Speed/Accurate,controllers,PSU

[thuffner3]

I'm wanting to build a new machine.
From any or all, I'd like to know what spped your achieving(consistantly),
Your accuracy, and (short and sweet)
If you'd change anything?
And with what control and power supply?

I have been looking and looking and the info is somewhat overwhelming.

Thank you for any input.

Peace
Neil

[Swede]

Neil, your question is pretty general, and it all comes down to the components installed, and these are driven by $$.

Any set of ground ballscrews will outperform rolled screws for accuracy and torque requirements, and these will usually outperform typical ACME screws.

Servos are generally better than steppers. Ways can be cylindrical bushing, THK-style rails/trucks, or iron dovetail. Each of these has advantages and disadvantages relative to speed and accuracy.

The spindle is really the heart of the machine and I don't think we "talk spindles" as much as we should on this site, maybe because there really aren't that many smaller spindles available. You go from modest Sherline setups, and mini-mill R8 heads, and jump right to big $$ HF spindles, and big iron Setco and Gilman ISO30 and up taper spindles. I really wish there was more in that big gap.

I recommend you analyze it backwards. Envision the product you want to make. Route wood? Make steam engines, or IC engines? Jewelry, molds? Or aerospace stuff which needs to be +/-0.0001"? Next, you take the $$ available. $500 is not going to produce a machine which will make the aerospace stuff at 200 ipm, so without the $$ available, you may have to reject certain machines and components.

So given the $$ available, and the product you want to make, I'd start at the spindle and build the machine around that, considering the performance of the installed components. I'd not bother installing ground ballscrews if your spindle has evil runout, or the ways have slop of .005". Match the precision of the components with other, mating subsystems.

Swede

[thuffner3]

Swede,
From a purely mechanical stand point that's exactly the way one should think. But from a purely hobbist point of view. I think most guys here are shopping for what is least expensive with alot of punch.
For me, I'm carving foam. So accuracy is not so important. My ability to final mill/lathe my casting will determine the quality of my part.
My question in a sort of general way is this. Because I see the same question asked over and over, with no real ground in stone answer.
What works for you? The best performance of components, and what are they.
At least with this or these idea's in mind one might say, well I like Swede's setup better, or I like Balsaman's setup better.
See where I'm going with this. I'm really only interested in the electronics end of it. Because the mechanics can be so varied. Those question are easily enough answered by looking through some oe the photo's of the many nice machines people here have built.

Thanks for replying.
Peace
Neil

[Swede]

Hmm OK I see where you are going. I think the problem that you will find is that many guys have a core set of software/electronics, and then vary the mechanical part, swapping motors, new leadscrews, etc.

For example, I am using a flashcut + servo system. I can comment on the speed and relative accuracy of Flashcut 2.0 on a Celeron 1 gig machine. I have no idea how that would relate to any other control and software package.

The fastest that I can push this sytem is about 250ipm at a resolution of 0.0002", using 1000 line quad encoders and 5TPI screws, directly driven. By slowing it down, I can resolve to 0.00005" in the software, of course the mechanics cannot achieve that accuracy.

[thuffner3]

Thanksso much Swede. Something just like this could really help a guy out. I my case I'm using stepper motors, -w- a basic transistorized board. Unfortunately I' only getting 10 IPM using 1/4-20 screws. -w- TurboCNC. Can't for the life of me get anymore than that. I just purchased the below board to see if I could change the speed, but no such luck.

http://cgi.ebay.com/ws/eBayISAPI.dll...=STRK:MEBWN:IT

I was thinking I might change the screws to 3/8-16. That might give me a little more?

Hell I might just have to shell out the bucks for Gecko's as controllers.
I am having fun at it, it's just so close. And although very slow it still carves and engraves must faste and nicer than I ever could.

Peace
Neil

[Swede]

Hey Neil, You'll get more speed with coarser screws if the torque requirements aren't excessive. Pretty much by definition, speed and resolution are inversely proportional. I think a given controller has a certain performance baseline, and the mechainics of the system further define whether the mill or router emphasizes speed or accuracy/resolution. One thing to keep in mind, it's pointless to shoot for fine resolutions if the mill cannot achieve them. For example, without ground ballscrews, 0 backlash nuts, and a tight bearing system, you won't get or need resolutions and accuracy of 0.0002". If I've built a router and I don't need better than .002", then it would seem that I could jack the speed up, and still keep the system within the inherent accuracy of the mechanics.

I don't know much about stepper systems. There has to be some guys here who can comment at length and with more intelligence on this topic of "theoretical speed vs accuracy".

[thuffner3]

As a carpenter/ceramic tile man, my accuracy is 1/64 - 1/32. Never more. Hell with the way my eyes are behaving lately it's a wonder 1" isn't close enough :-). With my machine, and the 1/4-20 althread. I've managed to get a pretty tight system. The backlash is at worst (loss of a 1/4") over 3600". That's on a block being carved. While the Z-axis is in motion as well. so I really wont compain. And I'm finding out that loss is from in large part due to the controller board not being setup correctly.

If you don't mind my asking. What controller and servo motors are you using?
The reason I ask is, I've been looking through a new catalog from IPS (Industrial Profile Systems), It looks as though they have the parts to make a really tight machine.

I'll be back
Peace
Neil

[Swede]

I'm using Flashcut 2.0 software, which outputs through the PC's serial port into a Flashcut signal generator box, which then outputs step/direction pulses to a LOGOSOL 3-axis servo driver unit. The Logosol drivers are good for 12A continous, 20A momentary. I'm pretty sure the signal generator box is primarily a buffer for the PC's output, and keeps Window's timing issues to a minimum.

The thing I like about these Logosol drivers is that they autoconfigure for brushed or brushless servo motors. I've picked up an assortment of both motor varieties off of ebay, and with rare exception they work great, regardless of the brand. I'm also a big fan of Renco encoders, as they are a real snap to mount and perform well. Most of the motors I have use 1000 line Renco quad encoders.

I'm not really experienced with CNC in general, but I've been very pleased at the flashcut software. It digests 2D DXF files with ease. For 3D stuff, I'm using Rhino and Visual Mill. The VM has a flashcut post-processor, which has been 100% so far, no probs at all. Flashcut isn't the cheapest, but it's been a good experience so far. I think they have done a great job with the software's ability to "look ahead" and blend the toolpaths. I know there's a term for that, but I'm not sure what it is!

Swede
http://www.5bears.com