, set the bar very low; build an embarrassingly cheap POC (Proof Of Concept... Piece Of Crud) that allows you to learn while not getting caught up in all the gumption traps that the quest for perfection invites. The next machine can be the good one.
There are over 7000 members here and just over 70 claim to have finished a machine...and it is the same theme on other forums.."I built two while he still plans the perfect one."

Okay, That's said, now for specifics regarding a first DIY machine.

Use gravity as your friend, design so that parts will "fall" to where they are supposed to be.
Remember basic geometry and physics. Design parts so that they are intrinsically stable; triangles are also your friend....and three points define a plane.
Make everything adjustable, getting it right on the first try is unrealistic so allow wiggle room.
Be realistic about your skill set and tool inventory.
High quality parts cannot be made with low quality tools.
Unless you have access to a machine shop, 0.001" accuracy is virtually unobtainable.
You WILL make mistakes (so start small and cheap)
Everything takes longer than expected so have patience, take the time to be careful; don't "Measure with a micrometer, mark with chalk, and cut with an axe"

Can't think of more right now but I address this a bit on my site as well.

Good luck!

12-07-2004, 09:49 PM
i got one.

it cost 10 times as much as you wanted and 2 times as much as it should have :P

12-07-2004, 10:06 PM
High quality parts cannot be made with low quality tools.

Not necesarily true.

12-07-2004, 10:19 PM
Yeah, I should have said "cannot be made quickly or easily..."
Ancient jewelry and sculpture always amaze me...they made it with just those basic tools?! Patience and perseverance at a level we can no longer imagine.

senor J.
12-07-2004, 10:44 PM
I got some advice that I learned the hard way... Trying to construct a z axis with precision bearings and ball screws and ball screw end blocks with minimal tools Ext Ect is 10 times harder and the exact same price as buying a premaid z axis assembly from parkerdaedal or thk on ebay. If you find a linear actuator that has the right stroke length you want but is $50 more than you think you could make it for buying the componets individually, shell out the extra fifty bucks you wont regret it and you'll have a machine up and running so much quicker.

12-08-2004, 03:04 PM
Great comments so far!

My personality is such that if/when I decide to do something, it will get done. To be honest, I'd really rather bypass the "build a small cheap one first" phase, and try to do as much homework/design up front for a good "keeper" first attempt. Even if I spend a year designing, to me that part is a lot of fun.

I like the z-axis suggestion...any other comments about ballscrews, slides, rails, motor couplings, drivers, power supplies, etc?

Regarding the budget, I'm trying to be realistic...I figure that to build the machine I have in mind (servos, 30" x 48" x 12" workspace, "very good" accuracy, and "not too slow"), I have to plan on spending around $3,000 USD.


12-10-2004, 03:34 AM
Suggestion number 1. Do your homework.
Learn about all the various components, brands, features. Chose your components based on this info and your wallet.
Chose a solid, basic design. Simple and solid rules!
Take your time and build true.

12-10-2004, 08:18 AM
Decide what components to build vs buy.
A breakout board has maybe $25 in parts in it but a long weekend worth of DIY time ... < $100 will purchase a good prefab that is tested and ready to go.
Motors can be picked up on Ebay but adding encoders, power supply and controllers that work well together can be a challenge. A package, or parts that are sold with good support, can make things MUCH easier, but it costs more of course...in money vs. time.
Same goes for just shopping around, again Ebay has good deals but the time it takes to find them may not be worth the savings. Some folks enjoy the window-shopping, others find it not worth the hassle and risk that goes with used goods.

Oh yeah, your tentative budget sounds realistic except that a good control box with cables will scream past $500 and approach $1000 really quickly. While on that subject; use a big box for your controller system, a computer cabinet is a decent inexpensive alternative. Things take more room than expected and it is a regal pain to work in a tiny enclosure. You know how I know. :rolleyes:

Again, I discuss this stuff in a show and tell on my website. Not trying to drive traffic that way, but it already addresses these issues a bit.

And good luck!

12-10-2004, 08:29 AM
Wow...didn't realize the controller (and cables) would be that much...good thing to ponder...

If you are allowed to (?) could you post your website address? If not, PM it to me.


12-10-2004, 08:41 AM
Click on his name in his post, and you'll see a link to his website.
Also consider that many people here have spent less than $250 on their controller setups. Sure you can spend a lot more, but it all depends on what you want.

12-10-2004, 09:41 AM
Chris, I can't help myself. You thread seems to be changing from a generic, what did I learn when building my first machine to what do you recommend for building a specific style of machine.

Anyway, here is what I learned.

1. Decide upon your application; type of materials to be cut, overall size, accuracy required, output needed, etc. I often see people say, "All I want to cut is foam, etc." They get advice accordingly and then wonder why their machine won't cut titanium alloys to .0000001" tolerance.

2. Turn this into a machine specification.

3. Don't try to do everything in the first machine. For example, I made a compromise in machine size and motion to try a build a flexible machine to do 3 different applications. I am now building 3 machines, one for each application. For example, I don't need a 30" x 48" machine to cut 4 axis parts the size of a pop can. Nor do I need 12" Z for cutting plywood parts, but I need a large bed.

4. Use as large of motors as your drive can handle.

5. Don't compromise on the linear slides. Especially on a 48" X axis.

6. Listen to everyone's advice, look at their comments in context and what they say to others. You'll get an appreciation of their skill and philosophy fairly quickly. Then do what you want.

With respect to your project. $3000 is an achievable goal. I would recommend you look at some plans for a machine this large.

You will have to look carefully for components as the cost could creep up to $4500 easily.

Xylotex now makes a 3 axis control box with cables for about $350.
You could probably make the same thing for about $200-$250. But I like the idea that its already done and ready to go. His controller will handle stepper motors up to about 280 oz-in and I've been happy with it.

I use the 5/8 ball screws from McMaster Carr and they can't be beat for price. The stepper motor is directly coupled to the ball screw using a spider coupling. I could not see adding in more complication of a timing belt, etc.

You can't beat EBAY for finding linear rails.

Another cost, You can easily spend $300 to $1000 on software. Don't forget about this cost

TurboCNC $60 or MACH 2 $150
CAD program $60 to $600
CAM program free (limited use) Sheetcam 2-1/2D work $150 Meshcam 3D work $150
Plus about a million other options.

12-10-2004, 01:13 PM
Yes (buscht), you are right...I wanted to keep this as much as a 'generic' lessons-learned discussion as possible, but it gets hard to go much further than "make it rigid, size your motors correctly, plan on spending a lot of money" -- without dropping a couple of hints as to the size of the machine.

Since we are in the 'woodworking' forum, I figured that my stated working envelope would be somewhat 'generic' for this audience...maybe a bad assumption on my part...

Your comments were very helpful...the software piece alone wasn't something I had put a lot of thought into! I think the gem you gave me was "don't try to do everything in the first machine". I really want to work out the specs, spend a LOT of time on the design, materials and component selection as possible -- and then set about building. I think that having a clear focus on what the machine will be intended to do, will go a long way to smooth the path.

Someone else mentioned using triangles in construction as much as possible...I had always known that in structural engineering a triangle is desirable -- but in thinking about it, it's given me a whole new set of ideas to think about for my bed design, and my gantry -- another gem.


12-10-2004, 02:10 PM
Chris, another thing I learned is that I don't want to be an electronics engineer. Soldering up do-it yourself circuit boards isn't my idea of a good time. So I spent extra money in buying a completed driver board. Just getting that interfaced to the PC was enough of a challenge for me.

My point is that my goal was (is) to produce products. I decided that a custom, low cost CNC was the best way to do it. The fun and excitement of building the machine/electronics is secondary to me.

Alot of guys enjoy the process and building the machine is the point. If that's your goal then your decisions will alot different than mine.

12-10-2004, 07:41 PM
Also consider that many people here have spent less than $250 on their controller setups. Sure you can spend a lot more, but it all depends on what you want.

I suppose this brings up another issue; steppers vs. servos (I assumed you were using servos, guess I confused this with another thread) but anyway with servos the power supply and controllers alone will be over $400 and this is with Gecko's current special sale price. Yes steppers can be cheaper but they introduce other issues that are addressed in servo vs stepper threads. So this choice is something else that has to be thrown into the equation....

12-11-2004, 06:45 AM
Here are a few things that I learned while building my router.
1: Do the design work from the inside out. Use your router for a size reference and work your way out to the x axis. I had to re-design my z because of space limitations, this is after I had already machined most of the componets. If i had worked my way out it I would have saved myself time and money. On top of that, i don't have as much z height as I would have liked.
All I had was 9 inch rails for my z so I thought that would be plenty of travel.

2: This was already said but I'll add to it.
if possible, make your componets adjustable but drill extra holes to install pins once you have it all squared up, make sure the hole is where you can get to it with a hand drill to install the pin. A little forethought goes along way here.

3. keep a notebook, write down everthing you can, there are alot of details that you will "do later" that might be forgotton if you didn't write them down.
I had to disassemble my z so I could tighten up the 6-32 set screw in the nook ballnut flange. Of course I had to realign everything.

4: Common sense....Loctite after everything is squared up....remove one screw at a time and add loctite.

5: You should not set your rails until you read 5bears website. His method works very well....it requires more effort but is well worth it.

6. Design your motor mounts so the steppers or servos can be bought and installed after the table has been constructed. One of the first things I bought off ebay were 4 nema 23 steppers. By the time they arrrived I knew I would need bigger motors....they are still in the box they came in...more wasted money.

7: If you use 6061 T6 alu. it will not be flat. I used alot of 1/2" stuff but the finished size will of course be less then that... .450 is what I used for most of the cad work on the componets I thought should be flat. I'm sure this is basic stuff to the eng. folks...I'm just a old e-tech maintence worker.

I'm sure there are more but thats all I got right now....hope this helped you in some way....good luck.


12-11-2004, 09:48 PM
Excellent...thank you...

Sol, I checked out your website, too - very nice & informative. This thread has turned out to be pretty info-dense...great pointers and foresight.

I was wondering about the aluminum selection. My background is mainly computer science, some electrical engineering and woodworking as a hobby-- so the metal working/fab aspect of building a router has me intrigued (and a teeny bit worried!). I checked aluminum prices, too - man that is stuff you don't want to have too many "do overs" with! :eek: