Linuxcnc with Teknic SST-1500-UCX

Yes looks like at least the cable is set up for differential signals. Maybe i'm remembering wrong and the motor I replaced the original with was the one that was single ended?

Trying to wrap my head around the photo of the other board. It looks like there are only 3 cables going to it. One to a section labeled "outputs" and two to Axis parts. Regardless it looks like far too many wires just for limits and step/dir signals. Could you show the cables going to the drives? It's hard to see that part in the pictures.

I haven't seen anything mechanical that looks like it would cause lost motion (no missing teeth on the rack or pinions) but I will be giving it a more thorough inspection to see if I can find something slipping.

The lost motion has only been in x and y thus far and seems to happen more often when simultaneous 3 axis moves are involved in the program. Although I have had an x shift in a program that I am fairly certain is only 2 axis moves.

I'll get some better pictures of that board when I get home later today.

It would be very unlikely that a mechanical problem causing lost motion would be with the rack and pinion it's self. But would more likely be something slipping in drive train between the motor and the pinion gear. Either the pinion is slipping on it's shaft, or one of the belt pulleys slipping on their shaft or the motor shaft, or the belt. The very first thing I would do is pull the covers off the belts and inspect them. Does your machine use racks for both the X and Y or only one of them? (Ours has a lead screw for the X, but still has a belt driving it.)

I don't think its a mechanical problem. I took the covers off and inspected the drive train, check all the screws and collars and belt tension. All the screws and collars are tight and the belt feels properly tensioned. This machine has dual rack and pinion for Y driven by a single motor and drive shaft that spans the gantry. The x is also driven by rack and pinion.

Here's some more pictures of the electronics in the box on the gantry: imgur.com/a/bKZPdLr

I gave them as descriptive captions as I could think of.

The three cables coming into the blue board go into three different sections of the board. One goes into "Outputs" and that section has only two pairs of wires coming out of it going to a couple solenoids (for oil and air). One of the beige cables goes to "Axis Set B" which has one six pin connector going to a servo drive and one four pin connector going out to limit switches. The last cable goes into "Axis Set A" which has two six pin connectors each going to one servo drive and two four pin connectors going out to limit switches.

Not sure why they decided to put things together this way, but at least I think I understand what is going on in this board.

I believe the three beige cables coming in are all the same in terms of wiring and have all wires hooked up to the plugs on their ends regardless of how many wires they actually need.

Well It looks like the drives only use single ended signals for the step/dir. I must have been remembering wrong. The Teknic drives I have on another machine have differential signals. You do have a copy of the manuals for the drives right? (If not I could send them to you.)

If it isn't a mechanical problem then I'd look at blaming noise on the step/dir lines. I'm guessing those beige cables are a relatively long run for TTL signals, and there is a good possibility that that blue breakout board could be introducing a big fat ground loop right in the middle of it. To explore that possibility you'd need to spend some time checking if the shield and signal grounds stay isolated and separate from each other as they pass through that blue board.

It is possible that those beige cables and their plugs could be easily reused as the new encoder feedback and step/dir cables from the servo drive to the Mesa card. But you might have to relocate a terminal or two (specifically the shield ground and another wire.)

Do you have the Teknic drive tuning software and the serial cable for connecting to the drives, to make tuing adjustments etc?

If you connect to the drive that is loosing motion with the tuning software you can monitor the step count, and see if and when the steps are getting lost, by comparing the step count (or position*step scale) in the control, with that on the drive.

I do have a copy of the manual I got online. It's been a good reference so far. Between that, and your suggestions, I am gaining a much better understanding of how this is all set up and feel like I'm narrowing down on the problem.

The beige cables run all the way back to the control panel, about 20-25 feet. I'm planning to put the Mesa board right in the box on the back of the gantry, basically where this blue board currently is, so I don't need anything that long.

I'll see if I can figure out if there is a ground loop in that blue board, but it is less important to me than getting the mesa boards working properly as that is where I want to be in the future anyways. I'll probably just get step/dir working open loop with regards to the controller first and then incorporate the encoder feedback after I've got that working.

I have the drive tuning software on the WinCNC computer, but I don't have the serial cable for connecting to the drives.

I haven't been able to reproduce the lost motion outside of longer programs, so I'm not confident enough yet to be able to catch the lost motion in action so to speak. I wrote a couple simple test programs that move in 3 axes and then return to the start point and plunge into a hole then repeat about 20 times in succession to see if I could cause it to happen, but no luck yet. Those were all feed moves however, so I am going to try rapid moves to see if that has any effect, because the geometry the machine cuts in the longer programs seems accurate, but the location is shifted part to part/feature to feature (I'm cutting 25 copies of a part out of a large sheet), which to me seems like it could be caused by issues with rapid moves.

I went ahead and pulled the trigger on the 7i76e+7i89 and those will be here today, so I'm going to start putting them together with a fresh install of LinuxCNC and see where I can get.

Do you know where to find your step scale for each axis in the Wincnc configs?
As well as max velocity and acceleration settings? Those will be helpful to know when you go to set up Linuxcnc.

I'm not sure where to get those, but I'm sure some poking around will uncover them.

I know that the max velocity is currently limited compared to what it is capable of (the previous owner mentioned he did so).

I'm using a different computer for Linuxcnc than the one currently running the machine, so I don't have to worry about losing any settings. I'm basically going to initially set it up so that I can just swap some cables over and run of either control system until I have Linuxcnc fully where I want it to be.

Wincnc has two config .INI files. One CNCSCRN.INI is just for configuring and customizing the UI screen, the other WINCNC.INI has all of the important info you will need. It is a rather terse and cryptic file but is still text based human readable. It can help to reference the Wincnc manual when trying to decipher what is what.

In the INI file there will be a section heading for each axis and they'll look something like this (for the X axis on our machine.)
The Wincnc manual may also mention a lot of settings that could be left out, for example there can be a different acceleration setting set for both rapid positioning moves and cutting feeds. (The example above does not have the feed acceleration specified so it defaults to the same as the rapid.) In the above the step scale is 5804.847 steps/unit (inches) an acceleration of 500 units/min/sec (divide this by 60 to get units/sec/sec for using in Linuxcnc.) Wincnc also claims to do S-curve acceleration. Which Linuxcnc doesn't do. But I've never tested these claims to see how well or if they actually work.

Thanks for the info. I poked around the .INI files. The values were set at 400 IPM rapids and 300 IPM feed maximums. I bumped these up to 1000 and 800 respectively after looking at the documentation from CNT motion that says the x and y axes should be able to hit 1500 IPM.

Haven't changed the acceleration and I don't remember off the top of my head, but I'll check that again tomorrow.

I've got my mesa cards and have linuxCNC installed on a separate computer. Going to pull 24v out of the current electronics cabinet and get the 7i76e speaking to linuxCNC sometime this week.

I ordered a flex rated, shielded ethernet cable from automation direct which should be here friday, so things are definitely coming together.