Announcement

Collapse
No announcement yet.

Project PID Silvia -Melbourne

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

  • #16
    Re: Project PID Silvia -Melbourne

    Hi, Sparky -

    Im the guy with the interference issue. I want to caution folks not to rush to any conclusions, good or bad, about the Auber PID based on my CG post. I am still trying to figure out the source of the interference. Based on the lack of others reporting the problem, I have to consider that there may be something unique to my test bench that is contributing. I can report, though, that 4 units of filtering made it go away completely.

    Now that I am tuned in to this a little more, I think I am now noticing a little jumping around with a 2nd, non-Auber controller. The instability with the non-Auber controller, if indeed thats what it is, has a smaller magnitude - usually varies by +/- 1F or 2F, instead of the 7F I was seeing in the Auber.

    I will pass along that I have a flourescent arm lamp plugged in to the same portable outlet strip as Silvia and the controllers. As soon as I have a chance, I am going to investigate that as a possible cause.

    Stay tuned...

    -- JGG

    Comment


    • #17
      Re: Project PID Silvia -Melbourne

      Hi all,

      Weve been out of town on our boat for the best part of the week and just got back. Still havent got my land legs, despite having left the boat 6 hours ago, and the computer screen seems to be swaying!

      Should pick up the PID from the post office mid-week. Looking forward to that, and will keep everyone posted.

      Yes, induced currents in ultra-thin thermocouple wire can be a problem. The wires have to generate such a tiny current that anything extraneous to it can give a false reading. Hope that wont be the case.

      Robusto

      Comment


      • #18
        Re: Project PID Silvia -Melbourne

        Right on the money again Robusto 8-),

        Hi JGG,

        You mention that you are using a "Portable Power Strip" that has other items connected to it, e.g. the fluoro light. Thats probably not a good idea where PID Controller with t/cs are concerned. A better idea would be to "Hard-wire" your controller via suitable in-line fuse, etc within the espresso machine itself. Also, if the t/c you are currently using is not of the shielded variety, then that would be something worth doing too, i.e. change it for a suitable shielded t/c cable and ground the shield at the espresso machine end and insulate the controller end.

        Both of these measures should go a long way to reducing the instability you are observing and hopefully negate the necessity for having to filter the t/c input. It may also just be that the fluoro light you have plugged into the same power board is emitting considerable EMI interference within the close proximity of your controller and t/c cable. Try removing the Fluoro from the vicinity altogether and see if this helps at all. It will really come down to the simple but sometimes annoying process of eliminating each possible EMI source one at a time until you identify the culprit. The measures mentioned above will also help by tying all power and ground cabling to the same locus within the case of the espresso machine, further assisting to reduce the effects of EMI interference. Hope you find the culprit,

        Cheers,
        Mal.

        Comment


        • #19
          Re: Project PID Silvia -Melbourne

          Hi, Mal -

          One of the great things about the way I have things wired in my shop is that there is always room for improvement!

          -- JGG

          Comment


          • #20
            Re: Project PID Silvia -Melbourne

            No worries JGG,

            Will await your forthcoming posts of complete success ;D,

            Cheers,
            Mal.

            Comment


            • #21
              Re: Project PID Silvia -Melbourne

              Project PID Silvia:  Part 3


              It’s Anzac Day, and having waited so long for the Auber PID ordered from the United States on Ebay, I can’t wait to open my present.

              But first, it has to be collected.

              It was hopefully delivered to my wife’s office during our brief holiday last week, and I can no longer wait for her to return to work tomorrow and collect it.

              If I don’t get it today, my schedule will preclude any work on it until the middle of next week.

              I need a head start, so we drive to collect it.

              Great, it’s arrived. In a padded parcel post bag  the size of A4 paper.  Inside are  11 pages of directions, and two small cardboard boxes.

              One, the size of a match box, contains the 25 amp SSR. Inside the bigger one is safely ensconced the  1/16 DIN Auber PID.

              It looks good. I want to know about the terminals --- they are the type where stripped wires are inserted and then clamped on by turning a bolt against a nut.

              Each terminal is numbered and corresponds with diagrams pasted on the PID and on the instructions too.

              Now the instructions.  Lots of diagrams. Good.   I’m fairly confident  I’ll be able to follow them when attaching the 240VAC wires, the thermocouple wires, and the output  12V wires to SSR will be fairly straight forward.

              There’s a recommendation for a “very fast blowing” fuse when output is to a SSR. On a later page, it suddenly becomes a slow-blowing fuse.  And high-voltage wiring should have  600V insulation.

              There’s a caution that thermocouples can suffer from noise from adjoining wiring, so it should be routed away from it.  (They must have read our CS posts)

              Or tightly twist the voltage wires so it doesn’t induce noise.  Or install a filter, but the higher the filtration the slower the response.

              Reading on, and getting way ahead of myself, I turn to instructions on the operation and tuning of the PID.

              Could be a little more tricky, and most of the language although in English may as well be in Chinese on my first cursory reading.  Hopefully it will all become self explanatory when the time comes.

              I like the section on fuzzy logic.  After 2 or 3 cycles, “the microprocessor in the instrument will analyse the period, amplitude, waveform or the oscillation generated by the on-off control, and calculate the optimal control parameter value. The instrument begins to perform accurate artificial intelligence control.”

              Wow! I like that!

              I have visions of the square-faced box just about smart enough to smell the coffee aroma  and use its artificial intelligence to act accordingly.

              For now, knowing what the gadget looks and feels like, I can start assembling wires, terminals and enclosure.  And pray at least one of the 2 thermocouple wires I already have will be long enough.

              To be continued.....

              Comment


              • #22
                Re: Project PID Silvia -Melbourne

                Hi Robusto,

                Glad to see that your hardware has arrived.... joy and jubilation .

                Just picking up on the statement in your new controllers user instructions about the t/c cables sensitivity to stray Electro-Magnetic Interference fields. As they say, an effective measure to counter this problem is to ensure that power cabling in close proximity to the t/c cable should be tightly twisted along its length. This is easy to do and to keep the cables twisted, you can sheath it in Heat-Shrink insulation of the right size and heat shrink away.

                It is also possible to obtain t/c cable that is tightly shielded with a conductive metal braid to reflect any EMI radiation that might be present. This shield is then connected to the main Earthing point in the Espresso Machine and left disconnected and insulated at the other end where the controller is. If you pursued both of these preventative measures, Id be very surprised if any EMI that may be present could cause you any trouble. Hope that is helpful,

                Cheers,
                Mal.

                Comment


                • #23
                  Re: Project PID Silvia -Melbourne

                  Part  4

                  Mission accomplished. Well, almost.

                  Until I get a flexible hose to cover all the wires, it won’t be a finished project.

                  Taking Bloop’s lead, I headed to Dick Smith for the shiny black  enclosure (just under $10). Many had scuff marks and scratches, so finding a decent one took some rummaging.

                  The store had none of the recommended 600VAC wire, so I decided to utilise some I’d removed from one of several disused power cords.  The PID draws .02 amps  so the current is not an issue, but adequate insulation may be?   For the SSR-to-heating element I used heavier cabling because that is where the maximum current of around 5 amps will flow.

                  I decided against soldering  terminals, and used some spare 6.3 mm female crimp ones.   No male spades, because I also had a boxful of nifty, insulated female-to-female joiners.  Since all of Silvia’s terminals are female, they slide on one end of the joiner, and the wires from the PID slide on the opposing end.

                  Bought a meter of some shrink tubing (smallest length sold) but really not mandatory. It’s to cover the piggy-back terminals on the switched side of Silvia’s power supply, from where the PID will take its power.

                  Following the makers suggestion,  I tightly twisted these power supply leads together to minimise electromagnetic interference being inducted into the thermocouple.

                  ASSEMBLY
                  First, unplugged from the wall socket and  removed the cup warmer (easy, 4 screws); the water tank; the splash shield (easy, 2 screws) above the drip tray, and the baffle separating the water tank compartment (need a stubby Philips head screwdriver, 2 bolts).    

                  I was going to work in the messy garage, but created a mess in the kitchen instead!  

                  Working out routes for the wiring, measuring, testing with scrap wire took the best part of an hour.   It’s time well spent, as you want to do a neat job on the inside as well as the out.  Like most Silvia PIDers, I opted for the handy exit hole at the bottom rear left –the side away from the steam.

                  The 25-amp SSR sits on the bolt which pokes through above the drip tray, and I had no trouble finding a suitable nut from my assortment box.  

                  A 45x45 mm hole has to be cut on the face of the Dick Smith enclosure for the PID. It’s a good idea to look at the inside as well as the out when locating where exactly to cut, just in case there are projections on the interior which will interfere with the body of the PID and especially the slider which holds it in place.

                  I scored the square cutout with a scalpel-like cutter, and cut carefully with a hack saw and fret saw.  Because  I wanted the hole located centrally, it meant cutting into the lid as well (the lid is parallel to Silvia’s side).  Any roughness was smoothed with a file.  Cutting and filing the plastic was very easy.

                  Drilled a 10 mm hole at the rear to route all the wires into the box.

                  Put the PID in the case  Used a 1 amp fast-blow fuse in a 12V fuse holder.

                  Looked at the wiring diagram a hundred times, and followed them (or so I believed).  Working out where the t/c went was educated guesswork because of the inadequate instructions.

                  Check, and re-check, and re-check, countless times.

                  OK, fingers crossed, hold the breath, and plug Silvia into the wall, and turn on its now dual-purpose power switch (turns on the PID at the same time).

                  Hallelujah. A beautiful sight. The PID lights up like a Christmas tree, red digits on the top line, green ones below.

                  I enter 108 and I can hear the boiler heating…..But there’s something wrong. The PV display seems to count backwards, and then  both lots of numbers start to flash.  

                  Switch off, consult the woefully written instructions. Did I say “instructions?”---a garbled mess of uncommunicative jargon and meaningless meanderings would be a better description.  None of the “Step one, do this. Then, do that…”.  Just random utterings which, without a logical sequence and explanation, mean litte.

                  In any case, flashing digits mean something is amiss with the wiring.

                  I’m baffled. Recheck wiring. Yes, all appears to be above-board.  Especially polarity.

                  Then I remember something about thermocouples where the red wire is negative, not positive. An internet search confirms I’ve reversed the polarity, so I rectify the situation.

                  But again, both lots of digits flash.  I try another terminal for the t/c positive wire –you’d never guess where to put it looking at the diagrams.

                  Switch on---It works!

                  I call Mrs Robusto in to have a look at  Silvia’s heater light pulsating for half a second or so instead of the usual 50 seconds.  It’s dancing in tune with the LED “output” display on the PID.  Joy.

                  Then,  things slow down. There’s the occasional pulse, but Silvia is barely above room temperature, and so is the group.  I connect my multimeter and its t/c and confirm everything is cold.

                  You have to realise that hours are ticking by with all this, and it’s now at least 8 hours since I opened up Silvia.

                  I also can’t get the display to show decimal places.  I know that decimals are only available in Celsius mode……Ah, that’s it. This American PID was pre-set for Fahrenheit.   Re-set it for Celsius, and away it goes again, heating.

                  Now to autotune with fuzzy logic…..very fuzzy indeed, because it overshoots by up to 6 degrees.

                  By now its close to midnight so Ill leave that for another day.

                  To be continued....

                  -Robusto

                  Comment


                  • #24
                    Re: Project PID Silvia -Melbourne

                    Geez, I’m on tenter hooks.
                    Looking forward to the next instalment.
                    Please don’t forget to add pic’s and, if possible,
                    A diagram of how you hooked up the PID.

                    Billi

                    Comment


                    • #25
                      Re: Project PID Silvia -Melbourne

                      Part 4

                      The PID is up and working, and as mentioned, auto-tuning wasnt a spectacular success.
                      It overshot the setpoint and kept a temperature 3 degrees above it.
                      To make matters worse, the steady reading was 10 degrees above my multimetre reading. Which was correct?

                      The disparity in temperature was confusing and annoying. I disconnected the DMM t/c from Silvia and dipped it in boiling water. It read 100 degrees, so no issues with its accuracy. Then I disconnected the t/c from the PID and connected it to the DMM ---now read the same temperature as the DMM, so the t/c accuracy was also OK.  

                      I adjusted the compensation parameter on the PID downwards by 10-degrees. Both it and the DMM now read in sync. Wonderful---but what the PID was now reading as  111 degrees, was in fact steaming hot water!  I pulled a shot, and it tasted burned, too.  

                      So, set compensation back to zero.  And yes, I do have the PID set to K-type thermococuple!

                      Some manual "auto-tuning" was in order. At first I worked totally by guesswork. Instructions talk about the P-I-D settings reacting to changes, compensating etc....  Perhaps it needs to react quicker....increase a setting. Nope, doesnt work.

                      Then it comes to me:  the ONLY compensation or reacting the PD can do is to HEAT or NOT to HEAT.
                      It doesnt decide: "Oooops! Overshot there, better pull it back a little by turning on the cooling." Its only choice in that scenario is do do nothing.

                      WIth that in mind, I decided ALL the parameters should be brought down. Drastically. Less time heating, more time doing nothing.

                      The P setting for instance was 180 on autotune. Its now 50. The I was 770 before, now 115. D 80 before, now 25.

                      I now have the boiler humming nicely 0.4 degrees C above set point.  Pretty good.

                      To be continued, with pictures

                      Comment


                      • #26
                        Re: Project PID Silvia -Melbourne

                        This is fabulous stuff Robusto, do you have any idea how much time and effort you are saving me!
                        Many thanks
                        Billi

                        Comment


                        • #27
                          Re: Project PID Silvia -Melbourne

                          Hi, robusto -

                          I have tuned at least a dozen controllers for Silvia. The two most challenging have been a Fuji PXR3 and the same Auber you are working with. Both fuzzy logic, both have overshoot problems after shots (although Fuji was rock steady at idle). Perhaps more tuning will solve...

                          In contrast, the old fashioned non-fuzzy controllers from Watlow, Omega, etc. tune wonderfully and generally have required almost no manual tuning. So I am still waiting to see the fuzzy logic advantage.

                          I am still not fully tuned with my Auber. My SP is 228F. But I cant get it to stabilize completely. I still am seeing 231F periodically while on idle.

                          After shots I am getting overshoot up to 237F.

                          FWIW, while I was manually tuning, I got very good performance with the controller in P-D mode (no integral term). Not sure yet what to make of this fact.

                          Please keep us posted as the magic numbers continue to evolve.

                          -- JGG

                          Comment


                          • #28
                            Re: Project PID Silvia -Melbourne

                            Still fine-tuning. Ive columns and columns of digits for different settings Im trying. If you get the steady-state setting right, it overshoots. And vice-versa. Its turgid and tedious but I wont give up.

                            Robusto

                            Comment


                            • #29
                              Re: Project PID Silvia -Melbourne

                              Hi All,

                              The thing with Fuzzy Logic enhancements to PID control is that you really need to know how the FL is written or at least have access to the Translation Table they have used to integrate it into the controllers own PID algorithms. Without this, it really comes down to a lot of guess-work in trying to understand the relationships between the FL Arguments and the PID integration.

                              Basically, most FL for PID control is written to enhance the controllers ability to react to Process Upsets, such as when you pull a shot after the Boiler temperature has stabilised. In Steady-State conditions it is usual for the FL to be desensitised and allow the PID Algorithms to do their stuff.

                              One way to check this is to allow your system to arrive at its Steady State condition, i.e. controlling around the Setpoint, start your FL Autotune function and then pull a shot (what ever is a typical shot for your situation, Double/Single/Ristretto, etc). Allow the FL Autotune to do its thing and just wait until the Autotune LED goes out or what ever the indication is on your particular controller. Hopefully the controller has returned your Boiler Temp back to the Setpoint and is maintaining same within acceptable limits (+/- 0.2deg C should be considered acceptable in this application). At this time, go back to your Controllers setup routine, then note and record what the PID parameters have been adjusted to, via the FL Autotune sequence.

                              If youre not satisfied with the outcome, try the above routine a couple more times and see how much, if any, the PID parameters are varying from one Autotune session to another..... If all is working as it should there should not be much variance.

                              If after all this, you are still not satisfied with the controllers performance then I would suggest contacting the controller manufacturers Technical Rep and seek their guidance. You have to keep in mind, that a PID Controller is designed to maintain a Setpoint within a given set of variable criteria such that when ever the Process (The Boiler Water Temperature in this case) is Upset, the controller commences to control the output such that the Setpoint Target is returned as quickly as possible and within the constraints of the PID parameters that have been set. It is not a glorified Thermostat and if this is all you expect it to be, then you dont need a PID controller but a simple Digital On/Off controller or Digital Thermostat.

                              Anyway, hope some of the above is helpful.

                              Cheers,
                              Mal.

                              Comment


                              • #30
                                Re: Project PID Silvia -Melbourne

                                Originally posted by robusto link=1144669688/15#24 date=1146179655
                                The P setting for instance was 180 on autotune. Its now 50. The I was 770 before, now 115. D 80 before, now 25.

                                To be continued, with pictures
                                The new numbers you have dialled in Robusto, certainly look a lot closer to reality than where the Autotune left them. Proportional Band still seems a little high though, as evidenced by the +ve 0.4deg C above Setpoint in Steady-State operation. You could try reducing P until you start to see an oscillation creeping in and then back it off a few points until this all but disappears. Its easiest to do this with Integral and Derivative disabled otherwise it will become very confusing. Once you have observed the onset of oscillation and reverted to a point where this is all but eliminated, you can enable Integral then Derivative and notice what effect each of these has on Process Upset control. I highly recommend consulting the info in this page here..... http://www.eurotherm.com.au/email/email_9_2.htm , it will help enormously

                                Cheers,
                                Mal.

                                Comment

                                Working...
                                X