[hive8]

I haven't bought the Parts to convert my G0704 yet before i ask a last question here about to go with Steppers or Servos...

I was going to go with the new keiling digital drivers based on the HOSS Heavy Digital Package...

Please can someone recommend for me if servos would be a lot better and why...

I have no problem in using EMC2 if that is the only reason... But i think i can also use Steppers with EMC2, is that right?

[SCzEngrgGroup]

If you design the system properly (properly size the motors, use compatible drivers, size the power supply properly - both voltage and current) there should be *no* difference in performance. But servos will cost you more, as the motors will be more expensive, the drivers will be more expensive, you will have the added cost of encoders, you will have to use belt reducers, and you'll experience the joys of tuning a PID loop. Once you eliminate all the mis-information about how much better servos are, and how unreliable steppers are, on a benchtop machine steppers make a lot of sense.

Regards,
Ray L.

[Mad Welder]

I haven't bought the Parts to convert my G0704 yet before i ask a last question here about to go with Steppers or Servos...?

You're getting good advice here from Ray and he definitely won’t see you wrong he’s given loads of us here excellent advice and help too….but as you mentioned the Digital drivers I’m using them in conjunction with Steppers on an Optimum BF20, similar build as yours and I’m pretty darn happy and as for the encoders Starleper1 is going to include an encoder modification on his build http://www.cnczone.com/forums/1084076-post328.html

[hive8]

So i made my decision going to use the Keiling Digital Stepper Drivers 5060D... now its up to the software since there are two options LinuxCNC (former EMC2) or Mach3, i have to say Mach 3 has a huge community which helps a lot... But LinuxCNC is free and i would safe about $149 (Software) and $189 SmoothStepper... but how well does LinuxCNC work with steppers?

[Connor]

So i made my decision going to use the Keiling Digital Stepper Drivers 5060D... now its up to the software since there are two options LinuxCNC (former EMC2) or Mach3, i have to say Mach 3 has a huge community which helps a lot... But LinuxCNC is free and i would safe about $149 (Software) and $189 SmoothStepper... but how well does LinuxCNC work with steppers?

EMC can use the Parallel port to generate the pulses to driver the drivers, or you can use MESA boards which are FPGA which can do it as well. FPGA offload the main CPU of the work, so work really well for machines which aren't up to the task of generating the pulses. I'm using EMC and it works great with 5056's (non-digital) and just finished phase 1 conversion. Computer wise, I'm running a Dual Core 1.8Ghz Intel Atom Mini-ITX computer. As for support, their is large community support for EMC as well, just not as much on cnczone.com. They have a IRC channel on freenote.net #linuxcnc

[hive8]

Thank you for this information, so what i will do is go first with the linuxCNC solution to see how it pans out... if not then i will do some MACH 3 work, its good to have options...

[SwampDonkey]

If you design the system properly (properly size the motors, use compatible drivers, size the power supply properly - both voltage and current) there should be *no* difference in performance. But servos will cost you more, as the motors will be more expensive, the drivers will be more expensive, you will have the added cost of encoders, you will have to use belt reducers, and you'll experience the joys of tuning a PID loop. Once you eliminate all the mis-information about how much better servos are, and how unreliable steppers are, on a benchtop machine steppers make a lot of sense.

Regards,
Ray L.

Great info Ray. Ive noticed that most (if not all) large industrial CNC machines run servos. Any reason they would be chosen over steppers on high-end machines?

[691175002]

It is kind of hard to properly convey the difference between the two systems since it comes down to a lot of subtleties about torque curves and control systems.

Some basic reasons why an expensive machine would prefer servos are reliability,speed, size and efficiency.


While a properly setup machine is never going to lose steps, stuff happens, and in large commercial machines that may be running unsupervised the consequences are far greater. A servo will also fault if you crash the machine whereas a stepper will be none the wiser.

Servos have a higher upper rpm limit. Servos scale up better to large sizes, and servos draw no current if there is no torque required (whereas a stepper will generally draw 100% or 50% of its rated current when holding a position).


Finally, a properly tuned servo is almost guaranteed to have higher positional accuracy, especially under load. We are talking a very small difference here, but a stepper that has no feedback has no idea how its axis is being loaded. Your Z stepper applies the exact same changes in current to move upwards (against gravity) as it does downwards, which will make its upwards movements react a tiny bit slower than its downwards movements.

A servo with a PID (or even more sophisticated control loop) will detect the load and automatically apply more power when moving upwards. Same deal with cutting forces.
The amount of error a stepper will have compared to a servo is essentially zero in a small mill but when you are selling a $50K+ mill things are different.

[SwampDonkey]

It is kind of hard to properly convey the difference between the two systems since it comes down to a lot of subtleties about torque curves and control systems.

Some basic reasons why an expensive machine would prefer servos are reliability,speed, size and efficiency.


While a properly setup machine is never going to lose steps, stuff happens, and in large commercial machines that may be running unsupervised the consequences are far greater. A servo will also fault if you crash the machine whereas a stepper will be none the wiser.

Servos have a higher upper rpm limit. Servos scale up better to large sizes, and servos draw no current if there is no torque required (whereas a stepper will generally draw 100% or 50% of its rated current when holding a position).


Finally, a properly tuned servo is almost guaranteed to have higher positional accuracy, especially under load. We are talking a very small difference here, but a stepper that has no feedback has no idea how its axis is being loaded. Your Z stepper applies the exact same changes in current to move upwards (against gravity) as it does downwards, which will make its upwards movements react a tiny bit slower than its downwards movements.

A servo with a PID (or even more sophisticated control loop) will detect the load and automatically apply more power when moving upwards. Same deal with cutting forces.
The amount of error a stepper will have compared to a servo is essentially zero in a small mill but when you are selling a $50K+ mill things are different.

Excellent explanation. Thank you.
So it seems the benefits of steppers are lower cost and more manageable complexity for the small machine hobbyist? Im aware of a guy who produces cnc'd circuit boards for RC and robotics applications in production runs with a stepper controlled x2.
Servos sound kinda awesome though....

[SCzEngrgGroup]

Great info Ray. Ive noticed that most (if not all) large industrial CNC machines run servos. Any reason they would be chosen over steppers on high-end machines?

It primarily has to do with power. Large steppers perform very poorly, and at a certain point, steppers become completely impractical. So, large machines will always use servos, because they can do the job, and steppers can't. That is not the case with benchtop machines. Steppers are MORE than capable of providing essentially the same performance, at lower cost, on smaller machines. Some may argue it has to do with positional accuracy or speed, but that is a red herring. On a benchtop machine, the primary sources of accuracy error are the lack of rigidity of the machine itself, and poor accuracy and fit of the ways and gibs, rather than the drive system. And a properly designed stepper system is MORE than capable of providing all the speed a benchtop machine can realistically make use of. As in all things, many people believe that more is always better, but it just ain't so. You can build a benchtop machine that runs 500 IPM rapids, but that is of ZERO practical value, and will not save you more than a few seconds of machining time in hours of work, if that.

Regards,
Ray L.

[hive8]

I am little torn now again i see the difference on the servo and the stepper one advantage i can really see in the servo is that it knows where it is all the time not like a stepper where it doesn't know hence closed loop, if i look at the difference of the system i want to use between the Hoss Heavy Digital Package and the Servo System the price difference is only a few hundred dollars which is not that much, what you guys think i should be using if i have the money for a servo system and dont really care about a few hundred dollars...

I am looking to get max between 250 and 300ipm rapids not looking to get 500ipm, more important for me is position accuracy and cut accuracy, i know it has also to do with the ways and so on, but if i get it right the servo is just on is just on part in the equation to get the errors down... or would a stepper be really more then enough.

On a other note is there a system that has feedback hence closed loop for a stepper system?

[691175002]

On a other note is there a system that has feedback hence closed loop for a stepper system?

Ray is entirely correct that there will be no accuracy improvement in the type of machines we are talking about.


You can create a closed loop system with steppers ( Closed Loop Stepper ), but you essentially get the worst of both worlds.

You end up paying for encoders (and individually they cost way more than they do with the motors) but you don't get many advantages because if you lose a step your stepper driver has also lost commutation. The system will try to correct but it will take a long time and overshoot (because you can only use the integral term in the control loop).

Keep in mind the marginal cost of a servo system might be higher than you assume because you will need timing belts (while steppers can direct drive).


This is a hobby though so in the end you can do whatever you want. I am planning on using dmm tech servos + linear encoders just because I think it will be cool and I like control theory.

[SCzEngrgGroup]

I am little torn now again i see the difference on the servo and the stepper one advantage i can really see in the servo is that it knows where it is all the time not like a stepper where it doesn't know hence closed loop, if i look at the difference of the system i want to use between the Hoss Heavy Digital Package and the Servo System the price difference is only a few hundred dollars which is not that much, what you guys think i should be using if i have the money for a servo system and dont really care about a few hundred dollars...

I am looking to get max between 250 and 300ipm rapids not looking to get 500ipm, more important for me is position accuracy and cut accuracy, i know it has also to do with the ways and so on, but if i get it right the servo is just on is just on part in the equation to get the errors down... or would a stepper be really more then enough.

On a other note is there a system that has feedback hence closed loop for a stepper system?

Go back and re-read my first post....

[hive8]

So do you think that with the hoss heavy duty digital package i be pretty set (I am pretty sure with Hoss well of knowledge he did a great job of putting the right components together)? Here is the link to it G0704 Electronics

Please let me know your thoughts...

Btw looking for like 0.01 to 0.02mm position and 0.025 to 0.05 cutting accuracy on aluminum (hope i am not dreaming here).

[Connor]

So do you think that with the hoss heavy duty digital package i be pretty set (I am pretty sure with Hoss well of knowledge he did a great job of putting the right components together)? Here is the link to it G0704 Electronics

Please let me know your thoughts...

I don't even think you need the digital package. I got the 3 x 570oz-in steppers, and the 5056's and I'm able to push around 100IPM on the X, 60IPM or so on the Y, and I'm still tweaking the Z, but I had it up to 80IPM a little while ago. All running 1/8 stepping. This is Phase 1 conversion, using STOCK screws--Direct Drive. No ball screws. I don't think you need anything more than that package. I don't even think you need the 960oz for the Z either. What are you planning on using this machine to do? It's good for Aluminum and steel.. Not sure I would try stainless on it.. I think servo's are WAY overkill and WAY more complicated than you need for this machine. Save your money and put that toward tooling and other cool things like a auto tool changer, 4th and 5th axis etc etc..

[hive8]

I will be using it mainly for aluminum and brass, some pastic... and some steel later down the road. I will be making mainly laser modules with it as seen in my post about building the machine, also some turbine designs that i have and been working on... Some smaller molds for injection molding.

Down the road i would like to make some watch parts if the accuracy permits...

[691175002]

On my current machine I have 570 steppers with digital drivers on the x and y and a regular driver on the z.

The digital driver is worth it in my opinion, you can clearly see and hear the difference in smoothness and they go faster as well.

Btw looking for like 0.001 to 0.0005 position and cutting accuracy on aluminum (hope i am not dreaming here).

It is extremely difficult to quantify a mills accuracy because there are so many ways it can influence a part. You are likely dreaming if you want 0.001" accuracy on a g0704 over more than a few inches though.

Just to give you an idea of how difficult 0.001" accuracy across a table is, the c7 ballscrews you will be buying are rated for 0.002" error per foot which can take any distribution. You will also find that almost everything on a stock g0704 will move a thou or so under even light pressure.

You can see a sample tormach inspection sheet here: http://www.tormach.com/uploads/18/TD...0112A-pdf.html
It shows the kind of tests that are made and the kind of errors that are expected. The tormach is an extremely nice machine as well, you aren't going to match it by sticking steppers on a stock g0704.

[Flexo!]

Btw looking for like 0.01 to 0.02mm position and 0.025 to 0.05 cutting accuracy on aluminum (hope i am not dreaming here).

Thats a little on the tight side, but not by much! I've seen my (WMD25/G0704 etc) mill get down to 0.005 backlash - with double nut ball screws, and the jibs loose however.

Realistically once you tighten the jibs to remove slop, the added friction drives that number to 0.035mm+. Everything starts flexing. No amount of precision ball nuts or pre-loading will save you!

With everything considered (flex, spindle run out, crap dovetails etc), I can usually make parts to 0.05mm in steel or alu - the material only changes the speed and amount you cut, not necessarily accuracy.

To get better, you need linear rails and/or a LOT more iron.

[lcvette]

alot of your tolerance will depend on tuning the steps per inch and what level quality of indicator you use for tuning it in. I have been using one that goes to .001" on the marks and i have been able to hold right near .001" on aluminum. the backlash is to blame for the lack of accuracy and of course machine flex.

I love my digital drivers from Keling, they are super quiet running the 570 motors. if you plan on running a cnc'd g0704 in a production envoronment thought, i would be a bit wary. first of all, the motor is going to need to be upgraded if it is anything like mine, i went through 2 factory motors before doing the belt drive and treadmill motor conversion. in retrospect i would have opted for a servos spindle motor but the cost would be significantly higher.

Steppers for the G0704 make the most sense. I have run several hour programs with only a few issues and those were caused by me trying to cut past the machine movement travel and stalling the steppers. the Y travel in stock form leaves alot to be desired. the spacer mode and travel increases while doing the ballscrew conversion is a must in my opinion, i now have 10" of Y travel and let me tell you, what a HUGE difference that 3.5" inches make.

If you do decide to get servos, I would get them large enough to run the next size up mill because you would always be able to upgrade mills if your requirements change with a lower initial cost. I wish i had done that in retrospect (with Steppers not servos).

Will you be doing ball screws in the beginning? if not beware the backlash on the stock screws is awful. I did the C7 screw upgrade from linear motion bearings, Fred Chai was great to work with and the price was ridiculously low for the quality of screws that showed up. if you do get them request extra balls, they apparantly show up missing a few from factory assembly. my reports on backlash are right at .001" with them and i am sure that is not in the screw but somewhere in the mill.

hope that was helpful and not too confusing.

[SCzEngrgGroup]

Btw looking for like 0.01 to 0.02mm position and 0.025 to 0.05 cutting accuracy on aluminum (hope i am not dreaming here).

Unfortunately, you're dreaming. The lead error in your ballscrews alone will far exceed that (typically 0.003"/foot, or 0.03mm/100mm). Flex in the machine will also exceed those numbers, as will backlash. A temperature change of a few degrees will cause the size of the machine itself to change more than that due to thermal expansion. Odds are, spindle run-out will also take you outside that range. Add all those errors up, and you're not even close, even over short distances.

Regards,
Ray L.