clough42
commented
1 year ago The ELS is behaving correctly in your video.
Both marks were spinning in the same direction, in sync, until you stopped the lathe at time code 6:44 in your video. At this point, both marks were pointing toward the front of the lathe. You then reversed the ELS and restarted the lathe, causing the marks to spin in opposite directions (i.e. they were only pointing in the same direction twice during each revolution of the spindle). At 6:46, you stopped the lathe again, this time with one mark pointing up, and one pointing down. You switched the direction again and restarted the lathe. At this point, the spindle and leadscrew were spinning in the same direction, at the same speed, but approximately 180 degrees out of phase, and this is exactly the state it was in when you checked and discovered it was out of sync. They're out of sync because it's doing exactly what you asked it to do. This is also exactly what would happen if you switched directions on a mechanical gearbox with a tumbler lever.
The ELS will properly retain synchronization when changing the direction of the SPINDLE. To cut a metric thread with an imperial leadscrew, or vise-versa, make your first pass, then leave the half-nut engaged and run the SPINDLE backwards to reverse the carriage back for another pass. Do not touch any of the buttons on the ELS. As long as you leave the ELS engaged, it will behave just like a mechanical gearbox, and the leadscrew will stay in sync with the spindle. (There is of course still backlash, so you have to back the tool out when reversing back for another pass.)
If you change any settings on the ELS (mode, pitch, feed, direction) it resets the synchronization internally and makes no guarantees with respect to any previous position. When you commented out this code, you discovered why it does this. The ELS continuously tracks the position of the encoder, applies a gear ratio multiplier, and calculates where the leadscrew should be so it can drive it to this position. If you press the REV button, it changes the sign of this multiplier, and suddenly has a new desired position far, far in the opposite direction. It must null out this value and start tracking from zero, or the code will try to actually drive the leadscrew to that far, far away position.
The "TOO FAST" message is a safety circuit breaker to prevent the servo from getting behind and continuing to spin after you stop the lathe. This doesn't normally happen unless you spin the spindle too fast for the servo to keep up, but it can also can serve as a safety measure, protecting the operator against something totally unexpected, like someone disabling the code that zeros out the accumulator when changing direction.
James
On Fri, Jul 29, 2022 at 1:27 PM Greg Reid @.***> wrote:
First of all, the obligatory but very sincere THANK YOU JAMES for all of your hard work on this project and for sharing it with the world. I've learned a great deal in following your excellent detailed videos and had a lot of fun building mine.
On my Chinese MX-210V mini-lathe, I don't have a dial/index on my lead screw. So whether cutting metric or imperial threads, I need to leave the leadscrew engaged, back out the cross slide, and reverse direction after each thread-cutting pass. No big deal once you have the muscle memory. But...
I've found that it will sometimes lose thread synchronization for the next pass, messing up the thread. I spent quite a while making sure that everything is "tight" in the lathe and leadscrew drive. Nothing is slipping. I spent a while longer ensuring proper shielding and grounding of the encoder and servo motor drive 'coz I'd assumed that the problem was most likely some lost pulses.
But I've found now that the REAL problem is when switching direction -- and of course I need to do this frequently when cutting threads on my lathe, lacking a thread dial/index. I found that I can now fairly easily recreate the problem so made and uploaded a video showing my setup and the problem recreation: https://photos.app.goo.gl/ri2bKXaHttoGKzpX7
It's very poor shaky-cam quality, unrehearsed and unprofessional! And the first 5-ish minutes just introduce my setup. You can skip to 5:30 to see the start of the problem recreation.
So then I had a quick look inside the ELS 1.4 source to check out places where change direction is handled, and found one interesting spot in the core.h ISR method where "if the feed or direction changed, reset sync to avoid a big step" and thought AHA, this is likely where the problem is happening. Well, I tried commending out that reset when the direction is changed, but it didn't work as expected. Now when I change direction, it tries to move the servo too fast and causes a "too fast / reset" error. Another short video of that FWIW -- oh, but I see now that it was accidentally truncated before you actually see the "too fast" error when I change direction: https://photos.app.goo.gl/gUQWFHanskAgDRoE8
I think that I'm on the right basic track, but a proper fix is going to need some more work. If James happens to read this and knows a solution, I'd love to hear it! If not, I'll dig in further when I get a chance and report back if I'm able to resolve the issue.
But I'm also wondering why nobody else seems to be running into this? Maybe because y'all have thread dials and always thread using the same imperial or metric as your lead screw so that you never have to do the back-out-and-reverse? Or maybe it's unique to my using this servo rather than a stepper. I'm using the ISV57T-180 "stepper killer" integrated servo: https://www.omc-stepperonline.com/servo-motor-nema-23-integrated-easy-servo-motor-180w-3000rpm-0-6nm-84-98oz-in-20-50vdc-brushless-dc-servo-motor-isv57t-180 .
More news as it happens...
Greg
— Reply to this email directly, view it on GitHub https://github.com/clough42/electronic-leadscrew/issues/224, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAZLSHD4IDU33DH7ZPVAWSTVWQWDLANCNFSM55BXHKGA . You are receiving this because you are subscribed to this thread.Message ID: @.***>
gregreid
Chucksbp
51mone
First of all, the obligatory but very sincere THANK YOU JAMES for all of your hard work on this project and for sharing it with the world. I've learned a great deal in following your excellent detailed videos and had a lot of fun building mine.
On my Chinese MX-210V mini-lathe, I don't have a dial/index on my lead screw. So whether cutting metric or imperial threads, I need to leave the leadscrew engaged, back out the cross slide, and reverse direction after each thread-cutting pass. No big deal once you have the muscle memory. But...
I've found that it will sometimes lose thread synchronization for the next pass, messing up the thread. I spent quite a while making sure that everything is "tight" in the lathe and leadscrew drive. Nothing is slipping. I spent a while longer ensuring proper shielding and grounding of the encoder and servo motor drive 'coz I'd assumed that the problem was most likely some lost pulses.
But I've found now that the REAL problem is when switching direction -- and of course I need to do this frequently when cutting threads on my lathe, lacking a thread dial/index. I found that I can now fairly easily recreate the problem so made and uploaded a video showing my setup and the problem recreation: https://photos.app.goo.gl/ri2bKXaHttoGKzpX7
It's very poor shaky-cam quality, unrehearsed and unprofessional! And the first 5-ish minutes just introduce my setup. You can skip to 5:30 to see the start of the problem recreation.
So then I had a quick look inside the ELS 1.4 source to check out places where change direction is handled, and found one interesting spot in the core.h ISR method where "if the feed or direction changed, reset sync to avoid a big step" and thought AHA, this is likely where the problem is happening. Well, I tried commending out that reset when the direction is changed, but it didn't work as expected. Now when I change direction, it tries to move the servo too fast and causes a "too fast / reset" error. Another short video of that FWIW -- oh, but I see now that it was accidentally truncated before you actually see the "too fast" error when I change direction: https://photos.app.goo.gl/gUQWFHanskAgDRoE8
I think that I'm on the right basic track, but a proper fix is going to need some more work. If James happens to read this and knows a solution, I'd love to hear it! If not, I'll dig in further when I get a chance and report back if I'm able to resolve the issue.
But I'm also wondering why nobody else seems to be running into this? Maybe because y'all have thread dials and always thread using the same imperial or metric as your lead screw so that you never have to do the back-out-and-reverse? Or maybe it's unique to my using this servo rather than a stepper. I'm using the ISV57T-180 "stepper killer" integrated servo: https://www.omc-stepperonline.com/servo-motor-nema-23-integrated-easy-servo-motor-180w-3000rpm-0-6nm-84-98oz-in-20-50vdc-brushless-dc-servo-motor-isv57t-180 .
More news as it happens...
Greg