[Whizbang]

Why use servos instead of steppers?
I understand that the 2:1 or 3:1 gives you more torque but you could do this with steppers aswell.
I have noticed servos could be used for positional feed back however does Mach even support it?
Wouldn't a closed loop stepper system with rotary encoder do the same thing as a servo and glass scale setup?

Many rf31 builds control z height by mounting a stepper to the fine feed knob on the head,micro kinetics also does this with their rf45 models opposed to mounting a stepper/servo to the column (most builds here as well as industrial hobbies). Is one of these systems more advantageous
On one hand you get greater travel with the column mounting however we are talking about a mill not a fdm machine no one mills with a 8 " long end mill anyway.
So using the fine feed seems more reasonable unless the ram isn't very sturdy.. this gives about 4.5" of travel. Bridgeport does the z travel this same way on their eztrak mills.
As far as ballscrews go most clones have trapozodial acme, ball screws are prolly a must even double nutted for back lash comp. Is chai good to go to or is there a better player in the game for45 style mills?

Noticing micro kinetics 45 XL has hand wheels in x and y travels still are there any problems with this a motor being turned is a generator so if I used a mill like that manually to make a part with the electronics off would this do damage to the electronics. My experience with both eztraks and causing kondias makes me more comfortable manually edge finding and finding centers of bores. Some operations and making of features are just easier to do manually.

I know that the 45 style mills come in many flavors but does anyone know of xy and z drawings out there to at least get ideas from for motor mounting options.

[109jb]

Why use servos instead of steppers?
I understand that the 2:1 or 3:1 gives you more torque but you could do this with steppers aswell.
I have noticed servos could be used for positional feed back however does Mach even support it?
Wouldn't a closed loop stepper system with rotary encoder do the same thing as a servo and glass scale setup?

A stepper motor has its highest torque at low RPM's and once past a few hundred RPM the torque drops off. A servo maintains constant torque from 0 RPM to RPM's much higher than you can achieve with a stepper. If you run a stepper with a reduction drive, the RPM at which the screw turns will be even slower when you get to the torque drop off of the stepper. At some point the torque is low enough that the motor stalls and there isn't anything you can do to increase the torque to overcome this. The limiting factor is quite often high rpm operation. A stepper can be made to stall with no load on it simply by trying to run it too fast. Now add the load of turning a screw to move a table and that stall speed comes down even lower. Generally direct drive is the way to go with a stepper.

A servo on the other hand has a constant torque to pretty high rpm, so you can get away with putting a smaller motor on, but with the reduction drive and the higher rpm, and the constant torque you can get much higher speeds. Also, if a servo starts to get to where the torque supplied is lower than the torque needed, the driver can increase current to provide an increase of torque.

You can certainly run steppers on a 45 size mill and people have, but you get to a point where the advantages of the servos overcomes the higher price. When you get into the larger steppers, larger stepper drivers, and larger required power supplies, the prices approach what you can do a servo based system for.

The Gecko servo drivers in particular close the loop at the stepper. To the control software (Mach3, LinuxCNC), a Gecko system looks no different than a stepper system. The gecko drives have a fault pin that can send a signal back to the control software to stop the program.

A system with a rotary encoder can never match a system that uses a glass scale for feedback. A glass scale provides the best positioning accuracy. The glass scale is not affected by things like the lead screw heating up and expanding during operation. A rotarty encoder system can't compensate for lead screw expansion no matter if it is on a stepper or a servo.

The new feedback stepper systems are appealing until you look at the price. I myself want to eventually get a 45 size machine and have been doing homework in this regard. To get a basic stepper system going (no feedback), a rough estimate puts it at $950 for the motors, drivers, and power supplies. A DC servo System looks like it would be about $1050. A feedback stepper system is about $1550. For about the same price you could do an AC servo system.

Many rf31 builds control z height by mounting a stepper to the fine feed knob on the head,micro kinetics also does this with their rf45 models opposed to mounting a stepper/servo to the column (most builds here as well as industrial hobbies). Is one of these systems more advantageous
On one hand you get greater travel with the column mounting however we are talking about a mill not a fdm machine no one mills with a 8 " long end mill anyway.
So using the fine feed seems more reasonable unless the ram isn't very sturdy.. this gives about 4.5" of travel. Bridgeport does the z travel this same way on their eztrak mills.
As far as ballscrews go most clones have trapozodial acme, ball screws are prolly a must even double nutted for back lash comp. Is chai good to go to or is there a better player in the game for45 style mills?

Putting the Z axis motor on the fine feed limits Z axis travel as you mentioned. The biggest problem I see with this is in regard to tool changes. Think about a drill bit in a chuck compared to a 1/4" end mill in an end mill holder. You quickly lose all of the quill travel. Also, the quill is not a very good fit in the head, so there is some slop right away unless you lock the quill. I ran in to both of these problems on my round column Harbor Freight mill. Neither is the end of the world, but having the whole head move like my G0704 I have now is a whole lot more convenient.

Noticing micro kinetics 45 XL has hand wheels in x and y travels still are there any problems with this a motor being turned is a generator so if I used a mill like that manually to make a part with the electronics off would this do damage to the electronics. My experience with both eztraks and causing kondias makes me more comfortable manually edge finding and finding centers of bores. Some operations and making of features are just easier to do manually.

You could never turn the handle fast enough by hand to do damage.

I know that the 45 style mills come in many flavors but does anyone know of xy and z drawings out there to at least get ideas from for motor mounting options.

I have searched for the same thing, but I have never got a good feeling that the different makes/models are made the same enough to allow making parts before hand..

[109jb]

Here is an example of a stepper converted 45 on youtube. His motors sound like way overkill, but it apparently works although he doesn't show a lot of miling in his videos

[ame=http://www.youtube.com/watch?v=FC_ok-zP99Y]CNC Conversion RF-45 Milling Machine, FOR SALE - YouTube[/ame]

[TiagoSantos]

Steppers can be easier to work with, there isn't much tuning to do, no encoders to fail, less wiring.. But you need fairly hefty steppers to match the potential of servos, and still no way to match the speeds. The tormach runs steppers, they work very well - but I believe their latest series machine moves at 140ipm or something? There are quite a few home converted RF45s with servos running 200ipm. The poster above made all the points I wanted to make about why you can't run steppers that fast. I know - I'm running steppers on an RF40! I'm also considering upgrading to servos. Or much bigger steppers, but I feel like for the money, I might as well go through the extra bit of trouble and get servos.

[arizonavideo]

It is not done often but direct drive steppers should be fine for the X and Y of the smaller table RF-45.

I like the 570-Oz Nema 23 (KL23H2100-50-4B).

The Z is a different story. Sometimes the head is just hard to get moving. A servo can start motion and fall behind a few steps and be fine, it will catch up and not loose position. A stepper might loose steps and your job could be toast.


A cost effective setup might be steppers for the X and Y and a servo for the Z.

[SCzEngrgGroup]

So much mis-information, so little time....

Steppers vs servos is a debate that goes back to the dawn of CNC, and there is a fantastic amount of just blatant mis-information that has become "fact" through sheer force of repetition, most often by people who simply don't know what they're talking about. The simple fact is, for most machines in the "benchtop" size class, EITHER will work very nicely, and it is entirely possible to design one machine using steppers, and another, identical machine using servos, and get *precisely* the same performance, reliability, and precision from BOTH. Servos ARE NOT inherently "better" until you get into pretty large machines, where the torque characteristics of steppers become a serious limitation. For any given cost, steppers will nearly always be cheaper in this size class (ignoring those who insist on comparing the cost of a new stepper system to the once-in-a-lifetime deal they found on used servos on E-Bay).

You will hear lots of anecdotal evidence from people who've built stepper-based machines, and suffered the dreaded "lost steps". Those people won't accept the simple fact that "lost steps" is, plain and simple, a clear sign of POOR DESIGN, and it absolutely NOT an inherent characteristic of steppers. Steppers are used by the millions every day of the year in high-precision, high-reliability industrial machinery, and they perform flawlessly for years on end.

Both steppers and servos are the same in one critical aspect - If you attempt to operate them beyond their capabilities, THEY WILL FAIL! Lost steps is, by definition, proof that the motor was being operated beyond its torque capability. Respect that limit, and steppers are 100% reliable, 100% of the time, just like servos. Ignore that limit, and you'll have failures with steppers or servos.

Whether stepper or servo, exceed the motors torque limit, and you WILL lose position. The only difference is in exactly HOW it will happen. A stepper will either slip several steps, or simply stall - stopping dead in it's tracks. In either case, you're no longer where you think you are, and the ONLY recovery is to re-reference the machine, and start over. With servos, they will tolerate *brief* excess loads, but THEY WILL STILL END UP OFF-POSITION for some period of time. They difference is, a servo MIGHT recover correct position once the load is reduced, *IF* the overload is brief enough that the servo following error is not exceeded. If the following error is exceeded, then the situation is precisely the same as with a stepper - you are hosed, and your work is probably scrap. In either case, stepper or servo, there WILL be a dimensional error in your part.

Contrary to common belief, servos are not a "magic bullet" that ensures you never lose position, and never create scrap. And, when using Mach3 as the machine controller, both steppers and servos are, for all practical purposes, open-loop, as the position loop CANNOT be closed back to the trajectory planner, and your only recovery from any position loss is an E-Stop, re-referencing the machine, and, most likely, scrapping the part in process. AFAIK, EMC also does not support feedback to the trajectory planner - it simply allows you to detect the failure, and stop execution. Feedback all the way to the trajectory planner it the only way to deal truly gracefully with overloads, by monitoring the following error, and adjusting the commanded position/velocity to reduce the load when the following error starts to increase. This IS how "real machines" operate, but it is NOT how Mach operates, even in v4.

However, if you PROPERLY DESIGN the drives for the machine, and always operate the drives within their capabilities, you will NEVER lose steps with either steppers or servos. That is a FACT. If you insist on running on the edge of the drives capabilities, you WILL occasionally lose position, whether you use steppers or servos. That is also a FACT.

Even for an RF45, there is not a reason in the world, if you design and operate it properly, you cannot achieve any reasonable level of performance with absolute reliability, with steppers. There are tens of thousands of old Bridgeport CNC mills out there, which are FAR larger and heavier than an RF45, and they all came from the factory with steppers, and have operated accurately, and reliably, literally for decades. If you're after bragging rights, and want to be able to brag that your RF-45 can run 1000 IPM rapids, then you'd better go with servos - you won't get there with steppers. But, in the real world, that level of performance is not only completely useless in a machine of that size, but it's actually quite dangerous. And, you'll spend far more on the drives than on the machine itself. If you want sensible real-world performance (100-200 IPM rapids) it can absolutely by done with steppers, with absolute reliability. Anyone who says otherwise, simply doesn't know what they're talking about.

Regards,
Ray L.

[TiagoSantos]

Thanks for the extensive post Ray - I think you confirmed what people were saying in this thread though. I didn't notice any posts above about losing steps, but I will say one thing that can be a decent advantage of servos - yes, Mach3 won't get feedback from the encoders, but losing position with steppers means you'll probably cut through something you didn't intend to cut or crash the tool into something solid. I've had this happen before, the tool slipped down in the collet (noob..) and I started taking a much heavier cut than expected, the X steppers started stalling and were behind by about 2 inches when the Y axis started moving for the following line of code Mach3 and the controller had no idea and it merrily ruined my part and broke my end mill. With servos, Mach3 still would have no clue, but the drives would have thrown the e-stop a lot sooner and I'd have an extra end mill and maybe a salvageable part!

Of course none of this would happen if I had any idea what I was doing and was the perfect machinist, mechanic and technician with many decades of experience. But I am not. And I will make mistakes, both in design and execution. I feel like there are enough servo packages out there for incredible prices that make the difference in price very attractive for guys like me who could definitely use the extra margin of safety with the encoder feedback to the drives.

Right now I'm looking at replacing my steppers (which I now know were a poor choice to begin with, but I didn't buy them with this machine in mind) with DMM Tech servos. The DMM 3 axis 400w package is now (with the new pricing) around $1300 for the full setup. I could maybe build a comparable stepper system for $900-1000. I'm not sure yet what I'll end up doing, but DMM being local and having a physical office where I can go and pick parts up from and get help when I need it is probably playing a big part in my swaying in the direction of servos.

[SCzEngrgGroup]

Thanks for the extensive post Ray - I think you confirmed what people were saying in this thread though. I didn't notice any posts above about losing steps, but I will say one thing that can be a decent advantage of servos - yes, Mach3 won't get feedback from the encoders, but losing position with steppers means you'll probably cut through something you didn't intend to cut or crash the tool into something solid. I've had this happen before, the tool slipped down in the collet (noob..) and I started taking a much heavier cut than expected, the X steppers started stalling and were behind by about 2 inches when the Y axis started moving for the following line of code Mach3 and the controller had no idea and it merrily ruined my part and broke my end mill. With servos, Mach3 still would have no clue, but the drives would have thrown the e-stop a lot sooner and I'd have an extra end mill and maybe a salvageable part!

Of course none of this would happen if I had any idea what I was doing and was the perfect machinist, mechanic and technician with many decades of experience. But I am not. And I will make mistakes, both in design and execution. I feel like there are enough servo packages out there for incredible prices that make the difference in price very attractive for guys like me who could definitely use the extra margin of safety with the encoder feedback to the drives.

Right now I'm looking at replacing my steppers (which I now know were a poor choice to begin with, but I didn't buy them with this machine in mind) with DMM Tech servos. The DMM 3 axis 400w package is now (with the new pricing) around $1300 for the full setup. I could maybe build a comparable stepper system for $900-1000. I'm not sure yet what I'll end up doing, but DMM being local and having a physical office where I can go and pick parts up from and get help when I need it is probably playing a big part in my swaying in the direction of servos.

Whatever cut caused the steppers to fail would have almost certainly caused the servo to fail as well. How extensive the damage is, is strictly a matter of how large the allowable following error. Even Gecko G320s, which have a notoriously small following error (128 counts) would have you off by several thou - enough to ruin any critical dimension. The G320X allows this to be increased, which most any reasonable person WILL do when tuning for best system system performance. This will, again, increase the magnitude of any error due to over-driving the motor. If all you're after is "Whoops!" detection, encoders on steppers with a Mach3 macropump to track the following error will get the *exact* same result as a servo setup.

Servos will rarely protect you from your own mistakes.

Regards,
Ray L.

[TiagoSantos]

If all you're after is "Whoops!" detection, encoders on steppers with a Mach3 macropump to track the following error will get the *exact* same result as a servo setup.

Sure - but by the time you add the encoders and whatever else you need, there goes most of the price difference to the servos again. And I guess you missed the part about the limited skills I'm sure there are a million different ways of doing things, but there is something to be said for following the herd when you have a limited budget, skillset or amount of time.

Also, I'm sure that despite the point you're trying to make, you can appreciate the difference between scrapping a part because you (hypothetically, more likely it would be me) messed up a cut by a few thou, or running the end mill into the vise or a clamp or whatever else ends up in the way when the steppers stall and the cutter takes a wrong turn. It should never happen, it would never happen if I was perfect, but I'll spend the extra couple hundred bucks to cover my shortcomings!

I understand your point and agree with you for the most part - like I said on my first post, lots of great machines out there are running steppers - Tormach being a great example on a machine similar to the kind of thing we're discussing. Hell, even my own machine - with crappy steppers and a G540, I'm easily and reliably running at 70ipm, with significant margins of safety. Like you said, I can push it up to 120ipm and it works most of the time, but it's too close to the performance limit to let it run unattended. But I'm at the limit of what I can do with a $500 stepper system, and the next step up is significantly more expensive.

[Whizbang]

This is a good discussion.
Many of my questions have been answered.
However this a over exaggerated question of what I'm after.
say I am making a pocket that is 5.500" but my glass scales are indicating my movement is 5.492 the system I am ideally after would continue to cut (no matter how long or through what) till the glass scale read 5.500"
I'm guessing this is called a closed loop system and from what was said above Mach 3 cannot do this nor emc2 so what software and hardware can?

I know this sounds like an awesome way to destroy and reak havok on a machine ,and your right however if I am going to be watching the operations the whole time it becomes less of an issue. The reason I want this ability is lets say I was a little over confident on some ambitious feeds and speeds that could bog down the motor and reduce accuracy this system corrects and fixes the issue. Also what if I have backlash error or non Crisco smooth ways the motor continuing to move until the exact coordinate is reached solves this problem as well.

[SCzEngrgGroup]

This is a good discussion.
Many of my questions have been answered.
However this a over exaggerated question of what I'm after.
say I am making a pocket that is 5.500" but my glass scales are indicating my movement is 5.492 the system I am ideally after would continue to cut (no matter how long or through what) till the glass scale read 5.500"
I'm guessing this is called a closed loop system and from what was said above Mach 3 cannot do this nor emc2 so what software and hardware can?

I know this sounds like an awesome way to destroy and reak havok on a machine ,and your right however if I am going to be watching the operations the whole time it becomes less of an issue. The reason I want this ability is lets say I was a little over confident on some ambitious feeds and speeds that could bog down the motor and reduce accuracy this system corrects and fixes the issue. Also what if I have backlash error or non Crisco smooth ways the motor continuing to move until the exact coordinate is reached solves this problem as well.

ANY servo system, if properly designed and calibrated, and operated within its capabilities, will do that, with no help from Mach3. That is, after all, the whole point of using linear scales. ANY stepper system, if properly designed and calibrated, and operated within its capabilities, will do the same, to within the physical accuracy of the machine hardware. The difference between the two is the servo system will operate, more or less, to the accuracy of the scales, regardless of changes in temperature of the machine, while the stepper system will operate to the accuracy of the screws and axis parameter setup, and will not compensate for errors in the lead accuracy of the screws, errors introduced by thermal expansion, etc. In general, neither should be counted on to produce parts to better than a few thou in any dimension, in the hands of an unskilled operator.

That said, in the real world, putting a system like that on an inexpensive Chinese machine is the very definition of "putting lipstick on a pig". There are countless sources of error in these machines, and addressing one, without addressing many of the others, will not gain much in terms of overall accuracy. What good is it to be able to position the table to within 0.00002", if there is 0.005" of flex in the column, or 0.003" of "spring" in the axis drives? Achieving high accuracy reliably requires that *every* part of the system be designed from the ground up for that level of accuracy. Addressing only a handful of the many sources of error may get you only a 10% increase in real accuracy, while costing you a great deal of time and money. It is simply not possible to get high accuracy without doing a LOT or hard work to understand first where the inaccuracies currently are in the specific machine in question, and their magnitudes, then figuring out how to mitigate each of them. Simply switching from steppers to servos will get you almost nothing in terms of improvement in measurable, real-world accuracy.

There two key questions you should ask yourself:

1) Why do I think you *need* such high accuracy? As an engineer for over 30 years, I have always found that there are very few applications that really *require* extreme accuracy. Far more often than not, I see designs which are over-complicated, that REQUIRE highly accurate parts to work, where a "better" design would have met the same design goals, at a lower cost, and with less effort, by designing in such a way that extreme accuracy is simply not required. In other words, more often than not, the requirement of extreme accuracy is a clear indication of poor design.

2) Am I skilled enough to achieve such accuracy, even on a dead-accurate machine? The work product will be only as good as the operator. Even the best machine, in the hands of an unskilled operator, will produce mostly scrap, albeit very fast, and at a very high cost. By the same token, a skilled machinist can easily produce parts to very high tolerance using a very sloppy, inaccurate machine. Give an amateur machinist a brand new Haas VMC, and odds are the parts he produces will be little better than those he would create using a home-built benchtop conversion. And, odds are, before the first day is out, he will have broken a very expensive machine and lots of expensive tools.

In reality, getting below a true 0.005" accuracy out of a machine tool takes a LOT of work, a lot of skill, and a LOT of money. No one component change will get you very far, and there are no shortcuts.

Regards,
Ray L.

[109jb]

Ray,

I don't see where anyone said a stepper system wouldn't work on a 45 size mill like you imply. They can and will work well if you live within their limitations. What has been said is that when you get to the larger size steppers like are needed on a 45, particularly for moving the whole head, you begin to get to a price point where the advantages of a servo system become more attractive. I quoted some prices in my post above and I just looked up the prices from Automation Technologies for all of the systems. I wasn't too far off on my estimates. Here are their prices and specs.

Stepper system with no feedback, 3-Axis NEMA34 CNC Kit (60V/20A/906 oz-in/ KL-8056D DIGITAL DRIVER) $899.34

DC servo system with rotary encoders, 3-Axis NEMA34 850 oz in 72V/20A PSU G320X Gecko Driver $1069.88

Feedback stepper system, NEMA34 1128 oz-in closed-loop stepper Motors 3 Axis CNC Kit (110VAC/220VAC) $1546.88

The above systems are all package prices for new components which are complete with power supplies, breakout board, motors, encoders, etc. where applicable. The only one I was really off on was the non-feedback stepper system where I was about $50 high, but I had a larger motor spec'd for the Z axis. Additionally, I think the servo system is overkill in the X and Y and the servo system could be reduced in price by going to smaller motors on those axes. This would further reduce the gap in price, but I am not done with my analysis on this aspect though.

While I agree that the non-feedback stepper system will run well when operated within limits, things can happen and pretty easily. A little too aggressive on feed rate, a dull cutter, the gibs a tad too tight, etc, can push an otherwise well performing machine over the edge. With no feedback to stop operation when some steps get lost the machine happily keeps running until the part is ruined, the vise jaw gets hit, the table gets drilled into, etc. I personally don't run "lights out", and always monitor my machine but that isn't true of everyone. If you add encoders to enable the machine to stop in the event of a following error, then you can add another $100 for the encoders. Now you are at $1000 for that system. For the $70 difference in price (plus maybe $100 for the belt drive components) I personally would go with the DC servo system. I think that is all that I or others have said. I can certainly say that that is all I was trying to say.

I agree that the whole system, machine, drives, screws, etc., have to be set up properly, but in my research, at these price points, the DC servo system would have an ample reserve of power at all speeds compared to the stepper system which has reserve at lower speeds but looses this reserve at higher speeds due to torque drop off. What I mean by this is if the machine is set for the same maximum speeds on each system (I used 200 IPM in my comparison), then the servo system will have a larger torque margin before stalling, particularly at the 200 ipm rapid speed. It then also has the benefit of already having a means of following error detection. Again, if the stepper system is set up right, you won't run out of reserve, but the higher margin on the servo drives is an advantage in my opinion.

The servo system with the Gecko servo drivers already has error detection that can send a signal back to the control software to halt operation. In LinuxCNC this is easy to do. I am not sure how easy it is in Mach3 as I haven't played around too much with Mach3, but I would think it would be possible.

As for the steppers with closed loop feedback, as was said, it is a matter of cost. I am sure these systems would work fine too, but at more than $300 over the DC servo system, I would opt for the servos again.

As for linuxCNC, it CAN close the loop directly back to the software using linear scales for position and rotary encoders for velocity. Not real easy to do from what I have read though. The stuff I have read indicates that to avoid problems, high resolution scales matched to the output of the rotary encoders is needed ($$$$$). However it can be done and the WIKI over at the linuxcnc.org site can has a lot of info on this. I personally don't think this is necessary for most applications. I would like to do it some day simply because I like a challenge and I already plan to buy at least one scale to temporarily mount so that I can map the screws on my G0704. Might as well put it to good use on the 45 when I get it.

On another note, I do plan to implement following error detection on my G0704. For this size machine I think it is a no-brainer to go with steppers vs servos. Excellent performance can be realized using steppers and the price difference is much larger for a machine this size. I was able to do mine for about $500 cheaper using steppers for the drive system than I could have using servos. This was using all new components but with a cheap server power supply. So with encoders for following error detection I will still be $400 cheaper than a servo drive system. This is enough of a difference that the simpler stepper system is more attractive to me.

Bottom line - In my opinion, it depends on the machine and the cost. Not just whether or not it can be done. It can certainly be done with any system you chose. In my opinion, the comparison of the different systems for a 45 size machine make the DC servo drive system more attractive to me. Others may feel differently.

[Whizbang]

As far as errors take machine rigidity out of it. I am an engineer as well and have made a set of ductile cast iron reinforcements for my 0704. I assure you there is no column deflection as the column and base are now one piece. I have also taken my machine to a professional scraper to get a quote as well as visually see where the machine is out of tolerance. Some people in the zone have seen all of these axis documented. That being said what I am looking for is a system that will overcome the inconsistency in the ways on either 704s or 45 and a over ambitious feed and speed so if i take too big of a bite or if I experience a bit of a extreme stick slip due to very tight ways that I can over come it and be positional accurate. Think solely geometry errors and aggressive feed.

[691175002]

With regards to linear scales as a supplement steppers or servos:

KFlop and EMC2 will both support it. Dynomotion even has a page on setting it up in their manual:
Closed Loop Stepper
(Rotary and linear encoders are very similar from an electrical/setup perspective).

Ray is correct in saying that no amount of feedback will save you if the machine is pushed too far. The sole advantage you get in such a situation is that your machine will fault instead of continuing to cut (potentially doing significant damage to itself).

Linear scales work to reduce sources of error that occur in and after the ballscrew, in particular variations in screw pitch, thermal growth of the screw, and to a lesser extent backlash.

There are two ways the loop can be closed, separately or together with the servo drive. Both KFlop and EMC2 support both methods, however the second is higher performance and much harder to set up. Note that a typical servo system will have a higher control bandwidth than a typical stepper system. This is of little importance in most machines, but plays a role when you are attempting to add additional feedback to an axis.

Like others, I don't really believe that scales add very much to a hobby level machine. Its just that they are cool and you can get them dirt cheap from china so why not.

[SCzEngrgGroup]

Ray,

I don't see where anyone said a stepper system wouldn't work on a 45 size mill like you imply. They can and will work well if you live within their limitations.

Just wait, someone will say it soon enough.... It always happens when this topic comes up. You'll also have those who insist servos are better because everyone just knows steppers *always* lose steps.

What has been said is that when you get to the larger size steppers like are needed on a 45, particularly for moving the whole head, you begin to get to a price point where the advantages of a servo system become more attractive. I quoted some prices in my post above and I just looked up the prices from Automation Technologies for all of the systems. I wasn't too far off on my estimates. Here are their prices and specs.

Stepper system with no feedback, 3-Axis NEMA34 CNC Kit (60V/20A/906 oz-in/ KL-8056D DIGITAL DRIVER) $899.34

DC servo system with rotary encoders, 3-Axis NEMA34 850 oz in 72V/20A PSU G320X Gecko Driver $1069.88

Feedback stepper system, NEMA34 1128 oz-in closed-loop stepper Motors 3 Axis CNC Kit (110VAC/220VAC) $1546.88

The above systems are all package prices for new components which are complete with power supplies, breakout board, motors, encoders, etc. where applicable. The only one I was really off on was the non-feedback stepper system where I was about $50 high, but I had a larger motor spec'd for the Z axis. Additionally, I think the servo system is overkill in the X and Y and the servo system could be reduced in price by going to smaller motors on those axes. This would further reduce the gap in price, but I am not done with my analysis on this aspect though.

While I agree that the non-feedback stepper system will run well when operated within limits, things can happen and pretty easily. A little too aggressive on feed rate, a dull cutter, the gibs a tad too tight, etc, can push an otherwise well performing machine over the edge. With no feedback to stop operation when some steps get lost the machine happily keeps running until the part is ruined, the vise jaw gets hit, the table gets drilled into, etc. I personally don't run "lights out", and always monitor my machine but that isn't true of everyone. If you add encoders to enable the machine to stop in the event of a following error, then you can add another $100 for the encoders. Now you are at $1000 for that system. For the $70 difference in price (plus maybe $100 for the belt drive components) I personally would go with the DC servo system. I think that is all that I or others have said. I can certainly say that that is all I was trying to say.

If you want to compare the TOTAL cost of a servo system, you have to add to your numbers a couple hundred $ for belts, pulleys, and additional metal for the belt reducers that are necessary for those servos, and more wire, cable, and connectors. Add that, and you've got easily a 30-40% system cost increase for servos over steppers. Many people would consider that quite significant.

I agree that the whole system, machine, drives, screws, etc., have to be set up properly, but in my research, at these price points, the DC servo system would have an ample reserve of power at all speeds compared to the stepper system which has reserve at lower speeds but looses this reserve at higher speeds due to torque drop off. What I mean by this is if the machine is set for the same maximum speeds on each system (I used 200 IPM in my comparison), then the servo system will have a larger torque margin before stalling, particularly at the 200 ipm rapid speed. It then also has the benefit of already having a means of following error detection. Again, if the stepper system is set up right, you won't run out of reserve, but the higher margin on the servo drives is an advantage in my opinion.

Torque reserve during rapids is nearly useless. At rapid speeds, the load is essentially constant, and not all that high. If you have enough torque to GET to rapid speed, you almost surely have plenty of reserve to RUN at rapid speed. Where you're FAR more likely to have problems is with highly variable cutting loads, where steppers are at their best.

The servo system with the Gecko servo drivers already has error detection that can send a signal back to the control software to halt operation. In LinuxCNC this is easy to do. I am not sure how easy it is in Mach3 as I haven't played around too much with Mach3, but I would think it would be possible.

It's quite simple.

As for the steppers with closed loop feedback, as was said, it is a matter of cost. I am sure these systems would work fine too, but at more than $300 over the DC servo system, I would opt for the servos again.

I'd have a hard time recommending closed-loop steppers to anyone for a CNC machine. It makes no sense at all.

As for linuxCNC, it CAN close the loop directly back to the software using linear scales for position and rotary encoders for velocity. Not real easy to do from what I have read though. The stuff I have read indicates that to avoid problems, high resolution scales matched to the output of the rotary encoders is needed ($$$$$). However it can be done and the WIKI over at the linuxcnc.org site can has a lot of info on this. I personally don't think this is necessary for most applications. I would like to do it some day simply because I like a challenge and I already plan to buy at least one scale to temporarily mount so that I can map the screws on my G0704. Might as well put it to good use on the 45 when I get it.

Please go back and read what I actually wrote. I said it can't close the loop **back to the trajectory planner**. Without that, you can only detect errors, you CANNOT correct them, because you can't modify velocity and acceleration on-the-fly. Once you detect an unacceptable following error, your ONLY recourse is to hit the big red switch. LinuxCNC simply puts the servo PID loop on the PC, instead of out in the Geckos. They are otherwise functionally identical, though LinuxCNC does support both position and velocity feedback, as do some Mach3 motion controllers, like KFlop. But, you CANNOT say "Whoa! This axis is starting to fall behind, so I better reduce the load so it can get caught up!". THAT requires re-calculating the trajectory on-the-fly, which LinuxCNC cannot do - it's FAR too compute intensive. Barring that, the motors, servo or stepper, have no choice but to either follow the commanded trajectory, or throw a fault and effectively E-Stop. There is no ability to "correct", on-the-fly, a commanded move that exceeds the drives capabilities, no matter what kind of drives or motors you use.
[/QUOTE]

On another note, I do plan to implement following error detection on my G0704. For this size machine I think it is a no-brainer to go with steppers vs servos. Excellent performance can be realized using steppers and the price difference is much larger for a machine this size. I was able to do mine for about $500 cheaper using steppers for the drive system than I could have using servos. This was using all new components but with a cheap server power supply. So with encoders for following error detection I will still be $400 cheaper than a servo drive system. This is enough of a difference that the simpler stepper system is more attractive to me.

Bottom line - In my opinion, it depends on the machine and the cost. Not just whether or not it can be done. It can certainly be done with any system you chose. In my opinion, the comparison of the different systems for a 45 size machine make the DC servo drive system more attractive to me.

Which was my whole point. Either can do the job, *IF* properly designed. Either will fail, if not. There is no "best", there are only trade-offs.

Regards,
Ray L.

[SCzEngrgGroup]

Both KFlop and EMC2 support both methods, however the second is higher performance....

I'm really curious as to your basis for that statement. The KFlop is exceedingly capable and powerful. Are you actually a KFlop user?

Regards,
Ray L.

[wizard]

Why use servos instead of steppers?
I understand that the 2:1 or 3:1 gives you more torque but you could do this with steppers aswell.

Sure you can gear steppers down but normally you don't want to do that.

I have noticed servos could be used for positional feed back however does Mach even support it?

Not directly from what I understand.

Wouldn't a closed loop stepper system with rotary encoder do the same thing as a servo and glass scale setup?

No and frankly a glass scale with a servo system isn't all joy either.

Many rf31 builds control z height by mounting a stepper to the fine feed knob on the head,micro kinetics also does this with their rf45 models opposed to mounting a stepper/servo to the column (most builds here as well as industrial hobbies). Is one of these systems more advantageous
On one hand you get greater travel with the column mounting however we are talking about a mill not a fdm machine no one mills with a 8 " long end mill anyway.

The length of the end mill means nothing in this discussion. If you have ever ran a manual mill you would understand why quill drives are useless. The constant cranking of the knee up and down is a real problem. On a CNC system quill drives would be a big mistake on a a square column machine.

So using the fine feed seems more reasonable unless the ram isn't very sturdy.. this gives about 4.5" of travel. Bridgeport does the z travel this same way on their eztrak mills.

Every "Bridgeport" style machine with a quill drive I've ever touched has been a pain in the a$$. If you do anything more complicated than machining flat plate you do not want a quill drive system. Quill drive systems aren't much more than glorified routers in my mind.

As far as ballscrews go most clones have trapozodial acme, ball screws are prolly a must even double nutted for back lash comp. Is chai good to go to or is there a better player in the game for45 style mills?

There are plenty of US based manufactures.

Noticing micro kinetics 45 XL has hand wheels in x and y travels still are there any problems with this a motor being turned is a generator so if I used a mill like that manually to make a part with the electronics off would this do damage to the electronics.

It could. Consult the drive manufactures documentation.

My experience with both eztraks and causing kondias makes me more comfortable manually edge finding and finding centers of bores. Some operations and making of features are just easier to do manually.

While I tend to agree that sometimes it is easier to manually machine a part, there is very little difference in edge or center finding via a CNC. In fact I'd have to say a CNC system can find centers faster than you can. The caveat here is that the CNC system has an edge finding probe.

I know that the 45 style mills come in many flavors but does anyone know of xy and z drawings out there to at least get ideas from for motor mounting options.

Nothing comes to mind.

[691175002]

I'm really curious as to your basis for that statement. The KFlop is exceedingly capable and powerful. Are you actually a KFlop user?

Regards,
Ray L.

I meant that both KFlop and EMC2 support separate and combined control loops, but combined control is higher performance.

After reading my post again I see that what I wrote was very ambiguous. IMO KFlop provides better tools to properly tune such a control loop which would make them easier to set up compared to EMC2.

[SCzEngrgGroup]

I meant that both KFlop and EMC2 support separate and combined control loops, but combined control is higher performance.

After reading my post again I see that what I wrote was very ambiguous. IMO KFlop provides better tools to properly tune such a control loop which would make them easier to set up compared to EMC2.

That I very much agree with.

Regards,
Ray L.

[wizard]

This is a good discussion.
Many of my questions have been answered.
However this a over exaggerated question of what I'm after.
say I am making a pocket that is 5.500" but my glass scales are indicating my movement is 5.492 the system I am ideally after would continue to cut (no matter how long or through what) till the glass scale read 5.500"
I'm guessing this is called a closed loop system and from what was said above Mach 3 cannot do this nor emc2 so what software and hardware can?

EMC and KFlop can do closed loop.

If you are closing the loop through the glass scales then you would get to your position. At least the position the glass scales indicate. The probable mistake here is believing that the glass scales would correspond to what is seen at the tool.

I know this sounds like an awesome way to destroy and reak havok on a machine ,and your right however if I am going to be watching the operations the whole time it becomes less of an issue. The reason I want this ability is lets say I was a little over confident on some ambitious feeds and speeds that could bog down the motor and reduce accuracy this system corrects and fixes the issue.

Maybe it will or maybe it won't. If you have overloaded the system any number of things outside of what the scales are seeing would throw off your machine. For example I've seen some of the following over the years. Tool posts knocked off lathes, tool posts twisted out of position, vises twisted out of position, tooling bent and a whole bunch of other nasties.

Also what if I have backlash error or non Crisco smooth ways the motor continuing to move until the exact coordinate is reached solves this problem as well.

These conditions may or may not be helped with glass scale feedback. It is far better to correct these problems if you want reliable operation. In fact if you do have backlash or other issues trying to close a servo loop through a glass scale may be more trouble than it is worth. Think about it this way, if the tool pushes the table around and then you get a servo trying to compensate for that movement. Depending upon your dead band and gain in the system you could get some unwanted performance. Often an encoder mounted on a lead screw can lead to better overall performance.