Robotic Arm

[happytriger2000]

Hi everyone,
I built a interface for my Mitsubishi RM series Robot Arm, the interface uses 6 rotary encoder + 6 ICs that receive encoder's signal and output to servo motor, so by turnng the encoder's shaft Left or right the servo motor turns respectively. Next step is to record the movement made by the robot arm, can anyone suggest any idea on how to do this?

thanks,

Freddy

 

[Ambient]

sounds like you are looking for a data logger to me. You could probably find plenty of designs using google. There may be some out there that are mostly software, with a PC recording in real time through the serial bus.

What kind of encoders are you using? Hopefully not optical, they can fail a lot. Do the encoders output serial or analog besides the PWM you seem to be using? If analog, you could modify a PC O-scope design. If PWM you could use a uC to change the PWM to serial data, and send to PC.

Hope that gives you some ideas.

 

[SeanL]

Magnetic encoders are best.

 

[happytriger2000]

Interface card.

sounds like you are looking for a data logger to me. You could probably find plenty of designs using google. There may be some out there that are mostly software, with a PC recording in real time through the serial bus.

What kind of encoders are you using? Hopefully not optical, they can fail a lot. Do the encoders output serial or analog besides the PWM you seem to be using? If analog, you could modify a PC O-scope design. If PWM you could use a uC to change the PWM to serial data, and send to PC.

Hope that gives you some ideas.

Hi Ambient,
Thanks for your ideas.
I did more research yesterday, and I found out that PCI I/O card is a possibility, but even better is the Incremental Encoder Interface, do you agree or maybe not?. There was once I thought the device I needed was a data logger, but didnt know what was a data logger at that time.

If I use a data logger then I also need a hardware like Incremental Encoder Interface to work together as 1 controller, right?. Most of the interface card I found does not include software.
I will keep looking for more info., and anyone who has better idea please give some ideas, thanks
Oh by the way Im using a rotary encoder, make = Omron.

Freddy

 

[happytriger2000]

Magnetic encoder

Magnetic encoders are best.

Hi SeanL,
Thanks for the link, is great.
How much do you know about magnetic encoder, and what is different between a magnetic enocoder and a rotary encoder?

Freddy

 

[SeanL]

The magnetic encoders come in some interesting varieties. The ones from Austria Microsystems are very nice, but not cheap. But they will not break down like an optical one will.

Basically, you have a strong button magnet (with N and S poles on each half, instead of top and bottom) which gets mounted on the motor shaft. the IC is then mounted as close as possible to the magnet (not touching, and secure). As the magnet turns, the field will rotate and the encoder will keep track of rotation. Some are 8-bit and have serial, use quadrature outputs, PWM, or analog. They also have incremental. You have many options.

You will not need an interface card at all. You can use a uC and the parallel port.

 

[SeanL]

The magnetic encoder is a rotary encoder. I would try to avoid the optical ones. They use spinning disks and light to see the direction of rotation and output ABI (quadrature) signals. Some may output analog too, but I never bothered to look at more. They start having problems and wear out. They also require a mechanical connection to the rotating shaft, while magnetic does not. Some magnetic encoders also sense when the magnet moves further or closer, and can be used for pushbuttons.

 

[dknguyen]

The magnetic encoder is a rotary encoder. I would try to avoid the optical ones. They use spinning disks and light to see the direction of rotation and output ABI (quadrature) signals. Some may output analog too, but I never bothered to look at more. They start having problems and wear out. They also require a mechanical connection to the rotating shaft, while magnetic does not. Some magnetic encoders also sense when the magnet moves further or closer, and can be used for pushbuttons.

WHy would an optical encoder wear out anymore than a magnetic encoder? Both are non-contact. If you mean FAIL, then yeah, an optical encoder is more delicate (due to things like dust getting isnide and obstructing light but not magnetic fields) but it also offers more repeatable readings with higher resolution when it does work. But failing is not the same thing as wearing out.

One thing to note is that magnet fields from the motor have the possibility of influencing the reading of a magnetic encoder, but not an optical encoder (especially if it is implemented poorly).

And exactly how does a magnetic encoder NOT require a mechanical connection to the rotating shaft while the optical encoder does? Both detect the rotation of the shaft across a air gap (magnetic fields or light) and both definately need something (magnet or striped encoder disc) connected to the rotating shaft so the detector has something to pick up from across the air gap.

What's the reasoning behind those statements? Because they don't make much sense to me.

 

[SeanL]

I should have worded it more carefully, sorry. The mechanical connection I was talking about was the main portion of the sensor itself. The optical encoders (that I have worked with) require the encoder to be mounted to the motor assembly. The magnetic encoders just need the magnet mounted to the rotating parts. So overall they are easier. And if the optical encoder is not mounted perfectly, the shaft can be under stress. We have seen the bearings wear and blow apart before.

The optical encoders we use pretty much looked like small motors. The disk was internal, with a shaft protruding out to spin the disk. I imagine the larger encoders are more robust. But we needed to use very little space inside the arms, as I imagine the OP wants as well.

The magnetic encoders I mentioned will give 4096 divisions per rotation. If mounted on the arm, you will have that many divisions. But mounting on the motor will give you the 4096*gearing. 4096 should be high enough, no?

We are probably speaking about entirely different encoders, dknguyen.

 

[dknguyen]

Oh, I see. You are referring to using any old external magnet on the rotating shaft and just sticking the magnetic encoder nearby. While on the optical encoder the striped disc and the rest, because of the dust and all, are enclosed in a single housing and therefore integrated bearing is needed to support the shaft and link the disc to shaft (even though the sensor itself doesn't wear out).

Do the magnetic encoder just give polarity readings (which would only give 180 degree resolution)? Or do they make a bunch of tiny magnets (which seems pointless and very time consuming and expensive). Or do they measure the magnituded and determine position from there?

Well, 4096 is pretty high resolution. But I think the optical encoder can still achieve higher resolution since I think you can more easily make the stripes on the encoder disc smaller than you can accurately measure the field strength of the magnet (and it's a bit harder to compare the two because the magnetic encoders' resolution and accuracy are not the same. It's easy on the optical encoder disc because it's digital and the accuracy = resolution). I just know the at least one of the US Digital encoders have two version- one is 1024 and one is 4096 steps. The note says they have the same accuracy (even though the 4096 step one has higher resolution at the expense of sampling time).

One thing is for sure...absolute magnetic encoders (different from incremental encoders which provide relative position readings and not absolute) seem to be much cheaper than absolute optical encoders. Which probably completely tip the scales in favour of the magnetic encoders for something like an arm.

 

[SeanL]

Oh, and as for the magnetic fields from the motor affecting the readings: No problems seen yet. I have been testing that companies newest IC with the magnets they provide. Even mounted about 1cm from the coil of a 80cm wide 3-phase motor, the magnet they provide is so much stronger that there are no problems. Without the magnet present the sensor picks up a lot of noise, even with no motor present. There may be motors out there that can interfere, though. In that case it can be mounted somewhere else, or further away.

 

[SeanL]

Ordinary magnets will not work, these are polarized with half the disk S, half N

 

[SeanL]

At the moment I am using the quadrature magnetic encoders. The other engineers are using the 4096 step. I am using the 360 count ABI output to replace the optical ones.

The magnetic encoders this company has (with ABI quadrature outputs) indicate which direction the shaft is rotating (A leads B one way, B leads other). You need to keep track of the position in software. At the very beginning, the robot automatically calibrates. But you can (if you use the serial versions) set the zero point. You can also have the encoder watch the strength of the magnetic field, like I mentioned before. That enables you to poll the chip to find out if the magnetic field is stronger than neutral (make a handy push button).

 

[happytriger2000]

Hi all,
More research on robotic Arm interface, check out Labview. US digital PCI-3E or USB 1.

Freddy