Resolver vs Encoder and Servo Drive selection

J

Thread Starter

Jon Oak

I am working on a single axis and a dual axis servo machines that are related. I am going to use a PC bus based motion card (galil, MEI, etc.) and some pc based control software. I have the option of specifying encoder or resolver but am not sure what the benefits of each technology. Being that these machine will be going into an automtive plant environment being robust both mechanically and electically are key. Cost is also a bit of an issue, but not as important as reliability.

Question 2: Same system as above. What should you look at in a servo drive? currently I am considering the following as the requirements

Digital Drive (no pot. tuning)
High voltage bus (460v Drive)
20 amp continuous 40 amps peak (min)
Communications port to talk to PC

What else should be taking into account. I tend to think of these as somewhat of a commodity item espceailly since i am using a PC motion card). I have seen a large price range $3500-2100 and would like to make sure the lower cost options will be perform well.
 
Jon,
Your questions cannot be taken separately because the type of drive you select (and the cost associated with it) will be related to the feedback options you select as well.

Short and sweet. . . fully digital 460V drive with high performance low cost with least expensive feedback option that is also robust . . . .Here is a combo that will not let you down . . .UniDrive 2403 will provide 25Amps continuous - and published 37.5Amps for a minute, 43.75 for 4 seconds, and we have instrumented it and seen 50 Amps for a second. Motors . . . CSM (MTS Automation) makes a great servo motor with 460VAC rated windings. Order it with a commutating encoder (Tamagawa). These are very robust and we have strung them on cables 300 feet long with no degradation in performance.

If you really want to cut costs . . . same motor + the powerblok from CSM (formerly a SemiPower product) combined with a UMAC with direct
digital commutation . . .you will not find a higher performance combination, however the learning curve is not for the novice motion
engineer. Position, velocity AND current loop are controlled by the motion controller = cheaper drive + better performance. We are doing several Anorad linear motor apps now with the UniDrive and the Powerblok head to head . . . haven't decided which one to put out in the field yet.

YOU SHOULD AVOID ANY 460V rated servo motor smaller than a 4 inch design . . . we have tried Baumuller, CSM, Control Techniques, Baldor and a few others . . . all have failed within a few hundred hours due to insulation breakdown from the PWM waveform. Short cables, long cables,
load reactors . . . it doesn't seem to matter. Use a 3 inch motor and you are toast. This is based on 3 years of field tested applications on
perhaps 30 different motors and it has cost plenty to replace all of them.

Resolvers are the choice for ultimate robustness (however, the digital drive cost is higher with a resolver feedback design). We have several 4, 6 and 8 inch servo motors running in apps that are literally beating the crap out of the motors (vibration, shock loading and heat) and the
commutating encoders are doing fine (oldest application is a set of high speed shears on a glass line shearing streams of molten glass into gobs - running now for over 2 years, this is essentially a 7x24x365 operation).

If you want a packaged deal look at the AB 1394 amps and motors or the products from Indramat, Baumuller has had problems with the higher
voltage servo drives, not too many players in the 460V market with much experience.

Good luck

Ken Brown
Applied Motion Systems, Inc.
 
S

Simon Martin

Hi Jon,

1) Encoder versus Resolver

Encoder is an optical device consisting of a translucent disk, scored with lines and 2 sets of LED and Photo Diode, each time a mark passes between the optical link, it is broken generating a signal, the signals are 90=B0 apart, generating the data for the A/B channels of the encoder. There is usually a Z mark once a rev as well. Usually the encoder is mounted on the back of the
motor.

Resolver is a magnetic device consisting of 2 static coils 90=B0 apart and one rotating one. The rotating coil is fed with a sinusoidal current, and the induced current in the static coils are measured. The relative intensities
of the signals gives you the position. Usually the resolver is part of the motor.

The encoder is used for most applications except the following:

1) Hard radiation (fogs the translucent disk)
2) Very high accelerations (damages the coupling)
3) Very high vibrations (damages the coupling, vibrates the disk)
4) Very high temperatures (damages the disk)
5) Very high resolution (optical resolution)

The resolver is not used in the following cases:

1) Cost factor
2) High speed applications (the lag through the position decoder means loss of accuracy)

I personally would go for a sealed encoder with differential output. MUST BE DIFFERENTIAL to avoid noise problems. Heidenhain is a make that springs to mind.

2) Drive selection

I think you are putting the cart a bit before the horse here. Start from your process, calculate your torque criteria (peak and RMS). If you have a
heavy duty cycle (lots of stops/starts, high accelerations, etc) then I would suggest an AC motor, as a DC motor will have excessive wear on the brushes making maintenance a problem. If the duty cycle is not too bad then go for a DC motor. Don't forget to balance the motor to the reflected
inertia of the mechanics, otherwise tuning the servo will be a pig.

Once you have selected a motor, then given the motor parameters you can calculate the peak and RMS current. Now you can specify a drive.

Just because you are using a PC based controller does not mean you can skip the electro-mechanical basics.


Debian GNU User
Simon Martin
Project Manager
Isys
mailto: [email protected]

There is a chasm of carbon and silicon the software cannot bridge
 
A

Anthony Kerstens

Just an additional tech note about resolvers. The two signals are a sine and cosine waves. Knowing both gives you, with a little math, the angle that the resolver shaft is at.

Anthony Kerstens P.Eng.
 
K

Kaufman, George

Jon:

Why don't you consider a control system using SERCOS (fiber optic connection between the controller and servo drives, see www.sercos.com for more info)? SERCOS will reduce your wiring cost and improve your reliability due to improved noise immunity and elimination of wiring problems.

You can look at pc-based controllers using AML software at www.MotionOnline.com. AML can work with the NT O/S on a PC of your choice or AML can run on the SMS-10 industrial PC with VxWorks. In either case, SERCOS is provided for connection to servo drives. You can also download a Handbook of AC Servo Systems at www.MotionOnline.com that provides a good overview of encoder operation.

One choice for SERCOS servo drives is PacSci at www.pacsci.com. They have a version of drives for 460 volts and SERCOS interface. One reason I mention PacSci is that they use a resolver for feedback but they have a very high resolution (patented approach) resulting in very good performance at a reasonable price. The resolver is more rugged than an optical encoder if this is necessary for your application.

George Kaufman
Automation Intelligence
 
Ken, thank you for the reply, it sounds like you have a lot of 'real world' experience with these types of systems... A few notes and comments of what I have found relating to this issue...

> Your questions cannot be taken separately because the type of drive you select (and the cost associated with it) will be related to the feedback options you select as well.

> Short and sweet. . . fully digital 460V drive with high performance low cost with least expensive feedback option that is also robust . . . .Here is a combo that will not let you down . . .UniDrive 2403 will provide 25Amps continuous - and published 37.5Amps for a minute, 43.75 for 4 seconds, and we have instrumented it and seen 50 Amps for a second. Motors . . . CSM (MTS Automation) makes a great servo motor with 460VAC rated windings. Order it with a commutating encoder (Tamagawa). These are very robust and we have strung them on cables 300 feet long with no degradation in performance.

I have spoke with Control Techiques (the 2403 was on of the systems I was looking at before posted the message, nice to get a little positive re-enforcemant!!) and there is an added expense for the resovler module but is is small in relation ship to the project. I am not sure of the cost difference between the physical resolver and encoder, but i assume it is not that much. I am leaning towards the resolver...

What I am looking at driving is an exlar actuator so motor selection is a given constaint for single axis application. I can choose motor on the second axis.

> If you really want to cut costs . . . same motor + the powerblok from CSM (formerly a SemiPower product) combined with a UMAC with direct

> digital commutation . . .you will not find a higher performance combination, however the learning curve is not for the novice motion

> engineer. Position, velocity AND current loop are controlled by the motion controller = cheaper drive + better performance. We are doing several Anorad linear motor apps now with the UniDrive and the Powerblok head to head . . . haven't decided which one to put out in the field yet.

>
I had a pMac board priced out with a semi-power power block a while ago and it was actually priced higher then the same card with a PacSci drive. This may have been done before CSM aquired them??? Originally the distributer thought this would be a low cost option and has sold it before for lower power systems... It also seem that the pMac boards start at four axis and the price per axis is low but i only need 1-2 axis. I don't think I will need super high performance and would trade for ease of use...


> YOU SHOULD AVOID ANY 460V rated servo motor smaller than a 4 inch design . . . we have tried Baumuller, CSM, Control Techniques, Baldor and a few others . . . all have failed within a few hundred hours due to insulation breakdown from the PWM waveform. Short cables, long cables,

I will make this a requirement for the 2nd axis!

> load reactors . . . it doesn't seem to matter. Use a 3 inch motor and you are toast. This is based on 3 years of field tested applications on

> perhaps 30 different motors and it has cost plenty to replace all of them.

Ouch


> Resolvers are the choice for ultimate robustness (however, the digital drive cost is higher with a resolver feedback design). We have several 4, 6 and 8 inch servo motors running in apps that are literally beating the crap out of the motors (vibration, shock loading and heat) and the

> commutating encoders are doing fine (oldest application is a set of high speed shears on a glass line shearing streams of molten glass into gobs - running now for over 2 years, this is essentially a 7x24x365 operation).



> If you want a packaged deal look at the AB 1394 amps and motors or the products from Indramat, Baumuller has had problems with the higher

> voltage servo drives, not too many players in the 460V market with much experience.

Used the AB 1394 (with AB servo controller) before i think this becomes more competitive the higher the axis count... using 1-2 axis i think it gets pricey... I will double check

Another drive I am looking at is Kollmorgen Servostar... They have a 460 drive at 20 amps continuos (and one that is coming out at 40 amps cont. is this a vapor drive???)... Any experience with this manufacture?

I have a couple more general questions that I am formulating to post in the motion group soon.

Again thank you for your insite.
 
G

Guy H. Looney

Jon,

Here's my 2 cents:

Several years ago, resolver would have been the way to go. However, the advancements in motor design & encoder technology have made the robustness delima a non-factor in automotive industry. Most all articulated robots use encoder feedback w/ great success. Simon is 100% correct when it comes to radiation & heat. The encoder won't hold up in those extreme environments. However, most encoders can handle ambient temperatures of 45 - 50 degrees Celcius (some go as high as 80). Be sure you know what the ambient temperature is going to be because torque/speed curves are based on a certain temperature. Those curves must be derated if the temperature is higher. I would say your environment is a non-factor in this application. Resolution may be a factor. Resolvers are typically 4096 ppr post-quadrature, while encoders can give you much higher resolution. Many standard motors have encoders with resolutions over 8,000 ppr. Both devices can have higher resolutions, but typically encoders have a higher resolution out of the box. Let the resolution you need dictate your choice. I must disagree w/ Simon's assesment of the where to use which device. Most modern servos have the encoder pressed onto the shaft, not coupled (just like a resolver). Therefore, a coupling is not used & does not limit the performance of an encoder in high acceleration applications. Further, I have had a 250,000 ppr encoder mounted to a servo motor to increase the resolution. High resolution is not a problem for encoders. The problem w/ the higher resolution encoders is making sure the motion controller can handle the input frequency. In regards to differential output, most modern servos provide that as a standard product.

Simon hit the nail on the head when he said you should size the system & determine the requirements before you choose the motor & amplifier. You said you are considering an amplifier with 20 amps contiuous & 40 amps peak. Why? Have you already determined the torque & speed requirments of the application? If so, do you know the torque constant of the motor? Your statement implies that you know all of the above. Digital amplifiers are definitely better than analog amplifiers for many reasons. I would recommend that you use a motor & drive from the same manufacturer if at all possible. It greatly reduces the headaches involved w/ trying to determine the torque & speed that is available and the wiring of the two.

One question I have is that you say you want to use a 460V bus. That's kind of an ambiguous statement. 230VAC servo systems operate w/ a 460VDC bus voltage to the motor. Do you mean you want to use 460VAC input? If so, there are many servo manufacturers that make such a product (Siemens, Allen Bradley, Indramat, Kolmorgan, Parker, Yaskawa, Pac Sci, just to name a few). Ken is right about the motor size IF you are talking about 460VAC input power.

Your comment about the motors & drives being a commodity due to the fact that you're using a motion controller is curious to me. All amplifiers are dumb devices. They require some sort of host controller to make them move. There is a varying degree of intelligence in servo amplifiers that give you bells & whistles, but they all require a host. If you're considering MEI or Galil, you'll want the amplifier to run in torque mode (amp closes torque loop, controller closes velocity & position loop) or velocity mode (controller closes position loop). Which mode should you use? I normally always you torque mode so I can set all the gains in the controller because that is what I'm used to. However, velocity mode can work equally well if the amplifier can close the velocity loop at an acceptable rate. Torque mode is perferred if you have to constantly adjust the gains or clamp the torque due to drastically changing loads.

A communications port on the drive will be only used to configure the amplifier. I'm not sure if you've been led to believe it will do more than that. The amplifier must be configured to work w/ the host controller, but after that is done the port will no longer be used at all.

The price range that you've listed is probably based on spec's. Some amplifiers can only remain in the peak regen for 1 second, some can remain for 5 seconds. Some amplifiers offer a step & direction input. Some allow you to scale the feedback that the controller will see. Some have more onboard I/O that others. Some will have a higher bandwidth than others. Some will have more diagnostics than others. Some will be rack mounted (allows you to share a common bus) others will be mounted indidually. Some have regen built in, some don't. Some have dynamic braking others don't. Some have an on-board oscillator. Some have a field bus. You application's performance dictates which spec's you need.

George's suggestion to use SERCOS is valid. AML is an awesome controller, but it is a stand-alone product that will require you to talk to it via Ethernet, serially, or w/ a field bus. I don't think that's what you're asking for. MEI makes a controller that can work with SERCOS (Galil does not) & plug into your ISA or PCI slot. SERCOS does have a lot of advantages, but there's a premium to pay for that. Further, it is my opinion that SERCOS is not for the "weekend warrier". If you're going to be doing a lot of motion control it may be worth investigating, otherwise I'd stick w/ a traditional servo system. Please don't read this as an attack on SERCOS. As I stated, it is a great product, but typically it's used in applications where there are many axes of motion and it does require quite a bit of start-up time.

I know that I haven't given you exactly what you've asked for, but to do that I need to know more about the application. If I can be of further assistance please let me know.

Thanks,

Guy H. Looney
Sales Engineer

Regan Controls, Inc.
475 Metroplex Dr.
Suite 212
Nashville, TN 37211
phone: (615) 333-1940, ext. 322
fax: (615) 333-1941
[email protected]
www.regancontrols.com
 
B

Brandon Ellis

Jon:

I have read your post and its replies. I want to point you to my companies web site:
www.roboticcontrol.com There is a hardware product (e.g. drives, I/O, motors) there that could be of use to you and that will integrate easily with your Galil PC based control through either an ISA or PCI connection kit.

I hope this info. serves you. Let me know if you have any questions.

Regards,

Brandon Ellis
 
Top