Using PLCs fur pure data acquisition

stube40

Member
Join Date
Aug 2012
Location
Perth
Posts
5
Hi folks,

I'm new to this forum and PLCs, so I'll take this opportunity to say hello now.

My company has been developing data acquisition systems for many years based on inhouse systems based on embedded controllers. I'm now currently investigating whether a system based around PLCs could present a much faster time-to-market and potentially present other advantages such as reliability, resistance to obsolescence, comprehensive built in diagnostics etc etc.

The application is for high-end simulators and thus the data reporting frequency requirements are low - 100Hz would easily be enough. When I say 100Hz, what I mean is that a PC on the end of the Ethernet cable needs to be updated at 100Hz - I appreciate that this is not necessarily how PLCs are designed to work.

One 'system' would typically have 40 - 50 digital inputs and around 20 - 30 analog inputs. we're typically measuring low voltages and certainly nothing over 24V.

Although not strictly a PLC, I did find MOXA's ioLogic E4200 system very intriguing and it seemed to tick alot of the boxes. Here is the link if anyone is interested:

http://www.moxa.com/product/ioLogik_E4200.htm

I guess my big concerns at this stage are:

  • Learning the SNMP protocol which I imagine has a steep learning curve
  • Whether getting the data to the PC with update rates of 100Hz wold be an issue or not
  • What happens if the PLCs we choose go obsolete
If anyone has any input I'd hugely appreciate it!!

Thanks,
Stuart.
 
Thanks Peter, strangely enough PC104 is on our list.

Since there has been no other responses, I can only assume that nobody thinks PLCs are a good idea. Or, possibly the suggestion is so crazy that people didn't think it even merited a response ha ha!!
 
I think cost is going to be the biggest problem depending on what you are competing against. If you need low cost high volume analog inputs, I think you can stay cheap and reliable by avoiding PLCs.

If cost is non-issue, getting data over the wire 10 times per second should not be a problem, but there really would need be no logic controller, just comms between cards and scanner. I think an industrial ethernet control protocol driver for the PC with a user selectable RPI rate would get what you want without the need for a CPU, just a "soft PLC" or i/o scanner in software and you have many choices for field bus inputs.

But I am speculating here, some of the PC based control gurus can probably pick out a specific protocol/existing interface.

If these are customized systems, then cost may not be a problem
 
I don't think that a PLC is what you are looking for. There are PLC's, that will handle the frequency of your measurements, but 100 mesurements per second is going to produce a lot of data.

The question then becomes; How many data points do you require per session? How long do you need to keep the data on the PLC. The memory structure of a PLC may be more of a problem than anything.
 
A PC/104 system is what I selected for a similar application. You may be able to use something like Matlab to program it if you need to do onboard processing.
 
Hi Brijm - The 100Hz update rate does look like it might be a problem - as Brijm said, using the standard PLC comms protocols, this could generate alot of data. Since we're effectively using the PLC to sample a real human interface control (eg a joystick) and provide some feedback based on the sampled position, latency is also an issue. Hence, storing the data on the PLC and then downloading it all in one go doesn't help.

OkiePC - so far, based on the estimates we've had, the cost doesn't present a huge problem for us. It looked to be on par with some of the PC104 stuff we'd looked at already (and others have suggested in this thread).

Binaural/Peter - PC104 is still on our list as a possible solution, however we do have concerns about obsolescence. If a board becomes obsolete then it could result in a significant re-write of the software and the OS image. Also, we worry that it's not that robust in terms of it's connectivity - you have fairly delicate connectors on the PC104 modules that you then need some sort of breakout board for to make them more robust (plus potentially add in higher voltage line drivers and/or protection that is difficult to get on PC104 modules). Wiring up these breakout boards would potentially be an expensive extra stage during production and a potential source of weakness in the overall reliability. These are some of the reasons that we started looking at PLCs in the first place.
 
With regards to PLC manufacturers, as a starting point I was going to look at Allen Bradley and Modicon. Is that a good starting point or would anyone suggest looking elsewhere?
 
I did some a test using a Siemens 317-2PN CPU:

Digital Data 64 booleans = 8 bytes
Analogue Data 32 integerss = 64 bytes
72 bytes required, round up to 100 bytes

100 bytes every 10ms = 10kBytes/sec

I created 25 data blocks, each occupying 10kBytes giving a total transfer requirement of 250kBbytes.

Using Simatic manager I downloaded the 250kBytes into an empty CPU using the Ethernet connection. The download took 7 seconds giving a transfer rate of 35kBytes/sec

I'm sure similar spec CPUs from AB etc. will give similar results.
 
With regards to PLC manufacturers, as a starting point I was going to look at Allen Bradley and Modicon. Is that a good starting point or would anyone suggest looking elsewhere?
You could look at the Modicon M340 using local io you will easily get 100 samples per second (I would expect an M340 with a small program to scan at about 5ms. It comes standard with a large (8M I think) sdram card that you can write files to from the plc code, and then you use FTP to read the files
 
Yeah, I am stuffing data captured at 10ms intervals into an array read as bigger packets by a G306, stored in chunks fit for a 5/04 and messaged over there every 0.1 second second (limited by the 5/04 through a digi with a compactlogix brick built way back in 2006.

So yeah, if you have a PLC running the machine, getting it to store and forward a boatload of data is not a problem. Converting that data to useful realtime information might be a problem at the other end, if that's the intent, but if you don't need to process the information that fast, if a few seconds or post analysis is desired, then I am sure you could use a modern Logix PLC or most any fast ethernet enabled one along with it's easy-to-work-with rugged I/O. It would be overkill, but hey, I like admire overkill...

If you need high speed data acq, for portability, maybe you ruggedize and rebrand a Lucent Technologies data collector. We pulled a months worth of analog data from 12 thermocouples in a spiral freezer with a rented one. It more of a benchtop device not suitable for the environment so we had to hang an enclosure and then add a UPS after we found a power blip had caused it to pause recording.

I think this is what we rented, but ours was not a high speed application and the model we had was only polling every 1/2 second, as we went and downloaded from it later. I do think I recall that it used one D/A converter per module for multiple channels but was very fast and accurate. It had a real time software interface, and some software available to interface with PC apps we didn't need or look into.
http://www.home.agilent.com/agilent....00&id=1000001313:epsg:pro&cmpid=zzfind34970a
 
Last edited:
Regarding PLC brand. Most of my work is outside Australia in Asia from India through to Korea. There it is usually a requirement that Siemens PLC's be used. It would be an uphill battle to try to change their minds. We purchase Siemens in Singapore and it is competitively priced there.

Not sure where your market is. AB did a brilliant marketing job in Australia getting their product into the technical colleges etc.
 
Wow!! Thanks everyone for taking the time to reply.

Some awesome feedback there and certainly nothing so far that suggests we shouldn't look at this in more detail.

I'm thinking that our next step could be to actually buy some modules and spend a few weeks playing around with them to get more of a feel what can and can't be done with them, and what learning curve we would be committing to.

So now we just need to decide which modules to buy!!! I have to say, for someone new to the technology it's pretty overwhelming since there seems to be alot of choice.
 

Similar Topics

Hi, Seeking consultation on an implementation matter, and have a question about Modicon Compact 984 communication through RS485: Three Modicon...
Replies
3
Views
84
We have a product we are now building and are planning on using RFID tags to ID trays of product as they move through the process (5 individual...
Replies
6
Views
1,363
Hi, I have S7 Profinet PLCs with local IP addresses in my plant. Can I use the 1783-NATR to configure a 1:1 NAT to get to the plant network...
Replies
7
Views
4,892
I've found an interesting issue on some code I've converted from rev 16 to rev 30. The original code has a SIZE instruction with an alias used as...
Replies
0
Views
1,856
Hello all, What is your thoughts about implementing BMS systems using PLCs and SCADA. We did a couple of projects like that with success. I also...
Replies
4
Views
2,026
Back
Top Bottom