Announcement

Collapse
No announcement yet.

Temperature control ?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Temperature control ?

    is PID controller (a proportional integral–derivative controller) the best way to control the temperature? Example: BGA Workstation.

    Do I have another option for more accurate & fast response temperature controller? maybe Arduino micro controller with AI?

    --------------------------------------------------------
    over 90% of the controllers used today use some form of P-I-D control.

    PID control dates back to the 1890 meaning that a majority of the vehicles used today in industry are using a 126-year-old controller
    methodology.

    Is there an alternative to PID control............a Fuzzy Logic Controller (FLC).

    PID temperature controller DIY Arduino
    https://www.youtube.com/watch?time_c...&v=LXhTFBGgskI
    Last edited by capwizard; 02-18-2019, 08:20 PM.

    #2
    Re: Temperature control ?

    Heard people talking about "bang-bang" control being different from PID control, where really IMHO it's just a special case of PID where the the gain is fixed no matter what the differential is.

    I doubt using "AI" would be much different unless you hook it up to a camera that checks out what you're soldering and adjusts power accordingly...ahh why am I leaking out patent ideas?

    Comment


      #3
      Re: Temperature control ?

      Originally posted by capwizard View Post
      is PID controller (a proportional integral–derivative controller) the best way to control the temperature? Example: BGA Workstation.

      Do I have another option for more accurate & fast response temperature controller? maybe Arduino micro controller with AI?

      --------------------------------------------------------
      over 90% of the controllers used today use some form of P-I-D control.

      PID control dates back to the 1890 meaning that a majority of the vehicles used today in industry are using a 126-year-old controller
      methodology.

      Is there an alternative to PID control............a Fuzzy Logic Controller (FLC).
      There are all sorts of control strategies (even PID can be implemented with many variations).

      You need to look at the "system" that is being controlled:
      • what sort of control "valve" do you have (if you can only say "HEAT ON" and "HEAT OFF" then you want a controller that thinks in those terms)
      • what sort of actuator delays (how long from when you call for "control" to when that control actually materializes)
      • what sort of sensor characteristics (how accurate, repeatable, etc.)
      • what sort of measurement delays (how quickly can you "see" the changes that the controlled variable is experiencing)
      • how sophisticated you want the control ALGORITHM to be


      PID is a very mature technology. A fair bit of research has gone into how to best "tune" a PID controller (which, in theory, must be "tuned" to match the transfer characteristics of the system being controlled) -- to get maximum performance from the system-in-control.

      But, PID harkens from the days of ANALOG controllers. Nowadays, there's a lot more smarts that can be applied that make the controller very nonlinear (virtually impossible to model in the analog domain).

      For example, your home thermostat can EVENTUALLY learn the characteristics of your home. Armed with data regarding exterior conditions (solar radiation, wind losses, etc.), it can keep your home "right on" desired temperature.

      OTOH, open the back door for 5 minutes as you haul stuff out to the barbeque and its learned model of the house is shot to sh*t! OToOH, let the controller (thermostat) KNOW that you've opened the door and it can switch to an alternate "house-with-door-open" model to improve its control despite the huge "leak" you've introduced.

      (think of applying the soldering station TO the BGA vs. just letting it sit idle)

      Comment


        #4
        Re: Temperature control ?

        The temperature controllers around us are everywhere. How do I know this temperature controller is controlled by PID?

        I have one more question?

        Industrial PID controllers look very similar in appearance, what is the difference between cheap and expensive one? Let me give you an example what is the difference between $8 PID from China and $135 PID from Japan or both functions are similar just the difference in quality?
        Last edited by capwizard; 02-19-2019, 05:08 AM.

        Comment


          #5
          Re: Temperature control ?

          Originally posted by capwizard View Post
          I have one more question?

          Industrial PID controllers look very similar in appearance, what is the difference between cheap and expensive one? Let me give you an example what is the difference between $8 PID from China and $135 PID from Japan or both functions are similar just the difference in quality?
          This is true for most things that the more money you spend on an item the better it is most of the time but greatly depends on where it was made a lot more of the time at least that is what I find to be true

          What I think is a better solution is have a PWM controller for heating element with a PID controller for the temperature all tied together in one unit


          Where I work we have a Break Press that has a PDI temperature controller for heaters in the hydraulic oil storage tank and it over shoot the temperature by at least 5 degrees F most of the time until it get closer to cut off temperature then swing is not so bad
          Last edited by sam_sam_sam; 02-19-2019, 08:57 AM.
          9 PC LCD Monitor
          6 LCD Flat Screen TV
          30 Desk Top Switching Power Supply
          10 Battery Charger Switching Power Supply for Power Tool
          6 18v Lithium Battery Power Boards for Tool Battery Packs
          1 XBox 360 Switching Power Supply and M Board
          25 Servo Drives 220/460 3 Phase
          6 De-soldering Station Switching Power Supply 1 Power Supply
          1 Dell Mother Board
          15 Computer Power Supply
          1 HP Printer Supply & Control Board * lighting finished it *


          These two repairs where found with a ESR meter...> Temp at 50*F then at 90*F the ESR reading more than 10%

          1 Over Head Crane Current Sensing Board ( VFD Failure Five Years Later )
          2 Hem Saw Computer Stack Board

          All of these had CAPs POOF
          All of the mosfet that are taken out by bad caps

          Comment


            #6
            Re: Temperature control ?

            That's true, my friend. I've owned so many multimeters, the one last is the Fluke.

            Comment


              #7
              Re: Temperature control ?

              Originally posted by capwizard View Post
              Industrial PID controllers look very similar in appearance, what is the difference between cheap and expensive one? Let me give you an example what is the difference between $8 PID from China and $135 PID from Japan or both functions are similar just the difference in quality?
              There are two aspects to a commercial controller: hardware and software (and a slew of other issues like support, build quality, etc.).

              The hardware is essentially an A/DC to sample the value of the field variable and a D/AC to drive the control valve. So, there's an issue of quality of each of these (accuracy, repeatability, etc.) as well as the design/layout of the circuitry (how far down is the noise floor?).

              Do you support 4-20mA inputs? Outputs? Direct thermocouple connections (cold junction compensated)? What happens if some nasty signal is impressed on the input (or output) -- will your controller emit magic smoke?

              Software is trivial (ISA form):
              Vo = Kc (E) + Kc (1/T1) Sum(E)dt + Kc (Td) d(E)/dt
              where:
              • E = error term (observed value vs. setpoint)
              • Kc = controller gain
              • T1 = integral term (inverse minutes)
              • Td = rate term (units of minutes)
              I.e., you can write the code to implement it in less time than it takes to read this post!

              But, implementing it CORRECTLY is a different story!

              Are you sampling at a fixed rate? How fast? How much jitter? Do you compensate for jitter or just let it muck with the transfer function?

              Do you implement wind-up protection on the integrator? Do you support a deadband? Do you clamp the contributions of each of the error terms to the final output?

              How do you TALK to the controller while its running? Do you implement a manual override mode? Do you support bumpless transfer?

              Do you apply any checks to the inputs (what happens if an input becomes "stuck" or disconnected)?

              How robustly have you validated the algorithm's implementation? (do you count on USERS to find your bugs?? what if they are using the controller in an expensive PROCESS... do THEY take the hit in terms of lost product?)

              Comment


                #8
                Re: Temperature control ?

                Originally posted by Curious.George View Post
                There are two aspects to a commercial controller: hardware and software (and a slew of other issues like support, build quality, etc.).

                The hardware is essentially an A/DC to sample the value of the field variable and a D/AC to drive the control valve. So, there's an issue of quality of each of these (accuracy, repeatability, etc.) as well as the design/layout of the circuitry (how far down is the noise floor?).

                Do you support 4-20mA inputs? Outputs? Direct thermocouple connections (cold junction compensated)? What happens if some nasty signal is impressed on the input (or output) -- will your controller emit magic smoke?

                Software is trivial (ISA form):
                Vo = Kc (E) + Kc (1/T1) Sum(E)dt + Kc (Td) d(E)/dt
                where:
                • E = error term (observed value vs. setpoint)
                • Kc = controller gain
                • T1 = integral term (inverse minutes)
                • Td = rate term (units of minutes)
                I.e., you can write the code to implement it in less time than it takes to read this post!

                But, implementing it CORRECTLY is a different story!

                Are you sampling at a fixed rate? How fast? How much jitter? Do you compensate for jitter or just let it muck with the transfer function?

                Do you implement wind-up protection on the integrator? Do you support a deadband? Do you clamp the contributions of each of the error terms to the final output?

                How do you TALK to the controller while its running? Do you implement a manual override mode? Do you support bumpless transfer?

                Do you apply any checks to the inputs (what happens if an input becomes "stuck" or disconnected)?

                How robustly have you validated the algorithm's implementation? (do you count on USERS to find your bugs?? what if they are using the controller in an expensive PROCESS... do THEY take the hit in terms of lost product?)
                Also nice and good but now it has become a lot more complicated in the design of what we are talking about sure you might get better accuracy but at what cost money and other wise
                9 PC LCD Monitor
                6 LCD Flat Screen TV
                30 Desk Top Switching Power Supply
                10 Battery Charger Switching Power Supply for Power Tool
                6 18v Lithium Battery Power Boards for Tool Battery Packs
                1 XBox 360 Switching Power Supply and M Board
                25 Servo Drives 220/460 3 Phase
                6 De-soldering Station Switching Power Supply 1 Power Supply
                1 Dell Mother Board
                15 Computer Power Supply
                1 HP Printer Supply & Control Board * lighting finished it *


                These two repairs where found with a ESR meter...> Temp at 50*F then at 90*F the ESR reading more than 10%

                1 Over Head Crane Current Sensing Board ( VFD Failure Five Years Later )
                2 Hem Saw Computer Stack Board

                All of these had CAPs POOF
                All of the mosfet that are taken out by bad caps

                Comment


                  #9
                  Re: Temperature control ?

                  you hook it up to a camera that checks out what you're soldering and adjusts power accordingly...ahh why am I leaking out patent ideas?........by Mr. eccerr0r

                  After a few nights of contemplation. I think PID technology is mature. There is no room for improvement. The best way is as eccerr0r said... use an infrared camera to detect the temperature of the object then feedback to the temperature controller, even laser thermometer only detects the surface temperature of solder joints only infrared camera can detect the center of the solder joints.

                  The temperature sensor should not put on the soldering iron tip, if possible it should be in the solder joins because the reality of the center of object temperature is different from soldering iron tip.

                  Anti-traditionally... solder iron first detects inside solder joint center temperature is about the room temperature then solder iron uses the full power maybe 800 hundred degrees until the center of solder joint reaches melting point then slow down. Isn't that same principle as my poorman's solder iron? my eyes are as good as infrared camera.
                  Last edited by capwizard; 02-20-2019, 07:29 PM.

                  Comment


                    #10
                    Re: Temperature control ?

                    While the infrared camera is one thing, and it doesn't even need AI, the fun thing is if AI can correctly calculate the thermal mass of the object you're trying to solder and adjust accordingly, before you touch the iron to the object

                    Of course if the AI is wrong and you then decide to solder that tiny little germanium diode after it heated the iron up for working for that big ground plane, oops!

                    Comment


                      #11
                      Re: Temperature control ?

                      an infrared camera or a thermal camera is good for temperature control?

                      Soldering Iron in Infrared
                      https://www.youtube.com/watch?v=7mgu46DL7P8

                      Does Louis owe thermal imager an apology?
                      https://www.youtube.com/watch?v=gRV0cmIj5Ks

                      Comment


                        #12
                        Re: Temperature control ?

                        Originally posted by sam_sam_sam View Post
                        Also nice and good but now it has become a lot more complicated in the design of what we are talking about sure you might get better accuracy but at what cost money and other wise
                        The bigger (in practical terms) problem with a full PID loop is that it's a challenge to tune it properly -- as it needs to be married to the response of the system it is controlling!

                        What you end up with is usually some ad hoc tweaks with no real "science" to defend your choice of parameters... it SEEMS to work but there's no guarantee that its working at its best (or that it is even working better than NO control!)

                        Comment


                          #13
                          Re: Temperature control ?

                          The problem is PID temperature controller uses complicated mathematics and simulated parameter and profile to approach desire temperature. The idea temperature controller is your phone adds the infrared thermal adapter, you just say...Alexa solders that BGA and plays Für Elise for me.
                          Last edited by capwizard; 02-21-2019, 08:04 AM.

                          Comment


                            #14
                            Re: Temperature control ?

                            Originally posted by capwizard View Post
                            The problem is PID temperature controller uses complicated mathematics and simulated parameter and profile to approach desire temperature. The idea temperature controller is your phone adds the infrared thermal adapter, you just say...Alexa solders that BGA and plays Für Elise for me.
                            The math is trivial:

                            Code:
                            input = read_process_variable()
                            
                            error = input - setpoint
                            
                            change = input - last_input
                            accumulator = accumulator + error
                            
                            output = ([B]proportional_gain[/B]*error) + ([B]integral_gain[/B]*accumulator) + ([B]derivative_gain[/B]*change)
                            [Neglecting sign conventions, term clamps, deadband, etc.]

                            The problem is the terms in bold. Pick them wrong and you can end up with a system that is chaotic and driven to oscillation even with minor perturbations.

                            OTOH, you can implement a self-tuning controller (but that's a lot smarter algorithm! )

                            Comment


                              #15
                              Re: Temperature control ?

                              I'm using Watlow series 965 PID Microprocessor-Based, auto-tuning controller, for my hotplate controller. So, I don't like it but I have to learn. But I will focus on PID turning.

                              Series 965 - Watlow
                              https://www.watlow.com/-/media/docum...965-eng-c.ashx

                              What is the difference between the bang-bang controller and the PID controller? he has a very good explanation on this at 17:11

                              A real control system - how to start designing
                              https://www.youtube.com/watch?v=Mbx5IMICS_Y


                              ------------------------------------------------------------------
                              PID Control - A brief introduction
                              https://www.youtube.com/watch?v=UR0hOmjaHp0&t=167s

                              Understanding PID Control, Part 1: What is PID Control?
                              https://www.youtube.com/watch?v=wkfEZmsQqiA&t=474s

                              Understanding PID Control, Part 2: Expanding Beyond a Simple Integral
                              https://www.youtube.com/watch?v=NVLXCwc8HzM&t=371s

                              Understanding PID Control, Part 3: Expanding Beyond a Simple Derivative
                              https://www.youtube.com/watch?v=7dUVdrs1e18

                              Understanding PID Control, Part 4: A PID Tuning Guide
                              https://www.youtube.com/watch?v=sFOEsA0Irjs

                              Understanding PID Control, Part 5: Three Ways to Build a Model
                              https://www.youtube.com/watch?v=qhIjIu-Zk10

                              Understanding PID Control, Part 6: Manual and Automatic Tuning Methods
                              https://www.youtube.com/watch?v=qj8vTO1eIHo

                              Understanding PID Control, Part 7: Important PID Concepts
                              https://www.youtube.com/watch?v=tbgV6caAVcs
                              Last edited by capwizard; 02-22-2019, 05:48 AM.

                              Comment


                                #16
                                Re: Temperature control ?

                                Originally posted by capwizard View Post
                                What is the difference between the bang-bang controller and the PID controller?
                                You can parameterize a PID controller to approximate a 2-step controller.

                                Referencing the previously posted math, you'd set integral and derivative gain to "0" to eliminate those terms from the output.

                                This effectively leaves you with:

                                input = read_process_variable()

                                error = input - setpoint
                                output = proportional_gain*error
                                A 2-step controller only has two output values (call them "on" and "off"). Obviously, the above has an infinite number of output values because the output is a linear function of the error signal. So, the output could be 0.1, 0.2, 1.4, 895, -376, etc.

                                If you set the proportional gain to "infinite", then you end up with just three output values: -INF, 0 and +INF.

                                In a practical implementation (of ANY controller), you put clamps on the output signal to restrict its range to something meaningful. If we say "ON" is +5 and "OFF" is 0, then setting these clamps to +5 for the maximum output and 0 for the minimum gives you the required 2-step output.

                                But, this controller is prone to oscillate as it reacts instantly to any "error" signals; they are directly reflected to the "output"!

                                A practical implementation implements a deadband -- some level of significance below which "errors" (the "error" variable) are ignored (or, handled differently). This desensitizes the controller to tiny (but inevitable) changes as the input hovers AROUND the setpoint (and peeks above it or pops below it).

                                Finally, many 2-level controllers implement hysteresis. This requires a memory element in the controller (e.g., integrator or direct feedback from output). Otherwise, even the use of a deadband just shifts the "region of sensitivity.

                                Hysteresis effectively alters the setpoint so that once the output goes "ON", the error must change (in a compensatory way) significantly before the setpoint will return to its original value -- the value that initially caused the output to turn "ON".

                                E.g., if the (temperature) controller had a setpoint of X and the output was OFF, an input detected at or below X-(deadband/2) would cause the output to turn ON. Simultaneously, the setpoint would be artificially increased (to X+hysteresis/2) to ensure the controller output stays on until the input has made significant progress above the original setpoint. Otherwise, the controller could tun OFF again with just a little bump in the input.

                                [There are college level courses devoted exclusively to control theory. And, more control algorithms than you can shake a stick at! Each has characteristics that are often absent in the others.]

                                Think about how you would implement controls on a PROCESS. E.g., say you're a robot welding together metal cases. How do you CONTROL the welding process so that every WELD is consistent with the design specifications for those welds? There are inherent variations in the materials being used. So, if you adjust your control based on THIS case's welds, will that translate to equivalent performance on the next case? What if the steel is marginally thicker? Or, has a different iron content? Or, the next case is positioned slightly differently wrt the welder?

                                Comment


                                  #17
                                  Re: Temperature control ?

                                  I have taken process controls in trade school about 10 years ago

                                  The person that did video was pretty good at explaining it in simple terms that why I made the comment that I did earlier in post # 8

                                  That just depends on how complicated you want your design be for how accurate you need the system to be if over and under shootings the set point is going to be a problem then you would need a more accurate controller to do the job that you need to do

                                  https://www.youtube.com/watch?v=Mbx5IMICS_Y

                                  The bang bang controller that he talks in his video has some uses but me personally I do not like this approach very much because you really do not have very much control over your process

                                  Like said earlier my approach would be to use a PWM controller so you control how fast it would heat your process and once it gets to the set point then reduces the power this is where the PDI controller would come in play would control the over and under shooting the set point and maybe adj the PWM controller ( with the PDI controller) to keep the temperature concent and keep the swings to a minimum which be ideal if you could do it this way I have not tried this approach yet but I think about trying on a project that I have in mind
                                  Last edited by sam_sam_sam; 02-22-2019, 05:08 PM.
                                  9 PC LCD Monitor
                                  6 LCD Flat Screen TV
                                  30 Desk Top Switching Power Supply
                                  10 Battery Charger Switching Power Supply for Power Tool
                                  6 18v Lithium Battery Power Boards for Tool Battery Packs
                                  1 XBox 360 Switching Power Supply and M Board
                                  25 Servo Drives 220/460 3 Phase
                                  6 De-soldering Station Switching Power Supply 1 Power Supply
                                  1 Dell Mother Board
                                  15 Computer Power Supply
                                  1 HP Printer Supply & Control Board * lighting finished it *


                                  These two repairs where found with a ESR meter...> Temp at 50*F then at 90*F the ESR reading more than 10%

                                  1 Over Head Crane Current Sensing Board ( VFD Failure Five Years Later )
                                  2 Hem Saw Computer Stack Board

                                  All of these had CAPs POOF
                                  All of the mosfet that are taken out by bad caps

                                  Comment


                                    #18
                                    Re: Temperature control ?

                                    Originally posted by sam_sam_sam View Post
                                    The bang bang controller that he talks in his video has some uses but me personally I do not like this approach very much because you really do not have very much control over your process
                                    Well, if you remove the hysteresis, you can improve your control -- at the expense of more control ACTION (it then approximates a PWM control -- turning the "valve" on and off rapidly as the error signal bounces around "0")

                                    Like said earlier my approach would be to use a PWM controller so you control how fast it would heat your process and once it gets to the set point then reduces the power this is where the PDI controller would come in play would control the over and under shooting the set point and maybe adj the PWM controller ( with the PDI controller) to keep the temperature concent and keep the swings to a minimum which be ideal if you could do it this way I have not tried this approach yet but I think about trying on a project that I have in mind
                                    You then have to worry about "filtering" the PWM output so the controller doesn't see all of that "ripple" and act on it as if it was real "signal". Your sampling rate and PWM switching rate need to cooperate.

                                    Comment


                                      #19
                                      Re: Temperature control ?

                                      Originally posted by Curious.George View Post
                                      Well, if you remove the hysteresis, you can improve your control -- at the expense of more control ACTION (it then approximates a PWM control -- turning the "valve" on and off rapidly as the error signal bounces around "0")

                                      You then have to worry about "filtering" the PWM output so the controller doesn't see all of that "ripple" and act on it as if it was real "signal". Your sampling rate and PWM switching rate need to cooperate.
                                      You have a good point about hysteria and filtering which I would have to take into account when do this concept in designing the project that I have in mind

                                      Thanks for sharing this
                                      9 PC LCD Monitor
                                      6 LCD Flat Screen TV
                                      30 Desk Top Switching Power Supply
                                      10 Battery Charger Switching Power Supply for Power Tool
                                      6 18v Lithium Battery Power Boards for Tool Battery Packs
                                      1 XBox 360 Switching Power Supply and M Board
                                      25 Servo Drives 220/460 3 Phase
                                      6 De-soldering Station Switching Power Supply 1 Power Supply
                                      1 Dell Mother Board
                                      15 Computer Power Supply
                                      1 HP Printer Supply & Control Board * lighting finished it *


                                      These two repairs where found with a ESR meter...> Temp at 50*F then at 90*F the ESR reading more than 10%

                                      1 Over Head Crane Current Sensing Board ( VFD Failure Five Years Later )
                                      2 Hem Saw Computer Stack Board

                                      All of these had CAPs POOF
                                      All of the mosfet that are taken out by bad caps

                                      Comment

                                      Working...
                                      X