Zeroing data lost

Welcome to the forum...............

Why people use backups to try and recover zero data is beyond me......just because it has worked previously, does not mean this is correct.

I personally will never support this type of method (for many reasons).

Kawasaki do not train, or provide ANY documented procedure for zeroing where they suggest using a backup to restore zeroing.

They have very specific procedures for re-zeroing, and these should be applied in all cases where re-zeroing is required.

Scribe lines are usually sufficient enough for re-zeroing, but in cases where they are not, then you should obtain the correct jigs and follow the specific procedures for the model and joint numbers that need re-zeroing.

These procedures, jigs and equipment are all outlined in the relevant documentation for your robot model.

Just re-zero JT1 and JT3 again using the scribe lines, takes less than a couple of mins.

Any re-zeroing will always impact on taught locations if the zeroing is not the same as it was when those locations were first taught.

Therefore, when people 'find' and 'use' they're own methods and do not follow OEM procedures, then tolerances and variations can occur.

So I cannot advise on 'alternative' methods as I don't use them.

The method I always tell my customers when an encoder battery warning occurs is:

- Do not power off the controller.

- Create a program and a precision point (if not already done) that has 0 degrees for each joint.

- Use CHECK and move the robot to the precision point - This ensures the robot is at the correct stored mechanical zero degree posture.

- Take a backup.

- Power off and exchange the battery (5 mins to do) which is less than maximum time recommended of 30mins without power.

- Then power on.

- No errors should appear, then check a couple of taught positions and confirm - they will be the same as no errors occurred.

- If any joints experienced encoder data loss, a 'joint abnormality error' would occur displaying the affected joints.

- Then as you were already at robots current mechanical zero, then the affected joints could be re-zero'd in the current position.

- Then no taught positions would be affected.

- Take another backup upon completion and delete the old backup.

In your case, I cannot say whether it is just JT1 and JT3 that require repeat zeroing until the end results are correct.

It could be that other joints have been affected too.

How do you know it is just JT1 and JT3?

Encoder battery supplies power to ALL joints, so for 2.5 Months of no Controller power, then battery must be of good condition to withstand duration.

If you are planning on powering the robot off for a lengthy period of time:

- Make sure any batteries installed are fresh, to ensure full power capacity is available.

- Park Robot in 0 degree position (if environment allows it).

- Try and park robot and make COG central, so to balance load across motor brakes and not bias on one joint to reduce brake slip.

- Perfect park position is position where gravity cannot act on any joint when just suspended by brake.

- Before power down, save position as a joint angle (precision point) and make all data backup.

- When returning, you have known joint angles joints should be at to compare to.

- If no gravity can act on joints, then if joint angle is different, you can use saved joint angle to re-zero at should problems have occurred like battery.

As explained, if ANY joint is providing different position when powered off to power on.

You will receive 'Joint abnormality error' referencing which joints are abnormal.

If this error occurs, this must be investigated as to cause and usually it is because of but not limited to:

- Backup has been used to restore Controller.

- Joint has physically moved, ie external force has been applied or brake wear has not enough holding torque for joint loads.

- Encoder battery is exhausted.

Just because joint angle is out of range, does not mean other joints that are within range are ok.

You really need to check and confirm ALL joints and scribe lines are as accurate as possible.

I am not a huge fan on zeroing with Kawasaki robots regarding accuracy when using scribe lines as some models have scribe lines that can 'move'.

In most cases though, they are sufficient enough to adequately zero the joints especially on larger manipulators.

So I can appreciate people adopt various techniques in order to improve on this.

Overall, though I do think some improvements can be made by Kawasaki to further improve zeroing accuracy on their robots.

From what I can recall (someone correct me if I am wrong):

- Each encoder has a backup capacitor fitted which can hold enough charge to power the encoder for up to 30mins upon electrical disconnection.

- The encoder provides values which the controller can only read.

- You cannot replace the data on the encoder itself by using any method I'm aware of (ie from a backup).

- When zeroing, these values are used to reference the encoder rotation counter and the zeroing of the at it's current location (joint angle) that you specify.

- Nominally, this is why you use scribe lines (these are the mechanical zero degree marks for the respective joint) or zeroing jigs/inclinometers etc.

- You can also re-zero using a degree value of your choosing if you already know the degree value the joint is currently at.

If power is not lost to an encoder, but during a controller off state, the encoder physically moves (ie the joint moves) then it is still tracked.

- Hence the 'joint abnormality error', which can be reset and the data from the encoder is still referenced correctly as it is still being tracked.

- Then a backup can be used to restore incorrect zeroing values if entered incorrectly.

If power is not lost to an encoder, then as long as the encoder has not been replaced or been misread, the encoder values will be the same.

- Then a backup can be used to restore incorrect zeroing values if entered incorrectly.

However, If power is lost to an encoder or is replaced, the values produced from the encoder and read by the controller will be different.

- Therefore the data in a backup cannot be used as you cannot rewrite the values in the encoder with old ones.

- The backup data will replace the old backup data (which is the same) BUT the new encoder read values are actually different.

- You then effectively re-zero the controller using the backup data values as the reference to the new encoder values = incorrect zeroing.

So if power has not been lost to the encoder, or the encoder has NEVER been replaced since manufacture or since the previous backup made:

- A controller could be re-initialized (emptying of data) and a backup used to restore it

- A backup could be used to re-zero if for some reason you were re-zeroing and entered the values incorrectly.

- The original supplied data sheet encoder values and offsets could be typed in Aux 050102.

- BUT as soon as power is lost in encoder, or it has been replaced, the backup data becomes redundant.

So the key is to identify the correct situation before applying a method.....I call this my OBA's

The main procedure for re-zeroing should always be carried out in the correct order:

1. Encoder Rotation Count Reset - Aux 050103 or ZZERO 10x command where x is the joint no.

2. Zeroing of the axis - Aux 050101 or ZZERO x command where x is the joint no.

A common mistake is to do one procedure and not the other, both procedures are always recommended.

In your instance though if you think JT1 and JT3 are the only problems:

1. Check to a known joint angle position within the program - JT2/JT4/JT5/JT6 will be correct (assuming JT1 and JT3 are only affected).

2. The location data stored (angle degrees) for that position will include JT1 and JT3 values.

3. Physically move JT1 and JT3 to the correct position visually.

4. Re-zero JT1 and JT3 using ZZERO commands and type in the values from the stored location (joint angles) for JT1 and JT3.

5. Then drive to 0 degrees and see just how far you were out.

6. Check back to the known position and check other positions......they should be corrected....unless other joints are also affected.

Alternatively, using a similar method:

1. Check to a known joint angle position within the program - JT2/JT4/JT5/JT6 will be correct (assuming JT1 and JT3 are only affected).

2. Take a note of the joint angles for JT1 and JT3.

3. Physically move JT1 and JT3 to the correct position visually - If you cannot achieve the position, then other joint(s) may also be affected.

4. If you can achieve the position, take a note of the joint angles for JT1 and JT3.

5. The difference between before and after values are the values you would need to advance/retard at your scribe line position.

6. Move JT1 and JT3 to 0 degrees then advance/retard the position by the difference you obtained earlier and re-zero there.

7. Check back to the known position and check other positions......they should be corrected....unless other joints are also affected.

Irrespective, it will probably take some time to 'tickle' your zeroing so that you won't have to 'tickle' your taught locations.

There is no extra charge on the encoders I know of. Once you power down a controller with low encoder batteries you are usually hit. Controllers come with the zeroing data from the factory inside the controller. If your programs were taught to this zeroing it should be good.Kawasaki are a non TCP Dependant robot. They record positions in joint values. Where you can run into issues is you have taught programs after re-zeroing. Have to remember that programs are taught to the zeroing data present and not always the factory zero.

At the factory, we often turn off the power of the robot after the work shift (16:30 to 7:00).

It's been using it for 3 years but has not reported low battery voltage error.

I am concerned that after the long holiday class it has problems and will lose zeroing.

So what do we need to do to prevent that?

Quote from phanthanhviet

At the factory, we often turn off the power of the robot after the work shift (16:30 to 7:00).

It's been using it for 3 years but has not reported low battery voltage error.

So the Robot encoders are being battery power supplied for periods of 14hrs.

This is the situation for 3 years.

Contact your local Kawasaki distributor and order battery assemblies and replace them.

Do not wait for alarms, always prepare before and create a schedule to replace.

You have a record now, that you're system will last at least 3 years before battery replacement is required, so create schedule now for regular battery change.

I always recommend to clients, irrespective of monthly power cycles, or extended shutdown periods:

- Replace battery every 12 to 18 months.

Car manufacturers using Kawasaki's often never power off the Controllers (only when required).

- They usually schedule between 24 and 30 months (I have also seen 36 monthly change too).

Remember battery is for 'backup' only, many reasons influence battery discharge conditions, so there is no definite time period you should make change, each system is different.

But in your case, I would considering changing soon.

Dear Sir.

Thanks about your reminder

How long does it take from the warning robot to replace the battery until the battery is completely depleted?

Because we do not have a backup battery for them right now.

Quote from phanthanhviet

How long does it take from the warning robot to replace the battery until the battery is completely depleted?

Because we do not have a backup battery for them right now.

That value is indeterminable due to heat/cold/electronic component current consumption.

So that warning should include:

- The battery should be replaced immediately.

- Do not power off the controller until you are going to replace the battery (less than 5 min procedure).

Considering your 'stated history', it is my opinion you should replace the batteries now.

You asked about 'preparation' - Prepare to prevent, not prepare to fail.

So, with a battery available, and 5 mins to do the job.

It's a no brainer.................prevention is better than cure.

Thank you very much!!

I will act now

You're welcome.

Your beer level in your glass becomes low (you can't see it), but know, so start to sweat a little (warning).

You still know some is in there so it doesn't stop you from drinking, but you know it is getting critical now.

You start to sweat more as the warning is still active so you can either:

- Order another and then seamlessly move from empty to full.

- Drink till it's empty and panic as you now notice the bar is closed.