Magnetic encoder woes

Karl_T

H-M Supporter - Sustaining Member
H-M Platinum Supporter
Joined
Oct 14, 2014
Messages
2,106
I scrapped out the Fanuc control on my Vectrax CNC mill and replaced it with Galil/Camsoft a couple years ago. I planned on CNCing the knee at that time but never got around to it. fast forward two years and I thought its about time. The pic shows the mill. there is a separate DRO for the knee visible in the pic that I have been using.


Vectrax.jpg





I had bought a magnetic encoder like this:







Got to work and machined a slot in a solid bar 20" long. Mounted this bar to the side of the knee with less than 10 thou run out in both X and Y end to end. Made a cover to fit over the top of the slot (removed in pic)





Fit the magnetic pickup to the end of a smaller rod to go the length of the slot and then extra to mount below the mill. made a shoe to hold the pickup just a couple thou above the tape. the top of the shoe also rides on the cover (removed in the pic) to hold it down close to the magnetic tape. The bottom of the small rod is mounted on a Heim ball joint and a slotted angle iron. trued everything up so the small rod never touches the slot in the big bar. paid a LOT of attention to detail here.


The whole idea of this design was to make it nearly impossible for metal chips to get near the scale and pickup head. With the cover on, the only opening is right at the bottom of the assembly.


scale.jpg
 
Nice install Karl, but what's the problem? What's the full part number of the read head?
 
OK, pretty proud of myself at this point. Thought this was a beautiful design well implemented.

I connected the eight wires to the galil board fifth axis: red and back to +5V and DC common. The other wires to +MAX -MAX +MBX -MBX +INX -INX on the second Galill 1900 ICM. This is a differential encoder hook up. checked with logic probe - all the wires flashed with movement as they should.

Spent a few hours reprogramming the software to have a knee readout and be able to re-zero as needed.

Then i started checking encoder counts vs. the DRO on the knee. it is not consistent nor is it possibly right. Seeing about 760 counts per inch going up (Note to Galil programmers - Counts for Galil are 4X more - they count edges - Camsoft counts full cycles) AND less than 1/3 of that going down

My conclusion is this encoder is missing all sorts of counts. Pretty sure this encoder actually does 1000 cts/in.


What is wrong here?
 
You are too fast Jim. :)

This is written on the read head
5Volt DC/0.0025mm
SN: 1904 1926
 
Depends on the resolution 1, 5, 25, or 100 um
1 = 25400 / inch
5 = 5080 / inch
25 = 1016 / inch
100 = 254 / inch

No matter what the manual says, clearance between the read head and the mag strip needs to be 0.005'' to 0.015''. These aren't as sensitive as the Renishaw units which will actually work at about 0.035'' clearance.
 
i will tear the unit off and double check clearance but almost positive this is good. Anything else to consider?
 
That is the only adjustment. I set mine at 0.010'' with a feeler gauge. They just work, I've never had a failure.

You went way overboard on the install, mine are just sitting in the open, covered in chips. :)

Here is the knee on my mill.

1620498661923.png

And the Y axis
1620498716843.png
 
Well, I did bury the scales on my lathe. :)

 
OK, tore it down and inspected. found no problems. Clearances were perfect. Added a conical spring and rebuilt the shoe to make the head ride just right.

put it back together and problem is WORSE. Put logic probe on galil channels quite a bit. Channel A+ is not always putting out a signal - just intermittent. The worst kind of problem.

Time to quit putting good money and time after bad. The whole thing goes in the trash bin.

Can you suggest a complete scale compatible with Galil inputs?

PS got a magnetic read head and tape free to good home. there is a chance problem is elsewhere.
 
Back
Top