Hi, i need create a sinusoidal signal. I want know what is the maximum frequency that the SM1232 could generate.

I think in various aspects:

1. The frequency output depend of scan time of CPU model. Is it correct?

2. If i create a FB for create a frequency signal it is deterministic?. I think that its depending of another parts of program.

3. May be i will need a stand alone signal generator.

The sinusoidal signal ever be constant.

Hello,

I'm not electronic expert, only think about your fine idea, in my opinion is this possible but only in positive area of voltage, it mean when we take range of Voltage 0-10V or Current 4-20mA, everytime is positive value. So probably with CPU scan cycle you can generate diferent sinusoide, but only in positive range. If it is suitable for you (make tests, play with this) I will be starting from 5V - mean start of amplitude and continue to generate signal with time to 10v and again to 5V - 0V -5V. This will be probably some sinusoidal rendering. Give feedback, in my opinion interesting

Edit: now I'm think with another, if will be CPU scan teoretically long the sinusoide will has ripple signal due not change Voltage value with time  maybe I'm wrong but this has some sence...

Edit2: heh perfect will be connect to that output Osciloscope to see real result  probably will try your interesting idea hehehe

Hello,

starting some tests and I must say some great news. I got my S7-300, CPU 313C with 2xAO and there can I set range from -10 ..+10V, so without problem can output this range  Of course I can set range from 0-10V, or 0-20mA but this is only positive range. Continue with next tests.

Edit: so taking back my previous info that only positive range can be set up. On your SM1232 same, you can set range -10..+10V, great

I think that range is no problem...

I think that the velocity of construct the signal is the problem.

I need use OB38 for generate 1 sample its may be good to lower frequency.. but its bad to faster frequency.. i dont know whats the best or maximun frequency for do it..

Hello,

heh interesting, for first I started very basically with 1 cycle - 1 point of change Voltage, messauring with my multimeter at output - no problem. Time to end one complete amplitude 40s. Then I started with add some offset and checking to not reach differentially to max/min values (+10/-10V), because some big offset will be up-step max/min V. When it have been closer to max, I was start minimize offset to reach so good, but you know it was slowing and time was not constant. Maybe I try your system. Thnx for info.

Edit: I try think to include SIN() somehow if it will be effective...

Hello,

can you say me your investigations? Did you implement SIN()? Give me your ideas, how much freq did you create? What is your set time interrupt of OB35? Pls say about that lower frequency too...

Edit: take we f.ex. freq 50Hz with +/-10V, what will be your solution?

cody_master

Hi, i need create a sinusoidal signal. I want know what is the maximum frequency that the SM1232 could generate.

I think in various aspects:

1. The frequency output depend of scan time of CPU model. Is it correct?

2. If i create a FB for create a frequency signal it is deterministic?. I think that its depending of another parts of program.

3. May be i will need a stand alone signal generator.

The sinusoidal signal ever be constant.

According to the S7-1200 System Manual, 06/2015, A5E02486680-AJ, page 1205

Settling time for analog output (95% of new value)
Voltage: 300 μs (R), 750 μs (1 uF)
Current: 600 μs (1 mH), 2 ms (10 mH)

..ignoring the PLC cycle - these values are the ceiling.

1. The frequency output depend of scan time of CPU model. Is it correct?
Yes the scan of the CPU will have an influence.. but not in the way you may expect. (more later)


2. If i create a FB for create a frequency signal it is deterministic?. I think that its depending of another parts of program.
You can use a cyclic interrupt OB3x. OB3x interrupts at a regular time base.

Good news: OB3x can interrupt the normal program sequence several times, if you need to.

Bad news: The problem is that between OB1 scans some tasks (communications / IO updates) that can hamper the OB3x to interrupt absolutely exact period. Measuring the internal PLC clock between OB3x calls may be done to try and compensate for deviations.


3. May be i will need a stand alone signal generator.
A PLC is not a signal processor (DSP / FPGA / CPLD). It would be a nice experiment to do but may not be good enough for the task.
Think about this.. 50Hz has a signal period of 1/50s = 20ms. So how many samples can you fit in 20ms?

But..
Rather ask how a PWM / PTO pulse can generate an analog signal. If I remember correctly search abbreviation are: FM    VCO     PLL    PFM   PWM   PCM   ... something like that... much to do with signal modulation and demodulation in the radio frequency field.

Does it help a little?
w

Hello William,

why did you answer all to him???  I want that he think about f.ex that 50Hz freq, how can he will solve it, how can count, what he will prepare. I feel that he know problematic, understand what he can do with PLC. Nevermind, agree with you that this can be perfect test project to understand what can I get from PLC. From start of this topic we all know that output from PLC can't be perfect, but can be interesting. Maybe he will answer

Duris_I

Hello William,

why did you answer all to him???  I want that he think about f.ex that 50Hz freq, how can he will solve it, how can count, what he will prepare. I feel that he know problematic, understand what he can do with PLC. Nevermind, agree with you that this can be perfect test project to understand what can I get from PLC. From start of this topic we all know that output from PLC can't be perfect, but can be interesting. Maybe he will answer

oops.. sorry my friend. 

I did not guess your strategy this time.
My PM is open to you if you wish to warn me next time.

Yes its a great help.

I go explain the project.

We need make a perforated plastic mesh. The hole is made with a servo valve.

Sinusoidal (setpoint) -> amplifier controller (closed loop) -> servo valve -> linear petentiometer -> to amplifier

The hole size on mesh is determined by the sinusoidal signal. You can see that frequency of sinusoidal signal it will determine the hole size. 

For this reason i think that sinusoidal signal generated by DAC module its not good for this aplication.

May be i could generate the signal but the hole size could be diferent.

cody_master

Yes its a great help.

I go explain the project.

We need make a perforated plastic mesh. The hole is made with a servo valve.

Sinusoidal (setpoint) -> amplifier controller (closed loop) -> servo valve -> linear petentiometer -> to amplifier

The hole size on mesh is determined by the sinusoidal signal. You can see that frequency of sinusoidal signal it will determine the hole size. 

For this reason i think that sinusoidal signal generated by DAC module its not good for this aplication.

May be i could generate the signal but the hole size could be diferent.

Now it is clear that this is an actual project and not pure theory.
Let me just say that I am impressed by your creative approach.

If you are married to the idea of using a (pneumatic?) servo valve - please consider getting a company such as Festo involved. Festo spends much time and efforts in RnD and might just be the guys to help you out with a controller. You may find pre-programmable / pulse driven / communication driven system that will do the job easily.

Personally I would go for another type of design if it was my project. A touch of old school. I rather like the idea of combining mechanical design into a control system.

A mechanical cam or crank might work nicely here. The linear movement that a crank gives is very much in a sinus wave form. With a servo motor (and positioning drive) driving the rotation, there is many possibilities opening up.

With the right servo motor system you can have a electronic cam controlling the movements of this slave axis synchronised with a master axis.

Please chew a bit on the idea.
w

Hi,

I have news:

good news: The signal command (setpoit) isn't sinusoidal. It is square signal.

Bad news: The frequency signal is 250 Hz. The square signal should be modify on Amplitude, Offset and frequency. 1 to 250 Hz.

Yikes, not sure if that is possible.

According to the data sheet, the Analog Output has a settling time between 300 and 600 microseconds, DAC's don't require encoding, so this should be fast enough.

250 HZ is 4ms, so you need to do a state change every 2 ms.  So, I guess you could set a base line cyclic interrupt at 2 ms.  Every other period you need would have to be multiples of 2ms with a direct write to the Analog output.  A 2ms (or even a 1ms) cyclic interrupt is theoretically possible, but your CPU may be devoting 100%, or more, of it's power to servicing the interrupt.

Since you need variable output, PTO is not really an option.  I suppose theoretically, you could use an external circuit using an analog opto-coupler with an analog output as the source and the PTO output as the gate. 

Good luck.

Yeah its very hard.

OB30 could be configurable to 1ms cyclic interruption and it configure to very high priority (using the 24 level) too. But the velocity of analog ouput module is not fast.

Definetely i need a electronic board based in OPAM where it's infinity resolution.

I´ll comment the solution and how its was coupled with PLC s7-1200

William_B

 

cody_master

Yeah its very hard.

OB30 could be configurable to 1ms cyclic interruption and it configure to very high priority (using the 24 level) too. But the velocity of analog ouput module is not fast.

Definetely i need a electronic board based in OPAM where it's infinity resolution.

I´ll comment the solution and how its was coupled with PLC s7-1200

FYI...

If you create OB30, you can only set the interrupting cycle in 1ms increments.

If however you use SET_CINT instruction once to configure OB30 or every cycle to manage OB30, you can set the interrupt cycle time in 1us increments.

I would encourage you to have a look at the SET_CINT instruction..... even if it is only academical.

Greetings,
w

 Thanks.. there are a little ligth to resolve the problem.

Now. I can use  SET_CINT to setup OB30 on 200 Us. But the Analog output (SM1232) maybe can't it be fast.

William_B

 

cody_master

Thanks.. there are a little ligth to resolve the problem.

Now. I can use  SET_CINT to setup OB30 on 200 Us. But the Analog output (SM1232) maybe can't it be fast.

Normally when we use the analog output we would use it as in the format of %IW64 or symbolically as "analogOutput1" for example.

Inside the interrupt you should add   [   :P  ] 
so that it becomes %IW64:P or "analogOutput1":P.
The   [   :P  ]   postfix forces the PLC to bypass the buffer
and immediately write to the physical output.

If the postfix is not there, the output will only be updated at the end of OB1 cycle.

Sorry for not mentioning this earlier.
w

 I dont stand. Whats this [   :P  ]  .

Please explain this.

William_B

cody_master

yeah i understood now.

You arethe boss. great help. i try this and i'll comment after.

first i have to buy all equipments.

(big smile)  Thank you for the compliment.
(bigger smile + wink) Next time I'll bill you my full consulting fee.

(serious)  I have no guarantee that the system will do what you want from it. Again... make the supplier part of the solution - request if you could borrow some demo equipment to do your tests.

PLEASE post a movie somewhere of the results.. I would love to see it.

Good luck.
w