Electrician Talk banner

1 - 20 of 24 Posts

·
Hackenschmidt
Joined
·
10,944 Posts
Discussion Starter #1 (Edited)
The site is kind of slow lately what with the banning and the election and all. In another thread @gpop was talking about taking out PLCs that turn out to do little or nothing in machines. How about a thread about where you got rid of a PLC?

I have a story about eliminating a PLC, not in a machine but in a convoluted stormwater system. I was subcontracted by a vendor of the engineering company that maintains the site for the owner, a mega-corporation. The companies that designed and installed the system were long gone, little had been done other than routine maintenance for over 15 years, which tells me they did a decent job. The engineers only had a general understanding of how the system worked. They had huge reams of records and documentation, but little of it relevant.

Anyway after a lightning storm the system isn't getting a pond level signal, which it gets from a 4-20ma transducer. They ordered an identical replacement transducer and sent me out to replace it. The I tested the transducer, found it indeed was dead, replaced it, tested the replacement, and it worked on its own not connected to the system. But the system still wasn't operating. I have no idea how to or if I can fix it but they have nobody else so they have me give it a go.

Some wire tugging reveals the 4-20ma level signal goes to a Koyo DirectLogic PLC, which I think is now the Automation Direct Clic. There is no documentation for the PLC and they can't get hold of the contractor that installed it. I have never programmed one of these, don't have any of the necessary cables or software.

But all this PLC has is one 4-20ma input and two 4-20ma outputs. What the hell. But anyway nothing at all was coming out of the outputs, so I declared it fried / dead, figuring the surge got it when it got the transducer. I tell them I am in the dark but it could be who knows how much work to reverse engineer the PLC. They say keep going.

More wire tugging and it turns out the 4-20ma outputs go to two separate 4-20ma inputs on a very nice fancy programmable pump controller. What the hell. Why would a nice fancy controller want the same information twice? I dig into the programming and talk quite a bit to the engineers about what the system is supposed to be doing, what that pump controller is supposed to do.

We determine that due to some limitations in the settings of the controller, for the automation and alerts they need, it needs the pond level on two inputs. There was no scaling or changing of the signal: it was looking for the raw unaltered tranducer output on both inputs. The PLC was being used to take the signal inputs and regenerate it on two outputs. So this won't be so hard to reverse engineer.

Another what the hell at this point - why not just loop the 4-20ma through both inputs? They had it looped through a display and another controller, whoever designed this understood looping 4-20ma. Well, I tried that, and that's when I learned that since those 4-20ma inputs on the pump controller are not isolated, you can't loop through two of them. I suspect the person that designed this learned the same lesson the same way, they planned to loop it through two inputs.

I think they put in that PLC as a kludge to get things going. Rather than saddle the customer with what would probably be another albatross someday, I searched around and discovered a 4-20ma splitter that was made for exactly this kind of task. Still a kludge, and not all that much less expensive than a cheap PLC, but zero maintenance. Hopefully less delicate with surges, but certainly quicker to replace with a surge.

I also installed 4-20ma surge protection on the signal, but it didn't work. Luckily the engineers decided to keep a spare transducer on site, because the transducer was destroyed again the next year and it also zapped the splitter. Luckily the splitter had four outputs, and only the two that were connected to the pump controller got fried, I was able to move to the remaining two outputs of the splitter and keep running. The customer was just happy the system wasn't offline for long the second time. I told them to buy a spare splitter as well, set it right in the DIN rail next to the one that's in production. With the Phoenix style terminals and a little slack it can be switched in about thirty seconds taking your time.

This was all T&E work for me. It was a lot of T and a fair bit of E due to drive time to a remote site. I think I earned their money, I had to do a lot of learning on their clock, but somebody had to do it. It sure isn't my fault that lightning strikes. I improved it a bit over what was there before. As a result of what I learned in the process, I was able to help them cheap and easy with a lot of unrelated issues in the future.

When the engineering company was bought a couple years later, they fired or moved everyone and I stopped getting calls from them, but who knows if I've seen the last of them. I should have pasted a business card sticker inside the control cabinet.
 

·
Senile Member
I make all the electrons line up for their Flu shots
Joined
·
32,178 Posts
Everything you did fixing the system was perfect , except for this one little thing.............

"it needs to the pond level on two inputs."
 

·
Hackenschmidt
Joined
·
10,944 Posts
Discussion Starter #3
Everything you did fixing the system was perfect , except for this one little thing.............

"it needs to the pond level on two inputs."
The best thing about the board software upgrade is, you have more time to go back and fix your typos :)
 

·
Band Member
DIYer Extrodinaire
Joined
·
7,007 Posts
Never removed a PLC to go back to relays etc ... but I know of a couple of places that the PLC was overkill, and just added complexity to a simple control circuit.

With sensors and the lightning issue though, I have put in good ol' blue balls in many overhead water tanks as a safety for WHEN the sensor fails , not IF it fails !
 

·
Registered
industrial E,I&C
Joined
·
2,822 Posts
I have removed lots of plc's, hmi's and vfd's simply because it was best option at the time. I have also installed plc's that someone in the future is going to be questioning why.

Anyone can switch out a plc but in some cases you are switching a old 1946 chevy side step with a new eco boost. The chevy might have been maxed out running the machine but the new plc is hardly noticing the code running in the back ground. So do you swap out every old chevy or do you transfer some of the work to the new plc.
Now there is always going to be a augment about putting all your eggs in one basket but modern plc's come with modern toys that allow a plc to act as a back up including looped Ethernet.
 

·
Registered
Joined
·
1,615 Posts
I had a customer request me to build a four ingredient batching system using relays. His logic(?) was that his employees did not understand PLC control, of course they did not understand relay logic either. My mantra was, If you want to pay me to do it, it is a great idea. The relays and loop controls cost many times what a small PLC would cost. It was expensive but it worked fine.
 

·
Bilge Rat
motors and controls.........
Joined
·
7,369 Posts
The one I remember most was about 20 years ago at a plant I worked at occasionally.

There were two air compressors factory mounted on the same skid and consoled by a PLC. For whatever reason the PLC failed one day, it would not start either unit. So it went to the nice green box out back and a simple alternating relay was installed.

I still work at that plant once in a while and the compressors still work just fine........
 

·
Registered
Industrial Electrician
Joined
·
82 Posts
A vertical form fill bagger that I was working on had a propriety computer rack with I/O cards which wasn't programmable per se.

Anyway, the cost of the cards was really ridiculous so I decided to gut it and go with relays.

The bagger used a prox tree instead of an encoder which made things pretty easy.

It was a fun project.

Before
151982


After
151983
 

·
Registered
Electrician
Joined
·
1,295 Posts
I haven't ripped out any for unnecessary reasons but have taken them out as they stopped working and a new PLC was much cheaper or I had some relays/timers on the van that could get it working "today". Funny you mention the PLC acting as a 4-20 splitter. I will be doing this on a job next month. I have a project that needs to verify 4 valve positions and turn on pumps accordingly with timers for cooldown of accessory's. I priced out the parts and decided on a $89 Click PLC, easy to program and will allow different operations based on the valve positions. I need to split a 4-20ma signal from one sensor and send it to two drives. Upgrading the CLICK plc to use as a signal splitter only cost an extra $45 so I will be using the plc as a signal splitter without doing anything to the signal and I don't have to add any other hardware.

Someone a few years down the road may not like seeing my PLC but I also get repeat calls as soon as a maintenance guy sees a PLC and says "WTF" and then notices my sticker. I get the call. So far the CLICK PLC has been 100% reliable for me. It's always a sensor or external relay that goes bad, a lot of maintenance guys get confused when they see a PLC and then they just call someone.
 

·
Chief Flunky
Field Service Engineer
Joined
·
1,800 Posts
I’ve done two. Maybe more. Got rid of a DirectLogic that was really just an interlock for a bunch of conveyors that was in a box that was damaged and corroding away until the PLC failed. Got rid of lots of them where they bred like rabbits when they needed just one.

I really like Toyo (AD). They have been around since the 1980s so not quite as old as Modicon and AB but pretty close. The thing with AB is you know it’s solid but it costs over $3,000 for the cut down limited software license PER laptop. To outfit all our techs with full licenses would cost us $40,000 without training. Never mind annual maintenance contracts and paying for phone support. That’s why programmers can charge $250-350 per hour to all these companies and get away with it.

So on most projects I can just put in a “pump controller” or whatever for maybe $500 or less for the PLC. I don’t even have to mention it’s there. The software is free. The training (just as good as AB classes) is free. Shipping is a day or two, about the same as AB. So the customers don’t question whether or not I have to use a PLC instead of relays. I can take care of the replacement issue easy. Just give them a spare preloaded with software.
 

·
Hackenschmidt
Joined
·
10,944 Posts
Discussion Starter #11
Funny you mention the PLC acting as a 4-20 splitter. I will be doing this on a job next month. I have a project that needs to verify 4 valve positions and turn on pumps accordingly with timers for cooldown of accessory's. I priced out the parts and decided on a $89 Click PLC, easy to program and will allow different operations based on the valve positions. I need to split a 4-20ma signal from one sensor and send it to two drives. Upgrading the CLICK plc to use as a signal splitter only cost an extra $45 so I will be using the plc as a signal splitter without doing anything to the signal and I don't have to add any other hardware.
Just out of curiosity, could you loop it through the drives and use one analog input? I am trying to think other than the $45 and a few extra lines in the program, if there's advantages / disadvantages to splitting it in the PLC versus looping it through.
I guess it would be more flexible in the future, you never know if you might want to manipulate the signal in logic rather than pass it right through, you could add a delay between the pumps, lead-lag etc. Also with it looped, if the 4-20ma signal wiring to either pump was damaged it takes both out, with them split it would tolerate that fault.
 

·
Hackenschmidt
Joined
·
10,944 Posts
Discussion Starter #12
Someone a few years down the road may not like seeing my PLC but I also get repeat calls as soon as a maintenance guy sees a PLC and says "WTF" and then notices my sticker. I get the call. So far the CLICK PLC has been 100% reliable for me. It's always a sensor or external relay that goes bad, a lot of maintenance guys get confused when they see a PLC and then they just call someone.
I would probably still be getting calls from them if they had gone with a PLC. It's a very funny thing what's happening in business. I was three steps away from the owners here, subbed by a vendor to the engineering company that does the work for the owner company. Something changes every couple years - the engineering companies buy each other, sell off contracts, etc., the salesman from the vendor company changes jobs and manages to bring the contract to his new company.

Between the layers nobody has the discipline to maintain the drawings and documentation and make sure it stays up to date and gets transferred end to end. When one outfit has things figured out, something changes and some new guy is there tugging wires and scratching his head.

If I had managed to make myself indispensable they'd probably still be calling me. Probably not with that little PLC though. My pitch was to replace that pump controller with a PLC that did exactly what they want, with all the logic in one brain, rather than hack around with the pump controller config to get as close as possible. And it would have been easy to incorporate a couple other aging systems run on relay logic into the same brain. Of course I'd still have to remember to put that sticker in the cabinet :)
 

·
Chief Flunky
Field Service Engineer
Joined
·
1,800 Posts
There are a lot of “non isolated” 4-20 mA inputs out there. By rights if it’s properly built you can just put everything on one loop until you hit resistance limits. But a lot of inputs are not isolated. Like a certain grossly overpriced PLC company who thinks that a $0.10 optocoupled input module deserves a $100 price tag, PER input. So a lot of problems go away if you tie all the return paths together even on unused input channels and then ground the DC supply negative. The easy/best solution analog wise is the same thing...use a 250 ohm precision input through a differential amplifier which in turn feeds an ADC chip on the other side, or a floating V/F converter feeding an optocoupler followed by a high speed pulse counter or timer on the other side. Simple, text book stuff here. This should be done by a first year engineer at the manufacturer as a training project. See Horowitz and Hill of MIT fame, Art of Electronics. But the likes of AB fail to do the basics making electricians have to clean up after the engineers.

What’s the solution? It’s a 4-20 mA isolator card. They are dirt cheap from once again, Automation Direct. Their FACTS engineering group really did their homework and made some excellent hardware that puts for instance Moore Industries to shame. They probably sell “splitters” too but with multiple FACTS you don’t need it. This is a little DIN box that has the circuits in it that AB and other cheap low quality vendors did not
Install in their controllers so that you can loop the inputs through a bunch of “isolators”, then ground returns and otherwise generally do things to fix the garbage hardware on the controller side.
 

·
Registered
Electrician
Joined
·
1,295 Posts
Just out of curiosity, could you loop it through the drives and use one analog input? I am trying to think other than the $45 and a few extra lines in the program, if there's advantages / disadvantages to splitting it in the PLC versus looping it through.
I guess it would be more flexible in the future, you never know if you might want to manipulate the signal in logic rather than pass it right through, you could add a delay between the pumps, lead-lag etc. Also with it looped, if the 4-20ma signal wiring to either pump was damaged it takes both out, with them split it would tolerate that fault.
Thats why I like to have the signal separate. Keep each signal independent, plus if I need to I can smooth out the signal with the plc if needed. If one drive goes down it won't effect the other, etc. Its only one line of code to output the signal now.
 

·
Registered
Joined
·
95 Posts
Most “4-20mA” inputs are really just voltage inputs with a shunt resistor. they are usually mA or voltage by a dip switch or jumper. The trick is to leave one as current, leaving the shunt in, creating a voltage across the terminals. Set the other input as voltage and parallel it with the first input. You can also leave both as voltage and provide your own shunt.
 

·
Chief Flunky
Field Service Engineer
Joined
·
1,800 Posts
Most “4-20mA” inputs are really just voltage inputs with a shunt resistor. they are usually mA or voltage by a dip switch or jumper. The trick is to leave one as current, leaving the shunt in, creating a voltage across the terminals. Set the other input as voltage and parallel it with the first input. You can also leave both as voltage and provide your own shunt.
That’s kind of strange way to do it.

ISA and most books show just run multiple receivers in series. As long as the power supply / transmitter can handle the additional resistor drop its not an issue.

Doing it the way you are doing it reduces load on the transmitter somewhat but has several disadvantages. The first is it’s not standard so may cause troubleshooting issues. Second depending on voltage drop in the wiring and connections you are losing accuracy. Third typically voltage inputs are 0-10 V or +/-10 V. You are using something screwy like 1-5 V which some receivers will do, some won’t. And if the current receiver is replaced and the resistance changes, your other receivers are out of calibration. Finally you can get into common mode issues since voltage inputs are rarely truly differential.

This sounds like a practice from back in the day when for instance on Honeywell UDCs you had to supply your own precision resistor. Those days are mostly long gone.

The reason for the whole shunt resistor thing is that ADC chips are almost exclusively voltage input.

The only time I really like voltage inputs is on high performance loops like servos where the 10-20 KHz of the high capacitance current loop is an issue compared to voltage loops. Calibration is less of a problem because the servo closed loop cancels out the calibration error.
 

·
Registered
Joined
·
95 Posts
Per Splatz's original post, the problem was that the inputs were not isolated. Looping the current signal was not an option. OK, leaving one set for current and the other set for voltage might confuse the next guy if he doesn't truly understand analog signals. That said, set them both to voltage and use a common shunt resistor. 250 ohm for 0-5v(1-5v actual) or 500 ohm for 0-10v(2-10v). The inputs when set to voltage will be some very high impedance so paralleling won't be a load issue as they are still significantly higher then the shunt resistor. The transmitter will never know the difference. It will still be controlling current. Per Splatz's comment, the inputs are on the same controller so they are probably not very far apart, so you will only have a voltage signal essentially between them. Hummmm, single, 25 cent component or $75 or more loop splitter? Since this is storm water control, my bet is NASA grade calibration is probably not necessary.
A surge arrestor anytime your transducer is outside is never a bad idea.
 

·
Premium Member
Working With the Tools
Joined
·
853 Posts
Never removed a PLC to go back to relays etc ... but I know of a couple of places that the PLC was overkill, and just added complexity to a simple control circuit.

With sensors and the lightning issue though, I have put in good ol' blue balls in many overhead water tanks as a safety for WHEN the sensor fails , not IF it fails !
Found one in a green house for a cistern pump that supplied their evaporator wall... it was definitely overkill.
 

·
Premium Member
Working With the Tools
Joined
·
853 Posts
I would probably still be getting calls from them if they had gone with a PLC. It's a very funny thing what's happening in business. I was three steps away from the owners here, subbed by a vendor to the engineering company that does the work for the owner company. Something changes every couple years - the engineering companies buy each other, sell off contracts, etc., the salesman from the vendor company changes jobs and manages to bring the contract to his new company.

Between the layers nobody has the discipline to maintain the drawings and documentation and make sure it stays up to date and gets transferred end to end. When one outfit has things figured out, something changes and some new guy is there tugging wires and scratching his head.

If I had managed to make myself indispensable they'd probably still be calling me. Probably not with that little PLC though. My pitch was to replace that pump controller with a PLC that did exactly what they want, with all the logic in one brain, rather than hack around with the pump controller config to get as close as possible. And it would have been easy to incorporate a couple other aging systems run on relay logic into the same brain. Of course I'd still have to remember to put that sticker in the cabinet :)
Oh man... you could have done that so easily. :LOL:
 
1 - 20 of 24 Posts
Top