Announcement

Collapse
No announcement yet.

Heat Treat Oven Accuracy / Precision?

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

  • Heat Treat Oven Accuracy / Precision?

    I picked up an old, 5 kW heat treat oven at auction for $40. Figured I couldn't go too badly wrong at that price point. It's got two different heater controllers on it, both of which are ancient. They are both largely mechanical, with one using a mercury filled tube to lift a plunger which actuates a needle and micro-switch. Neither seems to be working quite right and rather than tearing into a fussy box full of small parts and mercury filled tubes, I figured I'd make a new controller for it.

    Yes, I know. I could buy one off the shelf. But what fun would that be? Besides, I have experience building custom temperature controllers for scientific instruments from my day job so I don't think the learning curve will be too steep. I'll just be working in a different temperature range than what I'm used to.The first question, though, is how accurate and precise do heat treat ovens need to be?

    For absolute accuracy, is +/- 10 degrees good enough? +/- 50 degrees? +/- 100 degrees?

    What about for precision or the ability to maintain a constant temperature?

    The easiest control for a 5 kW heater would be bang-bang via a magnetic contactor / relay, which is what's on there now. Of course, I don't want my controller to kick the relay on and off every few seconds as it would wear out the relay. I don't know what the thermal properties of the oven are yet, but if I need to maintain a really tight control, I may want to use a more sophisticated control scheme. In my day job, I usually work with an absolute accuracy of +/- 0.01 degrees and, in some cases, a precision on the order of microdegrees. Obviously that isn't needed here, but I don't know what *is* needed here...

    Any thoughts?

  • #2
    There'll be a Murphy law here somewhere, as soon as you decide something is good enough, it won't be. Off the top of my head heat treating HSS would be one thing that requires high accuracy. I've never done it, maybe never will.

    50 or 100 seems too loose, but price sensitivity would come into the decision. if more accurate than I thought I needed was 1/10 more or 10x would affect what I bought.

    I'm also looking for a new PID controller, please keep us posted what you come up with. The developed worlds ones are so expensive, I think register a protest vote and go Chinese.....but not sure what one. You (imo) want ramp and soak, the cheapest PIDs dont offer that.

    A lot now are SCR. I bought a PID (wrong one though, no soak/ramp), under $20, and to my surprise it came with the SCR and thermocoupler. Any reason not to do away with the relay?
    Last edited by Mcgyver; 01-14-2020, 11:49 AM.
    in Toronto Ontario - where are you?

    Comment


    • #3
      I've got a heat treat oven build coming up and I'm using a cheap <$20 PID controller from banggood. While I've never actually used one, there are countless youtube build out there using the same hardware that seem to work just fine. Some other use an arduino for control too for different ramping and temp holding options, and I'd like to go down that road eventually too once I get it built and working. Coincidentally I just unearthed my old College texts and notes while cleaning up my Dad's basement this past weekend so I'll be re-reading them to try and relearn some metallurgy and heat treating stuff. I don't remember much and all my experience in industry since has just been torch heating 01, but I do remember some steels had specific ramp up and hold cycles but beyond that I don't remember much specifics. That's why I want that kind of control.

      Comment


      • #4
        Lots of info on line:

        https://www.youtube.com/playlist?lis...swrNli7VUzIEnE

        https://www.google.com/search?source...=1579017065195

        David Merrill

        Comment


        • #5
          I've been reading the Crucible catalog, its very informative about ramp times, soak times, temps and etc. http://www.crucibleservice.com/esele...eralpart3.html

          Comment


          • #6
            Hi,

            Did a rebuild of a similar oven many years ago. We decided that +/-10F was close enough for the pretty simple steels we were going to deal with. I can't remember the brand Pid we used anymore, but it was US made and worked well. Pretty sure I ordered it from Braas. We bought a lot of control and pneumatics from them in the 1990's.
            If you think you understand what is going on, you haven't been paying attention.

            Comment


            • #7
              I "grew up" using bang/bang controllers for ovens we made for chemistry stuff. A few years before retirement I worked at a medium sized engineering R&D company and they changed my outlook completely. We used PID controllers to maintain the test rigs +/- 0.1C. As you say the constant switching would wear out a mechanical relay, but a SS relay does just fine...

              The nice thing is that even relatively inexpensive process controllers can be programmed to do a cycle. We used to ramp up over a few hours, hold for while, change to a new temp, and then ramp down.

              However, do gird your loins before you go to program the thing. For some reason the interface seems totally obscure to me.

              Comment


              • #8
                Use a SSR and treat it like a PWM signal. Use a PID controller, or configure an output of an Arduino to function like a PWM signal, and use some coding to vary the cycling rate and on-time until you find a control scheme that works for your application. Much cheaper and easier than a mechanical control these days.

                Comment


                • #9
                  In the real world, you would find quite a large temperature range within the oven, especially from top to bottom. Even the fan ovens at my old workplace used to vary several degrees inside when 5 or more thermocouples were set up inside during recertification.

                  Comment


                  • #10
                    I gather this would be only for tempering and not the initial hardening?

                    For tempering to have a good and consistent control over the results I'd want to aim for something no worse than plus or minus 5F. That's a 10F range that it COULD vary over.

                    I'd guess too that as the walls will lag in their heating up that more of the cycling will occur during the warming up after switching on. Then as the walls and inside air reach some level of match the cycling would reduce. At least until the door is opened then it would cycle a few times until stable again.

                    If using solid state controls and something with some smarts could you incorporate a variable duty cycle that either controls the AC like a light dimmer by delaying the switching for each 60 hz cycle or that sets up a variable pulse timing to the element?
                    Chilliwack BC, Canada

                    Comment


                    • #11
                      Originally posted by old mart View Post
                      In the real world, you would find quite a large temperature range within the oven, especially from top to bottom. Even the fan ovens at my old workplace used to vary several degrees inside when 5 or more thermocouples were set up inside during recertification.
                      that's an interesting point. Any recollection on how much variance there was? You'd think it would quickly average in the work piece. Steel for example has around 800x higher superconductivity than air, but I'm surprised there isn't enough movement, convection I guess to keep the air temp more consistent.
                      in Toronto Ontario - where are you?

                      Comment


                      • #12
                        Originally posted by old mart View Post
                        In the real world, you would find quite a large temperature range within the oven, especially from top to bottom. Even the fan ovens at my old workplace used to vary several degrees inside when 5 or more thermocouples were set up inside during recertification.
                        This is a fan oven and my intent for my controller was to read out at least four channels. The custom controls I've designed for my company include 24 channel readouts for systems with independently controlled heaters, allowing the system to correct for gradients as the system experienced sometimes significant environmental perturbations (it was a mobile detector system that could go from baking in a parked vehicle to suddenly much colder as doors were opened or AC started).

                        But this is a useful data point. It sounds like, at your previous workplace, an accuracy on the order of 5-10 degrees was appropriate?

                        Originally posted by Mcgyver
                        A lot now are SCR. I bought a PID (wrong one though, no soak/ramp), under $20, and to my surprise it came with the SCR and thermocoupler. Any reason not to do away with the relay?
                        No real reason. Just that someone replaced the relay in the oven recently and I didn't want to over-complicate things if I didn't need to. That's why I'm asking the question; if having a tight control is really advantageous than doing a PID based solid state controller is not that much more difficult. Just the component cost / BOM cost goes up a bit.

                        Originally posted by Dan_the_Chemist
                        The nice thing is that even relatively inexpensive process controllers can be programmed to do a cycle. We used to ramp up over a few hours, hold for while, change to a new temp, and then ramp down.

                        However, do gird your loins before you go to program the thing. For some reason the interface seems totally obscure to me.
                        Absolutely! Even if I went with a bang-bang type control, I'd want the ability to do ramps and soaks. And part of the appeal of building my own is so that it makes sense to me. I've worked with a number of off-the-shelf process controllers and I agree with you regarding the interface. I'm sure it makes perfect intuitive sense to someone, but not me!

                        Regarding the vast amount of info on the internet regarding home brew heat treat ovens and cheap PID controllers... well I'm asking the question here because I'm hoping to get some more accurate information. If it really is as approximate as it seems, then great! This control can be very simple.

                        But if you really need +/- 5 degree absolute accuracy and control, well I highly doubt you're going to get that with a $20 controller. Just reading out a thermocouple accurately and compensating it correctly is not an entirely trivial matter. You're probably looking at an uncertainty of +/- 2 degrees right there, if not worse. There are just lots of places to go wrong reading a thermocouple... you're talking about microvolts per degree signal to read, which means it's susceptible to electrical noise, then you've got issues with long-term stability and corrosion at the weld, cold junction compensation, etc. A carefully designed system can maybe get to 0.5 degree accuracy with thermocouples, but that requires great care in the design and implementation. Setting aside issues with the sensors, you then have gradients inside the oven as Old Mart mentioned. The insulation in the oven, placement of the heaters, location of the sensors, etc. all become critical design choices when you're talking about really tight control.

                        It's one of those "down the rabbit" hole kind of things.

                        Dan Dubeau, keep us updated on your build and especially what you learn looking through those old metallurgy books! I'm sure that's where the answer lies; I'm just trying to mooch and get the info from someone who already knows without having to do a bunch of research myself!


                        Comment


                        • #13
                          Originally posted by Mcgyver View Post

                          that's an interesting point. Any recollection on how much variance there was? You'd think it would quickly average in the work piece. Steel for example has around 800x higher superconductivity than air, but I'm surprised there isn't enough movement, convection I guess to keep the air temp more consistent.
                          Tightest uniformity spec in AMS-2750 standard is +-3 Cel and lowest class of furnaces is +-28Cel

                          Average expectation could be somewhere between those two extremes but it could be also worse. 3 Celcius at 1200 Cel HSS hardening oven would really tough and hard to archieve.
                          Last edited by MattiJ; 01-14-2020, 03:53 PM.
                          Location: Helsinki, Finland, Europe

                          Comment


                          • #14
                            Question #1 is Are you paying Demand charges on the electric meter? If you are slamming heat elements on and off is dam expensive.

                            The mercury filled tubes you're so scared of are likely timers if they're sitting inside magnetic coils. They were equal to Hayden clocks in accuracy and superior in #cycles to replacement.

                            Precise temperature & time in the oven are far more critical to enginincompoops who never come within 100 feet of the part than they are to men working the oven and when engineering is giving orders a lot of parts either go back to the oven as rework or to the scrap pile.

                            Comment


                            • #15
                              Originally posted by MattiJ View Post

                              Tightest uniformity spec in AMS-2750 standard is +-3 Cel and lowest class of furnaces is +-28Cel

                              Average expectation could be somewhere between those two extremes but it could be also worse. 3 Celcius at 1200 Cel HSS hardening oven would really tough and hard to archieve.
                              This is exactly what I was looking for - thank you very much. Some Googling on AMS-2750 returned some great reading material.

                              Quick summary is here (for those interested): https://www.keithcompany.com/ams2750...fications.html

                              Comment

                              Working...
                              X