Servo motor VS Stepper motor
Posted by Jared Harvey
|
Re: Servo motor VS Stepper motor December 24, 2007 03:58AM |
I'm not a big fan of Servo systems just seems to much like hard work when steppers are so easy, all that said the Cyons 1001 looks excellent so i think you now have a convert 
with 3200 CPI < .01mm resolution and even the most basic giving 1150 CPI = 0.022mm well within spec
DIY soldering of QFN packages is do-able but quite hard work but probably worth it, at $11 - $13 not very cheap but with correct engineering you would only need 2 for all three axis. Being silicon it should get cheaper with Qty and time
Ian
[www.bitsfrombytes.com]
with 3200 CPI < .01mm resolution and even the most basic giving 1150 CPI = 0.022mm well within specDIY soldering of QFN packages is do-able but quite hard work but probably worth it, at $11 - $13 not very cheap but with correct engineering you would only need 2 for all three axis. Being silicon it should get cheaper with Qty and time
Ian
[www.bitsfrombytes.com]
|
Re: Servo motor VS Stepper motor December 24, 2007 10:08AM |
As it is a touch new I didn't really expect it to be that cheep (Shame realy), but for what you are getting compared to other technologies it is comparatively cheep.
Lowest cost source is probably going to be straight out of a cheep mouse in the short term. (The cost will have to be competitive with other optical solutions and these retail at the same or less than you suggest the chip on it's own would cost).
Unless they are bought in bulk rather than 1 or 2 at a time.
Interestingly enough if you compare the cost of the device with the cost of other technology which needs a precise drive train even at the price quoted it becomes comparatively cheep because a non precise drive train can be used. You are offsetting the savings made in one part of the system as a result of the technology change with the additional costs in the other part. I guess it goes without saying that where your final aggregate cost is less than you had before then the solution is relatively cheap.
Consider that a stepper on it's own's accuracy is a function of the accuracy of the stepper minus the inaccuracy's in the drive train.
Like wise a DC Motor with an armature encoder has the same considerations. Accuracy of the motor/encoder minus the inaccuracy's in the drive train.
The accuracy of a servo system using position sensing at the tool end of the drive train nulls out or corrects the inaccuracies in the drive train.
Providing you eliminate backlash (or at least reduce the backlash to a value significantly less than you require) you turn the source or actuator just that bit more or less until your tool end encoder says the position is good.
Unfortunately whilst it is a good sensor for what is currently being done, It would set me back on my personal end goal (although I am happy to help out with this one to bootstrap myself and others first).
For me personally a better RM/RP platform would not be a Cartesian Bot. More likely a turntable and articulated arm bot.
Cartesian's have some limitations as to where the tool head can get without fouling the workpiece.
I can though see a future for a reprap for me to build the components for the articulated arm and turntable though.>
<
aka47
Necessity hopefully becomes the absentee parent of successfully invented children.
Lowest cost source is probably going to be straight out of a cheep mouse in the short term. (The cost will have to be competitive with other optical solutions and these retail at the same or less than you suggest the chip on it's own would cost).
Unless they are bought in bulk rather than 1 or 2 at a time.
Interestingly enough if you compare the cost of the device with the cost of other technology which needs a precise drive train even at the price quoted it becomes comparatively cheep because a non precise drive train can be used. You are offsetting the savings made in one part of the system as a result of the technology change with the additional costs in the other part. I guess it goes without saying that where your final aggregate cost is less than you had before then the solution is relatively cheap.
Consider that a stepper on it's own's accuracy is a function of the accuracy of the stepper minus the inaccuracy's in the drive train.
Like wise a DC Motor with an armature encoder has the same considerations. Accuracy of the motor/encoder minus the inaccuracy's in the drive train.
The accuracy of a servo system using position sensing at the tool end of the drive train nulls out or corrects the inaccuracies in the drive train.
Providing you eliminate backlash (or at least reduce the backlash to a value significantly less than you require) you turn the source or actuator just that bit more or less until your tool end encoder says the position is good.
Unfortunately whilst it is a good sensor for what is currently being done, It would set me back on my personal end goal (although I am happy to help out with this one to bootstrap myself and others first).
For me personally a better RM/RP platform would not be a Cartesian Bot. More likely a turntable and articulated arm bot.
Cartesian's have some limitations as to where the tool head can get without fouling the workpiece.
I can though see a future for a reprap for me to build the components for the articulated arm and turntable though.>
<aka47
Necessity hopefully becomes the absentee parent of successfully invented children.
|
Re: Servo motor VS Stepper motor December 24, 2007 02:46PM |
Interesting methods
A friend of mine built a number of reflow ovens to order they were actually quite simple. (In the very early days of SMT components)
He used Long Wave IR to heat the boards up and they were passed under the heating elements on a chain/mesh conveyor.
Temperature sensing was done by a sensor just of to one side but still under the elements that overlapped the conveyor (big name for a little belt thing).
As the heating elements were controlled using distributed burst control the only other variable was the speed the belt ran at.
The great thing about this was that the process was continuous, if you have a number to do, once set up and up to temperature you just kept putting the preloaded boards onto the conveyor at one end and took them off at the other.
For heating elements you can use those quartz heater element/tubes that are starting to show up as budget heaters in those cheap shops.
I am not impressed with them as heaters but as reflow heaters they are great.
Looking at the links you sent Jared, for small scale limited runs I would opt for a heater plate with a halogen boost to finish off the reflow (Instead of the hot air thing). I like the idea of the pre and post heat. it minimises dry joints etc caused when crystals form in solder that is cooled too quickly.
The hot plate need be no more than a standard cooker hotplate with adequate heating control.
Most electric heating elements have a fairly high thermal hysteresis so burst control if distributed is normally quite sufficient.
Where your board is static ie on the hotplate a timed burst from a halogen element modulated for the ideal temperature should be enough.
If you want to eliminate ambient variation box it up (or at least all but one side. Grin)
All of the above is easily accomplished using a PIC and something sensible to detect zero crossing. (The same friend used dropper resistors and cmos gates direct off line, I am not sugesting you do the same). from this you switch an opto triac (driving a power triac if you need it for the relevant number of half cycles).
If you wanted something lower tec than a PIC then a counter (that counts to 100) and eprom as a pattern generator and counter that counts to 100 could be pressed into service. with the high order lines used to select one of 100 programs one for each percentage point.
100 out of 100 half cycles = 100%
50 out of 100 half cycles = 50%
etc
etc
etc.
Here in the UK at 50Hz the burst period is 1 second, ie 100 half cycles, for a reasonable 1% resolution.
In the US at 60Hz the burst period will be less than this.
If you make sure that the on cycles are distributed as evenly as possible through the burst period (ie not all on sequentially then all off) good control of a high hysteresis heating element can be acheived.
Advantages to this method are:-
1. Very low interference (Particularly compared to phase angle control) as you only ever turn the elements on when the voltage is low. (Close to Zero Point Crossing)
2. Easy and cheap to do, components do'nt have to be sized to cope with peak in rush current levels.
3. Lifetime of all the heating elements is preserved as you never switch it hard on at peak mains cycle when cold (inrush currents at this time normally mortal heating elements, particularly light bulbs, it's why they always go at switch on)
Thoughts for what they are worth.
Cheers
aka47
PS Merry Xmas everyone, I have avoided my daughter painting everyone's nails now and can go back downstairs. ><
Necessity hopefully becomes the absentee parent of successfully invented children.
A friend of mine built a number of reflow ovens to order they were actually quite simple. (In the very early days of SMT components)
He used Long Wave IR to heat the boards up and they were passed under the heating elements on a chain/mesh conveyor.
Temperature sensing was done by a sensor just of to one side but still under the elements that overlapped the conveyor (big name for a little belt thing).
As the heating elements were controlled using distributed burst control the only other variable was the speed the belt ran at.
The great thing about this was that the process was continuous, if you have a number to do, once set up and up to temperature you just kept putting the preloaded boards onto the conveyor at one end and took them off at the other.
For heating elements you can use those quartz heater element/tubes that are starting to show up as budget heaters in those cheap shops.
I am not impressed with them as heaters but as reflow heaters they are great.
Looking at the links you sent Jared, for small scale limited runs I would opt for a heater plate with a halogen boost to finish off the reflow (Instead of the hot air thing). I like the idea of the pre and post heat. it minimises dry joints etc caused when crystals form in solder that is cooled too quickly.
The hot plate need be no more than a standard cooker hotplate with adequate heating control.
Most electric heating elements have a fairly high thermal hysteresis so burst control if distributed is normally quite sufficient.
Where your board is static ie on the hotplate a timed burst from a halogen element modulated for the ideal temperature should be enough.
If you want to eliminate ambient variation box it up (or at least all but one side. Grin)
All of the above is easily accomplished using a PIC and something sensible to detect zero crossing. (The same friend used dropper resistors and cmos gates direct off line, I am not sugesting you do the same). from this you switch an opto triac (driving a power triac if you need it for the relevant number of half cycles).
If you wanted something lower tec than a PIC then a counter (that counts to 100) and eprom as a pattern generator and counter that counts to 100 could be pressed into service. with the high order lines used to select one of 100 programs one for each percentage point.
100 out of 100 half cycles = 100%
50 out of 100 half cycles = 50%
etc
etc
etc.
Here in the UK at 50Hz the burst period is 1 second, ie 100 half cycles, for a reasonable 1% resolution.
In the US at 60Hz the burst period will be less than this.
If you make sure that the on cycles are distributed as evenly as possible through the burst period (ie not all on sequentially then all off) good control of a high hysteresis heating element can be acheived.
Advantages to this method are:-
1. Very low interference (Particularly compared to phase angle control) as you only ever turn the elements on when the voltage is low. (Close to Zero Point Crossing)
2. Easy and cheap to do, components do'nt have to be sized to cope with peak in rush current levels.
3. Lifetime of all the heating elements is preserved as you never switch it hard on at peak mains cycle when cold (inrush currents at this time normally mortal heating elements, particularly light bulbs, it's why they always go at switch on)
Thoughts for what they are worth.
Cheers
aka47
PS Merry Xmas everyone, I have avoided my daughter painting everyone's nails now and can go back downstairs. >
<Necessity hopefully becomes the absentee parent of successfully invented children.
Sorry, only registered users may post in this forum.