Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Need to clean Optical high resolution encoder. Methods

Status
Not open for further replies.

fastline

Member
Hoping someone around here has experience in this? We are working on about a 2000 line encoder on a CNC machine that is giving us fits. i confirmed the wiring and power to it is just fine. So I removed it, gave it a little air clean and tested it and was able to fire the machine without errors and was able to jog it a little bit before it errored again. For this reason and my visual inspection, I believe the glass disc just needs cleaned. However, i probably only have about .100" in each side of the disc to get in there with something. Not sure if just alcohol will work? Concerned about streaks.

I won't be able to remove the glass disc without a complete teardown and risk because a bearing is somehow attached and all pressed together. hard to explain.

What I believe likely happened is machine was powered down for a while subject to some colder temps and condensation occurred, which did not damage the encoder, but over time has left a haze on the glass that is very obvious. I think it just spins to a bad spot and cannot see through the fog so throws an alarm.

Willing to order special materials for this repair but rather get some stuff locally and get it back together so it's protected.
 
You can get optical cleaner from local opticians, or failing that isopropyl alcohol does the job.
Max.
 
If it is really a laser etched glass, isopropyl alcohol is ok. If it is black ink on acrylic, you risk whipping the black ink off of the part and lose your opaque tick marks. I would use distilled water, q-tips and/or thin strips of microfiber cloth (low lint) as sold at optical shops (Costco).
 
I have an old strait one that the dust / light seal strips had degraded, letting light in.
If yours was in one position for a long time, the rubber / vinyl may be distorted there.
I had to make new strips.
 
Yes, the micro fibre cloth you can also get at Opticians.
Also there is a small piece of grating, the same resolution as the scale mounted to the receiving components (LEDs).
(Used to produce the Moiré effect).
Most scales of that resolution in CNC applications are photo-etched on glass.
Max.
 
Last edited:
I won't be able to remove the glass disc without a complete teardown and risk because a bearing is somehow attached and all pressed together. hard to explain.
i've seen the same thing in other machinery. use water and a piece of lint-free cloth to remove water soluble residue, then use alcohol on lint free cloth to remove any streaking.
 
Well, we are going to give it a test. Got it under the microscope and you could see a layer of particles over the entire disk. I used medical gauze and isopropyl. Under the micro, you can certainly see imperfections in the glass and appears made that way. There is just no way to clean the reader head without full disassembly so I will try this first I guess. There are multiple bands of segments on the glass to offer redundancy I would assume.
 
6 years I worked on encoders and kept all the factory machines running. Sometimes encoders can give you crazy problems and its not really the encoder. If encoder is belt drive sometimes belts stretch or wear enough to cause trouble. Machines usually have a ZERO starting point where you need to set the machine manually. Move your machine to the starting point which will be ZERO for the encoder. Adjust encoder to zero tighten belt and make sure encoder is still on zero, then see if the machine will run. Replace old belt with new belt test machine again. If encoder has no belt drive check the coupling. Check for bolts that are no longer tight. Check encoder cable. Swap cables if you have a spare. Swap encoders if you have a spare. Encoders are usually sealed up tight so nothing can get on the glass. I see you already tried blowing air on your encoder I hope you used canned air not air from air compressor hose. You can't use a Q-tip it might leave cotton pieces on the glass. What ever you clean glass with it can not leave residue on the glass. I always used 100% methyl alcohol in 1 quart can. Get a small diameter soda straw put it in the alcohol about 1" deep hold your finger on top end of straw so air is trapped then remove straw from can alcohol will not run out. Release alcohol onto the glass disc to wash it and let alcohol run off the side of glass disc. It is also possible the reader has a problem and glass disc is not the real problem. Make sure all alcohol inside encoder body evaporates before you put it back together. Call the factory service department ask for help they will talk you through the trouble shooting process.
 
Last edited:
First test was a fail. It is better but still will alarm after a little motion. This is not a belt drive. Every encoder is direct coupled with a plastic slotted key to the servo shaft. To complicate further, machine has factory installed upgrade linear encoders for true motion feedback so the servo encoders are only used to provide motor performance feedback.

I am unclear as to which part of the optics may now be of issue. I doubt I did a flawless job cleaning the glass but I have also seen people in the field clean these with MUCH less care than I used. I cannot tell which side is the light source and which is the "receiver" side of the device and which would give more issue or area of concern? The main PCB sits above the glass disc, then there is a small 4pin ribbon that connects to a flat component that runs extremely close to the disc. I know two of the 4 pins are identified as "Vss" and "Vcc".

It will be impossible to clean that device without full disassembly and that presents a challenge in that the disc seems to be glued to a shaft and that shaft goes through the housing and has the drive lug mounted to it. I know it will come apart but I need to get on the level of how it gets done.
 
You need to be careful when dismantling or removing the head as some are pre-aligned and some automatically align when mounting, if this is a horizontal scale
The critical side of the read head is not so much the emitters but the reading side which carries the very small Moiré grating.
Max.
 
This is a rotary optical encoder. The machine uses both linear and rotary encoders. Just to simplify, please disregard anything I ever said about linear encoders.

On the rotary encoder, the head that is just below the glass disc, that rides very close to the disc, is screwed in place and glued, and I believe is aligned. The PCB, however, just attaches to the housing with 4 screws and virtually impossible, IMO, to precision align it. for that reason whichever component requires less precision would be on the PCB. I am regretting not getting a picture!!! lol

From my limited understanding, so far is the bottom device is the LED light source and a Moire grating and would need to be rather close to the glass wheel for optimal precision? The device affixed on the PCB above the glass wheel is the photo detector or "read head" if you will? So are both just as critical for cleaning? Would sure seem like it! Is isopropyl still a good choice for this cleaning, assuming I can get it apart?

Also, something that is still bugging me is when I tested the cable, I had continuity but the resistance values were higher than I liked but I attributed it to poor meter lead contact. The cable that connects the CNC computer to the encoder is approx 12ft long and I was simply connecting two pins together at one end and tested both at the other so effective length was about 24ft. The wire size I am not sure yet but approx 18-20ga. Even at that, I was reading about 3ohms and 20ga should be about 10ohms/1000ft so even on a worst case, I still should be seeing below 1ohm?

It would seem like I should probably revisit my wire tests before anything else? I am also realizing that my meter will not induce any load on the wire so maybe I need a better method here. The main reason is this is a COMMON problem with CNC stuff! Wire has to move and will eventually break somewhere. I would like a more reliable method to test this anyway.
 
Three ohms sounds OK for the wiring, the fractionally higher than theoretical resistance is likely just contact resistance.
As long as it stays in low single digits when the cable is moved around, I'd not be concerned.

If it's a purpose made encoder cable the power cores are usually quite a bit heavier than the signal cores, sometimes massively so - so you may not get consistent readings across different connector pins.

Some are also double screened, with the overall outer screen to the machine metalwork & connector bodies, and an inner screen on the signal wires connecting through only to a pin on the CNC connector.

See the photo for an example - an offcut of Heidenhain cable for the 9 pin analog type encoders; the OD of the power cores is around the same as the jacket OD of each of the screened twisted signal pairs.
Heidenhain-Cable.JPG


Any dirt inside the encoder is a worry - they are supposed to be completely sealed & that points to a problem elsewhere, such as the cable, gland or connectors being messed up.

I've seen them before now where the casing is flooded with oil because a connector is resting in oil and its seals are poor, allowing oil to saturate the cable and reach the inside of the encoder through the cable. Once they get to that stage, they really need replacing - or at least a new cable and very thorough soak in a bath of solvent..


If it's only a trace of dust or oil then wiping the disc with a lint-free cloth and trace of alcohol should do it.

RS sell packs of small lint free cotton patches that are good for wet-cleaning the optics in those and linear scales.
**broken link removed**

What make and model is the encoder - a Heidenhain ROD series? I have a few good used 2000 line RS422 output ones kicking about, but some CNCs need the analog sine output type or digital serial.

The dual rotary + linear feedback system makes it sound as if it could be a Siemens drive system / CNC?
 
It has been many years since I serviced incoders technology has probably changed and some encoders are different. When working in a large factory and being in charge of servicing all electronics in the factory easy way to learn fast is, Call the factory service department ask for help they will talk you through the trouble shooting process. Once you learn service for that device you don't need to call for help. I never talked to a service department that refused to help or charged a fee, that would kill all future sales. Honeywell use to be a terrible company to deal with if you need a $1 part it will cost you $150 because you can't buy it from anyone else and they know it. We replaced all our Honeywell equipment so we didn't have to buy from them anymore. Honeywell were scoundrels they would remove names & part numbers from over the counter parts so you could not figure out where or what to buy a replacement part, then mark up the price many times higher.
 
Last edited:
From my limited understanding, so far is the bottom device is the LED light source and a Moire grating and would need to be rather close to the glass wheel for optimal precision? The device affixed on the PCB above the glass wheel is the photo detector or "read head" if you will? So are both just as critical for cleaning? Would sure seem like it! Is isopropyl still a good choice for this cleaning, assuming I can get it apart?

The grating attached to the read side of the head is also as important to clean as the scale itself.
Deposits on either can cause miss-reading.
I assume from the length of the cable, this is an extension, rather than the manuf supplied cable that is attached to the encoder.
Max.
 
Alright, for the guys that are asking, the the rotary encoders are optical serial type and are Fanuc, inside each Fanuc Alpha Red Cap servo. There are a total of 5 servos. As well, 3 of the axis (X, Y, Z) also have linear analog scales that report back true position of those axis. The linears use analog because the resolver types do not have glass in them and are more robust for the severe duty they are in.

The rotary encoders are factory installed in every servo Fanuc builds, and they will be very high resolution because normally these machines do not have linear scales and the servo encoders are used to report position.

the encoder in question happens to be in a servo with no chance of oil or water contamination. It sits basically on top of the machine and not subjected to coolant and such. There is an end cap on the servo, then another cap on the actual encoder. Yes, it should be sealed but could not have been a perfect air seal and I suspect over the many years, it got a little moisture in there that would condense and formed a tiny layer on the glass.

I really figured the cleaning would have worked.

The pic is the only one I have at the moment and does not show the read head arrangement. I was just getting a pic of the stuff on the disk. I did NOT touch that glass. I am not sure why there is a clean spot but once I saw that, I knew it needed cleaned for sure.
 

Attachments

  • 20200217_220710 mod1.jpg
    20200217_220710 mod1.jpg
    547.7 KB · Views: 127
The Alpha versions frequently used this type of encoder, the older Red cap motor encoder also had commutation tracks for the motor control.
There must be a system for this also on the Alpha versions.
Max.
 
Someone is going to have to explain the commutation of a servo as it relates to an encoder. This encoder can only go in 2 ways. there is a slot that the drive key goes in and I don't think there is ANY way, aside from marking it, to know which is which. I could be 180* out right now.
 
On the previous RedCap motors they had the regular quadrature pulse encoder, but there was also a separate track that consisted of a 4 bit code that would be sent to the drive in order to provide rotor position and commutation for the motor controller.
The 4 bit code was converted to a pseudo sine-wave.
I imagine with the Alpha serial, either this code is included or the motor uses a brief motion at power up in order for the rotor position to be obtained, from then on a encoder position would be used to determine the position angle.
.These motors are 3phase controlled, as opposed to the identical BLDC controlled motor.
Max.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top