Post by Roger on Apr 4, 2024 14:53:03 GMT
This is the 4-axis AC-Servo setup I've been running since 2008, and it's been totally reliable until a couple of weeks ago when it started throwing g-code errors that kicked the K-Motion control offline. Unfortunately, K-Motion doesn't tell you which axis caused the error, but commanding the X-axis to say X0, showed a value of around 0.3mm different on the DRO. You always get a few microns error between the desired and actual position on the axes with Linear encoders, because the DRO shows the encoder reading, while it's trying to control that position using the encoder on the motor, and they aren't perfectly connected. The drive hunts back and forth slowly, trying to reach the exact postion.
K-Motion clearly monitors this error and decides if it's too big, hence the error.
I don't have the Linear Scale on the Y-axis being used this way, because there's too much yaw on the bed due to wear.
The easiest thing to swap over was the Servo motor, and that made no difference. So I swapped the X and Y Servo amplifiers and the problem went away. However, that's only because the Y-axis wasn't being monitored by the Linear scale, and it was steadily losing position.
PXL_20240404_142230123 by Billy Roberts, on Flickr
So with nothing to lose, I took the offending drive apart to see if there was anything obviously wrong with it.
PXL_20240224_203651451 by Billy Roberts, on Flickr
It turns out to be reasonably simple, just two boards, one with the high voltage section and the switched mode power supply, and the other being the brain.
PXL_20240224_203915541 by Billy Roberts, on Flickr
Noting the chips used and downloading the datasheets gave me where the power pins were, and hence where to probe for the multiple voltages used. There's +3v3 +5v and +/-18V for the op-amps and opto isolators.
PXL_20240308_181436154 by Billy Roberts, on Flickr
Putting it back together and powering it up on the bench showed that the PSU was ok, so I changed the op-to isolators that interface the Step & Direction lines. Opto Isolators don't last as long as most other semiconductors, so it seemed worth a try. It's hard to lash up wires for a test, so I just changed them instead. Anyway, that made no difference, so I think it might be an issue with the motor encoder interface.
I'm a bit vulnerable with these old drives, so I thought I'd see about getting at least one working one while this one was down. Ebay had plenty of them, but they'd already been sold when I went to buy them. Having asked for all those listings to be removed, and asking if there were any more, I was assured that there weren't any more in China. These drives had clearly been stripped from machines that were being scrapped. We're talking about £700 for a new one by the way.
I left it a week and checked again, and found another drive in China, and bought that for £395 including shipping. As a long shot, I also googled the part number of that logic board, and found 5 from one supplier, again in China. I managed to buy one of those for $158 shipped, assuming this would be where the fault was.
Anyway, both arrived today, and after doing a factory reset and programming those, I now have five working drives. Happy days!
I've just made an offer for another couple of those logic boards, and hopefully will get those too. Sure, it's not been cheap, but the exercise has put me in a much better position to keep the machine running for the forseeable future.
Out of interest, while googling the logic board, it appears that the Delta brand also uses the same logic board, even though the Amplifier housing and power connections are different. Maybe this is why I managed to find these spare board, they're used on several makes of drives.
I'll keep a look out for a spare drive motor too. Those are very expensive though, around $700 from what I can see for a used one. I'm sure they can be repaired, but it's not the sort of thing I can easily do. You need to remagnetise them if you take them apart to replace the bearings. The encoders also need setting up, and that's not a trivial thing to do either. My friend Bill has a magnetiser, and I could possibly figure out how to do the encoder, but it's not something I'd undertake, bursting with confidence. Hopefully I'll never have to visit that topic, but you never know.
K-Motion clearly monitors this error and decides if it's too big, hence the error.
I don't have the Linear Scale on the Y-axis being used this way, because there's too much yaw on the bed due to wear.
The easiest thing to swap over was the Servo motor, and that made no difference. So I swapped the X and Y Servo amplifiers and the problem went away. However, that's only because the Y-axis wasn't being monitored by the Linear scale, and it was steadily losing position.
PXL_20240404_142230123 by Billy Roberts, on FlickrSo with nothing to lose, I took the offending drive apart to see if there was anything obviously wrong with it.
PXL_20240224_203651451 by Billy Roberts, on FlickrIt turns out to be reasonably simple, just two boards, one with the high voltage section and the switched mode power supply, and the other being the brain.
PXL_20240224_203915541 by Billy Roberts, on FlickrNoting the chips used and downloading the datasheets gave me where the power pins were, and hence where to probe for the multiple voltages used. There's +3v3 +5v and +/-18V for the op-amps and opto isolators.
PXL_20240308_181436154 by Billy Roberts, on FlickrPutting it back together and powering it up on the bench showed that the PSU was ok, so I changed the op-to isolators that interface the Step & Direction lines. Opto Isolators don't last as long as most other semiconductors, so it seemed worth a try. It's hard to lash up wires for a test, so I just changed them instead. Anyway, that made no difference, so I think it might be an issue with the motor encoder interface.
I'm a bit vulnerable with these old drives, so I thought I'd see about getting at least one working one while this one was down. Ebay had plenty of them, but they'd already been sold when I went to buy them. Having asked for all those listings to be removed, and asking if there were any more, I was assured that there weren't any more in China. These drives had clearly been stripped from machines that were being scrapped. We're talking about £700 for a new one by the way.
I left it a week and checked again, and found another drive in China, and bought that for £395 including shipping. As a long shot, I also googled the part number of that logic board, and found 5 from one supplier, again in China. I managed to buy one of those for $158 shipped, assuming this would be where the fault was.
Anyway, both arrived today, and after doing a factory reset and programming those, I now have five working drives. Happy days!
I've just made an offer for another couple of those logic boards, and hopefully will get those too. Sure, it's not been cheap, but the exercise has put me in a much better position to keep the machine running for the forseeable future.
Out of interest, while googling the logic board, it appears that the Delta brand also uses the same logic board, even though the Amplifier housing and power connections are different. Maybe this is why I managed to find these spare board, they're used on several makes of drives.
I'll keep a look out for a spare drive motor too. Those are very expensive though, around $700 from what I can see for a used one. I'm sure they can be repaired, but it's not the sort of thing I can easily do. You need to remagnetise them if you take them apart to replace the bearings. The encoders also need setting up, and that's not a trivial thing to do either. My friend Bill has a magnetiser, and I could possibly figure out how to do the encoder, but it's not something I'd undertake, bursting with confidence. Hopefully I'll never have to visit that topic, but you never know.