Home |
Search |
Today's Posts |
|
Metalworking (rec.crafts.metalworking) Discuss various aspects of working with metal, such as machining, welding, metal joining, screwing, casting, hardening/tempering, blacksmithing/forging, spinning and hammer work, sheet metal work. |
Reply |
|
LinkBack | Thread Tools | Display Modes |
#1
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
After the long discussion about one spindle following another and
using a Gecko 320 servo drive I have the first test results. The servo motor has a 500 line encoder mounted on it and the Gecko Drive reads it in quadrature for a resolution of 2000 counts per revolution. I'm using another of the same model encoder mounted to another motor powered by a PWM power supply. This motor is the master and the motor controlled by the Gecko is the slave. The master encoder is connected to an LS7184 quadrature clock converter. The LS7184 accepts the encoder signals and outputs step and direction signals which are fed to the Gecko 320. As the speed of the master motor is varied the slave follows PERFECTLY! There is no perceptible (to me) lag or overshoot. And when spinning the master encoder at approximately 1 rpm the slave follows perfectly with NO cogging. When the slave is commanded to move only one step it does so without jumping as would a stepper when operated with just switches or by just touching the wires to a power source. Obviously the Gecko ramps up to speed to make movement as smooth as possible. In fact, even though the Gecko is fed only discreet step by step motion signals and position feedback is only in steps it acts like the whole system is analog, when the motor is the only analog device in the system. It looks like I can now grab a part with two chucks at the same time without the worry of scuffing the part. Eric |
#2
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
consider reading up on Bode plots and Nyquest Stability Criteria - then take
some calibrated data via a frequency sweep from .01 hz to 100 hz and see what your mechanism is really doing wrote in message ... After the long discussion about one spindle following another and using a Gecko 320 servo drive I have the first test results. The servo motor has a 500 line encoder mounted on it and the Gecko Drive reads it in quadrature for a resolution of 2000 counts per revolution. I'm using another of the same model encoder mounted to another motor powered by a PWM power supply. This motor is the master and the motor controlled by the Gecko is the slave. The master encoder is connected to an LS7184 quadrature clock converter. The LS7184 accepts the encoder signals and outputs step and direction signals which are fed to the Gecko 320. As the speed of the master motor is varied the slave follows PERFECTLY! There is no perceptible (to me) lag or overshoot. And when spinning the master encoder at approximately 1 rpm the slave follows perfectly with NO cogging. When the slave is commanded to move only one step it does so without jumping as would a stepper when operated with just switches or by just touching the wires to a power source. Obviously the Gecko ramps up to speed to make movement as smooth as possible. In fact, even though the Gecko is fed only discreet step by step motion signals and position feedback is only in steps it acts like the whole system is analog, when the motor is the only analog device in the system. It looks like I can now grab a part with two chucks at the same time without the worry of scuffing the part. Eric ** Posted from http://www.teranews.com ** |
#3
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Fri, 5 Sep 2008 23:39:44 -0700, "William Noble"
wrote: consider reading up on Bode plots and Nyquest Stability Criteria - then take some calibrated data via a frequency sweep from .01 hz to 100 hz and see what your mechanism is really doing I don't have the equipment to do that. Maybe not even the mental equipment. Instead the next test will involve clamping and unclamping a cylinder mounted rigidly to one spindle. I haven't yet figured out how to do that though. But if the collet is closed slowly around the spinning cylinder, which will be painted with Dykem, it should be obvious from the marks if there is relative movement between the two spindles. It won't take much movement to damage cosmetically fussy parts. But lots of times I run jobs where I put the parts into the rotating collet and the part is not sctatched. The part is free of course to accelerate to the spindle's speed and not held rigidly at some speed that's supposed to be the same as the spindle. Eric wrote in message ... After the long discussion about one spindle following another and using a Gecko 320 servo drive I have the first test results. The servo motor has a 500 line encoder mounted on it and the Gecko Drive reads it in quadrature for a resolution of 2000 counts per revolution. I'm using another of the same model encoder mounted to another motor powered by a PWM power supply. This motor is the master and the motor controlled by the Gecko is the slave. The master encoder is connected to an LS7184 quadrature clock converter. The LS7184 accepts the encoder signals and outputs step and direction signals which are fed to the Gecko 320. As the speed of the master motor is varied the slave follows PERFECTLY! There is no perceptible (to me) lag or overshoot. And when spinning the master encoder at approximately 1 rpm the slave follows perfectly with NO cogging. When the slave is commanded to move only one step it does so without jumping as would a stepper when operated with just switches or by just touching the wires to a power source. Obviously the Gecko ramps up to speed to make movement as smooth as possible. In fact, even though the Gecko is fed only discreet step by step motion signals and position feedback is only in steps it acts like the whole system is analog, when the motor is the only analog device in the system. It looks like I can now grab a part with two chucks at the same time without the worry of scuffing the part. Eric ** Posted from http://www.teranews.com ** |
#4
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sep 6, 11:03 am, wrote:
I don't have the equipment to do that. Maybe not even the mental equipment. Instead the next test will involve clamping and unclamping a cylinder mounted rigidly to one spindle. I haven't yet figured out how to do that though. But if the collet is closed slowly around the spinning cylinder, which will be painted with Dykem, it should be obvious from the marks if there is relative movement between the two spindles. It won't take much movement to damage cosmetically fussy parts. But lots of times I run jobs where I put the parts into the rotating collet and the part is not sctatched. The part is free of course to accelerate to the spindle's speed and not held rigidly at some speed that's supposed to be the same as the spindle. Eric Another test would be to clamp a glass tube in both chucks. Dan |
#5
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
|
#6
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
Bob Engelhardt wrote:
wrote: Another test would be to clamp a glass tube in both chucks. I'd clamp a piece of rubber tubing in both & see if it winds up. Bob or a piece of stock in *each* (2 pieces) almost touching in the middle. Make a scratch across the "joint" & see if it stays aligned. -- Nota for President |
#7
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sat, 06 Sep 2008 15:50:30 -0400, Bob Engelhardt
wrote: Bob Engelhardt wrote: wrote: Another test would be to clamp a glass tube in both chucks. I'd clamp a piece of rubber tubing in both & see if it winds up. Bob or a piece of stock in *each* (2 pieces) almost touching in the middle. Make a scratch across the "joint" & see if it stays aligned. -- Nota for President Greetings Bob and Dan, The glass tube idea will only work if the tube is weak enough to break instead of forcing the spindles to keep in step. I think I have some glass tube so maybe I'll try that. As to the rubber tubing and the scratch I already know the spindles stay in line because of marks made on discs at the end of each spindle. It's cool to spin one spindle shaft by hand and watch the other keep up. It looks like the spindles are one shaft. The discs are right next to each other with only about ..020" between the faces. So one motor spins cw while the other spins ccw. So if the master spindle is unpowered and turned by hand the other one keeps up perfectly. It's actually kind of creepy watching it because I know there is no mechanical connection between the two. Eric |
#8
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
|
#9
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sep 6, 3:46*pm, Bob Engelhardt wrote:
wrote: * Another test would be to clamp a glass tube in both chucks. I'd clamp a piece of rubber tubing in both & see if it winds up. Bob I'd make two flywheels, the master thin and the slave thick, mark reference lines on them and look for the conditions where the slave starts to slip. |
#11
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sep 6, 1:51 pm, wrote:
Greetings Bob and Dan, The glass tube idea will only work if the tube is weak enough to break instead of forcing the spindles to keep in step. I think I have some glass tube so maybe I'll try that. Eric If the glass tube forces the spindles to keep in step, would not whatever you are using this for also force the spindles to keep in step. Dan |
#12
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sun, 07 Sep 2008 06:33:27 -0400, Wes wrote:
wrote: So if the master spindle is unpowered and turned by hand the other one keeps up perfectly. It's actually kind of creepy watching it because I know there is no mechanical connection between the two. Eric What is the ultimate purpose of your effort? I seemed to have missed the original thread. Wes So that with a part spinning in one spindle the second spindle can grab it without scratching it. Then the lathe can turnthe back side of the part. ERS |
#13
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sun, 07 Sep 2008 09:32:25 -0500, Don Foreman
wrote: On Sat, 06 Sep 2008 20:51:41 GMT, wrote: On Sat, 06 Sep 2008 15:50:30 -0400, Bob Engelhardt wrote: Bob Engelhardt wrote: wrote: Another test would be to clamp a glass tube in both chucks. I'd clamp a piece of rubber tubing in both & see if it winds up. Bob or a piece of stock in *each* (2 pieces) almost touching in the middle. Make a scratch across the "joint" & see if it stays aligned. -- Nota for President Greetings Bob and Dan, The glass tube idea will only work if the tube is weak enough to break instead of forcing the spindles to keep in step. I think I have some glass tube so maybe I'll try that. As to the rubber tubing and the scratch I already know the spindles stay in line because of marks made on discs at the end of each spindle. It's cool to spin one spindle shaft by hand and watch the other keep up. It looks like the spindles are one shaft. The discs are right next to each other with only about .020" between the faces. So one motor spins cw while the other spins ccw. So if the master spindle is unpowered and turned by hand the other one keeps up perfectly. It's actually kind of creepy watching it because I know there is no mechanical connection between the two. Eric There *must* be some relative torsional displacement because this is a feedback control system depending upon an error signal. It could be zero error at times of constant speed if the control system is PI (proportional-integral) but there must be some error during speed changes. Incorporation of torque sensing could minimize this: sense torque supplied by both drive motor and slave motor and have the control system operate to minimize the difference. You sorta have that now: torque difference is proportional to the difference in encoder signals and the "twist" in the workpiece. If the slave spindle need not provide any significant torque, then it may just follow the load like a live center. In that case, the only loads would be dynamic during accelerations and decelerations and the load would drive the slave motor rather than vice versa. In that case, if the torsional distortion in the workpiece is less than one encoder count from drive spindle to slave motor there'd be no excitation provided (or necessary) to the slave motor. Greetings Don, The two spindles will be at the same speed. What I need is to make sure that whatever speed the master spindle is turning at the slave will match. If the part swap from the master spindle to the slave spindle happens fast enough the part won't be scratched. So I really don't care if the slave lags one encoder count as long as it lags consistantly. And it appears that is happening right now. I will need to take some pictures with a fast shutter to see if the marks always line up. Eric |
#14
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
Greetings Don,
The two spindles will be at the same speed. What I need is to make sure that whatever speed the master spindle is turning at the slave will match. If the part swap from the master spindle to the slave spindle happens fast enough the part won't be scratched. So I really don't care if the slave lags one encoder count as long as it lags consistantly. And it appears that is happening right now. I will need to take some pictures with a fast shutter to see if the marks always line up. Eric Eric - what you need to consider is dynamics, not just statics - it is not clear from your limited requiements statement what kind of dynamic tracking is required. I will absolutely guarantee however that under dynamic conditions the two will not track. The amount of error depends on the specific servo loop and its response time (and power) - whether the error is a problem depends on your needs. I had suggested collecting the standard frequency response data so you could evaluate this mathamatically. If you'd rather not do that, then try this - mount an L shaped item (such as a heavy allen wrench) in the main chuck. set up a a steel block so the L will hit the block in about 1/2 revolution from "zero". Set the thing to zero, fire off your high speed camera and hit the L as hard as you can with a hammer (to make the main chuck rotate) - reveiw the high speed camera data to see how good tracking is. ** Posted from http://www.teranews.com ** |
#15
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
|
#16
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sep 7, 3:46*pm, wrote:
I will need to take some pictures with a fast shutter to see if the marks always line up. Eric A strobe light would probably be more useful to you. Make the two flywheels like I suggested. They can be thick and thin squares of steel plate balanced on a lawnmower blade balancer, nothing complicated or difficult. Press them onto shafts and put the thin one in the master chuck as a reference and the thick one in the slave to make it have to work hard to keep up. Line up the corners, then accelerate the master rapidly and slow it gradually. At or above some acceleration the slave will start to slip and the corners will separate. You could also watch the slave motor current as you twist it out of line with the master. Possibly the slave will hold lock until the current reaches a limit. If so, you could add a meter to see if the current nears that limit in operation. Then operate it below the maximum acceleration or current limit. |
#17
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sun, 7 Sep 2008 14:39:53 -0700, "William Noble"
wrote: Greetings Don, The two spindles will be at the same speed. What I need is to make sure that whatever speed the master spindle is turning at the slave will match. If the part swap from the master spindle to the slave spindle happens fast enough the part won't be scratched. So I really don't care if the slave lags one encoder count as long as it lags consistantly. And it appears that is happening right now. I will need to take some pictures with a fast shutter to see if the marks always line up. Eric Eric - what you need to consider is dynamics, not just statics - it is not clear from your limited requiements statement what kind of dynamic tracking is required. I will absolutely guarantee however that under dynamic conditions the two will not track. The amount of error depends on the specific servo loop and its response time (and power) - whether the error is a problem depends on your needs. I had suggested collecting the standard frequency response data so you could evaluate this mathamatically. If you'd rather not do that, then try this - mount an L shaped item (such as a heavy allen wrench) in the main chuck. set up a a steel block so the L will hit the block in about 1/2 revolution from "zero". Set the thing to zero, fire off your high speed camera and hit the L as hard as you can with a hammer (to make the main chuck rotate) - reveiw the high speed camera data to see how good tracking is. ** Posted from http://www.teranews.com ** Greetings William, The goal is to drive two spindle mounted chucks at the same speed so after a part is machined on one end in the master chuck the slave chuck can grab it. After the part is clamped in the slave chuck the master chuck releases it. Then the rest of the work can be done on the other end of the part. So the two chucks must be spinning at pretty close to the same speed when the swap occurs. So if the master is spinning at a certain fairly steady rate and if the slave can then spin at the same rate I'll be OK. I realize that the slave will always lag a little as far as angular position goes. But as long as the lag is unchanging then the two spindles can match speeds. Eric |
#18
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Mon, 08 Sep 2008 19:02:11 -0400, Wes wrote:
wrote: So that with a part spinning in one spindle the second spindle can grab it without scratching it. Then the lathe can turnthe back side of the part. Is this high production? Going slow or stopping would seem to solve the problem. I'm fairly our Mazak Multiplex's and Mori ZT2500Y's are using C axis indexing and are stopped during the transfer. Wes Greetings Wes, If I go to the trouble of putting an extra spindle on the lathe I want to change parts while the chuck is spinning. It's amazing how only a few seconds per part on a long run can add up. So If I can get the spindles to match speed without too much trouble I wanna do it. And not only that either. I just plain wanna figure it out. Cheers, Eric |
#19
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Mon, 8 Sep 2008 16:50:13 -0700 (PDT), Jim Wilkins
wrote: On Sep 7, 3:46=A0pm, wrote: I will need to take some pictures with a fast shutter to see if the marks always line up. Eric A strobe light would probably be more useful to you. Make the two flywheels like I suggested. They can be thick and thin squares of steel plate balanced on a lawnmower blade balancer, nothing complicated or difficult. Press them onto shafts and put the thin one in the master chuck as a reference and the thick one in the slave to make it have to work hard to keep up. Line up the corners, then accelerate the master rapidly and slow it gradually. At or above some acceleration the slave will start to slip and the corners will separate. You could also watch the slave motor current as you twist it out of line with the master. Possibly the slave will hold lock until the current reaches a limit. If so, you could add a meter to see if the current nears that limit in operation. Then operate it below the maximum acceleration or current limit. Greetings Jim, I already am using index marks to watch for slipping. And for sure a strobe is the way to go. Now I'm going to have to make one. Time for a google search for a simple circuit for one. One advantage I have is that the two spindles will not be loaded any more than friction and windage loads. They only need to match speeds during part changes. Cheers, Eric |
#20
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
In article ,
wrote: On Sun, 7 Sep 2008 14:39:53 -0700, "William Noble" wrote: Greetings Don, The two spindles will be at the same speed. What I need is to make sure that whatever speed the master spindle is turning at the slave will match. If the part swap from the master spindle to the slave spindle happens fast enough the part won't be scratched. So I really don't care if the slave lags one encoder count as long as it lags consistantly. And it appears that is happening right now. I will need to take some pictures with a fast shutter to see if the marks always line up. Eric Eric - what you need to consider is dynamics, not just statics - it is not clear from your limited requiements statement what kind of dynamic tracking is required. I will absolutely guarantee however that under dynamic conditions the two will not track. The amount of error depends on the specific servo loop and its response time (and power) - whether the error is a problem depends on your needs. I had suggested collecting the standard frequency response data so you could evaluate this mathamatically. If you'd rather not do that, then try this - mount an L shaped item (such as a heavy allen wrench) in the main chuck. set up a a steel block so the L will hit the block in about 1/2 revolution from "zero". Set the thing to zero, fire off your high speed camera and hit the L as hard as you can with a hammer (to make the main chuck rotate) - reveiw the high speed camera data to see how good tracking is. ** Posted from http://www.teranews.com ** Greetings William, The goal is to drive two spindle mounted chucks at the same speed so after a part is machined on one end in the master chuck the slave chuck can grab it. After the part is clamped in the slave chuck the master chuck releases it. Then the rest of the work can be done on the other end of the part. So the two chucks must be spinning at pretty close to the same speed when the swap occurs. So if the master is spinning at a certain fairly steady rate and if the slave can then spin at the same rate I'll be OK. I realize that the slave will always lag a little as far as angular position goes. But as long as the lag is unchanging then the two spindles can match speeds. Eric That's really neat and elegant. Hope it works! -j |
#21
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
You could watch both encoders on an oscilloscope. You might not catch
a single slip but you would clearly see multiple continuous ones. |
#22
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Wed, 10 Sep 2008 03:38:55 -0700 (PDT), Jim Wilkins
wrote: You could watch both encoders on an oscilloscope. You might not catch a single slip but you would clearly see multiple continuous ones. I already have digital readouts connected to each encoder. So I know how many counts each one is producing. The counts match exactly after everything stops. The numbers change way too fast to see how close they are during rotation. Eric |
#23
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Wed, 10 Sep 2008 00:02:07 -0400, JoanD'arcRoast
wrote: In article , wrote: On Sun, 7 Sep 2008 14:39:53 -0700, "William Noble" wrote: Greetings Don, The two spindles will be at the same speed. What I need is to make sure that whatever speed the master spindle is turning at the slave will match. If the part swap from the master spindle to the slave spindle happens fast enough the part won't be scratched. So I really don't care if the slave lags one encoder count as long as it lags consistantly. And it appears that is happening right now. I will need to take some pictures with a fast shutter to see if the marks always line up. Eric Eric - what you need to consider is dynamics, not just statics - it is not clear from your limited requiements statement what kind of dynamic tracking is required. I will absolutely guarantee however that under dynamic conditions the two will not track. The amount of error depends on the specific servo loop and its response time (and power) - whether the error is a problem depends on your needs. I had suggested collecting the standard frequency response data so you could evaluate this mathamatically. If you'd rather not do that, then try this - mount an L shaped item (such as a heavy allen wrench) in the main chuck. set up a a steel block so the L will hit the block in about 1/2 revolution from "zero". Set the thing to zero, fire off your high speed camera and hit the L as hard as you can with a hammer (to make the main chuck rotate) - reveiw the high speed camera data to see how good tracking is. ** Posted from http://www.teranews.com ** Greetings William, The goal is to drive two spindle mounted chucks at the same speed so after a part is machined on one end in the master chuck the slave chuck can grab it. After the part is clamped in the slave chuck the master chuck releases it. Then the rest of the work can be done on the other end of the part. So the two chucks must be spinning at pretty close to the same speed when the swap occurs. So if the master is spinning at a certain fairly steady rate and if the slave can then spin at the same rate I'll be OK. I realize that the slave will always lag a little as far as angular position goes. But as long as the lag is unchanging then the two spindles can match speeds. Eric That's really neat and elegant. Hope it works! -j Me Too! |
#24
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
On Sep 10, 4:49 pm, wrote:
The counts match exactly after everything stops. The numbers change way too fast to see how close they are during rotation. Eric Sounds like you are ready to try it out on some parts. I think it will work a treat. Dan |
#25
Posted to rec.crafts.metalworking
|
|||
|
|||
Homemade 2 spindle lathe preliminary results
|
Reply |
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Spindle nose adapter for Delta-Rockwell 11" lathe (#25-100) with L-00 spindle | Metalworking | |||
Homemade CNC 2 spindle lathe question | Metalworking | |||
FOLLOWUP -- Lathe and VFD -- Some results | Metalworking | |||
Homemade Lathe | Woodworking | |||
Results with Homemade Carbide Insert Lathe Toolholder | Metalworking |