Overwhelmed, have many parts... need guidance

[sansbury]

Since this is your first build my advice is to go with something like 1/2-10 acme rod and delrin nuts. With a good flex coupling this type of system will tolerate quite a bit of slop and misalignment, which is more likely than not on a first build. No matter what I do I can't get the Y-axis on my larger machine totally straight, it's probably off by a total of between 1/16-1/32" but it has minimal backlash on that axis and can position with around .005" accuracy. If I need better I will use the machine to cut new parts. I don't know how much misalignment a ball screw will tolerate without binding but there's a good chance you will find out.

Also just to reinforce what ahren said, the real cost of ball screws is in the whole system, not just the screw itself. McMaster has 5/8" screw for about $15/foot. The nuts are about $60 each while the mounting blocks are $300! I'm sure you can find cheaper but bear in mind that with precision machine parts, precision gets exponentially more expensive. A cheap ball screw system may not in fact be more accurate than a good acme rod. Ball screws are not just used for accuracy, they're also used because they're good for moving heavy loads very quickly due to their high efficiency.

[Darren]

Thanks sansbury, I think I'll stick with ACME for my first shot as you suggest. I just need to figure out how many starts now and what reduction I'd need for each start. I'm trying to understand what others have said but I'm not quite getting it.

[sansbury]

Here's the easiest way i can think of to break it down:

Diameter: the largest OD of the screw. Thicker screws can carry more load and resist whipping better. Whipping is a factor of the unsupported length of screw, its diameter, and how fast it's turning. At 48" you need to think about whipping. 1/2-10 going 2000rpm will definitely whip without intermediate supports. There are calculators out there you can use to determine what's acceptable.

Pitch: The number of turns per inch. Steeper pitch = more distance traveled per revolution and less mechanical advantage via reduction.

Starts: These act as a multiplier of the pitch. Standard screws are single start, so a 1/2-10 1-start will move the nut 1" for each ten turns. A 2-start will go 2", a 5-start will go 5". This also affects mechanical advantage; a 2-start screw will provide half the reduction of a 1-start screw with the same thread pitch. However, it will also move twice as far (fast) for the same number of revolutions (per minute).

Here is a thread talking about this for a larger machine than yours:
http://www.cnczone.com/forums/showthread.php?t=8138

[ahren]

As I mentioned earlier in this thread, I would get multi-start ACME screw -- you can get much faster linear speeds at the same screw RPM. If you're worried about resolution, gear down before the screw (which you wanted to do anyway to get your servos spinning faster). Per a previous post, I would go 1/2-10 5 start, with a 5:1 reducer between the motor and the screw.

[Darren]

Thanks again to both of you for being patient with me.

Ahren, sounds like a great recommendation and I have it noted from before and was seriously considering going that route. I am just trying to wrap my head around all of this so I understand everything completely. I really appreciate everyone's adivce on this. It has been a long time coming and I'm finally getting somewhere.

I have a friend that can turn the ends down on the Acme thread if I need. I just need to find good ratio reducer. Anyone have a preferred supplier? I'll be checking my post and checking the websites provided to me as well earlier in the thread.

The part I'm having trouble with is this:

It was recommended to me initially by the person I bought the servo's from to use a 4:1 reduction, that was considering a single start acme. So, if I went with the 5 start as recommended for reducing whip wouldn't that mean I would require a 20:1 reduction to get the same effect? I know a 20:1 reduction isn't really practical. Maybe I'm using incorrect math.

Darren

[sansbury]

Your math is basically right, but whether it matters really depends. The question proceeds from the resolution you want the machine to provide. I would guess that .005-.01" is a very good goal to aim for, especially if you're cutting wood, which will probably grow or shrink by several times that with humidity changes.

Motor resolution is typically discussed in terms of angular position change, since the motor turns in a circle. Steppers, for instance, are specified as being able to step in units like 7.5 or 1.8 degrees per step. Servos are electromechanically different in design, but the principle of commanding the motor shaft to turn a specified number of degrees is the same.

Now, assuming you used 5-start 1/2-10 acme screws, we can do the math:
1. 10tpi / 5 starts = 2 turns per inch, or .5" per turn
2. (0.5" / .005") = 100 or 1/100th of a revolution to move .005"
3. there are 360 degrees in a circle so 1/100th of a revolution is 3.6deg of angular position change

For reference, a typical hobbyist stepper motor will step in increments of 1.8 degrees. Whether you need to use reduction on the servo is above my pay grade as I'm not familiar with how torque curves on them work, but given the power of your servos versus your application, I suspect you may be able to do a direct drive with good results. At the least, you can try it without tears since it involves all the same parts (fewer, actually) you'd need for a reduction drive. Both machines I've built have been taken apart, adjusted, reworked, and reassembled a dozen or more times prior to "real use." No matter what you do it's going to have to come apart a few times.

[Darren]

Excellent explanation. I really appreciate that. I think I can proceed now that I understand.

Thanks again!

[ahren]

Everything Sansbury posted is correct, but you also need to take into account the feedback system you are using on your servo motor, since you aren't using steppers. Your feedback system is probably a rotary encoder, so your true resolution will be determined by the number of counts on your encoder disk. 100 CPR (counts per revolution) is quite low -- it's easy to get encoders with 512 counts per revolution or more. To calculate your effective resolution, just take your pitch in inches and divide by the number of encoder counts per rev. So, for 5 start acme with 512 CPR encoders, you'd have 0.5"/512 = 0.000976", or about 0.001". Any gearing you use will make this number even smaller.

Now, you may be thinking "0.001, that's better than I need!" Well, maybe it is, but maybe it isn't. Since you've decided to take the plunge with servos, you'll have to deal with tuning them. Steppers are easy, since they have discrete positions that they either make it to, or not. Servos have a continuous motion that is regulated by the feedback loop set up in your controller. For a DC brushed motor (typical of hobby servos, but I'm not sure what you have), this basically means the controller checks the position of the encoder, and increases the voltage if the actual motion is falling behind the commanded motion, or reduces (even reverses) it if the actual motion is getting ahead.

Why is this important for you? Well, even if you tune your system perfectly, under load your motor is unlikely to be able to track exactly the commanded position you give it. It's not uncommon to be 10-20 encoder counts off, and this is considered a pretty good tuning job unless you have an advanced controller. With 0.001" resolution, a 20 count error means a 0.020" error in position. Not so great. However, with 5:1 gearing, and a 512 CPR revolution encoder, your effective resolution is now about 0.0002", so that 20 count error would only be about 0.004". Furthermore, you'll have the benefit of running your servos at higher speeds, where they are more efficient (opposite of stepper motors, that have maximum torque at lower speeds).

OK, once you digest all that, I can tell you about quadrature decoding for your encoders, which is a trick that increases your resolution by another factor of 4. But let's leave that for another post. What I've posted above should get you some more knowledge about what to do with your system, and continues to support your original 4:1 or 5:1 gearing ratio. Of course, you could also use single start ACME to lower your pitch, but we've already discussed the foibles of this approach in terms of critical rotation speed.

Ahren
www.cncrouterparts.com