Hi,this is my first post so hello.I am doing a project where I want to keep a stainless steel container about the size of an alcohol measure at a constant temperature of 180c.It is wrapped in a 250watt band heater and I am trying to control it with a Rex C100 pid controller +SSR and a K type sensor.If I auto tune the pid it will keep the 180c temperature after it has settled down but if I pour a couple of mills of water into the container the temperature will over shoot to as much as 230c.Has anybody got a solution for this problem do I need different pid or sensor or is there something else I should be doing.My knowledge of electronics is almost nil any help with this would be greatly appreciated.
I am a beekeeper and the device I am trying to make is for vaporizing oxalic acid crystals which when injected into the hive are totally harmless to the bees but kills off varroa mites which are decimating honey bees world wide.

 

Can you tell us where the gain constants are set after the auto tune process is complete? The system, as tuned, does not handle a step function input properly. For information on step functions go here:

https://en.wikipedia.org/wiki/Step_function

The rapid temperature drop causes the heater to go to maximum heat to try and return the control variable to it's setpoint. The integrator stars integrating a large error, and because the response of the system has a large time constant there is a large input of heat before the temperature starts to respond, and the heater does not turn off in time to prevent the overshoot. I would start by reducing the gain of the integral term.

 

I am by no means an expert in PID controls, but have experience with a few projects.
There are definitely some PID parameters that are not appropriate to the task.

Can you post some pictures of the mechanics of your setup?
Where is your temperature sensor positioned relative to the heater and the liquid?
Do you have any figures on how much power is being supplied when the PID is regulated to the desired temperature?

 

How did you do the autotuning procedure?

 

Thank you for taking the time to reply.
These pics show the container with the heat band there is a gap underneath the container were the sensor will go tight up against the base.The completed project will be mounted on a hand held frame.If I can't get the temperature controlled properly there is no point in going any further.
I don't know what gain constants are or where to find them.Could you explain please "I would start by reducing the gain of the integral term."
Sorry for my ignorance but this would be a great help to me if I could get it up and running.

 

How did you do the autotuning procedure?

I set the auto tuning by using the set button till I got ATU(which looks like aru on the display)changed the range to 0001.Pressed set again and left the auto tuning to do its thing.

 

Tbh that Rex controller is a pretty lo-rent device and it's autotuning is really not wonderful. To set the constants yourself RTFM (link below), set the D value to zero then halve the I value. That should calm it down a bit, then play with P & I till it behaves itself. Leave the D at zero, you really don't need it for an application like this.

http://www.rkcinst.com/english/pdf_manual/imnzc18e3.pdf

edit: I see it has a deadband setting. It might be worth setting this to -2 or so to stop it responding unnecessarily to minor changes.

 

Page 6 of the manual http://www.fmfranklin.com.au/products/data/rkc/c100inst.pdf you have P, I, D, ARW and proportional cycle.

In order for autotune to work, you have to have step changes around the desired temperature. Auto-tune can get you close and then you can tweek.

The proportional cycle needs to be low, because you have a relatively low thermal mass.
ARW is almost like a "maximum power" control. It helps when your heater seems to be oversized for the job.

Temperature vs time graphs are your friend.

briefly - P is the gain or (Error * P) and the higher the gain, the better chance to oscillate and the tendency to be more stable.
The D term tends to reduce overshoot.
the I term tends to make the setpoint and the measured value agree.

ARW can be thought of as a Max Power, but it isn't.

I would guess that "proportional cycle" has to do with how long do you want 100% to be. The shortest burst is actually one half cycle of the AC line, but the gizmo doesn't really know that. This number relates to how fine you can vary the output. One manufacturer calls the output mode "Slow Cycle Triac".

There is a finer mode of control where the AC line is turned on for a portion of the power cycle. The output is generally proportional 0 to 100% of (V^2) of the resulting RMS waveform of a half-cycle. That may make no sense, but the 0-100% would be proportional to power if the load is resistive, That 0 to 100% would be used with the 4 to 20 mA output, for instance.

 

I set the auto tuning by using the set button till I got ATU(which looks like aru on the display)changed the range to 0001.Pressed set again and left the auto tuning to do its thing.

Was the liquid cold when you started the autotune?
Did the liquid warm up to the desired temperature after the autotune finished?

 

Was the liquid cold when you started the autotune?
Did the liquid warm up to the desired temperature after the autotune finished?

The liquid just a couple of mills was cold but steamed off the temperature takes a dramatic drop then over shoots by about 40c thats what it is registering on the pid.

 

Here are some more thoughts about the project:

I can not determine how well the temperature sensor thermal hold of the inner cup. However, it is important that this thermal path has considerably lower thermal resistance, than that of the heating element to the inner cup.

I have also done me some thoughts on what the difference in thermal energy between the amount of water you made the test, and the amount of oxalic acid crystals to be used later. Water is indeed an excellent thing to store and move energy, especially when you freeze or boil it.

 

Here are some more thoughts about the project:

I can not determine how well the temperature sensor thermal hold of the inner cup. However, it is important that this thermal path has considerably lower thermal resistance, than that of the heating element to the inner cup.

I have also done me some thoughts on what the difference in thermal energy between the amount of water you made the test, and the amount of oxalic acid crystals to be used later. Water is indeed an excellent thing to store and move energy, especially when you freeze or boil it.

Here is the commercial version in action using about 2 grams of crystals.

 

I'm wondering if you couldn't really simplify this project by using a vaping device. Let those guys do the heavy lifting in designing the device and getting the costs down. Just find a liquid concentration that works well with that kind of equipment. Maybe they don't get hot enough, but it seems worth a little research.

I've got several pounds of oxalic acid in my basement. It's great for bleaching wood and it's by far the best way to clean yellow-belly off a boat. It's magical.

It's also the same chemical that makes rhubarb leaves toxic.

 

I'm wondering if you couldn't really simplify this project by using a vaping device. Let those guys do the heavy lifting in designing the device and getting the costs down. Just find a liquid concentration that works well with that kind of equipment. Maybe they don't get hot enough, but it seems worth a little research.

I've got several pounds of oxalic acid in my basement. It's great for bleaching wood and it's by far the best way to clean yellow-belly off a boat. It's magical.

It's also the same chemical that makes rhubarb leaves toxic.

Tried the suggestions on all the posts still getting over shoot of 30c has anyone any ideas how I could solve this is there a different Pid system that would work.

 

Is the heater driven by a proportional control or a fully on/fully off arrangement? The reason I ask is that the fully on/fully off heater control introduces a non-linear element into what is supposed to be a linear system. It may be that there is no PID algorithm that will work under those conditions.

You might find the following discussion helpful:

http://math.stackexchange.com/quest...trol-a-nonlinear-system-with-a-pid-controller

 

Is the heater driven by a proportional control or a fully on/fully off arrangement? The reason I ask is that the fully on/fully off heater control introduces a non-linear element into what is supposed to be a linear system. It may be that there is no PID algorithm that will work under those conditions.

You might find the following discussion helpful:

http://math.stackexchange.com/quest...trol-a-nonlinear-system-with-a-pid-controller

Still can't stop the controller from over shooting can anyone suggest a better model of PID which I could buy which would control this or mabe another control system which would be more stable?

 

Overshooting means too much gain. So dial back p and d.

 

Any auto-tune MUST use a step greater then the proportional band or 1/Gain. That's fundamental.

The max output or how far the integral of I can get basically affects how fast the system will recover from an overshoot. The proportional cycle, would think you would want to be low here, It affects how finely you can divide the time.

P, I and D are all inter-related. P mainly affects stability or oscillations. I mainly affects the ability for the setpoint and measured value to agree. and D affects overshoot. The proportional cycle also affects overshoot. How far the integral of E can get affects overshoot and recovery.

here's http://www.nktechnologies.com/wp-content/uploads/2015/02/Monitoring-AC-Current-Powering.pdf a brief concept. The next one is more in-depth.

See here: https://www.chromalox.com/-/media/files/training-manuals/en-us/tm-pk501-scr-power.pdf

Your actually using the system defined on PDF page 14. The system described on page 15 which could use the 4-20 mA output is much more expensive allows finer control. I think it's supported on your controller, but you don't have the right output device,

You may also want to look here: http://www.omega.com/temperature/Z/pdf/z110-114.pdf

Here http://ccipower.com/products/controllers/microfusion-single-phase is an output device that will support 4-20 mA. They aren't cheap. There are probably some cheap ones out there though. I used Eurotherm and they ran hundreds of dollars.

You can try this http://files.redlion.net/filedepot_download/1300/4432 as a manual tuning procedure.

 

Any auto-tune MUST use a step greater then the proportional band or 1/Gain. That's fundamental.

The max output or how far the integral of I can get basically affects how fast the system will recover from an overshoot. The proportional cycle, would think you would want to be low here, It affects how finely you can divide the time.

P, I and D are all inter-related. P mainly affects stability or oscillations. I mainly affects the ability for the setpoint and measured value to agree. and D affects overshoot. The proportional cycle also affects overshoot. How far the integral of E can get affects overshoot and recovery.

here's http://www.nktechnologies.com/wp-content/uploads/2015/02/Monitoring-AC-Current-Powering.pdf a brief concept. The next one is more in-depth.

See here: https://www.chromalox.com/-/media/files/training-manuals/en-us/tm-pk501-scr-power.pdf

Your actually using the system defined on PDF page 14. The system described on page 15 which could use the 4-20 mA output is much more expensive allows finer control. I think it's supported on your controller, but you don't have the right output device,

You may also want to look here: http://www.omega.com/temperature/Z/pdf/z110-114.pdf

Here http://ccipower.com/products/controllers/microfusion-single-phase is an output device that will support 4-20 mA. They aren't cheap. There are probably some cheap ones out there though. I used Eurotherm and they ran hundreds of dollars.

You can try this http://files.redlion.net/filedepot_download/1300/4432 as a manual tuning procedure.

Finally have made some progress.I turned down a new stainless container made it thicker.
Setting the pid to 208c gives me 185 c in the bowl.The over shoot on the pid now goes up to 212c dosen't kick back in when water is added which steams of in ten seconds till the pid is at 190c
The pid readings are
AL1 45
P 65
I 122
D 30
AR 13
R 33
SC 10
Do you think it could be fine tuned even more eg.less over shoot and kick in faster when water is added?Now that I've got these settings if I auto tune and it is no better if I put these list of setting back in would you expect it to run as it is now again or if I had to buy another Rex c100 would putting these settings mean it would work the same?Thanks again for all the help.

 

Do you think it could ...

You do realize that by the time you get a guess from someone here, you could have completed a few rounds of experiments and be able to conclusively answer your own questions?