Testing panels?

Lyndon A

Member
Join Date
Sep 2014
Location
Minnesota
Posts
19
Hi guys:
I've been writing controls software for decades, but I've never used a PLC.

How do control panel shops do their testing? I mean, I understand simple testing like, activate an input and verify relay #1 turns on for 5 seconds, then relay #4 goes on, etc...

But if you're building a panel for someone halfway across the country and you have sensor signals faster than a human can deal with, like "must receive 100 pulses in a second and alarm if that threshold is not met," how do you verify the logic if you don't have access to the production hardware?

This thought came up because I build small Arduino (and microcontroller) based control systems and I have to verify a rotary switch sequence that happens in milliseconds. I normally keep a "test breadboard" around to simulate stuff like this. Would a regular control panel shop have something similar?

I've been hanging out here for a few years, but I've never noticed this aspect of PLC programming mentioned.
 
Hi guys:
I've been writing controls software for decades, but I've never used a PLC.

How do control panel shops do their testing? I mean, I understand simple testing like, activate an input and verify relay #1 turns on for 5 seconds, then relay #4 goes on, etc...

But if you're building a panel for someone halfway across the country and you have sensor signals faster than a human can deal with, like "must receive 100 pulses in a second and alarm if that threshold is not met," how do you verify the logic if you don't have access to the production hardware?

This thought came up because I build small Arduino (and microcontroller) based control systems and I have to verify a rotary switch sequence that happens in milliseconds. I normally keep a "test breadboard" around to simulate stuff like this. Would a regular control panel shop have something similar?

I've been hanging out here for a few years, but I've never noticed this aspect of PLC programming mentioned.

In my experience, real testing almost never happens until you get the production HW. Most panels I've seen usually need some kind of fix one they arrive. To be fair, it is often not the panel shop's fault; often the guy doing the drawings used a template and never went back and made the changes. Everything is design/build/test as you go, unfortunately....

However, there are software tools for logic validation. Some PLC systems have simulators, and there are a number of third party tools that can simulate the machine, or at least a people of it. Simulation & virtual commissioning are coming trends; if you hear anyone taking about Digital Enterprise or Industry 4.0, this is a big way in intersects with PLCs. You pair a logic Simulation with another package Simulating the IO responses.

Proper simulation is a big investment, though, and it isn't common yet. I've only seen it done in the process industry, when timelines are tight, risk of an error is huge, and the system response is well understood.
 
A panel itself has little to test apart from the electrical connections. On FAT's, I would like to get the drawings and if the revision didn't match 0 or something like that, I would ask for the change and test that.

Usually the entire system is small enough (like a machine) and can be factory tested... in which case there is little to simulate.
Or it can be a process or large enough machine that can't be moved in which case simulation is key.

As mk42 said, simulation is expensive... a global manufacturer of equipment I worked with invested in simulation early on and had very good results in reducing the amount of bugs and standardization of code (to link in the simulation code). But they had two teams on each project almost. One to build the code, the other to build the simulation.

Nowadays, particularly Siemens, has a wealth of simulators available that are incredibly impressive. As an example, they can now simulate Ethernet comms through simulators by opening two instances of PLCSim. You can also have sequences of reactions and so on... and that is where the market will go.
 
In my experience, real testing almost never happens until you get the production HW. Most panels I've seen usually need some kind of fix one they arrive.

True for us. We design our panels, but outsource the actual panel building. All of them are tested before we ship. The smaller ones are more standardized and usually need few if any fixes. The larger ones are more complicated and bespoke, more room for error, either by design or by the panel builder. It can take a week for one or two people to test and fix most of one panel.

We build machines that process bulk product, but we don't keep stock of the product itself. A standard dry run would test for mechanical and electrical issues only. It leaves a lot to be desired from a process control point of view. Therefore we have started using simulations to be able to go beyond dry running. That is where simulated machinery is a great help for us.

Simulation & virtual commissioning are coming trends; if you hear anyone taking about Digital Enterprise or Industry 4.0, this is a big way in intersects with PLCs. You pair a logic Simulation with another package Simulating the IO responses.

Proper simulation is a big investment, though, and it isn't common yet. I've only seen it done in the process industry, when timelines are tight, risk of an error is huge, and the system response is well understood.

That sounds like us, apart from the big investment. We are a small shop, so budget and staff is very limited. My first PLC project here was the process simulator that we are still using today to test our panels. It started out as a very limited simulation, has gradually been refined and extended as we keep coming up with more aspects of the processes that we want to test before shipping.

These days I aim for a full test on my desk of the PLC logic by itself. When the panel comes in, it gets linked with the machinery as far as feasible, rest of it is connected to simulator. For us it greatly reduces problems that would in the past have come up at commissioning or in the first weeks or months of real world use. Faster commissioning and less service intervention for us, better productivity for our customers.
 
I think I'm getting the missing link now.

In PLC-land, the person asking for a panel to be built has already designed the logic, but they want someone else to do the actual build. That means that the panel shop can test electrical connections, but they are following the logic from the customer, correct? That explains why the shop doesn't need to do full functional test: the customer can do that part.

In my world, it's different. The people asking me to automate something, may have little knowledge of electronics or automation, they just need it done. If it fails on site, they may not have the ability to debug & fix it.
That means, that if my customer is in e.g., Dubai, I need to be able to test my work before I ship it overseas.
 
Panel shops only scope of work is to build the panel according to your drawings, good ones will ask to deviate if it means saving money, panel space or labor to build. Better ones will tell you where you fudged up, because mistakes happen.

They make sure to test all electrical connections and such. They have ZERO responsibility to ensure the panel will do what the software logic asks of it. And it's up to the integrator to come to the panel shop for additional testing should the integrator want to test something beyond electrical. Such as networking between panels, or verifying a sensor will be detected as intended.

Most panel shops are going to have limited electrical capacity and floor space which can limit how much additional testing an integrator can do at a shop.
 
Panel shops often get better pricing because they're buying more components over the course of a year. Not uncommon for the contractor's total price, material and labor, to be less than what the hardware would cost from an infrequent buyer.

Panel shop owners got it tough, margins are thin and customers beat them up on price all the time. One of my best friends is closing up, not because the work isn't there, but because it just doesn't make sense to get a lower return on the cash he has to outlay on every panel job.
 
The system integrator I work for also builds panel from prints provided by others. When building panels from someone else’s prints we will typically just do continuity checks to make sure all hots and commons are tied together properly and same thing for I/O wiring. We usually don’t power anything up. That’s because they may have selected incompatible components or drawn something wrong. Of course, if we notice an obvious problem we will notify them. But they are paying us to wire a panel, not design it for them.
 

Similar Topics

Hi Guys, Can ControlLogix with Modsim32 for Modbus Server Testing? I have ControlLogix via Anybus Modbus TCP and I need to Test the...
Replies
6
Views
249
Hello all, looking for ideas on getting some basic training done, I need to practice on maintenance timers and bypass switches (with a Factory...
Replies
5
Views
1,275
I have a project in FTV ME (Studio v13) that I'm trying to apply security to a few buttons/setpoints. I have 4 users: Anonymous Logon, DEFAULT...
Replies
1
Views
764
I have a customer asking me to do some testing on non Rockwell I/O. They want to see if it will work. Thats simple hookup a PLC, put the IO in the...
Replies
4
Views
1,093
Hi, I’m just after a couple of pointers on testing comms between SLC PLCs in a test environment (i.e. not on a live plant). For a typical...
Replies
5
Views
2,718
Back
Top Bottom