STL Help again!!

[Gambrinus]

The original programmer mapped ASI Master 1 sensors to M140.0 to M159.7 then called the sequence program for the first station.
He then Mapped ASI Master 2 sensors to the same M addresses and then called the (copied) sequence program as station 2.

How he mapped it? In any case it's a bad programming style. Too complicated in debugging. Yes, it'l be better for you to map Master 2 into another place.

 

[STL???]

Simon,
The way the code works is that if a sensor is evaluated and the condition not met, It conditionally jumps to a label where a FC generates a protool textlist value.

So there is only one pointer value worked on each call of the sequencer.
The problem is there is a sensor number on the display for diagnostics, But no actual error message to Log which sensor was defective.

The operators currently are having a field day with this logging downtime as sensor faults!!

We need traceability.

I hope i can do it with 1 timer per FC Call, I prefer not to go down the IEC route.

Something along these lines

1. analyse the pointer
2. Identify the textlist no
3. Get the sensor address indirectly based on the Textlist value
4. Evaluate its condition
5. If not OK - start fault evaluation time
6. If Timeout
7. Set Error text for Indirect sensor ~ indirectly to correct Message bit

Ill see how i get on!!

STL

 

[SimonGoldsworthy]

You cannot use 1 timer to directly time more than one event. Is that what you are proposing ?

 

[Gambrinus]

Hm, pointers is one thing, timer is another. What does it mean "each call of the sequencer"? How often it occurs? Every OB scan or not? Main question: how pointer in AR1 points to concrete sensor, how it points to another after this? Now I don't understand this.
For example, if in scan 1 pointer points to sensor1, in scan 2 to sensor2 - timer will not work properly because it's called with different sensors.

 

[STL???]

OK,

The Step controller for each Station is broke down into individual
sequences:

For example - Part control

The code is as follows:

 

 

AN "Part Control"

JC M042

LAR1 P#M 146.4

A "ST1".SENSOR._8S5

JCN M004

LAR1 P#M 146.5

A "ST1".SENSOR._8S6

JCN M004

LAR1 P#M 146.6

A "ST1".SENSOR._8S7

JCN M004

LAR1 P#M 146.7

A "ST1".SENSOR._8S8

JCN M004

AN "TRANS-ENDE"

= "TRANS-ENDE"

M042: NOP 0

Part Control is set active by the sequencer,If part control isnt active, the sensor evaluation doesnt take place.

When part control is active each one of the above sensors must be ok to achieve "TRANS-ENDE" which tells the sequencer to write the next step which is in this case Clamp Parts.

Example
LAR1 P#M 146.4
A "ST1".SENSOR._8S5 //DB5.DBXxx.x
JCN M004

1. load Pointer ID for text list generation
2. load sensor address for evaluation
3. if ok move to next sensor and load pointer, If not ok jump to label M004

4 At label M004 there is a FC (not my effort) which uses the AR1 value to generate text to the user (M146.4 = 52)
52 in the text list will read "Waiting for Sensor xxx = On "

This is called every scan providing the station is set active

I hope this is clear, i'm terrible at explaining things.

I'd post the whole code but its big 120 networks +

 

[Gambrinus]

Uff, I don't understand why to do it so complicate?! Why to use AR1 instead of direct message number? All sensors must be on in this code part, I see. Can you explain structure of "ST1".Sensor? Are sensors _8Sx defined as BOOL and placed in chain in that DB, without any another variables between?

 

[Gambrinus]

It seems you can write it much more simple than it is now...

 

[STL???]

Hi Gambrinus,
You are right, i would like to scrap it all, But i just dont have enough time allocated to this, Before production restarts

You are correct the Memory bit Order, Sensor Order and Error Message Order all match sequentially (i made sure of this when i changed the sensor mapping). They are all BOOL Data format.

I'm working on a FC now, Its ****, but i think it will show best what i need to do.

Thanks again for your support!

 

[Gambrinus]

I think another problem can occur. Timer starts counting from signal front. But if the signal doesn't change state, timer doesn't start counting...

 

[PeterW]

seems an interesting one to have missed, work always gets in the way.

My 2peneth (late), I wouldnt use timers from this sort of thing, I woould usually create one shot bits from the CPU clock memory bits and use these as time bases for counting, fall back frpom S5 days of not trusting timers.

 

[Gambrinus]

Timers can work if adressed indirectly. Unfortunatly I cann't find working example, but I have it. But 160 timers... It's too much. I wait for more information from STL. It seems he must use counting as you described, I agree.

 

[PeterW]

The other way for timers is to create the sub-routine as an FB instead of an FC, then inbed a timer FB in the STAT's.

 

[Gambrinus]

Yes, it is. But using of incrementing int is more useful in this case. Less memory.

 

[SimonGoldsworthy]

Here's an idea - if the code already generates messages that say "Waiting for Sensor xxx" why not look in the message area (be it a DB or M bits), find the first bit on and then start timing. If the first bit on changes, restart the timing. If the timer times out, set your fault bit according to where you found the first on bit.
This code will run all the time at the end of your existing processing.

Assumptions:
Only 1 bit will on at any time,
A bit is on to indicate waiting for sensor

 

[STL???]

Assumptions:
Only 1 bit will on at any time,
A bit is on to indicate waiting for sensor

Good idea, but only an Int value is generated between 0 + 160 to display the message.

Right, ive had a good go at this, and although not perfect - PLCsim will run it.

My test program
i copied all the blocks i needed to simulate the process
FC82 - Criteria analysis(text value generator) original block
DB5 - St1 Sensors
DB11 - Alarm Messages
DB1 - Fault Timers
DB41 - Sequence DB for Station 1

OB1 Code

1. Opened DB41 which must be open for FC82
2. loaded a test pointer of M140.0
3. called FC82 - value generated and stored in DBW30 0f DB41
4. call my FC and use DBW30 as basis for all offsets
5. DB5,DB1 and DB11 are used within my function.

I need to do something, so if a sensor changes the accumulated value is zero'd.

Please feel free to Let me know what you think