Are these assumptions correct?

[Ubarch]

Hello,

From what I gather, posts of this sort are pretty regular around here, so I'll try to keep it entertaining. I'm new to CNC; myself and some partners are looking into starting a business that would require prototyping of shapes, possibly some that I do not believe a human can make (precise, nonlinear surfaces). It's a robotics company, so naturally we anticipate the need for many strange parts, some of which would become rather expensive at eMachineShop. If you're wondering why a bunch of people that cannot machine parts think they can start a robotics company, I can answer that. We are all electrical engineers; to build products we have skill sets covered up until and excluding a machinist. I'm hoping that a desktop CNC system, plus hours put into learning how to use it will allow us to prototype what we need, when we need it.

That being said, I was doing some research. Sherline mills looked very cool at first, and the "ultimate machine shop" would just fit in our budget, but I actually found someone that had a Sherline and they expressed less than %100 confidence in it. The Taig systems looked very nice, but it bothered me greatly that the servo retrofits seemed to be grabbing the encoder data off the back of the motor shaft, when the actual motion you are measuring goes through a belt. Why not just buy servo motors with attached gearboxes as needed, and affix an encoder to the shaft, post-gearbox? We've done that before for different applications of motion control and it worked beautifully. With this logic, I set out to find some of the motors we were used to using (TRW and Maxon), but finding a reseller of those things is kind of hard (ours were surplus). I figured this would be a good option should we be able to find the motors, because we have already designed PID controllers for them and they would require minimal redesign. I'm assuming that servo CNC systems use some form of PID, and aside from power and a nice H-bridge, we would need only worry about the interface between the software and the PID microprocessor. Is this correct?

Anyhow, gettign the the real set of questions. I found this place called industrial-hobbies that made a servo kit that mounted the encoder on the ball screw, which sounded like a much better idea. I called the place up, and it seemed like the guy had some practiced.... gospel... to impart on me. I wanted to run some of the relevant things I took away from the phone call by you guys:

-- Desktop mills (such as the Taig), for the most part, are not up to the task of cutting steel.

-- The same class of mill can cut aluminum, but it cannot cut it accurately (the term he used was 'chattering'). This seems odd, because I've seen a lot of seemingly accurate aluminum work in the internet, some of it involving steel.

-- The above two are marginally improved by purchasing a larger mill such as the Sieg-X3.

-- Again, for the same class of mill as the Taig, the performance of a servo CNC system is negligably different from a system that uses steppers. However, he did say that servos over steppers were "nice". This seems to be in conflict with the above statement, and I was wondering if anyone had some quantifiable data that would seperate the performance of the two motion systems.

-- Proper CNC machining is very, very hard.

Apologies for the length of the post. Thanks in advance for any help you can offer.

[HuFlungDung]

It sounds to me, from what you wrote, that you guys need a real machining center, not a toy. Something like a Haas VMC, which come in many sizes. Because these machines are "real cnc's", they take a lot of the "hard" out of proper cnc machining.

What size range are your components?

[Ubarch]

Not very large at all, although I'm hesitant to predict very far into the future. In the forseeable future, we will not require anything that cannot fit within a 5" cube.

I found some VMC's on eBay, and these look like they are far, far beyond what we have for startup budget. It looks like buuying a used Haas ($30k) would be more than sending every prototype we need to eMachineShop.

I guess what I really don't know is where the dividing line in performance is, between 'toy', and 'real' cnc. What kind of parts are on the borderline between manufacturable and not-manufacturable for, say, $5000 of homebuilt system?

[HuFlungDung]

Well, you did say that you were thinking of starting a business, so I assumed that productivity is a bit of a concern. Prototyping is slow enough going, with good equipment.

Typical homebuilt takes quite a lot of time to do correctly, although I'm sure you guys could design something. It can be frustrating to find out that you've built it "short" of all the features you will need, and that is why a commercially made machine is a good choice: someone else has killed most of the bugs off already.

What kind of accuracy requirements will your work have?

[Ubarch]

Well, you did say that you were thinking of starting a business, so I assumed that productivity is a bit of a concern.

This is a good assumption to make, but I didn't mean to give the impression that myself and my friends are cashing in our life savings and renting property in the art district. Currently, we all have day jobs, so I guess you can think of it as a business-as-a-hobby. Several of our colleagues from college have really nice little businesses of this fashion. The only thing they really manufactured themselves was intellectual property, and dealt with subcontractors if they had to make many of some physical part. They did some soldering, tech writing, and handled the e-store. When dealing exclusively with electronics, you can afford to pay for many one-offs (there are fabrication companies with $30 deals that include 24 hour turnaround time), but this is not the case with parts that need to be machined. Since we no longer have access to our lab's machine shop, the CNC option seemed like a good way to prototype.

I'm sure we could build a lot of stuff, yeah, but I'm afraid we'll end up tinkering on the wrong things! As for accuracy, I was hoping for a couple thousandths tolerance at least, with hardened aluminum, and the ability to make things like snap-together ball and socket joints from Delrin or nylon.

Is this unreasonable for a Taig?

[HuFlungDung]

I hope someone else with actual experience with those will chip in.

Personally, the smallest machine I would bother owning is a mill with about 10" of Y travel, and 16" Z travel. This is simply because of fixturing: its nice to be able to hold stuff in a "regular" 6 inch vise, and it may surprise you to learn that it takes a machine this large to really have room to mount a "standard" 6" Kurt vise. It also takes a fair bit of Z travel, because the vise is about 4" high, and then you need room for your 5" cube plus ordinary jobber drill lengths (maybe 6" for a 1/2" drill), so that is 15" of Z travel, more would be better.

[nervis1]

I had a Taig for a while that was CNC with a flashcut system. If you are fooling around on Sat afternoon with a few hours to burn they are fun. As soon as you want to make anything bigger than a watch case in steel, or aluminum, with 1-2 thou tol, you will be dissapointed. 1/4 EM will bog the motor and feeds are really limited due to the 1/4 hp motor.

Listen to Hu, get a used VMC and be done with it, even an old bridgeport retrofitted would be a hundred times better than a little desktop.

I remember taking almost three hours to cut a little part on my Taig in 6061 that was 4" in diameter, bunch of 0.5 inch pockets, had to stand there the whole time with a squirt bottle of coolant and a brush, yuck. When I got my little Fadal I cut the same part in less than 5 min, with a much better finish. Plus I could walk away and do something else while it cut (not that I did my face was glued to the lexan window .

I've seen plenty of used CNC knee mills for 5k. If you get into the 12-15k range you can get a decent VMC if you look around.

As far as CNC being hard, it depends on the complexity of your part, and a lot on how good your CAD/CAM software is. I still have a lot to learn but I've done things this year that I would have thought were impossible last year, never really felt overwhelmed, just learn as you go and ask lots of questions (the set up is the hardest part sometimes).


What are you planning to use to program / run this mill?

[ViperTX]

Well you mentioned ball & socket made from delrin & nylon...depending on the dimensions and the need for the ball to snap into the socket and be retained by the socket, I'm thinking that any VMC would be used mainly for roughing a cavity I believe what you'll end up needing is a mold to form the outside surface of the socket, the ball will most likely be a type where it breaks into several parts...your socket will mostly likely be a press mold...I would recommend drawing up the parts or products and then approaching a company that produces prototypes and get recommendations on how best to mass produce the parts. Are you all thinking of medical implants?

[nervis1]

I think he's makin robotic parts.

[ViperTX]

Also, when you start looking at molds with intricate cavities...you'll probably need to use EDM for those cavities.

Thanks Nervsi1, so these are most likely rotating joints which can be assembled from pieces which will reduce the complexity of the socket component...they can piece it like the artifical human ball & socket joints.

[Ubarch]

The ball and socket is actually something I've made before by hand from both materials, but it was absolutely arduous, and it took about four prototypes to zero in on the correct dimension that woud allow it to "snap" into place without a.) cracking the bushing, or b.) giving too much friction.

The socket bushing was made on a lathe with a special tool. I ground out a half-circle on a bench grinder, and just sat there holding it up to the light against a proper half-circle cut into a template until no light shone through. It ended up looking like the letter P. I then used this tool to 'drill' into a piece of Delrin/Nylon (I experimented with both), and dug out the cavity, leaving a constriction at the mouth.

With a 4th axis and the ability to change the angle of the column, I think that structure is millable. I found both materials to be very easy to machine.