Servo Encoder Troubles

Hi.

When I embarked on a retrofit a while ago, I always thought my first post would be asking for help with servo tuning (that will come!), not help with trying to understand an encoder issue!

Part of switching to LinuxCNC and a Mesa 7i77 including swapping out the servo encoders. I bought some Chinese "Omron" E6B2-CWZ1X for each axis - Y and Z worked fine but the X encoder only worked in one direction, but lost about 5% of the counts in the other. Go figure!

I found a reasonably priced Heidenhain ERN470 5000 PPR encoder that arrived last week, seemed alright on the bench, so after hours spent making the adaptor plate and shaft, fitted it to the machine.

When I hooked it up, I tested again using the Halmeter, and for some reason I can't fathom, I'm getting 200,000 counts per revolution, for what I am guessing is 50,000 PPR. I can't remember if it was correct when I bench-tested it, or I misread the count as 20,000.

I was hoping that someone on this forum is able to explain why I might be getting 10x the counts out of this encoder? The counts are accurate - I can turn it several revs in one direction, and the same in the opposite direction returns to zero, so I have no qualms about the accuracy of it.

I can't find any info specifically for the 470, so I've used info for the ERN400-series (the ERN420 being the most similar) from their rotary encoder manual:
Heidenhain ExN 400 Rotary Encoders

My other question is, if I use it in the machine, will it derive the correct position as a 5,000PPR, or will I need to dial it in as a 50,000PPR? I'm a bit confused about what should be fairly straight-forward, I would have thought!

you shoudt get only the edgecount so 4 signals per encoder count

as haienhain are most resolver encoders
how did you make your adapter
can you maybe post your pcb schematics

the work is right as you see all signals clear
but maybe the wave trigger is to pricice so you see 10 insted of 4 per wave
THAT woudt be amazing work on adapter

Four edges per count - exactly what I would expect. 200,000 counts = 50,000PPR, should be 20,000 counts for 5,000PPR!

The encoder is wired directly to the Mesa 7i77 encoder input as differential TTL. Manual available at mesanet.com

My concern is that I've done something wrong. Do these Heidenhain encoders require a dedicated interface? Is this more of a Mesa or LinuxCNC question?

This is a nice little encoder, it would be a shame to have to ditch it.

The ern 470 has an odd marking on the encoder TTLx10 (found a picture after a while) picclick.com/Heidenhain-Ern-470-5000-01-...02.html#&gid=1&pid=3 Guess you have that on your encoder as well. I dont't know, but it sure looks like it gives out 50000 ppr. But has 5000 physical lines?

Yeah, that struck me as odd, also. I was going to mention that in my original post, but couldn't figure out how to add a pic (noob!).
I don't understand why it is marked as 5000PPR, but then outputs 10x? 50,000PPR seems overkill for any application, and sounds expensive!
And, I thought TTL referred to the 5V circuitry? But you're right; that "x10" seems suspect!

Thank you pl7i92 and Hakan; I've come to the conclusion that you are right and the encoder internal circuitry multiplies the lines tenfold. I've ordered another encoder where the label does NOT have "x10" after the TTL.
Thanks again.

andypugh wrote:

Is there any problem with the 10x feature? The 7i77 is unlikely to run out of counting rate

You're probably right; I did a quick calculation and it works out to be a max of 4.8e+6 counts for the entire 3m length of the X-axis (125mm per revolution).
It's just, being new to this, and already struggling with the servo tuning, I didn't want to complicate it more.
Can I assume the deadband value will help with the over-accuracy? I plugged in 0.0001 for the other axes, and it worked great. Is this in the machine units (mm in my case), and will work for my X-axis?