PID Temp Controller

[leadmonkey]

I just finished the PID temperature controller for my Lee Pro 4-20 pot.
99063

99064

99065
The controller and SSR cost $32.99 and the thermocouple cost $6.39, both from evilbay.
The single ac outlet was about $6 from Lowe's.
The fuseholder and switch cost about $10 from a mail order surplus dealer.
The cabinet is an old printer a/b switch.
The rest of it is junk box salvage stuff, but I'm betting it wouldn't have cost more than $20-$30 if you shopped around.
I could have added additional features like a PID bypass switch for heatup, and a maybe a few indicator lights, but I wanted to keep it down and dirty, and simple.
A heat sink has not been necessary. I mounted the SSR directly to the bottom of the cabinet with thermal compound. The cabinet only gets slightly warm. If this controller was driving a higher current pot, heat sinking might be needed.
I did not bypass the thermostat in the pot. It serves as some protection against temperature runaway, if something goes wrong with the controller.
The most difficult and time consuming part of the project was fabricating all the hardware items.
Wiring was dirt simple.
The only issue is that you have to use crimp type connectors on the chromel/alumel wires from the thermocouple. They won't take solder. So hooking them up to the connector took a little extra doing.
99067

It heats up nicely, and maintains temp within about 20 degrees. That may improve, as I don't think it has "learned" the environment yet. It's fairly easy to tell when the element is on, it buzzes pretty noticeably.

Now a couple of questions. Note the difference in indicated temperatures. According to the PID indicator, the lead liquefied at around 635 degrees F.
Which indicator would you believe?
99068

[Mike W1]

Well for starts I don't think I'd believe the stem thermometer! My TC stays in the pot and I've watched the temp as it's coming up to casting temperatures. Somewhere in the 540° to 580° is when my pot does it's dripping. Only a small portion of the TC actually produces the signal and that's down there where the heat is. I'm guessing here but with the top not melted yet and obviously the bottom part of the pot is, contraction or expansion is going on and causes the leaks. Pure lead melts at around 621° and if there's tin and antimony in the alloy the melt point is lower.

My PID keeps my melt ± 2° once the auto tune has finished. FWIW

[leadmonkey]

I've been measuring temperatures in the pot as the lead solidifies. In testing this thing out, I've made a few discoveries.

The lead in the pot is stick-on wheel weights, and I would guess that it's pretty close to being pure lead, so I would expect it to melt or solidify at around 620 degrees.

When the lead in the bottom of the pot near the thermocouple begins to solidify, the PID indicates around 615 degrees, and the stem thermometer indicates 545 degrees. I'm thinking 615 degrees is a whole lot closer to being correct than 545 degrees is. Sounds like the thermocouple readout is fairly accurate. (And the stem thermometer is not.)

The lead on the surface of the pot begins to solidify when the temperature at the bottom of the pot is around 650 degrees... about thirty degrees difference. A handy thing to know.

The controller's first temp overshoot was twenty degrees. After that, it stayed within 5 degrees of the target temp.

Also another question.
If you study the thermocouple theory, you'll see that in the junction between the chromel and alumel thermocouple wires forms a thermocouple that produces a voltage in proportion to the temperature of the junction. But, there are also junctions between these wires and the copper conductors that they connect to, which also form thermocouples, which also produce voltages in proportion to the temperatures of those junctions. There are several standard ways of compensating for the errors that these "cold junctions" produce. Some are simple, some are more complex, depending on the accuracy needed. Has anyone seen any indication of what these PID's use?

[RobS]

The PID should have a function to adjust your controller should the thermal coupler probe be off +/- certain degrees and then there should also be a function to set so the controller can "learn" your pot.

[RobS]

Here is a quick read on melting point of different alloys:
http://www.lasc.us/castbulletalloy.htm

[dikman]

The miniscule voltages that might be generated by using dissimilar metals in the probe electrical connectors has been mentioned before. I'm of the opinion that if the connections are good and solid, regardless of the type of connector used, then for our needs any such concerns are irrelevant. After all, those "official" probe connectors that you can buy are pretty cheap (both in cost and manufacture) - does anyone really know what type of metal is used in the connectors? I'll be surprised if it's anything special.

As for soldering the wires, it can be done as that's what I did when I used BNC connectors. It took a bit of doing, though, I had to scrape the surface of the wires to ensure they were clean and then it took a bit of heat to tin them. But it worked, fortunately, as I'm not a great fan of crimped connections.

I also used a switch box for my case. Nice and compact - and cheap!

[bangerjim]

I've been measuring temperatures in the pot as the lead solidifies. In testing this thing out, I've made a few discoveries.

The lead in the pot is stick-on wheel weights, and I would guess that it's pretty close to being pure lead, so I would expect it to melt or solidify at around 620 degrees.

When the lead in the bottom of the pot near the thermocouple begins to solidify, the PID indicates around 615 degrees, and the stem thermometer indicates 545 degrees. I'm thinking 615 degrees is a whole lot closer to being correct than 545 degrees is. Sounds like the thermocouple readout is fairly accurate. (And the stem thermometer is not.)

The lead on the surface of the pot begins to solidify when the temperature at the bottom of the pot is around 650 degrees... about thirty degrees difference. A handy thing to know.

The controller's first temp overshoot was twenty degrees. After that, it stayed within 5 degrees of the target temp.

Also another question.
If you study the thermocouple theory, you'll see that in the junction between the chromel and alumel thermocouple wires forms a thermocouple that produces a voltage in proportion to the temperature of the junction. But, there are also junctions between these wires and the copper conductors that they connect to, which also form thermocouples, which also produce voltages in proportion to the temperatures of those junctions. There are several standard ways of compensating for the errors that these "cold junctions" produce. Some are simple, some are more complex, depending on the accuracy needed. Has anyone seen any indication of what these PID's use?

If you want to get down in the weeds with accurate temp measurement/control, you have to use real t/c extension wire with isothermic plugs on your t/c wire and in the box. Not just any old copper wire!!!!! They are designed of contrasting metals to eliminate the "other metal to metal" junctions you form.

You know, you can make a t/c with two dis-similar metals of any kind. Dis-similar would be the screw terminals on the back and a piece of copper wire. Type K is the one you should be using, as it gives the most mv output in the ranges you are working with. I use a paperclip on the screw terminals of the controllers I sell and check out. Not NIST calibrated, but close enough for proof of function.....and government work. It reads my lab temp close enough.

If you have copper junctions in there, everyone of them forms another t/c junction that measures ambient temp and either adds or subtracts from what the REAL t/c is reading. That is why it is imperative you use isothermic connectors exclusively in thermocouple measurement systems.

That is why I do not mess with PID controllers in something as simple as a lead melting pot! I sell them and engineer control systems around them.

banger