Hi,

I'm using an 840D sl with Sinumerik Operate 4.4. My goal is to monitor tool wear by analyzing spindle and axis power/torque. I can read $AA_POWER[x] values and write them to an R-parameter using synchronous actions:

ID=1 DO $R[100]=$AA_POWER[X]

However, I would also like to save the values to a file every IPO-cycle to form a graph. I can write to a file using the WRITE function, but that doesn't work in synchronous actions.

Is there any way to log the power value to a file/usb device?

Thanks very much in advance,

Jasper

Hi Diego,

Thanks, that's a great idea. This is how I worked it out in SinuTrain.

Will test tomorrow in the machine.

DEF INT _ERROR

$AC_MARKER[1]=0

ID=1 EVERY $A_MSECOND MOD 10 == 0 DO $AC_MARKER[1]=$AC_MARKER[1]+1 $AC_PARAM[$AC_MARKER[1]]=$AA_POWER[X]

<ACTUAL PROGRAM>

CANCEL(1)

$AC_MARKER[2]=0

WHILE $AC_MARKER[2]<$AC_MARKER[1]

$AC_MARKER[2]=$AC_MARKER[2]+1

WRITE(_ERROR,"TESTLOG",$AC_MARKER[2]/10<<" | "<<$AC_PARAM[$AC_MARKER[2]])

ENDWHILE

 To not clog the processor I decided to first test with a frequency of 100 Hz (1000ms/10). Had to increase 28254 $MC_MM_NUM_AC_PARAM. 20000 is the maximum amount, so with this frequency I can log for 200 seconds.

Thanks again,

Jasper

So I tried it at the machine. It sort of worked, but had 2 major flaws.

First, it should save about 150 seconds cycle time * 1000/50ms = 3000 measurements. It only saved 398. Within 9 measurements (135-143), I recorded 5(!) tool changes.

My guess is that the processor cannot output every single $A_MSECOND, which results in the skipping of $AC_PARAM[$AC_MARKER[1]]=$AA_POWER[X]. So I need to find a way to make this more robust.

Second, the WHILE loop that writes to the file takes way too long, over 30 seconds. We can't lose so much time when every second counts. I thought of putting all data in a string first and then writing it to the file, but a string apparently has a maximum length of 200 characters.

Thanks for your response. I don't have much experience with $AC_TIMER. The Synchronized Actions manual states the following: 

Function Manual Synchronized Actions

Incrementing

The value of the timer is incremented by the duration of the set interpolation cycle (MD10071

IPO_CYCLE_TIME) each interpolation cycle.

$AC_TIMER[<index>] += <interpolation cycle>

 Do you expect that in practice the timer will increment with exactly IPO_CYCLE_TIME? If so, I would be able to just multiply the timer value by the IPO_CYCLE_TIME to determine the real time.

I will perform some new tests on monday.

Thanks again!

Hi Diego,

I'm sorry for my late reply. I tried your code with an interval of 16 ms. The real interval should then be 0.016064 s.

The result is great, I have a very nice graph. I also included the current G block, by adding $A_GG[1] to the spindle power. In excel I could easily extract it, since the power value always ends with a 0. Saving still takes a long time, but well..

Thanks!

Jasper