Thread: Deadlines

  1. #31
    (?<!re)tired Mario F.'s Avatar
    Join Date
    May 2006
    Location
    Ireland
    Posts
    8,446
    Quote Originally Posted by swgh View Post
    I'm guessing automation has something to do with transport?
    Not really. Automated and semi-automated processes. Usually factory production processes like production lines. But also all other sorts of other operation processes, like controlling a nuclear plant, developing the world's largest automated warehouse and distribution center, managing an oil rig, controlling an airport traffic, scheduling traffic lights, managing a public transportation network, etc...

    (*) Transforming warehouse automation - YouTube
    Last edited by Mario F.; 09-24-2016 at 01:31 PM.
    Originally Posted by brewbuck:
    Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.

  2. #32
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    Yes -- basically specialized industrial equipment. The rule for automation as a whole is basically to do custom microcontroller boards for things where the production size is larger (and therefore justifies the development cost vs. hardware per unit), or PLCs and similar equipment for things with smaller production size. In my (more specific) line of work, I'll program a PLC (programmable logic controller, basically an MCU with hardened I/O) to control oil & gas production measurement skids (where programs are used on anywhere from 1-20 units usually).

    So the (my) problem is the PLC programming software. It's usually total crap that is poorly designed and buggy. And the programming languages for the PLCs pretty much suck too, the only thing I can really stand is structured text, which is basically Pascal. I think there's room for CS people with desktop app dev experience to make much better software. Other problem is making that happen. These PLC companies obviously don't want to pay someone to change their software for the better when they can continue to sell their crappy software for outrageous prices ($1,200/seat/yr roughly for Rockwell products last I checked).

  3. #33
    (?<!re)tired Mario F.'s Avatar
    Join Date
    May 2006
    Location
    Ireland
    Posts
    8,446
    That's a coincidence. I've been working for the oil industry too, Epy. At a higher level of programming, though. We have one system in place; a work shift management program that was test-driven two years ago on an oil platform and is expected to expand to 3 platforms still this year. We've been contracted to also present by June next year a stock (warehouse) management solution for individual rigs, which is the reason I delayed my going back to Portugal until next year.

    As for your post, I have a thing for the automation software industry and I consider it the most exciting field in the whole software development industry, for someone who wants a field that puts CS to the test. Other than the research fields, it's the field that best represents the task of finding computing solutions to real world problems. It's amazingly exciting from a software development perspective and I envy you. Other fields like web development, game development, or my own of business development, are not comparable in any way. We are basically LEGO players, reshaping and reusing the same old and well-documented solutions, over and over again, with only very rare and very small moments of true problem solving. Whereas in the automation industry, every project is a new CS challenge.

    I didn't know the PLC hardware was gamed. Which is a shame, really. However, after some time I stopped understanding that market, to be honest. I think that around the early 00s, PLCs became a very targeted solution, but many systems around the world are still being designed around this type of hardware without a need to do so. PCs and the explosion of easily accessible micro-controllers do scale down very well these days to harbour automation systems and they seem to me result in cheaper contracts. I learned from a documentary on TV that the Heathrow Airport moved from a PLC solution to a micro-controller and PC-based automation solution for all their operations, including a massively complex algorithm to manage its runways (10,000-15,000 planes a day).
    Last edited by Mario F.; 09-25-2016 at 05:30 AM.
    Originally Posted by brewbuck:
    Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.

  4. #34
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    Lol, with what I'm doing I'm pretty much doing the LEGO work. I like to try out different PLCs so I'll have to figure out comms and I/O on that PLC, but the programs are pretty much ported and modified as needed. A lot of our well measurement skids are the same, so. Every once in a while I do get to mess with something new though.

    In a nutshell, software for the automation industry sucks. And it's a lot of simple stuff too -- I just used a package called FlexiSoft -- lot of places where you can tell 1 thread is being used to do the work and also update all windows, so when it has to do anything, it's unresponsive. Too many PLC companies only care about if it works or not, not if it works well or is user friendly. A company called Red Lion has a package called Crimson that is currently the gold standard in my eyes. Founder was a CS guy and it shows. The software is so super nice and easy to use. Just wish their hardware wasn't so expensive...

    Point I'm driving at, if I were a software guy, or maybe if I wasn't lazy, there's plenty of room here for better software. If you put out high quality stuff, you'll be recognized right away by control systems and automation guys like me.

  5. #35
    Its hard... But im here swgh's Avatar
    Join Date
    Apr 2005
    Location
    England
    Posts
    1,688
    Epy - why not attempt to write your own? I mean at least a working template or demo.
    Language probably does matter in this case - I would personally use ADA for such an
    industry spec task - but t think C/C++ will get the job done.

    Why not assemble a small team or at least solo part of it? Worth a shot. As you said,
    "Good Software = Excellent Recognition"
    Double Helix STL

  6. #36
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    Believe me I've thought about it. And I also came to the conclusion that Ada would be a good language (designed for realtime systems, very strict, etc.). IEC 61131-3 is the standard for PLC programming languages and one is structured text, which is pretty much Pascal, which is very similar to Ada.

    Problem is that each PLC brand has its own package, can't just write your own software and use it with any PLC you want. It's dumb because even though the languages are standardized (which no one really follows appropriately), every PLC is still different -- there's device specific code of course. So to do what I want, I'd have to make my own PLC, which I have also thought about, but I'm not an electrical engineer and it would take a lot of money to get proper approvals etc. (UL, FM, CSA, etc).

  7. #37
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    Related, also thought about making a source to source compiler. Some packages take XML files of variables, code, etc. Could come up with my own software package that transcompiles to another package, but seems kind of lame to me.

  8. #38
    (?<!re)tired Mario F.'s Avatar
    Join Date
    May 2006
    Location
    Ireland
    Posts
    8,446
    I wasn't aware the PLC market was in such a state. I always assumed the problem was price. But that surely helps explain the interest in homebrew and alternatives like single-board computers with modbus drivers.
    Originally Posted by brewbuck:
    Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.

  9. #39
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    Quote Originally Posted by Mario F. View Post
    I wasn't aware the PLC market was in such a state. I always assumed the problem was price. But that surely helps explain the interest in homebrew and alternatives like single-board computers with modbus drivers.
    Price is a big problem to be sure (cough Allen-Bradley), but everyone being all cathedral doesn't help either. I've been looking hard at the SBCs and what it'd take to get UL approval etc just for that reason. See that's the outrageous part about it that kills me, these high-end PLC CPUs probably have the computational power of a smartphone or a pentium 4 etc, yet they are thousands of dollars...robbery. People can argue that you get more reliability, and they have to have that money to maintain their software etc. Rubbish. A lot of these hobbyist SBCs are made with I guess you would call it commercial grade components (0-50/60 C?), can easily get the same chips for -40 to 85, put a fatty heat sink on it, adequately space components (not trying to belittle electrical or thermal engineers who put a ton of effort into this, just making the point that it's overpriced, I know it's not this simple).

    With the ESP32 coming out recently (dual-core 32-bit 240 MHz processor, more program memory than most all PLCs available), and the new Teensy 3.6 (180MHz Cortex-M4 with FPU and DSP, plenty of flash memory as well), just kills me inside to be buying these PLCs. Again, I know it's not apples to apples, but for instance, $2500 for a PLC with 768KB of program memory? Please...

  10. #40
    Registered User
    Join Date
    Apr 2015
    Posts
    16
    With regards to deadlines, you should handle it with patience and be honest enough to managers and superiors who gave you the task if you aren't sure that you can finish it right away. Compromise is a keyword to settle everything with ease.

Popular pages Recent additions subscribe to a feed

Similar Threads

  1. deadlines....
    By goran in forum A Brief History of Cprogramming.com
    Replies: 5
    Last Post: 11-21-2001, 07:07 PM

Tags for this Thread