Any electronics engineers in the house? Linear encoder repair questions.
 

I need help with the read head of an obsoleted linear encoder. We believe a portion of it is damaged and need to know if it can be repaired. I know it's a long shot but we're trying everything over here. If this makes sense to anyone I'll go into the details. Thanks.

Re: Any electronics engineers in the house? Linear encoder repair questions.
 

I have a paper that says "BSEE" but most of my work is in the RF field. Depending on what you have/need I might be able to help.

Well, it's 50" long linear encoder with a "read head" that travels along it and lets the servo drive basically know where it is(via magnetic pulse). There is a .004" air gap between the read head and the 50" scale that it reads from and debris between the two has eaten into the read head and it no longer works correctly. I am wonderingif these heads can be repaired or not. Or if diagnostic testing can be done to see what is actually wrong with the head.

We have a new updated encoder/scale on the way but it is 12-14 days out, so I'm looking to limp this thing until then if possible.

Can you get a new/used read head for your model?

Nope, they are obsolete. I even badgered the mfg. into telling me who he sold the last few to and I called them in hopes that the still had them but no dice.

FWIW the read head consists of three stacked boards and I believe them to be fine. Those boards plug into the actual magnetic reading head that is encased in a hard epoxy via eight pins. My wondering is if there is a likely culprit in these that I could try to repair. Or even a way to diagnose via the pins as to what's actually wrong with it. The mfg. is ZERO help... which is nice.

I know this is highly specialized stuff but hell, who knows it's worth a try at this point.

Is the sensing tip extended from the rest of the assemb or is it flush? Also, can the mount spacing be adjusted?

I once worked on a similar system that counted spacing on an assembly line track, the exposed tip became damaged from debris. We re-ground the tip to the proper shape then adjusted the mount so the spacing to the track was correct. Not sure it will work for your situation, but I thought I'd throw it out there.

-Nick

::edit:: Sounds like this wont work. Your post came through as I was typing:p

Thanks Nick, this does not have an exposed area. It is encased in clear epoxy and all you can see are two metallic pads with three brass pins on each side (again encased in epoxy) We have been playing with the spacing so you're right on with that idea. The funny thing is that it will read the scale smoothly but will not find the registration mark (when you "home" the drive it needs to find this reference point so it knows where zero is). I have double checked all connections and that has me believing that by wearing through that epoxy we have damaged or contaminated something.

Lendaddy, I am going to be out of town tomorrow, but Friday I will check with our shop to see if they have any advise as I seem to recall similar issues in our machine shop in the past and we have some pretty sharp troubleshooters in our shop. We may even have some old encoders laying around. Is this off a milling machine? Just off the top of my head, but it seems if it reads movement, but will not home, maybe the problem is not in the actual scale unit?

Is there a way to use an external "home" switch and send the signal to where the read head home switch is connected? (bypassing the read head home switch)

Automation Direct sells all kinds of limit switches that aren't too terribly expensive. I was thinking something like this:

http://web2.automationdirect.com/adc/Shopping/Catalog/Sensors_-z-_Encoders/Double-Insulated_(PBT)_Limit_Switches/One-way_Lever_with_Polyamide_Roller_Actuator

Would this happen to a Balluff or MTS unit?

Len, this doesn't sound like a problem with the reader head being worn out, especially if it is functioning properly except for the homing. I would guess that there is a separate electrical path that is supposed to trigger when the registration mark is found. I would suspect a broken solder connection on a pin.

My guess is that the sensor is a hall effect. This would produce a small signal pulse when the reader passes over the reference mark. This needs to be amplified for processing. All of this would be the same for the other read process, I suspect. My guess is that there is some special output line that is reserved for the zero reference. This would be on one of those three boards and feedback to the machine.

You don't happen to have any schematics? It would be possible to suggest some things to check if you have this.

Thanks Tim, it's actually on that huge tube shear I showed you. The application is VERY specific as it must work with the Unico drive, which requires a specific type of input signal. I can buy read heads and encoders all day, but they are either absolute or multiple ref. or don't produce the same output signal, etc.....

I really need to hope for a fix on my current head. Maybe I'll get some pics up this A.M.

Your thinking like us:D We called Unico and asked if there was a way to trick the drive by supplting a pulse at the right moment (it's not that crfitical really) and they claim there is not...but I'm with you, it's looking for a signal of some type so why can't we just manufacture it? I'll call them again today. Thanks

And we use Automation Direct all the time, great company!

No, it's a Dynapar but Unico does recommend the MTS unit as a replacement so you've got the right idea.

We're on the same page. I have the ref. line located in the control box and get continuity all the way back to the read - head connector (I will check from the read-head board this A.M.). I have checked all the lines but I agree it seems like a connection thing. I don't have a schematic, and like I said the mfg. is useless but I can try again and hope for a different idiot. Thanks.

I took the cover off and checked for continuity from the board to the control box and it's good. We are checking with unico on creating a signal via a device and trigger to "fool" it but the guys here think that will be expensive and difficult...we'll see.

Here is a brief on the read-head. It is the non-guide quatrature TTL type.

http://forums.pelicanparts.com/uploa...1176387578.jpg

From this I don't think it is a problem on the sensor head. Those 8 pins you mentioned would be the 4 different coil used to pick up the changes in magnetic field. They are what generates the A & B (also C & D exactly opposite of A & B) signals. I think that you have a problem somewhere on the circuit boards.

Because this is a TTL system, they are doing some processing before feeding an output to the motor. It should be very easy to "fake" the signal depending on the timings. It did not show from this picture the frequency of these signals. I could calculate it from the 025 code shown if I knew the speed of the head. (Is this fixed or changing?)

For example, the 025 code is 25cycles/mm. If the speed was 1mm/sec (really slow), then the frequency would be 25cycles/sec or 25 hertz. All the other signals seem to be a phase shift (time delay) of the same frequency. Then if you knew the pattern that the motor is looking for when the registration mark is found, you could recreate this using a simple micro-controller.

Good luck. I hope that you can get this working. I wish I was closer and could lend more of a hand. I love trying to engineer a workaround.

Interesting, would there be a way to test those four coils via the eight pins just to be certain?

Also, the speed does vary.

Would the surface of the coil getting exposed and possibly "shorted" via debris possibly cause damage to the board? Thanks!

A few pics might help.

http://forums.pelicanparts.com/uploa...1176398376.jpg
http://forums.pelicanparts.com/uploa...1176398406.jpg
http://forums.pelicanparts.com/uploa...1176398429.jpg
http://forums.pelicanparts.com/uploa...1176398451.jpg

And a picture of the other half...

This is a similar setup wih both parts shown:

http://rsf.reachlocal.net/bilder/off...0_b1w_0204.jpg

And the track that provides the quadrature pulses is undamaged?

Can you replace the linear pulse train with one of dynapar's rotary units, driven from a drive shaft or motor on the machine? Or add ona small rack and mount a pinion gear on an encoder? that's how the old boring mill displays worked.

You could test them by measuring the resistance of each coil from one end to the other. If three of them measure very similar and one is way off, you might have found something. The hard part is going to be finding out which two pins are the ends of the same coil.

A better measurement would be to measure the inductance of each coil, but this would require special equipment.

I looked up a similar system and they are using a 10kHz magnetic field. If yours is the same, those coils should have a capacitor on the PCB that is used to create resonance at the 10kHz. This allows them to be more "sensitive" to small signals at that frequency. If one of these was damaged, it could cause the signal to be missed. This only make sense if one of the coils is devoted or on used for the registration mark.

Could you take a shot of the PCB where the connectors attach straight on? Are there any components on the top side of that board?

The scale or "track" is fine we believe. We have two of them and they act the same. There are multiple problems with a rotory in the output and reference mark departments.

Here are both sides of all three boards. They come apart on pins:

http://forums.pelicanparts.com/uploa...1176405216.jpg
http://forums.pelicanparts.com/uploa...1176405240.jpg
http://forums.pelicanparts.com/uploa...1176405268.jpg

Do you have a scope or digital probe?

No, so I'm SOL best I can guess. Like I said earlier, this is a long shot.

I assumed by digital proobe you did not mean a multimeter, I do have a nice multimeter.

This:

http://forums.pelicanparts.com/uploa...1176407254.jpg

Not this:

http://forums.pelicanparts.com/uploa...1176407274.jpg

http://forums.pelicanparts.com/uploa...1176406899.jpg

I suspect that these are the trim capacitors for the coils. This is to tune the resonant frequency like I mentioned. There are more on the other board. I am not sure how this play in.

I do not see any visible damage to the PCB or circuits showing. Without a schematic, I can only speculate on the circuits. Also without an oscilloscope to look at time-based signals this could be very difficult to figure out.

I think that maybe your best bet is to try to "hijack" the signal (that should be) sent to the motor when the registration mark is found. Do you know if one of these dedicated for that purpose?
http://forums.pelicanparts.com/uploa...1176407292.jpg

Yes, the pink and grey leads are labled "ref" but I have no idea how to introduce the faked signal nor even what it would be.

The hard part is to find out what it should be. I would think these unico guys would be able to tell you this. I suspect it is just a high or low. You could do that with a switch. If it is a more complicated signal, it could still be done but you might have to use a micro-controller.

Let me know if you get any information from them. I am intrigued now.

Thanks, will do. We have been tracking down the last 11 of these sold back in '03-'04 and we are hoping to locate one tomorrow in overstock but they are closed for the day. I have to be honest in saying I hope I don't have to fix this one:D