Hi all!
I hate to start this new Post but I really would like to know.
I have been surfing the Net and reading articles about complete kits, but turns out there are non. Or at least I can't find them. Problem is that I have limited funds. I would like to stay under 4 grand.

Please someone "in the know" explain something to me. I'm looking at putting together a 1000 to 1200 in/oz Servo machine. I know about stepper setups but can't quite grasp the Servo setup.
Want List:
3 1000 to 1200 in/oz Servo's with 2000 line Encoders
1 600 in/oz Servo for 4th Axis
4 G340 Gecko Drives
Mach3 Software and Pendant with Feed override.

Here are my Questions:
Are all Servo systems closed loop. From what I read these systems will not "unlike Steppers" lose position at any time. Is this correct?

Will I need a Power supply for each Motor?

What about a control Unit is there anything that work with USB? Or would parallel be better?

I would greatly appreciate any Input.
Thanks.

To answer your servo questions

Servo systems, in this kind of application, are always closed loop.
This is the reason that they don't lose position. The controller is always checking and compensating for the position of the axis.
So, current is being applied to the motor, even when it is in position.
Needless to say, with the cost of servos being much more competitive, I wouldn't use steppers.

The power supply for the servos, really depends on which ones you are using.
In the case of the Gecko drives, they typically all use a common supply.
But in the case of more industrial drives it is common for them to have individual supplies.

I don't know which would be better (USB or Parrallel). But it seems to me that parrallel would be faster because it is not a serial communication.
However, it all depends on how it is being used on a very low level.


Hope this helps.

HI Rally,
I'll try to give in my input from my experience. Servos never loose steps for the reason that the encoders tell to motor when and where to stop, so unless the final destination (as inputted from the computer to the driver cards) has not been reached the motor keeps on going until it does. As regarding to the power supply, as long as the voltage of all motors are all the same, having one common supply is the economical way to go. On the other hand if you have different voltages you iether require a transformer which is tapped to the required voltages or else separate supplies. I prefer the first option which requires a tailor made transformer. Important if you intend to build it yourself consideration is to be taken to compensate step up in voltage when rectifying (1V AC is equivalent to approx 1.414V DC. Power of the transformer is rated in VoltAmpers (VA) not Watts and it's power is to be the total power of the motors alltogether. I'd like to add up another 20% to that in case I would require to hook up anything in the future. Important is to cater for 5, 12 and 24V DC, only if you will be putting in a tailor made transformer. To know what power you require find the max current rating (Ampers) and multiply with the rated Voltage (Volts) of each motor and add them up. Take that value and divide it with 1.414 and you should get the required voltage output of the transformer.
Regarding the paralell port, I can use 2, however it seems that the way to go now adays is USB from what I here and see.
Hope I have been of any help.
regards

Hi Rally,
A servo is by definition a closed loop system. In the case of Mach3 and Geckodrives the loop is closed at the servodrive (G340 in your case). What that means is that Mach3 doesn't really know where the machine is but relies on the servodrive to make it go there. If it ends up to far off target due to overload or whatever the drive shuts down and (optionaly) lets the computer know that something is wrong.

Now, regarding the lost steps.... It is true that underpowered stepmotor systems can lose steps when driven outside of there specification. But to say that a servo system can't lose steps is not entirely true. If for example there is noice on the encoder signal there can be a problem and if the step/dir signals doesn't meet the specification there is exatly the same risk of losing steps as with a step motor.

There are pros and cons to both types of systems. An overloaded step motor stalls (loses steps) but no harm is done to the motor. An overloaded servo motor may burn up if not protected. A step motor is generally more appropriate for low/medium power CNC due to it's torque/speed curve while a high power servo generally needs some kind of reduction (belt drive, gearbox etc) to make use of all the available power. They (the servo system) can alos be more problematic to tune but it doesn't have to be. The servo system can also run away in case of lost encoder signal or shorted outputstage of the drive.

As to your powersupply, if all motors are the same voltage then a single powersupply will be the best/cheapest option but you can use one for each if you like. If you want absoulte max power from your motors you need to carefully select the voltage of your transformer. In that case you need to concider the max voltage of the motor, max voltage of the drive and the PWM modulation of the drive. Some drives have a max PWM modulation of ~85% which means that if you have 100V powersupply the max voltage the motor will "see" is 85V which would be quite a good fit for a 90V motor. On the other hand IF the drive should fail with a shorted output transistor the motor COULD see the full 100V which MAY be more than the absoulte max on the motor and could make permanent damage to it.

All this needs to be concidered when designing a servosystem.

There is a new device, currently undergoing beta-testing called the Smoothstepper. What I've read is that it should act and look just like two ordinary LPT-ports but it interfaces to Mach3 via USB. I don't know much more than that but if you have a look at the Mach3 forum there should be some info there, just search for Smootstepper. There's also the NCPod but again I'm not really up to speed on the status of it.

Hope it helps!
/Henrik Olsson.

don't forget about options including gearing. You can easily reach your required torque range, and encoder range by reduction gearing. Just have to give up some speed. (always give and take) 600in/oz 1000lpr encoders and 2:1 reduction gearing would give you 1200in/oz and 2000lpr on the lead screw.

Some drives have a max PWM modulation of ~85% which means that if you have 100V powersupply the max voltage the motor will "see" is 85V which would be quite a good fit for a 90V motor.

Hi HO,
As you have mentioned regarding the PWM, I have some difficulties of how this effect the motor voltage can you be a bit more elaborative. I am using Rutex drives indicated in the following link and running servo motors at 60V DC ( http://www.rutex.com/us/home/index.htm ), it says 19.5KHz PWM but I really don't know what it means. Can you tell from this information if I actually have 60 Volts across the motor terminals and how to calculate this (if there is any calculation!!)?
Thanks

Hi H.O,
Actually the correct link to the drive that I'm using is the following:
http://www.rutex.com/pdf/R990H.pdf

Thanks

Hi Martin,
Actually I made a little mistake but more about that later....

I can't seem to find info about the max modulation in the Rutex datasheet you linked to. The 19.5kHz number is the switching frequency of the drive. The transistors in the drive works like very fast switches and turns on and off the current to the motor 19500 times per second. The PWM modulation or dutycycle is how long the transisitor is ON relative to how long it is OFF. In other words if the duty cycle is 50% the transistors are ON 50% of the time and OFF 50% time. If you have a 60V supply the average voltage to the motor will be 30V.

The reason most drive can't reach 100% modulation is that the bootstrap capacitors for the highside drivers of the H-bridge needs be refreshed. As I said, I can't find any info on the Rutex drive and I can't seem to remember seeing any info about the Gecko series of servo-drive either. But if you look at the datasheet for the VSD-A (http://www.granitedevices.fi/index.php?q=servo-drive-vsda) from Granite Devices you'll see that it has a max duty cylce of 88%. So with a 100V supply the average voltage seen by the motor would be no more than 88V. If the motor has a voltage constant of 30rpm/V you'll "only" get 30*100*0.88=2640rpm from that motor even though you might think you'd get 3000rpm because you have a 100V powersupply.

If your system works then don't worry about it. If you want to find out you need to ask Rutex what the max PWM modulation or duty cylce is and then multiply your supply voltage by that percentage.

The error in my previous message is that if the absoulte max voltage rating of the motor is 90V then it wouldn't be advisable to use 100V supply even with a drive with 85% max duty cycle. The reason is that even though the average voltage would never be more than 85V the actual voltage applied to the motor during the On-time of the switching cycle IS 100V, or very close to it.

Hope it makes some sense, if not let me know and I'll try again.

/Henrik Olsson.

Most stepper motors are LOW SPEED motors and Servo Motors are HIGH SPEED motors. You will probably need to use gearboxes between the Servo Motor and the whatever you will be driving. If your servo motor is 0 to 3000 RPM and you need a much lower RPM you will also increase your torque by the same ratio as your gear box.

10:1 gear ratio = 10 Tames the Torque.
100:1 gear ratio = 100 times the torque.

Some stepper applications drive directly from the stepper motor, most servo applications run through a gearbox or belt reduction.

The other part of the system to look at is where the encoder is attached. If the encoder is attached to the motor you will have finer resolution, it the encoders are attached to the output you will have lower resolution. With high ratio gearboxes with the encoder mounted to the motor, you may get more encoder counts than the system can handle. In other words - the farther from the load the encoders are the lower the encoder counts you will need for a given resolution.

What max output speed do you really need from your drive?
What max speed are the servo drives?
that gives you some numbers to determine your gear reduction. Once you have your gear reduction you can then determine what torque you really need from your motors.

Hi H.O:
I am using Reliance 690 servos for the X and Y axes of a B'port mill. I have measured the watts used by these motors when making 1/2" deep cuts in mild steel, using almost the entire diameter of the 3/4" roughing end mill. The power used is tens of watts, something like 40 or 50 watts. This is small portion of the power available with these 480 watt motors. They are rated something around 1,200 inch-ozs. While they look like they are the right size for a standard B'port, they are in fact huge.

While i had the power meter (Brand Electronics E-1850) on there i confirmed that the motors pull zero watts when they are on location. That's what servos do, they only draw power when there is a difference between commanded and sensed position. Position is sensed by the encoder and the circuit that spends its time counting up and counting down. For your planning purposes, 10 counts off location brings on modest power and 40 counts off location brings on significant power to restore position.

You mentioned 2000 line encoders which are the same as 8000 count per revolution. This allows extremely fine position control, finer than is needed for milling but might be ok for a cnc jig grinder. And your PC controller can output perhaps 30,000- 50,000 steps / second so you might be stuck with a machine that positions very precisely but has a top speed of ten inches/ minute.

Generally the USB port is by design not a real-time device. The parallel port is. To get real time pulses out of the USB port the pulses are generally stored in an external box which then parses them out on a real time basis. Extra complexity but it can do so faster than the parallel port.

If you are using 2:1 belt and pulley step down then you might want to look at 1,000 or even 500 line encoders, depending on the pitch of your lead screws. Do not set up for too many pulses/ inch! 20,000 - 40,000 per inch is fine for general purpose milling.

Pete

You mentioned 2000 line encoders which are the same as 8000 count per revolution. This allows extremely fine position control, finer than is needed for milling but might be ok for a cnc jig grinder. And your PC controller can output perhaps 30,000- 50,000 steps / second so you might be stuck with a machine that positions very precisely but has a top speed of ten inches/ minute.

Generally the USB port is by design not a real-time device. The parallel port is. To get real time pulses out of the USB port the pulses are generally stored in an external box which then parses them out on a real time basis. Extra complexity but it can do so faster than the parallel port.

Pete

Mach3 can output 100,000 steps per second through the parallel port, and up to 4,000,000 with the soon to be released SmoothStepper which uses USB. www.warp9td.com Projected price is just over $150

You can also use a Gecko G100 with Mach3 to get up to 4,000,000 steps per second via ethernet.

Hi Rally,

I am a machinery importer. I have recently started using the Mach3 software. We have assembled a machine with a 12" x 54" table and DC servo motors & Gecko drives. It works okay but I wanted more. So, I called overseas and found a 750kw AC servo motor with drives. This is about 20 inch pounds STALL not PEAK. The endcoders are 2500 count per rev. Enough torque that it could be direct mounted without belts or reduction. I bought 12 of them. They will be here in 2-3 weeks. I can sell a motor, drive, endcoder & 10' cable for $625. plus shipping. The torque alone was enough for me to go shopping for AC drives, but I was amazed how close the cost is by the time you add the DC motor, drive, endcoder, cable & Power supply up.

Thanks !!
Kirk Brown
Majestic Machinery, Inc.
626-336-2350

Hi Pete,

Just for the record, It wasn't me asking the question or mentioning the 2000 line encoders... but anyway...

I'm very pleased to read your numbers on your E690 servos! What drives are you driving them with and at what voltage? I'm in the process of rebuilding a mill that has quite large Indramat DC servos and I'm having trouble deciding which servodrives to use as I may end up a bit short on current with some of my choises. The motors are rated at 19A and 24A continous, 160V. Reading your numbers tells me I'll probably be fine even though 50W does sound extremely low.

Regarding 0 watts when stationery it may differ between drives but I know that the Geckodrive dithers between adjacent encoder counts which means the motor move back and forth surely it must take power to do so?

There is one advantage in using a higher resolution encoder than what you need for positioning though. Let's oversimplify this a bit and for arguments sake say we need the motor shaft to position in 90° increments. In this case you could use a 2-line encoder producing 4 counts per revolution, right. The downside with this is that the servo wont know that the shaft is out of position untill it's 90 or more degrees out of position. A higher resolution encoder would help make the servo "stiffer" even though we don't actually "need it". As I said, oversimplified but I hope you'll get what I'm aming at.

So, a high(ish) resolution encoder can be combined with a drive with built in step-multiplier to get both "enough" resolution, high speed and with very good control of the motor even though you have a limited pulse-frequency. The downside is that the drive needs to be able to cope with tha frequency of the encodersignals and it gets more sensitive to noice.

/Henrik Olsson.

Thanks for all of the Good Information and Opinions, that defiantly helped out.
There is however something that I would like to add to this discussion. It was suggested that for example a 600in/oz Motor on a 2-1 reduction would produce 1200in/oz of torque
for all practical purposes that is correct but the actual number would be more like 80 or 90 percent depending on the type of reduction used.
Again thanks for all the Input.

Hi Martin,
Actually I made a little mistake but more about that later....

I can't seem to find info about the max modulation in the Rutex datasheet you linked to. The 19.5kHz number is the switching frequency of the drive. The transistors in the drive works like very fast switches and turns on and off the current to the motor 19500 times per second. The PWM modulation or dutycycle is how long the transisitor is ON relative to how long it is OFF. In other words if the duty cycle is 50% the transistors are ON 50% of the time and OFF 50% time. If you have a 60V supply the average voltage to the motor will be 30V.

The reason most drive can't reach 100% modulation is that the bootstrap capacitors for the highside drivers of the H-bridge needs be refreshed. As I said, I can't find any info on the Rutex drive and I can't seem to remember seeing any info about the Gecko series of servo-drive either. But if you look at the datasheet for the VSD-A (http://www.granitedevices.fi/index.php?q=servo-drive-vsda) from Granite Devices you'll see that it has a max duty cylce of 88%. So with a 100V supply the average voltage seen by the motor would be no more than 88V. If the motor has a voltage constant of 30rpm/V you'll "only" get 30*100*0.88=2640rpm from that motor even though you might think you'd get 3000rpm because you have a 100V powersupply.

If your system works then don't worry about it. If you want to find out you need to ask Rutex what the max PWM modulation or duty cylce is and then multiply your supply voltage by that percentage.

The error in my previous message is that if the absoulte max voltage rating of the motor is 90V then it wouldn't be advisable to use 100V supply even with a drive with 85% max duty cycle. The reason is that even though the average voltage would never be more than 85V the actual voltage applied to the motor during the On-time of the switching cycle IS 100V, or very close to it.

Hope it makes some sense, if not let me know and I'll try again.

/Henrik Olsson.

Hi H.O.
It seems to make sense from the little electronics that I know, but the thing that made me say "huppps I missed that one out" is that I am running servo motors at their correct voltage 60V Dc but for some reason or the other one of my axis is sometimes moving slower when jogging to a position and having the motor lagging the atual position thus keeps turning for a few seconds after releasing the joystick until it catches up to the right position. The symptom is like it's a bit overloaded but I'm sure it's not. What do you think it could be?
Regards

have you tried connecting (rewiring) the motor to a different axis and see if the same motor exhibits the same symptom?

if its not loaded, it may be a lag in the system. but a few seconds is quite high. or it could be a deceleration curve if all axes experience the same thing.

I am not familiar with all of the drivers and controllers in the market though I am doing my own projects in servo and stepper.

on Parallel versus USB, USB is definitely much faster. the real problem with Parallel port is timing. Unix based OS has better control of the Parallel port than Windows (close swirl patterns at the top speed of the Parallel port look a little bit better on Unix). the better solution is getting away of the CPU timing and having an external driver which has much better Step Pulse timing (I think SmoothStepper does the same). the best solution is just sending away the position, speed, trajectory, etc to the controller but that needs a different software and controller and are much suited for high speed machining.

most controllers do tend to have 0W (or near 0W) consumption on standstill (a reason why ballscrew on Z-axis must be supported or the servo will experience ripple in the setpoint), unless there's a constant disturbance on the setpoint. current is proportional to torque, no disturbance at setpoint = no torque to compensate for and no current to use.

machining at low speed tend to use very low power compare to the actual capacity of the motor, though high speed machining will tend to use up lots of power.

higher encoder rate = lot more room for precision. you can even make the motor set in 50 line interval and even have good accuracy. it is always harder to tune when you are trying to achieve precision in every step of the encoder. but going too high an encoder rate gives the controller a hard time and the budget unreasonable for some purposes.

on the motors, as long as you keep the windings on its current limit (too much current over the limit on continuous use will affect the magnets on the motor if there are) there's not much problem on voltage, the voltage rating is the uncontrolled voltage where the motor may experience its full speed at which some of the mechanical parts are rated. the nameplate ratings are also for continuous use where temperature rise is acceptable to the design. some motor manufacturers even give data on how their motor performs when you exceed the nameplate ratings. though it is not always advisable to do that. as long as you maintain the average of the motor use to the nameplate ratings it would be fine and the motor would last long enough until you have to change brushes.

have you tried connecting (rewiring) the motor to a different axis and see if the same motor exhibits the same symptom?

if its not loaded, it may be a lag in the system. but a few seconds is quite high. or it could be a deceleration curve if all axes experience the same thing.

I am not familiar with all of the drivers and controllers in the market though I am doing my own projects in servo and stepper.

Hi rdpzycho,
No I have not connected the motor elsewhere, however originally I had a 400oz/in torque motor and replaced it with a 750 oz/in one and still have the same effect. In a way I was thinking the same (i.e. lag in the system) but since I noticed it during jogging mode, I thought to come back to it and observe it when I will do some actual work to see if this reflects the same while on production. I have given it some dry runs and doesn't seem to do it but there again since it is not loaded with the opposing cutting force, I'll have to wait and see.
Regards

Martin,
If you're not trying to jog it near its full speed then the voltage probably isn't the problem. What do you have the following error limit set to on that Rutex drive and roughly how much does the motor move between you letting go of the button and the motor actually reaching position? 180°, 360°, 3600° ?

Hi, H.O.
I'm not quite sure if it's running up to the max limit cause when I increase the jog rate over a certain limit the speed stays the same no matter how high the setting is. On the other hand I have the following setup so that you understand what I'm talking about:
motor speed: 4000rpm
pulley speed ratio:4:1
lead screw pitch: 10mm
This means that if I had to hook up a supply directly onto the motor terminals I should accieve a speed of: (4000*10mm)/4=10 meters/min or 160mm/sec.
Currently the jogging speed is aprox 30 mm/sec, so having said that I presume that in jogging mode the max speed is automatically reduced to the lower levels for safety reasons. As I said I have to do an actual run to verify this.
Regards

Hi H.O.
I forgot to answer your question. I don't have any "following error limit" setting and the ammount traved between letting go of the button till the motor stops is about 50mm (aprox 1800°).
Thanks

Hi Martin,
Something sounds really wrong with your setup. If the motor lags 1800° the servo should stop it. I don't know what the default following error limit on the R990 is but it says it's adjustable between 1 and 20000 - they call it Lock Range. It is the number of encoder counts that motor is allowed to lead or lag commanded position before the drive shuts down.

If you're using Mach3 you hold down the shift-key to jog at the max speed that is set up in the motor tuning dialog. If you don't hold down shift it jogs at setable percentage of the max-speed.

Check your powersupply voltage at the drives terminals. Then check the voltage that is actually applied to the motor when you jog. Also take a look at the PID parameter in case you have something wrong there, like really low P-gain or something.

When the motor finally reaches the commanded position does it stay there or is it easy to turn the motorshaft making it deviate from commanded position? If it is easy to turn it it's probably a tuning thing.

/Henrik Olsson.

HI Henrik,
In fact I am using Mach3. I don't think it's the power supply as I have made sure to have ample power to hook up another motor in case I would want to do some cnc lathe work. I'd more suspect that the PID setting are a bit sluggish as I have had difficulties with setting them up on several attempts. From what I have read others say it's not difficult but for some reason or the other I can seem to find a compromise with PID settings. I have tried many setup proceedures but was not sucessful. Mainly for 2 reasons: A) that after thinking that I have found a good setting and re-test it again and again with the same settings, I get different over shoot values and curves (naturally using the Rutex tune setup software) and B) after starting out with accieving some sort of settings with Kp and Kd, then after gradually adding Ki the other 2 settings (Kp & Kd) require readjusting without having any clue from which to start due to variations described in A. So I have come to a conclusion not to waste any further time with it (unless I find a concrete proceedure which worth a try) and see the result when running a cutting path. For your information I have 500ppr encordes on all motors which give me a resolution of 0.005mm per step. Do you know of any particular setup prceedure?
Regards

Hi Martin,
The only procedure I know of except trial and error is the Ziegler Nichols method but I'm sure there are others. If you do a Google search on that you can read up on it.

If you have 500 line encoders and the motor is 1800° behind that equals 10000 encoder counts. That is A LOT, the Gecko G320/340 drive has a fixed maximum following error of 128 counts before it shuts down and only now and then people report problems with the drive faulting out.

If I've understood correct it's only one axis out of three that behaves like this. If they are setup the same in Mach3 (steps/unit, acceleration and speed) but it doesn't behave the same when you jog it has to be the tuning or a bad drive (not likely). Try changing the drives around and see what happens.

I'm sorry I can't give you any better advice!

the method (Ziegler-Nichols) is an educated guess and actually it is a good one. try searching the net on the method. there are variations on the method.

when tuning, take more than one sample of the curve with one setting. sometimes the curve may vary depending on other disturbances.

my control system mentor can do these things intuitively. maybe he just got used to tuning every once in a while. :)

try rewiring them first so the motor is using another axis, like invert x and y for the moment.