Pm25-mv Conversion Planning

Charlton, you're not hijacking at all. The more input the better, as this post was mostly to bat around ideas for the conversion of this mill.

Yes, it is the newer one, purchased this summer. The primary differences are the belt drive, brushless motor and a larger base to column mount (column bolts down to base vs G0704/Weiss bolting to the side). I still need to do some more research on exact differences, but do believe that Brav is right about different brackets being required.. I also thought that the spindle bearings were upgraded as well but not sure on this one. I'll worry about power after the main conversion is up and running.

I am thinking of going with a kit for ballscrews and mounts, mostly due to 2 factors.
1) My south bend 9A Lathe can't fit the ball screws through the headstock to machine the ends of the most commonly used size for the Z (20mm), and I don't own a stready rest, nor do I understand how I'd hold it and machine the ends.
2) Due to the lack of documentation on converting the newer PM25s.

There is a guy on another forum that was going to put a kit together, so I may go this route. I also may try to figure out the Z dimension and order it machined for me or pay a local guy and machine the rest myself using the Hoss plans as a rough guide... Just not sure yet. I'll pass on info for a kit if I can get more info on it.
 
Would that person be ArizonaVideo (Dave)? I PMed him last night to get some feedback from him about the mill and to see if he had plans to put together a kit. I know there were some kits about the TM25's having holes that seemed to be drilled rather randomly (possibly with a hand-drill) though this also seemed to be the case with some of the Weiss machines as well...this would presumably make kits harder to make. Perhaps we can suggest a kit that's machined but that doesn't have the mounting holes drilled.

Cheers,
Charlton
 
Charlton, Yep. Same guy. There is a post on there showing an installation of one of them in the benchtop section.. I wanted to email him to get some more info. The brackets look real nice, and probably way better than I could do now without a DRO on my machine. I emailed him once and he confirmed he had a kit, but haven't yet followed back up. His Kit was exactly for the PM version per his words in the one email we've exchanged. I didn't specify because I wasn't sure of the details yet.
 
My take on ball screw drive, I use a 3:1 drive ration since my servos spin 5000rpm max, don't need to spin them ball screws that fast. It also multiplies the torque by 3, it also increases the resolution per step into the millionths of an inch. I could have used MUCH less powerful servos to get the job done, mine are powerful enough that it will easily break stuff if asked to lol, my motor drive combo is good for 1hp, 3hp at the screw, and around 12lb-ft of torque at the screw. All that is nice, but belt drives are not backlash free, very close with the gt2 profile. I have yet to see backlash from my belt system. Also belts wear out, I put a good amount of tension on mine. If you go direct drive you need to make sure tell motors you choose are going to be more powerful than you think you need, you risk losing steps with steppers being underpowered.
 
I can shed a bit of light on some of the stuff on this thread.
I had a PM25-mv, it is a great little mill - a very impressive machine for it price point and size. Sadly, the column in mine was tapered making it impossible to raise the head when the gibs where properly adjusted at milling height. Matt ordered me a new column, then offered me to exchange or even upgrade it so I would not have to wait. I opted to buy the PM450G which has about the same size table, but is an entirely different class of machine. So the first major difference between Grizzly and QMT, is that as good as Grizzly customer service is, Matt will probably always do better and with less trouble.

On the subject of Weiss, there really is no such animal. I also thought some of these machine where simply Weiss machines that are rebranded, when I asked Matt about it last year, the response was surprising. It turns out that Weiss is just a handful of guys in an office. They just buy machines from Chinese factories and resell them just the same as Northern Tools does. Apparently there is nothing special about the brand beyond what ever custom options that ask for.

The PM25-mv has ABEC5 roller bearings in the spindle. They are pretty massive for the size of the machine and will handle pretty high RPM. The quill though is not of the same quality. You will lose a great deal of the accuracy of the bearings under the heavy pressure that CNC produces because the quill is not tight. I would work out locking the quill before I worried about the bearings.

The y-axis trunnion on the PM25 is very narrow. This means that the mill's work envelope is really effectively about 7" x 7". When you get to about 5" off center, the table sags and the either you lose z-axis accuracy if the gibs are loose, or the table becomes hard to move if they are tight. This is not really a big deal, since this is really a small mill and meant for small parts. You will want to consider counter weighting the table if you are concerned about table tilt on wider parts.

My understanding about Arduino is they lack the bandwidth required for good milling. Not only do they not have enough head room for closed loop operation, they are slow. Better to stick with one of the chips from Texas Instruments. TI has a really amazing line of chips that are designed solely to make motion control cheap and simple. I am talking stuff in the <$10.00 range for a chip. For a bit more you can get a full fledged motion control chip with a full library on board that will do far more than you can imagine.

My own opinion (that will probably get shot down), is don't waste your time and money on an open loop system. I do not for a moment buy the argument that a well designed and build open loop system does not loose steps. This is an engineers fantasy in my opinion. The cost of closed loop systems has come down so much in the last few years that there is simply no reason to not use closed loop.
Check out these folks, I know many people like them. They certainly have good prices and they do respond decently well to emails.
http://www.automationtechnologiesinc.com/
 
Ok, so you've officially put a bug in my ear about the closed loop. I had it in my head already, but its worth thrashing the details around.

So, "closed" loop seems to be a bit of a gray area with many ways of going about it.. I know some systems are fully and redundantly closed. What I think this means is the servo systems have both encoders at the motor, and also glass or other scale systems employed for absolute position feedback. I don't think I need to go that crazy for this mill conversion, but am interested in adding some closed loop, or what I think may be towards that direction.

So, in a simple closed loop I think I would need a Servo or Stepper with an encoder at the motor, and a different driver.

Automation Technologies has a couple options. What looks like a Servo with Encoder and also a stepper with an encoder. So, if I were interested in adding a closed loop capability, what else do I need to be thinking?
 
IMHO, the only way to go is put the encoder on the load (table) rather than the motor/lead screw. What you want to position accurately is the load, not the lead screw. This automatically compensates for some backlash and any inaccuracy's in the lead screw. Linear glass or magnetic scales work very well as the feedback device. Personally, I prefer the Renishaw 1 micron magnetic scales, these give you a theoretical 0.000039 position accuracy.

It is possible to close the loop with a stepper system, but a DC or AC servo system is the best.
 
I 2nd Jims idea. I would like to do this on my Mill at some point at least on X & Y to start. If you use the Renishaw (or Equiv per a post I did here) you will still need a controller that can accept and use the feedback for positional control. If you find a Mach3/Stepper setup that can do this I'd be very interested. Here are two I investigated as full kits. .... http://machmotion.com/ http://www.ajaxcnc.com/ Hope you keep us up to date on what you find / buy!
 
Countryguy - Check out Mesa Cards. Its not Mach/Windows but rather Linux CNC. Which I am planning on going with. They have what seems to be a large number of different FPGA IO cards along with Daughter cards that can take servo encoder inputs. At least from what I have looked into so far. I think I can go either stepper or Servo with encoder feedback using these card combos.

I'm still digesting all of this so, will probably just post here and there as more questions come up. I am intrigued by adding the full closed loop. Just don't want to break the bank. Its is just a PM25 after all. Trying to blend the cost/performance ratios and see where it all comes out. At some point accuracy is going to be maxes out by the mill's capabilities itself. One day I would love love to have a true VMC to retrofit. But for now, I just want to dip my toes into it with this little mill.
 
I have an Ajax DM45nc mill which is about 10 years old. It was factory built by Ajax using the Centroid CNC10 controller and software. It is a brush type DC Servo system based on Linux CNC. One problem they had with machines was vibration causing read errors on the hard drive when the PC was mounted to the column of the mill, so they used a Compact Flash card as the hard drive. Just mention that in regards to how you package the electronics. I love the DC servo system. The servos are belt drive to the ball screws, about a 2 to one reduction (just guessing). 2000 line encoders are what they recommend for the feed - and the encoder is on the motor shaft. I actually have a G0704 mill I'd like to convert as well, so I'm thinking about the same things. I'm a PC guy so the Linux CNC is a bit intimidating to me, but I do like the way it operates in the Ajax form. I just looked at the Automation Technology website and they have some very nice looking prices for DC servo motors. The Gecko G320 drive lets you use it just like a stepper motor. I need to look into this to see how it tells the controller the drive didn't move properly, but it kinda looks like there is an error output that may take care of that. Looks like the encoder feeds back to the gecko drive. More research needed on my part. It looks like a Nema 23 drive might be big enough for the x and y axis with one of the nema 34 for the Z axis. Prices look like about $200 per axis and up. Not sure how a hybrid stepper works, but with an ecoder on a stepper, it would seem possible to get some odd feedback unless the encoder was closely matched to step size, so I'd vote for Servo. Just my thoughts on this.
 
You must log in or register to reply here.
Back
Top