It would be great to gather some experience and feeling from you about it.
A machine builder has the need to store cyclically data at a rate of 50KB per second from a PLC into an external USB mass storage. In that case, it is the 0TGF016.01.
I’ve look into the datasheet of the 0TGF016.01 and I determined that if we continuously write 50 KB per second, your USB storage is estimated to last around 25 years, based purely on the 40 TB endurance rating of the datasheet.
So basically for the storage perspective quite good.
But in terms of the application, is it possible/recommended with FileIO to write cyclically data every seconds? What could be the stumbling stone or the limitations?
To be more specific the need for cyclic writing refers to variable of type MC_ENDLESS_POSITION to store all information necessary for restoring the axis position after a warm or cold restart. They want to store this critical data in an external memory.
the use case sounds a little weird to me, to be honest. Considering the delays those writes introduce, the positions will likely not be correct whenever you really need them (-> stumbling stone)
Both, ACP10 and mappMotion talk about using a remanent variable for that and this already ensures being written whenever needed (especially on power failure).
Please elaborate further why this data should be on a thumb drive instead.
While I am no motion expert I agree with Michael.
What happens when during write you have a powerfail ? The data is then most likely invalid/damaged/not from the current position.
If this is critical data then I would not rely on something slow and potentially damageable as a file (in case of a powerfail)
Thanks for your feedback. It is actually because they want to avoid to loose the data when the remanent memory gets initialized du to some problem or crashes which the actually had some time in the passt.
The thumb drive will basically never hold the actually correct data so it’s not a proper replacement, as outlined above by both oliver and myself.
Maybe, if there is distrust in the remanent storage (which is a whole different discussion on what really happened in the past and if this could potentially still be an issue or not), maybe they could use a whole other method, e.g. limit switches or just a dedicated homing switch?
