-
Need help brainstorming a solution for this mill turn spindle
Im wanting to build a horizontal spindle for my mill that will double as both a lathe and a 4th axis, mostly for smaller diameter (roughly up to 2 inch max diameter) aluminum parts, turning with tools mounted to main milling head and index milling with main spindle. At this point im leaning towards a foundation of a premade dunham 5c headstock/spindle. Seems to be pretty good quality, very low runout, 5c taper is ground after mounted, 1700 bucks. The challenge is choosing a proper servo to drive this thing and come up with a solution that will give enough rpm as well as enough holding torque and resolution for 4th axis operations. These are the goals:
1-At least 2500rpm for turning, more is better, 4k would be great
2-At least 1hp for turning. Smaller aluminum parts so don't need crazy power
3-A minimum resolution of .01 degree step for 4th axis operation. 90 percent of what I'm doing doesn't require this kind of resolution, but one exception to the small aluminum parts is hobbing large helical delrin gears, about 4inch diameter so .01 degree step will be about 3 tenths on the edge of a gear. Higher resolution would be better, but I figure this is a minimum.
4-enough holding torque for indexing a mini horizontal tombstone with tail stock support, smaller tools up to 1/4inch machining out at a max radius of 2.5 inches, only light machining, rarely more than 1 cubic inch mrr, no drilling larger than 3mm.
5-budget of 3500 max including the 1700 dollar headstock /spindle
At first i was hoping to find a powerful enough servo to do all this without changing gearing, but after some research it doesn't seem too realistic. A big 1200 dollar clearpath servo will give far more than enough power, 2700rpm, and probably just enough holding torque (4800oz/in) with direct drive, but not quite the resolution i need. This would give .03 degree resolution which would be about a thou step at the edge of a 4inch gear, don't think that's gonna cut it for precision hobbing. So the main challenge in my mind is figuring out how to give this thing 2 gearing options that can be swapped fairly quickly, better yet an automated changed over. If I choose a more realistic size nema 34 dc servo, it wants to run somewhere between 1 to 1 and 2 to 1 ratio for turning, and between 5 to 1 and 10 to 1 for indexing. 10 to 1 or more would be preferred. Pulleys and belt is fine for the turning, but a 10 to 1 ratio with a belt for indexing isn't really optimal. Not sure what kind of rigidity and backlash I can count on with that. Then I was thinking maybe somehow use harmonic drive for the 4th axis reduction and belt for turning, but I would still need to place the harmonic drive next to the spindle and drive it with a 1 to 1 belt since it's a 5c spindle and harmonic drive can't couple directly to the back of it.
So, any ideas of the best way to tackle this? Would be great if there was a servo capable of what I want with only one gear ratio and within the budget, but I don't think that's realistic, so the main puzzle is coming up with how to give this thing 2 quickly swappable gearing options (automated would be great) to meet the goals.
-
Re: Need help brainstorming a solution for this mill turn spindle
An interesting dilemma. Speed, accuracy, cheap, pick any two. :D I'm going through somewhat the same thought process on my lathe which right now does not have spindle indexing capability, but does have live tooling. 7.5 KW (10HP) spindle, just to make it more interesting.
In your case, why not put the spindle motor of your choice on the spindle, any cheap motor would do for turning. Then a separate small servo motor with a timing belt drive, driving a large sprocket on the spindle. But the sprocket freewheels until you lock it to the spindle. The sprocket lock mechanism could be as simple as screwing in a bolt, pushing in a pin, or for full automation could use an automotive air conditioner clutch (those will transmit an immense amount of torque, around 10 HP worth). For additional holding power, you can always add a disk brake, motorcycle or go-cart parts.
For servos, take a look at DMM Tech products also.
-
Quote:
Originally Posted by
Jim Dawson
An interesting dilemma. Speed, accuracy, cheap, pick any two. :D I'm going through somewhat the same thought process on my lathe which right now does not have spindle indexing capability, but does have live tooling. 7.5 KW (10HP) spindle, just to make it more interesting.
In your case, why not put the spindle motor of your choice on the spindle, any cheap motor would do for turning. Then a separate small servo motor with a timing belt drive, driving a large sprocket on the spindle. But the sprocket freewheels until you lock it to the spindle. The sprocket lock mechanism could be as simple as screwing in a bolt, pushing in a pin, or for full automation could use an automotive air conditioner clutch (those will transmit an immense amount of torque, around 10 HP worth). For additional holding power, you can always add a disk brake, motorcycle or go-cart parts.
For servos, take a look at DMM Tech products also.
Not a bad idea. Probably even cheaper. To get a servo that will provide the needed power for turning, it's 600 bucks minimum. If I separate it into a servo and a bldc motor, the servo doesn't need nearly as much power for the indexing as long as it's geared down enough, more like 300 dollar servo. One hang up I have on the heavily geared down servo though, 10 to 1 with pulleys is gonna be a huge pulley and a tiny pulley, i dont picture the tiny pulley holding onto the belt with total rigidity during indexing. I can imagine there will be a tad bit of give there. Maybe not as bad as I think. I guess there's still the option of putting the servo on a 50 to 1 harmonic, then link that to the spindle with 2 decent size pulleys. Then i would have all the resolution i could ask for and still probably 100rpm after 50 to 1 gearing which is plenty snappy for a 4th.
Actually that gives me a good idea. Only 2 pulleys, one on the spindle, one next to it. The pulley to the side is supported by it's own bearings with a shaft protruding out both sides. One end of that shaft will have a link to the harmonic reduced servo, other end linked to bldc motor. Some kind of detachable coupling for both ends or clutch system. I'll take a look at the a.c. compressor clutch mechanism. I can also picture some kind of 2 in 1 actuated system. Dowel pins passing through that offset pulley. When pushed one way or the other, they will disengage one and engage the other in one single mechanism. Hmm.i think I can start playing around on cad now, see how this will lay out. Thanks for the dual motor idea, might make the most sense.
So along the lines of picking a bldc, what would you recommend? Does 2hp come in a pretty small package? How do you control it with mach3? Guess I can probably figure that out if I look. Never used them before
-
Re: Need help brainstorming a solution for this mill turn spindle
Glad I could get the creative juices flowing :)
Maybe this spindle motor would work, as far as I know it does not have indexing capability. It is compatible with Mach3.
https://www.automationtechnologiesin...or-and-driver/
Automation Tech has a number of steppers and servos available also.
-
Quote:
Originally Posted by
Jim Dawson
Glad I could get the creative juices flowing :)
Maybe this spindle motor would work, as far as I know it does not have indexing capability. It is compatible with Mach3.
https://www.automationtechnologiesin...or-and-driver/
Automation Tech has a number of steppers and servos available also.
Yep, something like that would be perfect. Looks like it takes 0-10v analog input so I assume I just need an encoder pulse on the spindle and mach3 can take it from there. That pulse could also double as a home switch for A.
I also just realized I could probably use something I already have. My main hobby that drives the machining hobby is rc helicopters. Have a few spare motors for those. They are pretty impressive little motors, smaller than your fist, capable of 3kw continuous easily, 10kw burst. Would just need a dc power supply and an analog 0-10v to pwm converter to control the esc with mach3. Kind of mind blowing how such a tiny motor could put out that kind of power. Only thing is, those escs don't like low duty cycle with heavy load so would need to be cautious of that but definitely an option
-
Re: Need help brainstorming a solution for this mill turn spindle
I'm no expert in Mach3 and breakout boards, but I think Mach3 outputs a PWM signal that is converted to 0-10V with a low pass filter on the board. So all you would have to do is bypass the filter to get a PWM output. Maybe this would work with your RC hardware.
Worst case, since you already have the parts on the shelf, is that if it doesn't work you just switch over to a known system. Might be worthwhile to see if you can control a RC motor with Mach3 on the bench.
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
Im wanting to build a horizontal spindle for my mill that will double as both a lathe and a 4th axis, mostly for smaller diameter (roughly up to 2 inch max diameter) aluminum parts, turning with tools mounted to main milling head and index milling with main spindle. At this point im leaning towards a foundation of a premade dunham 5c headstock/spindle. Seems to be pretty good quality, very low runout, 5c taper is ground after mounted, 1700 bucks. The challenge is choosing a proper servo to drive this thing and come up with a solution that will give enough rpm as well as enough holding torque and resolution for 4th axis operations. These are the goals:
1-At least 2500rpm for turning, more is better, 4k would be great
2-At least 1hp for turning. Smaller aluminum parts so don't need crazy power
3-A minimum resolution of .01 degree step for 4th axis operation. 90 percent of what I'm doing doesn't require this kind of resolution, but one exception to the small aluminum parts is hobbing large helical delrin gears, about 4inch diameter so .01 degree step will be about 3 tenths on the edge of a gear. Higher resolution would be better, but I figure this is a minimum.
4-enough holding torque for indexing a mini horizontal tombstone with tail stock support, smaller tools up to 1/4inch machining out at a max radius of 2.5 inches, only light machining, rarely more than 1 cubic inch mrr, no drilling larger than 3mm.
5-budget of 3500 max including the 1700 dollar headstock /spindle
At first i was hoping to find a powerful enough servo to do all this without changing gearing, but after some research it doesn't seem too realistic. A big 1200 dollar clearpath servo will give far more than enough power, 2700rpm, and probably just enough holding torque (4800oz/in) with direct drive, but not quite the resolution i need. This would give .03 degree resolution which would be about a thou step at the edge of a 4inch gear, don't think that's gonna cut it for precision hobbing. So the main challenge in my mind is figuring out how to give this thing 2 gearing options that can be swapped fairly quickly, better yet an automated changed over. If I choose a more realistic size nema 34 dc servo, it wants to run somewhere between 1 to 1 and 2 to 1 ratio for turning, and between 5 to 1 and 10 to 1 for indexing. 10 to 1 or more would be preferred. Pulleys and belt is fine for the turning, but a 10 to 1 ratio with a belt for indexing isn't really optimal. Not sure what kind of rigidity and backlash I can count on with that. Then I was thinking maybe somehow use harmonic drive for the 4th axis reduction and belt for turning, but I would still need to place the harmonic drive next to the spindle and drive it with a 1 to 1 belt since it's a 5c spindle and harmonic drive can't couple directly to the back of it.
So, any ideas of the best way to tackle this? Would be great if there was a servo capable of what I want with only one gear ratio and within the budget, but I don't think that's realistic, so the main puzzle is coming up with how to give this thing 2 quickly swappable gearing options (automated would be great) to meet the goals.
This is one you may get some Ideas from WebPageMain, he uses a servo motor with a direct drive ratio and a backgear for low speed
-
Re: Need help brainstorming a solution for this mill turn spindle
Not sure of the power etc, but I'm waiting on a mill turn machine at the moment that uses hefty hobby king sized BLDC motors running off an o-drive board. I'm not sure whether I'm going to have to change the A spindle out for something similar to what you're looking for but the designer of the machine reckons (and his videos seem to show) that the BLDC off the O-drive gets the job done nicely.
https://www.kickstarter.com/projects...ur-machine-sho
-
4 Attachment(s)
Re: Need help brainstorming a solution for this mill turn spindle
Cheap 5C spindle harvested from a spin indexer. Deep groove bearings in an aluminum housing. Bearing seats were line bored in place with the 4th axis.
Taper has essentially no runout measured with a 0.00005 B&S indicator. Threaded the outboard end for a clamping lock nut with two holes to mount a GT2 pulley.
Using it as a tailstock for my 4th axis, but could be powered with a servo down the road.
Way, way cheaper than the spindle you're considering.
https://www.cnczone.com/forums/attac...d=415058&stc=1https://www.cnczone.com/forums/attac...d=415054&stc=1https://www.cnczone.com/forums/attac...d=415056&stc=1https://www.cnczone.com/forums/attac...d=415060&stc=1
-
Re: Need help brainstorming a solution for this mill turn spindle
BTW, go check out the InTurn from Cube Studio. Lots of Youtube videos, and the early ones show some pretty basic stuff like I built.
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
So, any ideas of the best way to tackle this? Would be great if there was a servo capable of what I want with only one gear ratio and within the budget, but I don't think that's realistic, so the main puzzle is coming up with how to give this thing 2 quickly swappable gearing options (automated would be great) to meet the goals.
One example: https://www.instagram.com/p/BsCHjVMoCQv/
Another vote for DMM. The 1.8kw 120-DST-A6HK1 with a DYN-4 and cables will be around $700. Resolution under .0001 degrees and a 24VDC brake.
-
Quote:
Originally Posted by
footpetaljones
That servo looks very impressive. Much more compact than the clearpath I was looking at, also much cheaper. When you say a resolution of .0001 degrees, are you referring to the machine in the Instagram post, with some kind of gearing? I'm having a hard time finding what resolution the motor is without gearing. Says 16bit encoder, but I don't know what that means, totally new to servos. I would be running this with mach3 and ethernet smooth stepper.
-
Re: Need help brainstorming a solution for this mill turn spindle
16 bit = 2^16 or 65536 positions per revolution, about 0.005º resolution.
-
Re: Need help brainstorming a solution for this mill turn spindle
(Anyone else feel free to chime in if I get something wrong here)
If you're rconsidering ESS/Mach/DMM, then the limiting factor for angular resolution is speed vs. pulse output to the servo drive.
Lets say you have a 1:1 motor to spindle pulley ratio. If you want to spin the spindle at max motor RPM (3000 for the DMM 1.8kw), then you have to figure out how many pulses per second that is at direct resolution in the servo.
The servo has ~65k encoder positions per rev. That means you need to send 65k * 3000RPM /60 (sec/minute) = 3.2Mhz (millions). The ESS is theoretically capable of outputting that number of pulses per second, but I'm not sure Mach3 or your computer is. And the DMM DYN4 drive is limited to 500khz input.
Enter electronic gearing.
The servo drive is capable of moving the motor various number of encoder counts per input pulse. You set the gearing to "4" (or whatever) and 1 input pulse (from Mach/ESS) results in a motor move of 4 encoder counts. At the hypothetical "4" electronic gearing you will need 3.2Mhz /4 = 812khz pulses to move the motor at 3kRPM. Still too fase for the DYN4 drive.
Change that to 8:1 gearing and you're down to 406khz(ish): In the ballpark for the DMM maximum pulse input speed.
The tradeoff with gearing is that you lose resolution. At 1:1 electronic gearing, one pulse will move the spindle 0.005 degrees (based on dharmic's post above), or 0.00026" at the outer edge of a 6" circle (your chuck or whatever). At 8:1 gearing, you get .002" of movement per input pulse at 3000RPM with a direct drive motor to spindle.
Bringing us to the motor to spindle pulley ratio. You'll want, as you discussed, two gear ratios - one for indexing and low-end torque, and another for turning small stuff at high speed. If we set the two ratios at 3:1 (high) and 12:1 (low) you'll have a top speed of 1000RPM and 250RPM.
Assuming the 8:1 electronic gear ratio, at 1000RPM you'll have three times the resolution as the 1:1 ratio. So about 0.0007" per pulse. At low ratio, you'll be back down to around 0.0002" per pulse.
And that's at the outside of a 6" diameter part. Divide by 3 for your intended 2" (and smaller part) and you're down in the 0.00006" range per pulse output.
This is just the math, of course. At that fine a resolution everything is flexible - the belts, the collet, the part, the temp, the weather. You aren't going to be able to repeatably index in the millionths of an inch, but you should get the point.
Even if you change the pulley ratios to get 1500 or 2000RPM on the spindle for small stuff, with a 'backgear' you'll still have a mechanical (theoretical) resolution much finer than I expect you need from a home-brew dingus.
To sum up... The DMM/DYN4 combo will get you the speed you'd like for turning small aluminum parts, as well as very fine angular resolution for cutting splines and other critical features (rotationally speaking) - assuming you set up the pulley ratios appropriately.
Again, check out the Cube Studio stuff. He has a brilliant double-reduction back gear + direct drive setup. Frankly, if you were considering $1700 for a spindle by itself just BUY one of his Mega 4th axis packages. Even comes with a control system that looks amazingly slick.
-
Quote:
Originally Posted by
dharmic
16 bit = 2^16 or 65536 positions per revolution, about 0.005º resolution.
Wow! That is impressive! I could gear the thing 1 to 1 and have plenty of accuracy. What's with the 24v brake? Can't find much detail about that either. If it allows for a really locked position, it kind of sounds perfect for my application.
So .005 resolution, that's 72000 steps per rev, whuch is 3.6mhz at 3k rpm. smoothstepper can be set up to 4mhz. Looks like I'm right at the limit for using full resolution up to 3k rpm, unless I'm not understanding that fully. Really curious about this servo now and need to learn more about it.
-
Quote:
Originally Posted by
spumco
(Anyone else feel free to chime in if I get something wrong here)
If you're rconsidering ESS/Mach/DMM, then the limiting factor for angular resolution is speed vs. pulse output to the servo drive.
Lets say you have a 1:1 motor to spindle pulley ratio. If you want to spin the spindle at max motor RPM (3000 for the DMM 1.8kw), then you have to figure out how many pulses per second that is at direct resolution in the servo.
The servo has ~65k encoder positions per rev. That means you need to send 65k * 3000RPM /60 (sec/minute) = 3.2Mhz (millions). The ESS is theoretically capable of outputting that number of pulses per second, but I'm not sure Mach3 or your computer is. And the DMM DYN4 drive is limited to 500khz input.
Enter electronic gearing.
The servo drive is capable of moving the motor various number of encoder counts per input pulse. You set the gearing to "4" (or whatever) and 1 input pulse (from Mach/ESS) results in a motor move of 4 encoder counts. At the hypothetical "4" electronic gearing you will need 3.2Mhz /4 = 812khz pulses to move the motor at 3kRPM. Still too fase for the DYN4 drive.
Change that to 8:1 gearing and you're down to 406khz(ish): In the ballpark for the DMM maximum pulse input speed.
The tradeoff with gearing is that you lose resolution. At 1:1 electronic gearing, one pulse will move the spindle 0.005 degrees (based on dharmic's post above), or 0.00026" at the outer edge of a 6" circle (your chuck or whatever). At 8:1 gearing, you get .002" of movement per input pulse at 3000RPM with a direct drive motor to spindle.
Bringing us to the motor to spindle pulley ratio. You'll want, as you discussed, two gear ratios - one for indexing and low-end torque, and another for turning small stuff at high speed. If we set the two ratios at 3:1 (high) and 12:1 (low) you'll have a top speed of 1000RPM and 250RPM.
Assuming the 8:1 electronic gear ratio, at 1000RPM you'll have three times the resolution as the 1:1 ratio. So about 0.0007" per pulse. At low ratio, you'll be back down to around 0.0002" per pulse.
And that's at the outside of a 6" diameter part. Divide by 3 for your intended 2" (and smaller part) and you're down in the 0.00006" range per pulse output.
This is just the math, of course. At that fine a resolution everything is flexible - the belts, the collet, the part, the temp, the weather. You aren't going to be able to repeatably index in the millionths of an inch, but you should get the point.
Even if you change the pulley ratios to get 1500 or 2000RPM on the spindle for small stuff, with a 'backgear' you'll still have a mechanical (theoretical) resolution much finer than I expect you need from a home-brew dingus.
To sum up... The DMM/DYN4 combo will get you the speed you'd like for turning small aluminum parts, as well as very fine angular resolution for cutting splines and other critical features (rotationally speaking) - assuming you set up the pulley ratios appropriately.
Again, check out the Cube Studio stuff. He has a brilliant double-reduction back gear + direct drive setup. Frankly, if you were considering $1700 for a spindle by itself just BUY one of his Mega 4th axis packages. Even comes with a control system that looks amazingly slick.
Ok, so Im definitely not gonna be running this thing up to 3k rpm at full resolution. Too much for the driver to handle. I'm really curious about the 24vdc brake. Not sure what exactly that does, but it might still be possible for me to just keep this thing at a fixed ratio for most work which would really be nice, but I guess I would at least have to switch the electronic gearing back and forth from high resolution at low speed indexing, to lower resolution at high rpm turning. I guess it's probably not any more hassle to leave electronic gearing alone and have different actual gear ratios to really get optimal torque and speed for turning/indexing. I'm really curious how the 2 stage gearing was handled on the inturn. Need to look into that more.
-
Re: Need help brainstorming a solution for this mill turn spindle
The inturn mega looks like a great machine, but it's still a pretty large price gap even if I do build this around the dunham spindle. The mega is 3 grand and that's no servo included, more like 3800 when you're done. I think I could do this for about 2600 using the dunham spindle. It won't have the disk brake like the inturn, but I don't think I'll need it for index work on 2 inch aluminum parts as long as I have a decent servo and gear down for the index work. Using a 5c indexer spindle and building my own headstock is an option, but I don't think I would get as good of results. For one I don't have capability to line bore the whole assembly with the equipment I have. The dunham spindle is really pretty sweet. Nice solid cast headstock, spindle is assembled into headstock, bearings preloaded, then spindle taper is ground. They guarantee runout less than half a tenth. Also it has a 4 degree taper nose to allow use of hardinge closer rings for big step chucks.
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
spumco
The tradeoff with gearing is that you lose resolution. At 1:1 electronic gearing, one pulse will move the spindle 0.005 degrees (based on dharmic's post above), or 0.00026" at the outer edge of a 6" circle (your chuck or whatever). At 8:1 gearing, you get .002" of movement per input pulse at 3000RPM with a direct drive motor to spindle.
You don't loose any resolution using electronic Gearing, the Encoder resolution does not change
DMM also have a 32Bit Encoder if someone wants an even better positioning resolution
-
Quote:
Originally Posted by
mactec54
You don't loose any resolution using electronic Gearing, the Encoder resolution does not change
DMM also have a 32Bit Encoder if someone wants an even better positioning resolution
You don't lose accuracy of position, but you lose steps per unit, each step it receives from smoothstepper will be multiplied by gearing. Right? Only way to make higher rpm manageable by the drivers frequency limitation
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
Wow! That is impressive! I could gear the thing 1 to 1 and have plenty of accuracy. What's with the 24v brake? Can't find much detail about that either. If it allows for a really locked position, it kind of sounds perfect for my application.
So .005 resolution, that's 72000 steps per rev, whuch is 3.6mhz at 3k rpm. smoothstepper can be set up to 4mhz. Looks like I'm right at the limit for using full resolution up to 3k rpm, unless I'm not understanding that fully. Really curious about this servo now and need to learn more about it.
Your Step / Per you can control with Electronic Gearing like if you used DMM and set 500 for the Electronic Gear number it would only be 2,000 Steps / Per rev
The Brake on a Servo Motor is not suitable for positioning, they are only used for when you power off the Servo Motor, if you want to activate a brake then it has to be independent of the Servo motor Brake
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
You don't lose accuracy of position, but you lose steps per unit, each step it receives from smoothstepper will be multiplied by gearing. Right? Only way to make higher rpm manageable by the drivers frequency limitation
You don't loose anything, you are calculating what Steps are needed Per Rev
-
Quote:
Originally Posted by
mactec54
You don't loose anything, you are calculating what Steps are needed Per Rev
by increasing electronic gearing and reducing steps per unit, you are losing resolution, correct? Specifically i mean the minimum amount of rotation you can input. If you double electronic gearing and half steps per unit, your minimum amount of movement doubles
In other words, if I want to command the servo to rotate 0.005 degrees, I can't if there's any electronic gearing used, right?
-
1 Attachment(s)
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
by increasing electronic gearing and reducing steps per unit, you are losing resolution, correct? Specifically i mean the minimum amount of rotation you can input. If you double electronic gearing and half steps per unit, your minimum amount of movement doubles
You have to calculate the correct number of Step / Per Rev just like setting up any new control or you won't get the correct rotation or positioning
You don't loose any Resolution no matter what way you want to do it, the Encoder Resolution does not change only how you drive it
If you want to command the servo to rotate 0.005 degrees it will do it or any position you want to move it
-
Quote:
Originally Posted by
mactec54
You have to calculate the correct number of Step / Per Rev just like setting up any new control or you won't get the correct rotation or positioning
You don't loose any Resolution no matter what way you want to do it, the Encoder Resolution does not change only how you drive it
If you want to command the servo to rotate 0.005 degrees it will do it or any position you want to move it
I'm still not getting it. I understand you need to set correct number of steps per rev based on electronic gearing, how does this not reduce resolution? Let's say with no electronic gearing it's 72,000 steps per rev. With this a single step coming from mach3 will rotate the servo .005 degrees. If I set the servo to 10-1 electronic gearing, steps per unit goes down to 7,200 steps per rev. At this setting a single step from mach3 will rotate the servo .05 degrees and I am no longer able to move in smaller increments than that, resolution has been reduced 10 fold. What am I missing?
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
I'm still not getting it. I understand you need to set correct number of steps per rev based on electronic gearing, how does this not reduce resolution? Let's say with no electronic gearing it's 72,000 steps per rev. With this a single step coming from mach3 will rotate the servo .005 degrees. If I set the servo to 10-1 electronic gearing, steps per unit goes down to 7,200 steps per rev. At this setting a single step from mach3 will rotate the servo .05 degrees and I am no longer able to move in smaller increments than that, resolution has been reduced 10 fold. What am I missing?
I don't know what you are missing, your electronic gearing is just a ratio so lets say you did have 7,200 step/per rev setting which is highly unlikely. The Encoder16,384 has not changed only how you are driving it ,so if you commanded .0001 it is going to move .0001 or any number you want it to move
Here is a video for you https://www.youtube.com/watch?v=fyKB...Uj6f4Q&index=9
-
Quote:
Originally Posted by
mactec54
I don't know what you are missing, your electronic gearing is just a ratio so lets say you did have 7,200 step/per rev setting which is highly unlikely. The Encoder16,384 has not changed only how you are driving it ,so if you commanded .0001 it is going to move .0001 or any number you want it to move.
Here is a video for you
https://www.youtube.com/watch?v=fyKB...Uj6f4Q&index=9
It still doesn't make sense to me. Mach3 can not send a partial step to the servo drive. If you have 7200 steps per rev, how is it possible for mach3 to tell the servo drive to rotate the servo less than .05 degrees? It can only send full steps. Even with no electric gearing, an actual move of .0001 degrees at the servo isn't possible. A single encoder pulse is around .005 degrees.
Watched the video, isn't really relevant to resolution vs electrical gearing.
Edit: I got mixed up on that last part, a move of .0001 rev is possible for the servo, but not .0001 degrees, but still not a possible move with 7200 steps per rev coming from mach3. The part that doesn't make sense to me is how mach3 can send less than 1 full step
"If you want to command the servo to rotate 0.005 degrees it will do it or any position you want to move it"
This is only possible with at least 72000 steps per rev in mach3, which would be one single step. If electronic gearing is changed and steps per rev reduced, how can you command that small of a move?
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
This is only possible with at least 72000 steps per rev in mach3, which would be one single step. If electronic gearing is changed and steps per rev reduced, how can you command that small of a move?
No it does not work like that, how do you think all controls work you have a handwheel that can move your machine with different resolutions, Mach3 is the same way and can move in .0001" increments the readout in Mach3 has 4 decimal places 0.0000 there would be not point if it can't do that
The video is just one of many that shows accurate positioning with inertia
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
The inturn mega looks like a great machine, but it's still a pretty large price gap even if I do build this around the dunham spindle. The mega is 3 grand and that's no servo included, more like 3800 when you're done. I think I could do this for about 2600 using the dunham spindle. It won't have the disk brake like the inturn, but I don't think I'll need it for index work on 2 inch aluminum parts as long as I have a decent servo and gear down for the index work. Using a 5c indexer spindle and building my own headstock is an option, but I don't think I would get as good of results. For one I don't have capability to line bore the whole assembly with the equipment I have. The dunham spindle is really pretty sweet. Nice solid cast headstock, spindle is assembled into headstock, bearings preloaded, then spindle taper is ground. They guarantee runout less than half a tenth. Also it has a 4 degree taper nose to allow use of hardinge closer rings for big step chucks.
Seriously, for an extra grand you get:
Heavier duty than you are likely to build
All controls sorted out and able to fully integrate in to your machine
Spindle brake - very important for milling operations or drilling off-axis
Tech support and soem sort of warranty
5C collet and D1-4 spindle nose built-in - dime a dozen chucks will fit
Your machine won't notice half a tenth of spindle runout - the spindle belt vibration will induce that much or more
Spindle taper ground in place
And you'll have it working and making parts much sooner than if you build this thing yourself, even with a preassembled cartridge spindle.
A grand for all that? Holy smokes. The only reason I built my own is that I could do it for an order of magnatude less - $400 for mine vs. $3800, not $2600 vs. $3800.
And you can bore your own housings. Make the frame, pre-bore the bearing seats with an end mill. Then clamp the assembled frame to an angle block and use a boring head to put the final bearing pockets right where you want them. If they don't come out exact, shim the components until it's dead on. Like clamp a 1" ground rod in a v-block and indicate it parallel with the X axis and perpendicular with the Z. Then lock the spindle to the bar with a 1" 5C collet and use epoxy to shim the components while the spindle is held in alignment. Epoxy sets up, tighten screws, and you're all set.
For the total DIY with the spindexer route, you're looking at about $900 all in, including a 1kw servo. That 1.8kw is likely overkill if you're sticking with a fixed pulley ratio. $600 for servo, $60 for bearings, $50 for the spindexer, Some bearings and figure another $100 for belts and pulleys.
-
Re: Need help brainstorming a solution for this mill turn spindle
I don't know what type of motor he's using in the instagram post.
By <.0001 degrees, I was referring to the DMM servo. Servo encoders are usually classified by 2 to the power of "XX" steps per revolution, with "XX" being the 16 in 16 bits, or 17 bits, or whatever. So a 16 bit encoder is 2^16=65536. 1/65536=0.0000152 degrees per encoder pulse.
-
Re: Need help brainstorming a solution for this mill turn spindle
1/65536 revolutions? per encoder pulse. Multiply by 360 to get degrees per encoder pulse.
-
Re: Need help brainstorming a solution for this mill turn spindle
Doh. Apparently I can't math this week.
.005 degrees per pulse.
-
Quote:
Originally Posted by
mactec54
No it does not work like that, how do you think all controls work you have a handwheel that can move your machine with different resolutions, Mach3 is the same way and can move in .0001" increments the readout in Mach3 has 4 decimal places 0.0000 there would be not point if it can't do that
The video is just one of many that shows accurate positioning with inertia
I understand that mach3 has 4 decimal places on the dro, it's irrelevant, it can actually use up to 12 decimal places (maybe it's 10), 4 is just what's shown on the dro, that doesn't mean it can send partial steps to a servo drive. The limitation for resolution is based on steps per unit. The minimal output it can give (actual movement) is limited by the distance of a single step. In most machines, running in inch mode, moving 0.0001 is not usually a problem. On my machine for example, I have 5mm pitch screws and 2000 micro step, comes out to about a tenth of linear axis movement so moving .0001 in inch is fine, however I do most of my work in metric, so mm for units. In this case a single step is .0025mm (400 steps per unit in mm) and I assure you, mach3 does not allow a smaller move than that since I only have 400 steps per mm, it would if i had finer resolution motors and higher step per unit. I can take a video to demonstrate if you like. If I set the step jog to .0001mm, I have to give 25 clicks of the handwheel before mach3 will actually send a microstep to move the motor. Obviously servos can have finer resolution than steppers, but a minimum move is still limited by a single step. I'm sure any machine on servos and most on steppers has a fine enough step to move .0001 inches, but we are talking much smaller moves, .005 "degree" at the servo shaft, not .0001" (inch linear) on a linear axis which is a MUCH larger move. Mach3 will not output a partial step, if 1 step is larger than .005 "degrees" at the servo shaft, mach3 will not do it. You would need 72000 steps per rev minimum (revolution, not inch) to command the servo to make that small of a move. Im not sure where we are getting mixed up, i think it's because you are thinking linear inches, in which case .0001 inch moves is not usually limited by a single step. I'm talking much smaller moves.
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
I understand that mach3 has 4 decimal places on the dro, it's irrelevant, it can actually use up to 12 decimal places (maybe it's 10), 4 is just what's shown on the dro, that doesn't mean it can send partial steps to a servo drive. The limitation for resolution is based on steps per unit. The minimal output it can give (actual movement) is limited by the distance of a single step. In most machines, running in inch mode, moving 0.0001 is not usually a problem. On my machine for example, I have 5mm pitch screws and 2000 micro step, comes out to about a tenth of linear axis movement so moving .0001 in inch is fine, however I do most of my work in metric, so mm for units. In this case a single step is .0025mm (400 steps per unit in mm) and I assure you, mach3 does not allow a smaller move than that since I only have 400 steps per mm, it would if i had finer resolution motors and higher step per unit. I can take a video to demonstrate if you like. If I set the step jog to .0001mm, I have to give 25 clicks of the handwheel before mach3 will actually send a microstep to move the motor. Obviously servos can have finer resolution than steppers, but a minimum move is still limited by a single step. I'm sure any machine on servos and most on steppers has a fine enough step to move .0001 inches, but we are talking much smaller moves, .005 "degree" at the servo shaft, not .0001" (inch linear) on a linear axis which is a MUCH larger move. Mach3 will not output a partial step, if 1 step is larger than .005 "degrees" at the servo shaft, mach3 will not do it. You would need 72000 steps per rev minimum (revolution, not inch) to command the servo to make that small of a move. Im not sure where we are getting mixed up, i think it's because you are thinking linear inches, in which case .0001 inch moves is not usually limited by a single step. I'm talking much smaller moves.
Are you comparing Steppers to Servos there is no comparison between the 2 systems ( 1 ) click of the handwheel would mean the servo would move what ever that resolution was set at inch or metric makes not difference
Even it something was to move .005 degrees you would not have anything you could verify the measurement with, the mechanic's that you posted for your build would be the deciding factor not what you are driving it with
I have a Megatorque motor that I use for a 4th axes it has a 20Bit Encoder which is over 1million CPR and it can move and hold any position you can dream up, these motors only do 500 RPM though
-
Quote:
Originally Posted by
mactec54
Are you comparing Steppers to Servos there is no comparison between the 2 systems ( 1 ) click of the handwheel would mean the servo would move what ever that resolution was set at inch or metric makes not difference
Even it something was to move .005 degrees you would not have anything you could verify the measurement with, the mechanic's that you posted for your build would be the deciding factor not what you are driving it with
I have a Megatorque motor that I use for a 4th axes it has a 20Bit Encoder which is over 1million CPR and it can move and hold any position you can dream up, these motors only do 500 RPM though
I think you are missing my point. Answer me this, mach3 can only communicate with the servo drive with full step pulses, true or false? I think we can agree that's true right?
We are talking specifically about a 16 bit encoder and a servo drive capable of a max frequency of 500khz. This means it will not be capable of BOTH 3000rpm and .005 degree steps(single encoder pulse) at the servo shaft. For the servo drive to handle 3000 rpm at less than 500khz input, it needs electronic gearing of roughly 10 to 1 to handle that. This means a single step coming from mach3 is going to rotate the servo 10 encoder pulses. The resolution will in fact be reduced by a factor of 10 and a single step move becomes .05 degrees. Mach3 can not output a tenth of a step to tell the drive to rotate a single encoder pulse. Which part of this are you saying is incorrect? It is not dependent on the mechanics of the machine or mechanical gearing. I am talking purely resolution of rotation at the servo shaft vs electronic gearing and steps per revolution
-
Re: Need help brainstorming a solution for this mill turn spindle
Mactec: although the servo is capable of infinitely small resolution and the driver is capable of crazy small resolution to the point where the limit is usually the encoder resolution, when you're driving with Mach3 AFAIK you're limited to a step pulse. Which can be configured to be anything, .000001º if you like, but then the limit becomes how fast Mach3 can reliably pulse those step signals which determines top speed. If you set up the electronic gearing in the servo loop to increase the speed, you're saying "when Mach3 sends 1 pulse, move n steps". Which means every move is a multiple of n steps in that arrangement, regardless of the resolution of the components.
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
dharmic
Mactec: although the servo is capable of infinitely small resolution and the driver is capable of crazy small resolution to the point where the limit is usually the encoder resolution, when you're driving with Mach3 AFAIK you're limited to a step pulse. Which can be configured to be anything, .000001º if you like, but then the limit becomes how fast Mach3 can reliably pulse those step signals which determines top speed. If you set up the electronic gearing in the servo loop to increase the speed, you're saying "when Mach3 sends 1 pulse, move n steps". Which means every move is a multiple of n steps in that arrangement, regardless of the resolution of the components.
Most people are not using Mach3 with a PP these days, the servo loop is closed in the servo drive the control can be anything you want it to be in most cases Mach3 is doing very little when you use a external CPU that sends the commands to the Servo Drive, Mach3 today does very little when connected to an external CPU
The whole idea of having electronic gearing is so the supply CPU does not have to be maxed out to match any encoder resolution
For those that don't know a SmoothStepper, there default I is 256 KHz this is more than enough to run any servo motor at max RPM this also depends on the voltage of the Servo Drive as to the max RPM
The lowest electronic gear setting for a Dmm servo drive is 500 Gear number this equals to 2000 Steps/Per Rev using this setting the motor has no trouble reaching it's max RPM for any of there 120v/240v Servo Drives motor combinations
The UC100 will run this setup at max RPM also with this 16Bit encoder resolution
-
Quote:
Originally Posted by
mactec54
Most people are not using Mach3 with a PP these days, the servo loop is closed in the servo drive the control can be anything you want it to be in most cases Mach3 is doing very little when you use a external CPU that sends the commands to the Servo Drive, Mach3 today does very little when connected to an external CPU
The whole idea of having electronic gearing is so the supply CPU does not have to be maxed out to match any encoder resolution
For those that don't know a SmoothStepper, there default I is 256 KHz this is more than enough to run any servo motor at max RPM this also depends on the voltage of the Servo Drive as to the max RPM
The lowest electronic gear setting for a Dmm servo drive is 500 Gear number this equals to 2000 Steps/Per Rev using this setting the motor has no trouble reaching it's max RPM for any of there 120v/240v Servo Drives motor combinations
The UC100 will run this setup at max RPM also with this 16Bit encoder resolution
Using a parrallel port has nothing to do with it. The issue in question remains the same. In your case of 2000 steps per rev, the smallest move mach3 can command to the servo is 0.18 degrees, whether your using a PP, uc100, ess, or whatever. Mach3 is still communicating with that device using FULL step pulses, one step at a time. In your example, it's gonna send 2000 step pulses for each rev. It doesn't chop them up into something smaller. For the type of accuracy I would want on my 4th, that's far too low of a resolution, even if I have 4 to 1 mechanical gearing, that's still .045 degrees per step which is too much, that would be about a thou and a half step on the circumference of a 4 inch gear.
-
Re: Need help brainstorming a solution for this mill turn spindle
Quote:
Originally Posted by
QuinnSjoblom
Using a parrallel port has nothing to do with it. The issue in question remains the same. In your case of 2000 steps per rev, the smallest move mach3 can command to the servo is 0.18 degrees, whether your using a PP, uc100, ess, or whatever. Mach3 is still communicating with that device using FULL step pulses, one step at a time. In your example, it's gonna send 2000 step pulses for each rev. It doesn't chop them up into something smaller. For the type of accuracy I would want on my 4th, that's far too low of a resolution, even if I have 4 to 1 mechanical gearing, that's still .045 degrees per step which is too much, that would be about a thou and a half step on the circumference of a 4 inch gear.
So then why are you fixated with Mach3 if it is not going to do what you want
I guess you are never going to build this, and are just wasting everyone's time, as there is nothing out there that you could buy or put together that is going to satisfy your needs, you seem to have all the answers but can't figure it out how to do it
There is nothing new about a 4th axes, manufacturers have been doing this for years and with even more accuracy than what you are trying to do, my 20Bit 4th Axes are what dreams are made of, 1,048,576 PPR or another one I have at 23Bit 8,388,608 PPR
You want a better resolution without spending much money, then start with a harmonic drive than you can add a stepper motor and you would be done just as may hobby users do
As I said you won't even be able to build the mechanical part accurate enough to do what you want and you are talking about step/per what about +/-10v or digital using fiber optics
-
Quote:
Originally Posted by
mactec54
So then why are you fixated with Mach3 if it is not going to do what you want
I guess you are never going to build this, and are just wasting everyone's time, as there is nothing out there that you could buy or put together that is going to satisfy your needs, you seem to have all the answers but can't figure it out how to do it
There is nothing new about a 4th axes, manufacturers have been doing this for years and with even more accuracy than what you are trying to do, my 20Bit 4th Axes are what dreams are made of, 1,048,576 PPR or another one I have at 23Bit 8,388,608 PPR
You want a better resolution without spending much money, then start with a harmonic drive than you can add a stepper motor and you would be done just as may hobby users do
As I said you won't even be able to build the mechanical part accurate enough to do what you want and you are talking about step/per what about +/-10v or digital using fiber optics
Not sure where that came from. When did I say mach3 was not going to do what i want? This back and forth debate about resolution started when someone else and myself stated that resolution is lost when electric gearing is increased and you replied with "you dont lose anything", which is incorrect. Resolution is lost proportionally to the increase in electronic gearing. Has nothing to do with a shortcoming of mach3, it's the same with any control software communicating with step/dir. It simply means a balance needs to be chosen between speed and resolution along with other possibilities like multiple gear ratios. This whole time ive just been trying to understand what I'm missing since you insist resolution isn't lost with increase in electrical gearing, but I don't think I'm the one missing something.
Building a 4th axis with high resolution is not the problem. If that's what I was trying to do, I would give it 1 to 1 electronic gearing and have 5 times the resolution i could ever need and mach3 would have no problem doing it. The challenge is building a mill turn spindle which is what this thread is about. Finding a balance of accuracy, holding torque, and max rpm. The options for achieving this were being discussed until it went off the rails starting with your comment of "you don't lose anything". Not sure what the attack is about. Just trying to get the facts straight. You started off saying you don't lose resolution by increasing electronic gearing and now your switching to it doesn't matter since you cant build something that accurate anyway.
-
Re: Need help brainstorming a solution for this mill turn spindle
Lol, mactec's interpersonal skills shine upon us once again. QuinnSjoblom: I don't think he even means it as a personal attack, forget it and move on. There's nothing to be gained by continuing down that dead end.
Still, whilst we're on the Mach3 hatewagon... Will it even support a mill/turn? Looking at the config section, it doesn't look like it supports multiple independent spindles. Or have I missed something? So from what I can see the best you could hope for is you'd need to switch back and forth between C and A spindles using its SwapAxis() function and have either the mill spindle or the lathe spindle running and the other one stepping as a 4th axis. Seems a PITA and is probably why the machine I'm looking at is using GRBL instead (may get an upgrade to LinuxCNC). Just a thought.