# Axis 3 absolute position offset by 360deg after power loss **URL:** \ **Category:** Technical Questions\ **Created:** [January 15, 2022, 2:37pm UTC](https://forum.universal-robots.com/t/axis-3-absolute-position-offset-by-360deg-after-power-loss/18694 "2022-01-15T14:37:06Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![hi11](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.universal-robots.com/hi11/32/2179_2.png) [@hi11](https://forum.universal-robots.com/u/hi11)\ **Post date:** [January 15, 2022, 2:37pm UTC](https://forum.universal-robots.com/t/axis-3-absolute-position-offset-by-360deg-after-power-loss/18694/1 "2022-01-15T14:37:06Z")
Our robot was mistakenly powered off when someone pulled the power cord unknowingly. I restarted the robot, it said the robot had moved so I had to visually confirm that the 3d robot model matched the real model and confirm that. Once done the axis 3 is now offset by 360deg of the original position and I can only jog it from -180deg to -360deg, instead of the normal ±180 (i know the joint can do ±360 but axis 3 can only reach ±180 before colliding with the base). Seems to me that somehow the encoder position has been offset during re-sync. How can I manually go and perform an axis calibration again in this case? Best, Thibault ---
**Author:** ![mmi](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.universal-robots.com/mmi/32/691_2.png) [@mmi](https://forum.universal-robots.com/u/mmi)\ **Post date:** [January 17, 2022, 11:55am UTC](https://forum.universal-robots.com/t/axis-3-absolute-position-offset-by-360deg-after-power-loss/18694/2 "2022-01-17T11:55:36Z")
Move arm to a different pose, power arm off, delete _.urcontrol/.last\_seen\_joint\_positions.conf_, and power arm back on. Possibly it was near the point where position wraps around. ---
**Author:** ![hi11](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.universal-robots.com/hi11/32/2179_2.png) [@hi11](https://forum.universal-robots.com/u/hi11)\ **Post date:** [January 19, 2022, 4:15pm UTC](https://forum.universal-robots.com/t/axis-3-absolute-position-offset-by-360deg-after-power-loss/18694/3 "2022-01-19T16:15:01Z")
Great thanks, that’s also what I suspected. ---
**Author:** ![hi11](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.universal-robots.com/hi11/32/2179_2.png) [@hi11](https://forum.universal-robots.com/u/hi11)\ **Post date:** [January 28, 2022, 12:18pm UTC](https://forum.universal-robots.com/t/axis-3-absolute-position-offset-by-360deg-after-power-loss/18694/4 "2022-01-28T12:18:41Z")
So, I finally got time to try the procedure, the robot was not accessible beforehand. I booted up, powered the arm, jogged a bit, powered off arm, logged via SSH and rm’d the .last\_see\_joint\_positions.conf, powered up arm and the recovery mode came up, but the axis 3 is still offset by 360deg, indicating -355deg when it should be 5deg. I then tried the same procedure but deleting the .last\_see\_joint\_positions.conf file before rebooting the controller, also after rebooting the controller, to see if the file was maybe already loaded in RAM, no change. I then looked at the content of the .conf file and the `revolutions` line indicate the axis 3 as -1 revolutions. So i tried changing it to 0, but it seems to be ignored, powering up the arm shows again -355° on axis 3 when it’s straight up. also changing the `angles` value of axis 3 does not trigger a recovery mode error, so I’m not sure how to proceed next, I would have thought that a simple update of the `revolutions` value would have fixed it since that’s the actual issue, but it seems like the controller overwrites it when powering down, and loads it in RAM on boot as if I change the value over SSH, powering off-on the arm does not seem to take into account whatever I write in this file for axis 3 position. So on this assumption I manually overwrote the `revolutions` line for axis 3 but triggered a hard shutdown to prevent the file from being overwritten with whatever was in RAM, but on boot and power up the file was overwritten and back to -1 revolutions… I then proceeded to power off the robot, take out axis 3 cover and manually move the arm past the -360deg point by unclutching the brake, powered up, it said there was a large movement while off, went into recovery but indicates -400deg. I powered off, removed the .last\_seen\_joint\_positions.conf and powered up but it is still thinking it’s at -400deg. Basically something remembers the revolution count, as I’d assume it would be starting at 0 otherwise when reset? Is there some other file that keep track of that? ---
**Author:** ![hi11](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.universal-robots.com/hi11/32/2179_2.png) [@hi11](https://forum.universal-robots.com/u/hi11)\ **Post date:** [January 28, 2022, 12:49pm UTC](https://forum.universal-robots.com/t/axis-3-absolute-position-offset-by-360deg-after-power-loss/18694/5 "2022-01-28T12:49:08Z")
Alright so after some thoughts I figured a way to trick the robot 🙂 I figured that the robot will “guess” its revolutions offset when it has moved while off to the closest previous seen position. Positioning it to -350deg (almost straight up with 360deg offset), I initially moved it while off and releasing the brake manually to the normal -10deg, but it still booted up as -370deg. I assumed it would snap back to the closest ±180deg position, so I placed it to -200deg (the normal 100deg on axis 3), powered off, then manually moved the axis 3 all the way to -100deg so that the total movement would be larger than 180deg. I booted up, it complained of a large movement, but then thought I moved through the shoulder since the distance was shortest, so it snapped back to -100deg, and not -460deg. Problem solved on my side, but then I’d still need to know why did the procedure you mentioned didnt work?