CNC'ing The Knee....

[SCzEngrgGroup]

I've recently completed a CNC conversion on my BP clone, and it works great. I took the quick and lazy approach to the Z axis, by replacing the knee crank with a servo motor and belt reducer, leaving the bevel gears and leadscrew in place for now (to be replaced with a ballscrew at some point). A friend has an identical machine, so I made him a set of drives and other parts identical to mine, which he has just fitted. But, his machine apparently has a higher torque requirement on the knee than mine, and as a result, his knee does not work reliably. Mine is perfectly reliable, albeit not real fast, at only 25 IPM.

So, we're looking for options as to how to resolve the problem. Some things that have come up are getting a larger servo motor (currently 850 oz-in, with 4.8:1 reduction), getting a big stepper motor (2550 oz-in), adding gas springs, going ahead and converting to ballscrew, and eliminating the bevel gears, to reduce friction.

So, some questions for anyone who has experience with powered knees:

1) How have others done this? What is the most common drive configuration for powering the knee? What kind of motor? What reduction?

2) What kind of gas spring is appropriate? I bought a pair of 200 pound springs from McMaster (not yet mounted), and am surprised to find I cannot make them budge, even putting my full weight on them. Makes me wonder if they're TOO stiff.

3) Can a large stepper be used here? My impression is probably not, as given the high reduction (48:1), 1250 RPM would be required to even reach 25IPM. That seems high for a stepper, especially a large one.

4) Any other ideas or suggestions? Pictures/info of how others have done the knee?

Regards,
Ray L.

[Al_The_Man]

I have a couple of Excello's that power the Knee but it is used for initial positioning only, the quill is CNC, the system uses a Ball screw Jack, similar to the ones Nook can supply, these offer worm to ball screw activation, and in the case of the knee, any backlash is usually not a problem as the weight of the table etc, keeps a downward pressure.
The right angled shaft that goes through the jack has a servo motor on one end and a brake on the other.
The whole knee is counterbalanced by use of two pneumatic cyclinders, one either side of the knee and is controlled by a constant pneumatic pressure regulator for balance.
This system could also be used to CNC the knee if so wished, I prefer the CNC quill though.
Al.

[SCzEngrgGroup]

Al,

Any idea how much up-force the pneumatic cylinders provide? What pressire does it run, and what is the cylinder (actually piston) diameter? Where does one find a constant pressure regulator?

I'm thinking a ballscrew arrangement could be done on the knee quite easily by replacing the leadscrew with a ballscrew anchored to the knee, and replacing the nut with a tapered roller bearing with a pulley and rotating ballnut.

[TOTALLYRC]

I've recently completed a CNC conversion on my BP clone, and it works great. I took the quick and lazy approach to the Z axis, by replacing the knee crank with a servo motor and belt reducer, leaving the bevel gears and leadscrew in place for now (to be replaced with a ballscrew at some point). A friend has an identical machine, so I made him a set of drives and other parts identical to mine, which he has just fitted. But, his machine apparently has a higher torque requirement on the knee than mine, and as a result, his knee does not work reliably. Mine is perfectly reliable, albeit not real fast, at only 25 IPM.

So, we're looking for options as to how to resolve the problem. Some things that have come up are getting a larger servo motor (currently 850 oz-in, with 4.8:1 reduction), getting a big stepper motor (2550 oz-in), adding gas springs, going ahead and converting to ballscrew, and eliminating the bevel gears, to reduce friction.

So, some questions for anyone who has experience with powered knees:

1) How have others done this? What is the most common drive configuration for powering the knee? What kind of motor? What reduction?

2) What kind of gas spring is appropriate? I bought a pair of 200 pound springs from McMaster (not yet mounted), and am surprised to find I cannot make them budge, even putting my full weight on them. Makes me wonder if they're TOO stiff.

3) Can a large stepper be used here? My impression is probably not, as given the high reduction (48:1), 1250 RPM would be required to even reach 25IPM. That seems high for a stepper, especially a large one.

4) Any other ideas or suggestions? Pictures/info of how others have done the knee?

Regards,
Ray L.

Both of my bport machines take 12 pounds on an 8" (.66') handle to raise the knee. Thats 8 ft/lbs = 96 inch/lbs = 1536 in/oz input to the knee shaft.
They have a total reduction of 10 turns to the inch based on the dial.


your 2550ozin stepper is going to lose torque as soon as you start to spin it and while it will move the knee, I am not sure if it will be fast enough for the z. although a g203v will give you the best speed when it morphs.

I would send Gecko an email and ask them what they suggest.

[SCzEngrgGroup]

Interesting. My knee requires about half that torque - I measured about 45-50 oz-in. My friends is a bit higher, but not a lot. Keep in mind, we have an additional 4.8:1 belt reduction as well, so I don't think torque on the stepper will become a limit until we get up some speed, but I'm not convinced we'd be able to get enough speed for a reasonable rapid before the stepper saturated.

I would expect eliminating the bevel gears and going to a ballscrew would reduce the torque significantly.

Regards,
Ray L.

[Al_The_Man]

Al,

Any idea how much up-force the pneumatic cylinders provide? What pressire does it run, and what is the cylinder (actually piston) diameter? Where does one find a constant pressure regulator?

I'm thinking a ballscrew arrangement could be done on the knee quite easily by replacing the leadscrew with a ballscrew anchored to the knee, and replacing the nut with a tapered roller bearing with a pulley and rotating ballnut.

The diameter of the single ended cylinders is 2.5", the regulator pressure gauge reads 60lb. when set.
The Ball screw jack is a nice answer as it has ball screw and reduction all in one.
But remember, with the weight of the knee, table, and a heavy work piece, there is quite a large inertia required to move it all at an appreciable acceleration speed.
Break away torque is quite low, but inertia comes into play as soon as you try any rapid moves.
There have been many past posts on this very subject over the years and a search should bring a few up, including a source of the pneumatic regulator.
Al.

[TOTALLYRC]

Interesting. My knee requires about half that torque - I measured about 45-50 oz-in. My friends is a bit higher, but not a lot. Keep in mind, we have an additional 4.8:1 belt reduction as well, so I don't think torque on the stepper will become a limit until we get up some speed, but I'm not convinced we'd be able to get enough speed for a reasonable rapid before the stepper saturated.

I would expect eliminating the bevel gears and going to a ballscrew would reduce the torque significantly.

Regards,
Ray L.

I was assuming direct drive of the knee shaft with the stepper instead of the handle. No 4.8 reduction.

[SCzEngrgGroup]

I was assuming direct drive of the knee shaft with the stepper instead of the handle. No 4.8 reduction.

BTW - I mis-spoke - The measured torque on the knee handle was 45-50 in-lbs, not oz-in. So, about 800 oz-in.

Regards,
Ray L.

[SCzEngrgGroup]

Could I be barking up the wrong tree here? I just got done installing a pair fo 200 pound air springs to my knee. I had expected that would reduce the load enough that I could perhaps get at least a little more rapid speed out of it. But, it did absolutely *nothing* to improve rapids. That got me to thinking (usually a dangerous situation....):

I have 10:1 total reduction on X & Y, and can achieve only a little more than 100 IPM rapids. I have 48:1 total reduction on Z, and can achieve only about 25 IPM rapids. In both cases, the motors are turning only about 1000 RPM. The motors are rated for 4200 RPM, and I am running the, at rated voltage and current. My encoders are 500 lines/2000PPR, and I'm running Gecko G320s, which will fault with a 128 count following error. For Z, 128 counts is only 0.0013 inches!

Would I benefit from lower line-count encoders? Or, perhaps, LESS reduction? On X and Y, I don't seem to have any problem whatsoever with torque - I've snapped off 1/2" endmills without losing position....

Regards,
Ray L.

[polaraligned]

Would I benefit from lower line-count encoders? Or, perhaps, LESS reduction? On X and Y, I don't seem to have any problem whatsoever with torque - I've snapped off 1/2" endmills without losing position....

Regards,
Ray L.

Hey Ray,
I remember when you were asking about what to use originally back in spring or thereabouts on the BP list. I recommended 200 line encoders and 2:1 reduction ratio when using Gecko drives. This is a combination that has performed well for people. The SS may allow use of higher resolution encoders though because of the clean output from it. I also recommended using 28 in-lb continuous torque motors.
As for your knee, I know a guy that used a 40 in-lb continuous servo on the knee with a 2:1 reduction and he uses it in a commercial setting. He has drilled holes in 3/4" plate all day with that setup. Oh yeah, he is driving the 40 in-lb servo with a Gecko. Sorry, I don't know what his rapids are but I will bet they are scary.
Let us know what resolves the problem.

Good luck,
Scott

[SCzEngrgGroup]

Hey Ray,
I remember when you were asking about what to use originally back in spring or thereabouts on the BP list. I recommended 200 line encoders and 2:1 reduction ratio when using Gecko drives. This is a combination that has performed well for people. The SS may allow use of higher resolution encoders though because of the clean output from it. I also recommended using 28 in-lb continuous torque motors.
As for your knee, I know a guy that used a 40 in-lb continuous servo on the knee with a 2:1 reduction and he uses it in a commercial setting. He has drilled holes in 3/4" plate all day with that setup. Oh yeah, he is driving the 40 in-lb servo with a Gecko. Sorry, I don't know what his rapids are but I will bet they are scary.
Let us know what resolves the problem.

Good luck,
Scott

Scott,

Wow! You have a good memory! Back then I was not planning on doing the knee. I actually made that decision spur of the moment about 6 weeks ago, and did the whole thing in one day. I'm actually no unhappy with how mine works. But, my friend with an identical machine is in a pickle, since his doesn't work like mine. So, I'm trying to sort out his problem from cross-country. I think we're going to try an encoder swap.
My motors are 850 oz-in/53 in-lbs and with 2.5:1 reduction they would barely move at all - faulted almost immediately.

Regards,
Ray L.

[TOTALLYRC]

Could I be barking up the wrong tree here? I just got done installing a pair fo 200 pound air springs to my knee. I had expected that would reduce the load enough that I could perhaps get at least a little more rapid speed out of it. But, it did absolutely *nothing* to improve rapids. That got me to thinking (usually a dangerous situation....):

I have 10:1 total reduction on X & Y, and can achieve only a little more than 100 IPM rapids. I have 48:1 total reduction on Z, and can achieve only about 25 IPM rapids. In both cases, the motors are turning only about 1000 RPM. The motors are rated for 4200 RPM, and I am running the, at rated voltage and current. My encoders are 500 lines/2000PPR, and I'm running Gecko G320s, which will fault with a 128 count following error. For Z, 128 counts is only 0.0013 inches!

Would I benefit from lower line-count encoders? Or, perhaps, LESS reduction? On X and Y, I don't seem to have any problem whatsoever with torque - I've snapped off 1/2" endmills without losing position....

Regards,
Ray L.

I would try lowering you acceleration in mach3 down to 6 or 8 and see if you can rapid faster. The 128 steps of error is really noticable when the control say go from 0 to full speed in a very short time. The electronics respond instanly but the motor, being mechanical, can only respon so fast.
The 4200 rpm rating will be at less than max as a motor makes its max power at ~half of its no load rpm.
I had a similar problem and then I decreased my accel and was able to gain about 25 ipm, even with my amps being misadjuasted.

1000 rpm is 16.6/sec x 2000 counts = 33,200 counts/sec.= 33.3khz
2000 rpm is 33.2/sec x 2000 counts = 66,400 counts/sec.= 66.4 khz
What is your kernal set to?
If your computer will go faster, turn it up or use the step multiplier board to convert to g340's or a lower count encoder.
20,000counts/in is a resolution of 1/2 of a tenth or .00005. if you went to 250 encoders or half of what you are currently running, your resolution would "drop" to .0001. and the machine theroeticaly would go twice as fast, if it is a step generation speed problem, all else being equal.

I have the same gearing and encoder counts on my mill, but the dspmc board has a 2MHZ limit= 2000khz. 600ipm max possible if my math is right. but my motors and stuff are rated for 300ipm but i have it set for 125ipm and right know as I am learning it is fast enough for me. I will tune the drives and try for the 300ipm in the future, just to see it go that fast.

Scott,

Wow! You have a good memory! Back then I was not planning on doing the knee. I actually made that decision spur of the moment about 6 weeks ago, and did the whole thing in one day. I'm actually no unhappy with how mine works. But, my friend with an identical machine is in a pickle, since his doesn't work like mine. So, I'm trying to sort out his problem from cross-country. I think we're going to try an encoder swap.
My motors are 850 oz-in/53 in-lbs and with 2.5:1 reduction they would barely move at all - faulted almost immediately.

Regards,
Ray L.

Is the 850oz-in continuous or peak. Peak is only good for 1 second or so.
On you friends machine, did you try loosing the gibs and add extra lube to see if is drag related.

Mike