[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math

View post   
File: 42 KB, 720x389, plc-automation.jpg [View same] [iqdb] [saucenao] [google]
10791776 No.10791776 [Reply] [Original]

These things are used in just about every industrial environment. They seem outdated af. Is there some way to replace them? Maybe the Internet of Things?

>> No.10791798
File: 14 KB, 340x255, da_vincis_inquest.jpg [View same] [iqdb] [saucenao] [google]
10791798

>>10791776
>Maybe the Internet of Things?

>> No.10791837

they're used because they're simple and reliable. they're also not that antiquated, Siemens PLCs which I use a lot are extremely powerful

>> No.10791858

They are already being replaced.

>> No.10791861

>>10791858
With what?

>> No.10791870

>>10791776

They will not be replaced.
There are many processors in the world, but PLCs are not just about the processor; they are about the standardization of things you can hook up to them. They are about the interface to the equipment and the motor drivers and whatnot.
They are about matching the speed of commands and standardized buses....

You are not going to replace that because there is no need to. Automation just doesn't need to operate that quickly to require any kind of improvement...

>> No.10791874

>>10791776
PLCs are about reliability. Not features.

>> No.10791886

>>10791776
>They seem outdated af.
No they don't.

>Is there some way to replace them?
With what? They do what they are meant to do.

>Maybe the Internet of Things?
Nice buzzword

>> No.10791895

>>10791837

I mean I respect the PLC usage but let's be real, they are just arduinos that cost a lot and use a shitty programming language.

This whole reliability thing is just sales talk to justify the insane prices these 8 bit micros cost.

>>10791886

Maybe we can do it in the cloud?

>> No.10791905

>>10791895
5/10 almost got me

>> No.10791911

>>10791905

> He drank the koolaid

Anon look deep inside and you will realize ladder logic blows and you're just using a expensive micro with easy to program gpio

>> No.10791919

>>10791776
The fuck? FPGA are both cheaper and faster. Why would any industry fall for this bullshit 'logic card'. It's like you fucks want to go back to punch card computation.

>> No.10791923

>>10791895
>This whole reliability thing is just sales talk
>t. pinto driver

>> No.10791926

>>10791776
if it's not broken, why fix it?
>>Internet of Things
great your factory has been destroyed and took the buildings surrounding it with it because a crafty hacker set the temperature controls way too high starting a fire

>> No.10791934

>>10791926

> Suxxnet

>> No.10791941

>>10791919

What do you need a fpga for? This shit mostly drives motors and stuff. You don't need much more than a 8bit micro though arm chips are probably getting cheaper than 8bits and replacing those.

Fpga would just complicate things. Idea is this shit be dead simple.

>> No.10791943

>>10791919
>FPGA are both cheaper and faster.
Uh...

>> No.10791948

>>10791861
I don't know.

>> No.10792017

>>10791776
>nuts and volts seems outdated af. Can we replace them?

>> No.10792047

>>10791941
Why would you want anything mechanical to have 'that' much processing power? Any point of kinesthetic expression requires mechanical engineering, because all computation can be offset by a conductive line back to whatever CPU is required.

FPGA > PLC

PLC is for fucking engineers that don't even know these things exist: https://en.wikipedia.org/wiki/Aeolipile

>> No.10792777

>>10792017

Maybe IoT and synergy with the cloud could modernize this.

>> No.10792783

>>10791926
PLCs have no protection or security at all. If you have an ethernet cord you can do anything you

>>10791911
To be fair, ladder logic is great for programming a bunch timers and counters. To do it in c++ you'd need to know multi threading real well.

>> No.10792853

>8-bit
>8-bit
>8-bit
>8-bit
You adruino losers are cute

>> No.10793007

>>10791895
That shitty programming language is for electronics technicians because compsci kids don't want to get their hands dirty.

>> No.10793011

>>10791911
Modern PLCs can convert ladder logic to function block code or script, it's not that bad.

>> No.10793017

>>10792783
If you have an Ethernet cord and plug it into the right jack you can take anything down, that's why physical security exists.

t. substation automation tech

>> No.10793322

>>10791895

>I mean I respect the PLC usage but let's be real, they are just arduinos that cost a lot and use a shitty programming language.

Their languages use 1/10th the space and 1/100th the power ardunios use, they're built to be optimized in process machinery.

>This whole reliability thing is just sales talk to justify the insane prices these 8 bit micros cost.

Most PLCs are run by people without college degrees and who sometimes don't know how to use a landline. If the machine breaks it could take anywhere from 12-48 hours for a guy to show up to reset it, and 2-3 weeks for an actual repair job. This is an unacceptably large amount of downtime for any business, PLCs are simply and easily swappable without needing to get technical.

Simplicity is key. You're not using your imagination and thinking about the extremely broad spectrum of people who use PLCs. For example illiterate farmworkers running a sorting machine, or convicts stamping license plates.

>> No.10793331

>>10792047

>Why would you want anything mechanical to have 'that' much processing power?

So when the dead man's switch is released, the machine only stops operating and doesn't fully power down. This basic safety feature requires a PLC unless the factory wants to risk large amounts of downtime due to power cycling, which also harms the machine's lifespan. Ditto for any machine that only does one or two movements, but across different settings controlled by a central desktop PC. There's no reason to give it a dedicated machine.

>> No.10793428

>>10791776
I'm triggered. That's the dumbest thing ever.

>> No.10794393

>>10792783

You clearly have never done embedded programming......

>> No.10794733

>>10791776
I'm an engineer who develops PLCs. IOT stuff is working its way into building automation really hard. Everybody wants to control their lights and heaters with an app and send energy usage data to the "cloud". However the core ladder logic still remains because that's what technicians understand and they still needs to support 80 year old protocols because there's so much old equipment still in use. We released a model with only USB, Ethernet, and WIFi a few years ago and six months later we were shipping it with ugly USB to RS232 and RS485 adapters hanging off.

>> No.10795466

PLCs are amazing, fuck you op

>> No.10795598

>>10791941
just get a cheap $10 dollar FPGA then

>> No.10795856

>>10791776
They exist because dummies can program them of necessary. You can replace them with hard realtime systems but then you would meed a realtime programmer.

>> No.10795970

>>10795598
>FPGAs are magic

>> No.10796013

>>10795970
>implying copy and pasting verilog from stackoverflow is hard

>> No.10796648
File: 12 KB, 258x245, 1505950035706.jpg [View same] [iqdb] [saucenao] [google]
10796648

>>10793322
>people without college degrees and who sometimes don't know how to use a landline

>> No.10796925

>>10796013
>implying all sorts of implications
FPGAs have their place, but this is not it.

>> No.10797110

>>10796925
What the fuck do PLCs even do? Not the guy you're replying to, but I know a lot about FPGAs so curious

>> No.10797150

>>10797110

PLC's are used to control inputs and outputs from sensors, motors, actuators, and other industrial equipment. They have ladder logic which is simple but good for intensive parallel programming, and multiple inputs, outputs, etc.

>> No.10797207

>>10791776
PLCs are definitely not outdated. I don't know why you think so. IoT doesn't even make any sense as a viable candidate. You are either ignorant of industrial needs, or you are shorting ROK.

Do you know about remote i/o?
Black box troubleshooting?
SFCs?
Safeties?
Industry standards are made at a very simple level so people don't get killed and machines stay running.

>> No.10797212

>>10797150
are we talking mostly analog signals?

how is the ladder logic implemented? in an ASIC or just with a few simple as fuck ICs?

>> No.10797240

>>10797207
Pretty much this. OP is a faggot as always.

>>10797212
Both digital and analog signals. Some PLCs need an extra module for analog inputs and outputs, but some of them come with digital inputs and outputs, and few analog inputs and outputs. It all depends on the manufacturer, and some of them offer a better thing against the other. Also, you can also program the controller with functional blocks, or even grafcet, if you don't want to deal with ladder logic.

>> No.10797242

>>10791895
>I mean I respect the DAC usage but let's be real, they are just diodes and resistors that cost a lot and have no programming functions.
>This whole reliability and signal accuracy thing is just sales talk to justify the insane prices these 8 bit micros cost.
>>10791919
>The fuck? NAND logic on a piece of silicon are both cheaper and faster than dedicated hardware circuits. Why would any industry fall for this bullshit 'video card'. It's like you fucks want to go back to punch card computation.
>>10792047
>Why would you want anything mechanical to have 'that' much processing power? Any point of kinesthetic expression requires mechanical engineering, because all computation can be offset by a conductive line back to whatever CPU is required. As a matter of fact I suggest we wire everythnig up to a satellite for cloud computation so we can amplify the delay from signal to actuation.

>> No.10797264

>>10797240
>how is the ladder logic implemented? in an ASIC or just with a few simple as fuck ICs?
what about this?

>> No.10797277

>>10797264
Transistor, and logic gates. That's about it. As for analog signals, Digital to Analog converters.

>> No.10797310

>>10797277
>Transistor, and logic gates
So you're saying it is very simple logic with no need for a high level of data processing?

>> No.10797335

>>10797310
It does have a level of data processing, but it is not as high or complicated as you might think. Sure, it depends if it has another features like working on a network or a ModBus, but that is handled by another driver with communication with the PLC's main processor and specific structure of data for communication only.

>> No.10797356

>>10797335
>with the PLC's main processor
so the PLC does have a processor like a CPU? what kind of data processing does it do?

>> No.10797393

>>10797356
It does, but its processing power is something closer to a microcontroller, and it depends of the manufacturer. As for the data processing, again it depends of the manufacturer offers and options but most of it is just memory storage, logic operations, enable flags, some aritmethic operations, possibility to display text, signal processing and sensors data scaling (converting a analog signal from 4-20 mA to a real number), and even some have a PI controller programmed, you just need to adjust the controller parameters.

Remember that the PLCs offer reliability and robustness for the process control and machinery control signals. And their programming, as well some other things like networking is standardized already with a lot of protocols behind it, make it easy to implement (exceptions apply) on most industrial process out there.

I recommend you to, if you want to learn a bit more, play around with some software for PLC programming or automation system simulations. On the first case I recommend you using LogoSoft from Siemens, just the demo version to play around, and for the second one Automation Studio or anything other Open Source alternative you can find.

>> No.10797456

>>10797150

> But this is totally not like a cheap micro!

Just be honest, it's a well tested micro that has a good tolerance for abuse / retard and is easy to program for people who shouldn't program.

>> No.10797460

>>10797456
Yes. One of the things about industrial automation is simplicity and robustness. Your point being?

>> No.10797469

>>10797460

You're (well your company is) paying ALOT for a micro that costs a few cents in qt1. Y'all getting hustled by Siemens and friends.

>> No.10797473

We use PLCs for the same reason bank software doesn’t run on consumer-grade chips and because they’re more retard resistant than a general purpose PC

>> No.10797474

>>10797469
If you think you can provide the same quality at a lower price point, why not make a business out of it? You probably wouldn’t convince big companies to switch but you could sway mom and pop shops

>> No.10797477

>>10797469
Right, does a micro comes with a case to support the harsh environment of a industrial process of any kind? Does a micro has a ready to go control logic or a PID controller ready to use and easy to adjust? Does a micro has support from the company that sells you the equipment if you have a problem (unless you tamper it)? Does a micro focused on industrial process is retard resistant and has a great reliability for work a long as time with minimal checking and maintenance as it is? Until something better, cheaper and that accomplish all of the above I will stick to PLCs, no matter the manufacturer.

>> No.10797532

>>10797393
Ok, that all makes sense. I'm just failing to see the difference between a PLC and a typical microcontroller

>> No.10797536

>>10797473
>bank software doesn’t run on consumer-grade chips
What does it use? Industrial grade? In my experience the difference between comsumer-grade and industrial-grade chips is typically things like shock/vibe and temperature rating. not sure how that would be needed for bank software

>> No.10797566

>>10797536
Mainframe processors, such as IBM’s z13 and z14, which in general are a lot more robust and fault tolerant than x86 chips. They also have more redundancy built in, if an x86 chip fucks up a calculation it just results in something like a miscolored pixel, if a mainframe processor fucks up a calculation in a banking setting it could mean millions of dollars lost

>> No.10797575

>>10797566
I get what you're saying but I have to make a semantic/pedantic statement. I would still refer to the z13 "chip" itself as consumer-grade. It has a more fault-tolerant architecture (RTL level), but there does not seem to be anything different about the way the chip is manufactured. these things are probably shit out of the same process as other 22nm "consumer" chips (i could be wrong though, too lazy to find this info).

>> No.10797578

>>10797532
Look at it this way. A PLC is a uC packaged in relatively standardized hardware/mechanical package that can be easily swappable. Instead of using text based programming language, they use ladder logic, a form of visual programming that's easier for process/production engineers to comprehend.

>> No.10797583

>>10797575
Well, yes, at the end of the day you’re laying down silicon. The difference is in the specs they have to perform to, I’d be willing to bet that even chips NASA uses for space applications are made using largely the same processes as x86 chips. For example, I work in aerospace manufacturing and the biggest difference between making commercial and military engines is the specifications, we use the same mills/CNCs/etc to make parts for both engine.
What would a higher-than-consumer grade chip look like to you?

>> No.10797595

>>10797583
>ven chips NASA uses for space applications are made using largely the same processes as x86 chips
Maybe for some smallsats that only need to survive in LEO for a couple years (and even then they will add extra reliability with the board design and other components) but that is absolutely not true when it comes to longer-term missions or harsher orbits or deep space / Mars. They have chips specifically made radiation-hardened with a completely different manufacturing process. Look up "radiation-hardened processors"

Take a look at Xilinx chips. They have consumer grade, industrial grade, military/defense grade, and space grade

>> No.10797620

>>10797595
Neat, I’ll take a look at Xilinx
Anyways, PLCs need to operate in harsh environments but not space harsh, think Arizona in a summer day near a large furnace harsh. A microcontroller built with standard processes is good enough(tm).
All a PLC needs to do is provide an application-agnostic control loop based on input/output modules, ladder logic lends itself well to this application because it’s easy to debug since it’s visual, simple for boomers coming from a relay-logic background to understand, and just powerful enough to do process control.
Also keep in mind that when ABB sells you a PLC, you’re also buying the environment and standardization around the PLC. Could you potentially build a process control computer with an Arduino? Probably, I wouldn’t trust it to be as rugged as a full PLC solution.

>> No.10797643

>>10797620
Xilinx has basically given up on making radhard chips but their other chips are still pretty widely liked by the space community for missions that don't need that high of a level of reliability. Companies like BAE Systems have a much higher track record of making radhard chips. I work in this area so I know quite a bit about the state of the field. There is also a pivot to using more commercial parts leveraging shit from automotive and other industries but this comes with its own challenges and there are certain missions that will always need radhard and not commercial

Cool discussion though. It's always weird to see some of the distinctions in how certain types of systems are classified. sometimes it can be pretty arbitrary

Do you happen to know which processor chips are typically used in PLCs?

>> No.10797663

>>10797643
How does one go about radhardening? After hearing about the Chernobyl miniseries I have been scouring for the few good documentaries on the event and the failure of the Joker rover was really interesting. Obviously you could cover everything is a thick wall of lead but I figure weight has to be a huge concern.

>> No.10797676

>>10797663
I have limited understanding of the actual physics behind it but there are several techniques

>rad hard by process
use an insulating oxide layer in the chips which helps somehow
>rad hard by design
redundancy within transistor layout and something called guard rings
>rad hard by architecture
basically doing extra shit in your board design or software design to support the chip
>shielding
this works but does not protect perfectly against everything just extends your lifetime. also adds size/weight

I assume there are differences in the hardening based on the environment (space would be different than nuclear plant and that would be different than nuclear plant after meltdown). I have no idea what they put in the Joker rover but i'm bet me left nut it wasn't designed for such an event

>> No.10797707

>>10797676
Thanks anon.

>> No.10797725

>>10797707
I should add

rad hard by arch is the least well defined here. this could really be any fault tolerance technique like adding triple modular redundancy into your FPGA circuit, using error correcting codes, or anything. only the first two i mentioned are actually dealing with how the chip is fabricated

>> No.10797729

>>10797707
>>10797725
ok i keep adding shit.

none of these will allow your chip to last forever in a radiation environment. they are rated by several factors such as Total Ionizing Dose (cumulative effect that alters silicon physically) and Single Event Effects (bit flips), and there is basically always a limit with TID. SEEs can potentially always be mitigated (i think) with a few rare exceptions. as transistor sizes shrink TID is becoming less of an issue with newer chips but SEEs are becoming more of an issue

>> No.10797732

>>10797643
No clue, I know Allen-Bradley gives full spec sheets on their PLCs, maybe the info is in there
RSLogix 5000 series is their most modern product line iirc

>> No.10797749

>>10797732
Looks like they use Intel Atom. Always wondered what these things were used in cause they have basically no compute performance

>> No.10798353

>>10791934

>Stuxnet

The most sophisticated software in human history, custom-tailored for unique ASICs that needed decryption keys from two separate parts of the most secure building of the worlds largest microchip manufacturer - stolen by commandos from the intelligence apparatus of the sole world super-power, who dedicated its best and brightest programmers to design this worm.

https://www.quora.com/What-is-the-most-sophisticated-piece-of-software-ever-written-1

>> No.10798369

>>10791776
>These things are used in just about every industrial environment
Yeah.
>They seem outdated af.
For what? For PLCs the primary goal is predictable, repeatable, consistent and reliable performance. If you shell out thousands for a control system you want it to work. They exist to replace boxes full of switches, relays and buttons for basic control logic, not to do fancy processing.
>Is there some way to replace them? Maybe the Internet of Things?
You can offload the processing to a less reliable device where the connection might be interrupted, sure. But why would you want to? You probably want a bit of basic, reliable control to remain on site, physically accessible. Which just takes us right back to PLCs.

You need to understand that control systems have layers. PLCs are local. They manage IO. They serve as the modular core of control systems that are often unique or only made in small numbers, so each time you have to design from scratch. They can interact with more complex systems as is. So they're here to stay. Some might include things like REST APIs and HTML servers, but the basic concept is gonna be around as long as small batch manufactured industrial equipment is.

t. worked in industrial automation for a while, designing and programming control systems for factory equipment.

>> No.10798381

>>10797356
PLC is a term for the end product package, they can use any number of different system architectures so long as the programming process complies with that IEC standard. There are some on the market that are literally just arduinos with a big ass shield board - so 16bit AVR microcontrollers you flash directly - while others might have full Linux OS running on ARM with a gigabyte of RAM.

>> No.10798862

>>10797566
>built in redundancy
active or passive replication?

>> No.10798884

>>10797676
>rad hard by architecture
Theoryfag here working in critical systems. What fault model do you guys assume in industry(byzantines, crash, omissions, etc), and do you assume synchrony in your devices?

>> No.10799943

>>10797477

> you make the purchasing decisions and aren't a slave to certified vendors

Kek

>> No.10799947

>>10798884
I'm not in industry. I'm a government researcher. I don't totally understand your question but I'm thinking they would typically cover all types of faults and that Markov models are often used

>> No.10799987

>>10791776
They're simple, reliable, basic and get the job done. Despite not looking like the sci-fi tier automation technology that engineering undergrads want to see, there's no reason to think they'll be replaced any time soon, if ever.

>> No.10800002
File: 24 KB, 636x213, c_fit,fl_progressive,q_80,w_636.jpg [View same] [iqdb] [saucenao] [google]
10800002

>>10797595
>>10797620

Lot of shit is running 486's I believe. I guess it takes a super long ass time to actually test and vet this stuff. Space has some pretty crazy standards. (Read their cabling / crimping shit they don't fuck around)

Their programming stuff for newer probes is pretty neat, I think they are using Lua or maybe it was erlang due to how important hot code reloads are and being able to reliably fall back if something fails. I can't imagine the stress probably takes years off your life.

>> No.10800294

>>10797469
>paying for a micro
I'm not though. I'm paying for the PLC system, of which the micro is a part of. With a PLC you also get in buil IO, code libraries for process control and communication, guaranteed low failure risk, a whole ecosystem of plug and play expansion modules, etc. It's painfully obvious that you've never had to actually make something where money is on the line. Most PLC systems are one offs, the cost of labour in design and programming, not to mention all the IO easily swamps the cost of the controller itself. If you can rapidly put together a system THAT YOU KNOW WILL WORK you make back whatever you could've saved by jewing on one bit of hardware.

Bear in mind that if anything fails you will have to go back and fix it.

>> No.10800430

>>10799943
Eat a duck, lad. I don't have time to build a controller for every job.

>> No.10800683
File: 160 KB, 1362x1004, LNT-main.jpg [View same] [iqdb] [saucenao] [google]
10800683

This is why American manufacturering is shit.

>> No.10800688

>>10800683
PLCs are used in every mainstream manufacturing facility, world wide.

>> No.10800691

>>10800683
Don't you mean Chinese manufacturing?

>> No.10800705

The only realistic way of replacing them that I see, is using similar devices (in terms of IO) that look less shit (which is being done little by little from what I have seen) and doing everything via function-block or something similar, all be it nicer. That way it has the crucial advantages from before without the problems that go with reading shitloads of ladder (I think it should slowly be faded out since FB is just as good). Most importantly, it needs to be easier to use when connecting (they always take forever to set up). Also Rockwell > Siemens and Schneider

>> No.10800738

>>10800683
I have personally set up PLC controlled systems in a whole bunch of third world shitholes.

>> No.10800764

>>10800705
Most of it is done in function block, or has logic compliant with IEC 61131, which means the logic can be converted back and forth from function block no problem. They use ladder logic a lot because automation techs know it.

>> No.10801484

>>10800683
Germany is the standard for PLCs, retard

>> No.10802128 [DELETED] 

>>10791776
Because it's made up of multiple species of single cell parasites that go through a life cycle in and out of the body
The fact that we know anything about how it works on the low level is pretty amazing let alone a 40% effective vaccine

>> No.10803335

It's all about stability and simple