Is the algorithm and processing speed too much for a PLC?

[piscis]

A local factory will be closing its door soon and offered be at rock bottom prices six of their “MultiHead Weigher”. (see link below). All the electronics; the display, boards and motor drivers were completely removed from them. This is a small 10 heads model. –meaning it has 10 weighing buckets with load cells in it, capable of 60 dispensing a minute.

I’m seriously thinking about buying all six of them and installing a new AD 260 PLC to use them on my own food packing -sideline business-. Question: Can the control of these MultiHead Weigher be done with a PLC instead of the PC it came with? Has anyone on this forum programmed one of this before? Is the algorithm and processing speed needed too much for a PLC and that’s why a PC was used?

http://en.wikipedia.org/wiki/Multihead_weigher

Thanks for your answers.

 

[Clay B.]

Processing speed is leat of your problems

Since the controls were removed you are first going to need to build an interface to get the load cell data to whatever CPU you use (PLC or PC).

Before you buy I would stongly suggest that you figure out what your budget is.

You will need 10 strain guage inputs. Discrete I/O for valving. An Operator interface.

Without knowing the exact algorithm I can not say for sure but I do not believe you couldn't use a PLC.

That said I would probably dedicate a PLC to each unit.

Can you post the algorithm you want to use?

 

[piscis]

I was planning in using a PLC for each unit with a small C-More and analogs/discrete cards. They didn’t remove the small servo motors which open and close the gates to allow the products to drop into the packing bags.

I will also need at least 10 stepper motors drivers. I’m planning to retrofit a single machine first and see how it goes but I need to find our first if this is an application for a PLC, that is why I’m posting the question here.

 

[Clay B.]

It is but it isn't

This is one of those famous gray areas. The problem is really how fast do you want to go. A PC can handle complex math faster than a PLC but I am not sure it is so much faster you could not do this on a PLC. There are too many unknowns.
First of is how fast does the math need to take place. I mean if the PLC can gather its data and make its decision in 1 to 2 seconds I would say a PLC should have no issues. If it has to be faster than that I think you may need a PC to "share the load" so to speak.

If it was me I would build it as PLC controled system then add hardware as I needed for speed.

One issue I have not seen you address is how are you going to get the weight data. The AD line of PLC's do not come with weight modules.

 

[piscis]

Thanks for the answers. I will find out if there are any gateway or converters to Modbus TCP so the weight data can come into the AD PLC as fast as possible.

 

[Clay B.]

Thanks for the answers. I will find out if there are any gateway or converters to Modbus TCP so the weight data can come into the AD PLC as fast as possible.

I would look at Red Lion's Data Station Plus. You can get 2 strain guages per card and they would go across a back plane to the Data Station Plus CPU that could then transfer everthing directly to your PLC.
Here is some info: http://www.redlion.net/Products/ProcessControl/ModularController/ModularControllerModules/CSSG.html

I would also suggest the G3 HMI over the C-more. If you need the power of the PC you would be pretty close with the G3 HMI from Red Lion. You could write your algorithm directly into the G3 and run it there. It also gives you a ehternet port to your drive network.

 

[shooter]

I would go for the PLC:
if speed is not important you can start with one strain gauge card and some relays to switch for every bucket.
use several output cards for the motors. just simple on/off digital cards.

If speed gets important get 10 strain inputs.
the plc is calculating fast enough. 1 every second is no big problem.

example:
filling weight is 1000 grams
use 5 buckets so fill every bucket with around 200 grams. do this with a timer. make this timer variabel for each bucket.
add buckets and find out best result.
open them to bag.
fill the opened buckets again. If last result was over 200 grams put timer smaller if under increase timer.
If fill time gets longer increase shaker.
if fill time is good again increase machine speed. (temp of sealing higher)
etc.
Yes buy them, even running manual is good.
other way is to fill each bucket up to 1000 grams and run in circle. easier and higher speeds are possible, however precision is lower. depends on cost of product.
you can even put two products in one bag, like nuts and screws, divide the disc in two zones. or up to 10 products in one bag.

 

[ojz0r]

If you want more speed you could always go for a softplc.
Otherwise this would be fun to use a FPGA for.