Electronics > Beginners
Getting a good reading with a thermocouple
(1/2) > >>
Peabody:
I'm working on a reflow hot plate project, and am using a TM-902C K-type thermocouple with digital readout to monitor the temperature.  Picture 1 shows the setup, with a circular saw blade 3/8 inch above the bare-element hot plate.  I adjust the power with a controller driving an SSR that switches the power on and off per an adjustable duty cycle every 2 seconds, with duty cycle granularity of 20ms - i.e. about one AC cycle.

The problem I'm having is that even after running at the same duty cycle for a long time, so the temperature should be completely stable, the reading moves back and forth by as much as 10°C over a period of 30-60 seconds.  The movements are always gradual, so it never jumps directly by 10°C, but that's just too big a variance to be useful.  None of this happens at room temperature.

The second picture shows the exact same setup, but with a teaspoon of sand dumped on the tip and board.  Doing that almost completely eliminates the variations, so that they never exceed one degree in total, and usually stay right on the money for extended periods - plenty good enough for this purpose.

So I think this means my thermocouple is functioning properly, and the temperature of the blade is in fact constant, but something is making the reading vary when the sand isn't there.  Since this is out on the kitchen counter, and the air temperature is much cooler than the hotplate, the variations must be caused by air cooling off the wires near the tip, thereby cooling the tip itself by varying amounts.

The third picture is the thermocouple tip.  I don't know whether the wires should touch ONLY at the tip, or if they are ok the way they are.

What I'm really curious about is all the toaster oven projects that use a thermocouple for feedback.  From what I've read, you have to place the tip of the thermocouple right on the board, preferably in a plated-through hole.  But even so, based on my experience you might easily get that 10°C variance, and that could substantially affect the controller's actions.  Why would they not experience the variations that I see?  Or do they, but just not know it?  My understanding is that with quartz heating elements, heating is primarily radiant, so the air doesn't get all that hot, but I guess there would be less differential than with my hot plate setup.

Is my explanation of the cause of the problem correct, or is it something else?  Since I can't be dumping sand onto the solder paste, is there some other way to solve this problem?

By the way, while I haven't given up yet, so far it seems that a hot plate just has too much thermal inertia to permit following the reflow protocol.  You can't go from soak to reflow quickly enough without way overshooting.  This makes me wonder why so many people prefer hot plates.  Anyway, I may well end up with a toaster oven, and would need the thermocouple to work properly there.  So I need to figure this out.

Any ideas would be appreciated.

jmelson:

--- Quote from: Peabody on December 14, 2018, 05:47:41 pm ---
The problem I'm having is that even after running at the same duty cycle for a long time, so the temperature should be completely stable, the reading moves back and forth by as much as 10°C over a period of 30-60 seconds.  The movements are always gradual, so it never jumps directly by 10°C, but that's just too big a variance to be useful.  None of this happens at room temperature.

--- End quote ---
First thing I'd do is get it running as it does, and then disconnect the heater from power, and continue recording the temperature reading.  If it suddenly becomes smooth, then it may be that the AC coupling from the heating element is disturbing the thermocouple-reading chip.  Possibly, grounding the saw blade will stop this.

Your sand may be offering electrical isolation, or it may just be slowing down the time constant of the thermocouple.  But, it seems the saw blade can't change temperature all that fast.

Jon
Andy Watson:
It looks like your probe has bare wires from the end of the cable sheath to the fused blob at the end. They also appear to be twisted together - so you have no idea where the actual measurement point is. The wires should be insulated/isolated up to the point of measurement. I think your pile of sand is stabalising the temperature along the whole length of the twisted section. I would start by tidying-up the probe such that there is only one, well defined point at which the wires connect.
Peabody:

--- Quote from: jmelson on December 14, 2018, 07:35:49 pm ---
--- Quote from: Peabody on December 14, 2018, 05:47:41 pm ---
The problem I'm having is that even after running at the same duty cycle for a long time, so the temperature should be completely stable, the reading moves back and forth by as much as 10°C over a period of 30-60 seconds.  The movements are always gradual, so it never jumps directly by 10°C, but that's just too big a variance to be useful.  None of this happens at room temperature.

--- End quote ---
First thing I'd do is get it running as it does, and then disconnect the heater from power, and continue recording the temperature reading.  If it suddenly becomes smooth, then it may be that the AC coupling from the heating element is disturbing the thermocouple-reading chip.  Possibly, grounding the saw blade will stop this.

Your sand may be offering electrical isolation, or it may just be slowing down the time constant of the thermocouple.  But, it seems the saw blade can't change temperature all that fast.

Jon

--- End quote ---

Yes, I had the same thought about unplugging it.  Like maybe even the mains voltage was wavering.  So I tried that, but it didn't change anything.  Of course the reading went down over time, but still wobbled as much as before until it got closer to ambient temperature.  The thermocouple display is battery powered, and the hotplate has a two-conductor power cord.  So there aint no stinkin' grounding anywhere.

Peabody:

--- Quote from: Andy Watson on December 14, 2018, 08:34:12 pm ---It looks like your probe has bare wires from the end of the cable sheath to the fused blob at the end. They also appear to be twisted together - so you have no idea where the actual measurement point is. The wires should be insulated/isolated up to the point of measurement. I think your pile of sand is stabalising the temperature along the whole length of the twisted section. I would start by tidying-up the probe such that there is only one, well defined point at which the wires connect.

--- End quote ---

I'm not sure I understand exactly how thermocouples work, but it does seem that the nearest point of contact between the two wires is where the reading takes place.  So I think you're right about that.  I don't think I can add insulation, but I can un-twist the wires so they don't touch except at the weld at the end.  But let me ask this:  if the two wires were very close to each other so they were at the same temperature, but were not connected by that weld or anywhere else, would they still function as a thermocouple?  Wouldn't the voltage differential still be there?  I watched Dave's video again, but still don't get what the weld does in electronic terms.
Navigation
Message Index
Next page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod