PLC + Load Cells

[mjamil]

Hello Friends,

after a long break I am back, most of you guys may recall me, Im from the Cigarette Industry, 'been asking too many questions in the past.

I've been playing with Siemens S7-200 series (microwin), slow and unsteady!

I've been given a project, that is to weigh a product being dropped in a chute, after the desired quantity is inside the chute, the feeding has to be stopped and down comes the bailing press (jeranamooo!)

My FIRST questions was, since I've been playing with Microwin (and am comfortable with it), I preffer S7-200 plc to handle this task.
What do you guys think about it? and no there no data storing, or hi-tech task in this project. just opening the chute closing it, telling the press to come down, blah blah. only thing new the customer showed me in his existing press, is that the bottom plate of the platform, can be dis-engaged by use of a piston, so that when the Press comes down with tonnes of load, the load cell is far away. (probably b'coz it does'nt go Boink!)

I was checking Siemens, SIWAREX, apart from the Cost that may be involved in it, it requires an S7-300, I've had some experience, rather a intensive course inducted by Siemens its self for S7-300, but I'm not comfortable even thinking to use it, ofcourse the adventure I may go through learning it is my type, but was wondering what you guys would advise?

 

[JesperMP]

Hi mjamil,

I am doing weighing/batching just as you suggest with a standard PLC.
Ask Siemens what are the specific benefits for using Siwarex (I too would like to know).

Anyhow, this is what I do:
Weigh hopper is standing on 3 load cells. A loadcell transmitter generates a 4-20 mA signal. The transmitter is placed locally at the weigh hopper. Easy to troubleshoot by one man, and only a 3 wire cable from the transmitter to the control panel.
In the program, I do the following:
I filter the weighing signal to make it less sensitive to mechanical vibrations and electrical noise.
When weighing starts, I store the current weigh signal as a "tare" value.
During weighing, the net weigh signal is calculated like this: NET = ACTUAL - TARE.
I continue to add material untill the setpoint is reached, that is a tail-compensated setpoint. The "tail" is the change in the weigh signal after weighing stops:
SP_WITH_COMP = SP - TAIL.
Weighing continues as long as NET < SP_WITH_COMP.
The new TAIL value is calculated like this, some time (5 sec) after weighing has stopped:
TAIL = NET - SP_WITH_COMP

I achieve an accuracy of +/- 0.1 kg, with a weigh hopper for 100 kg material. And this is enough for my purposes.

Also, S7-200 and S7-300 are both from Siemens, but thats about it when it comes to commonality between the two series. It will require quite some effort for you to learn the difference.

 

[Bob O]

I filter the weighing signal to make it less sensitive to mechanical vibrations and electrical noise.

JesperMP,

How do you do this? I have a scale system that I am just starting back on and this may help me.

Thanks,
Bob O.

 

[Mark Buskell]

I too was looking at a job that had 3 load cells outputing a 4-20 ma signal. They use 2-wire if that makes a difference. I was going to take the 3 cells and run them each back to an AB 5/20 analog input module and then add together and scale the combined results.

Is there a better way to do this? Perhaps someone could explain how they tie the three cells together and have only one set of wires coming back. I imagine ohm's law comes in here.

 

[ushidayo]

Mark,

Load cells don't output a signal such as 4-20mA or 0-10v. To get a useful signal, a transmitter is required. These weight transmitters are often in the form of a programmable weight indicator with a built in analogue output. Most of these weight transmitter/indicators can be connected to 1, 2, 3, maybe more, load cells. The weight indicator can then calculate the load being measured, from the load cell(s) and output an analogue signal representing the weight. They're really easy to use, even for me.

 

[JesperMP]

Bob,

a simple but effective filtering:
(new)FilteredVal := (RawVal - (old)FilteredVal)/Factor + (old)FilteredVal
This formula is triggered in my program every 0.25 sec.
FilteredVal is the filtered signal that you use further in the program.
RawVal is the input from your load cells (scaled).
Factor can vary between 4 and 20.
I use floating point to avoid rounding errors.

I used to use a transmitter that summed up several load cells into one 4-20mA signal.
Today the load cells are digital, and I use a digital transmitter that generates a 4-20mA signal. I could have chosen Profibus or DeviceNet, but see no compelling reason to do it at the moment.
I think that a real drawback of conversion in a PLC module, is that you have to pull screened multicore cables for each load cell all the way. And it is very difficult to troubleshoot locally only one person. And its yesterdays technology. And who needs esoteric PLC modules anyway ?

 

[BobB]

JesperMP
That is hard work. I just program my analogue input card to return the average of 16 reads.
beerchug

 

[JesperMP]

OK, some cards can do that but not all.

 

[labingtone]

we have a systems like this.it uses wieghing controler to be connected to load cel.The Weighing controler is connected to a Simens S7 300 via profibus DP.this system is very accurate and reliable.


Lovington

 

[BBitters]

Last year, we retrofitted a very large weigh scale that sent its weight signal from the summing box to a process meter, through a BCD board, finally ending up in an A-B PLC-5/20. The system was extremely tedious to calibrate, and not very responsive. The only good news is that we could use a 20-year-old backup scale if the plc 'went stupid', which has almost never happened.
In the revised system, we took the raw signal from the load cell summing box into a millivolt card on the plc. We then set up the scaling right in the plc, and allowed a calibration mode where we'd store zero and span values for the scale. We used one of our SCADA computers as the interface - hit a button for zero pounds when the scale is empty, hit another button when the calibration weights are added (in this case, totalling 10025 pounds) and a third button when the process is complete.
One thing we considered but never implemented was running each load cell to its own channel on the millivolt card. That would allow us to identify a dead load cell quickly.

 

[Stephen Luft]

mjamil,

One way to simplify your application, both in cost and complexity, is to bypass the amplifier (transmitter). This would require finding a PLC that can accept a millivolt input.

We recently did an article in our news letter, dealing with PLCs and load cells. (specifically Entertron)

http://www.entertron.com/einl/1-5.htm

With any load cell and PLC, an amplifier is required, because the load cell generates a millivolt output that is based on millivolts / volt of excitation. Amplifiers add at least $250 to the cost of your application / per load cell. The load cells that have the amplifier built in are a little less expensive than both components sold separatly.

Siemans makes a very good product. Some of the comments we have received from our customers, have been the complexity of their programming and the cost of their software as being issues.

Note - (I work for the company whose product information I have provided, in direct response to the content of this thread)

I hope this information helps with your applications.

God Bless,

 

[this is mjamil too!]

Conclusions & more questions

Conclusions:

1) I can still use a PLC-226 (Siemens) but other items required are Analog input module (which I already have) + four load cells and an amplifier.

More questions for the pros.

1) Any reason why I shouldnt be using a CPU-226? as mentioned above.

2) When weighing a platform with the help of four load cells, the value from each load cell is added up within the amplifier or controller? or do I have to add all these four values independantly into to the PLC and then invent some new formula (within the PLC) for removing errors from the total.

3) I've come to know, temparature has its own effects on the read-out? any suggestions, as in my application, ambient may run up to 60-80 Centigrade with some high moisture contents in the air. ( a higher safety class I suppose, but how to handle the deviation in the readings)

4) I have been searching net for loadcell information, I see there are plenty of them, each with thier own advantages mentioned on the website, (and dis-advantage yet to be discovered by each user.)
in General, I would like to know what you guys prefer? platform types? button types, etc etc? and why?
(I want to be damn sure what I am buying and why, and I think this is the best way to communicate with the users rather than relying on the salesman of the particular product.

and yes, I am not restricted to have too much accuracy, and the whole equipment is going to be manufactured locally, (fabricated actually), so I currently have no restriction of shapes and types of loadcells.

 

[JesperMP]

Hi mjamil,

2) Can be done both ways. But most common is to let the amplifier do the summing up.

3) The loadcell should have its own temperature compensation.

4) I use Eilersen load cells.
Main reason is that they are REALLY robust. They can take more abuse than any other brand that I know.
Not too expensive, and easy to setup.

 

[BobB]

mjamil
The list of advantages and no or very few disadvantages is known in the sales game as the "Benjamin Franklin" rule/close. It is only surpassed in the sales gane by the "Hard Close". Had 25 years of that nonsense. Now sales reps hardly ever call on me because I give them as hard a time as they give me, and everybody else. Send me a genuine technical guy anytime. They get an interview. You have to sort out the sheep from the goats, as we say in Ozz.
I have used M-System transducers with great success. Have used MIL spec load cells in an offal treatment plant also with great success. I am trying to recover some old information and find the brand name. If I do I will reply. I might add that these things spent half their time under tallow!!! The temperature ranges were 50+ in summer and
-8-10 in winter. Degrees c that is.
beerchug

 

[ejessen]

Hi mjamil,

My experience with load cells is limited, and have never been the one to choose the system, however, I have made PLC programmes for several different systems.

I have the greatest knowledge of Mettler-Toledo weighting systems and you find them both as analog and digital (intelligent) systems.
The analog ones you choose by the range and the accuracy necessary. As accuracy increases you go from a 5V to a 15V system and a central multiplexer collects a number of load cells and send a unified signal to the weghting unit, in which you setup the behavior of the system (range, accuracy, correction for geological location).
The digital ones are even more accurate and will challenge the resolution you will find in analog PLC modules (most of them are limited to 13 digits). The digital ones can setup the load cells induvidualy to compensate for any unliniar mechanical behavior, and will download this setup when a cell is replaced. As I recall it, the digital ones compensates automatically for temperature variations.
In a simple system as you discibe above, you will find it an "overshoot" situation to compensate for location and probably also for unliniarity, but if it is for medical or other documental-required use, you would need this.

For filtering the signals, the method JesperM describes can be further developed. In a very noisy enviroment, I have seen it necessary to build a 10 register stack. At a sampling rate, which during SAT was changed several times to fit the enviroment the most, a new reading was written to the stack. Then the greatest and the least value was neglected and the average of the excisting 8 was calculated. The result was then entered into a 3 register stack and these 3 was averaged to the new operating value. Yes, it was a very noisy place, and we had to position a inverter controlled motor

A comment to the use of load cell signals directly: In a noisy enviroment, current signals prove to be less sensitive to noise than voltage signals. Filtering signals from external source implicate the use of ferrite materiales and "T-filters", depending of the nature of the noise, seen on the signal cable.