Sciencemadness Discussion Board

DIY temperature probe for hotplate, help!

p4rtridg3 - 23-11-2020 at 11:51

Hello all,
I am not much of an electrician so I hope someone here has some ideas. Here's the situation:

I have a Corning PC-420D Hotplate/Stirrer. It is compatible with an external probe that will allow the plate to hold a solution at an exact temperature. Pretty useful. However, Corning thinks it's worth one hundred and seventy five dollars. I kid you not. For a themometer. Utter BS.

Anyway, I've taken to building one myself for far cheaper, and I've almost succeeded. I ordered the real one for testing purposes (going to send it back later), and so far I've determined the following from testing it with a multimeter:

The probe connects to the back of the hotplate through a standard PS/2 Port. The probe itself is nothing more than a type K thermocouple, the positive end corresponds to pin 4 and the negative end corresponds to pin 3. A PS/2 pinout diagram is attached at the bottom for reference.

Pins 1 and 2 are shorted (connected), and shorting those pins is what tells the hotplate when the thermocouple is connected.

Wiring it up just like this, it almost works. It hold temperature all right, but in my experiments, it was exactly 17C too high. The next day it was 20C too high. I think I know what's going on here:

There is a circuit I have neglected. Clearly, the hotplate believes ambient temperature to be 0C right now, that explains the temperature discrepancies. Of course they would put the cold-junction compensation device in the probe, as if they put it in the hotplate it might be influenced by the heating element and give inaccurate temps. Because I haven't wired it, the hotplate assumes it is zero, and therefore it goes too high.

Testing the official external probe multimeter, there is indeed one more circuit. With the one lead placed on pin 6 and the and on either pin 1 or 2, I get a resistance measurement of 42.44K ohms. In diode check mode, a voltage drop of 2.150V is read only with the positive lead on pin 6, not the other way around - current flows in one direction. When I put the probe in a cold place, the voltage drop increases - it went to 2.200V after being in my freezer for 30 minutes. Heating it with a heat gun makes it drop.

I'm feel like I'm close to understanding what the cold junction compensator is. It checks the boxes for being a diode - the right resistance, current flows in 1 direction only, and the voltage drop increases as temperature decreases, and vice versa. However, the forward voltage is really high for a diode.

I guess it could also be something else, but I don't know.

This is where my electronics knowledge dries up. I don't know enough to identify the ambient temperature sensor. It's the last circuit - if we can figure out what it is, then anybody with a Corning hotplate can have a fully functioning temperature probe for a few dollars. Ideas, things to test that would narrow it down, and all other suggestions are much appreciated. If this never works out, my probe is still mostly functional, and I can deal with it, but it will be annoying to adjust for room temp during sensitive reactions in a lab where temperature fluctuates so wildly!

Thanks for any and all help!

robertmach_6-pin-din.JPG - 36kB

chemship1978 - 23-11-2020 at 12:07

Maybe that, adapted, could be of use? https://www.youtube.com/watch?v=ORBgemwMLS4&ab_channel=N...

p4rtridg3 - 23-11-2020 at 12:47

Quote: Originally posted by chemship1978  
Maybe that, adapted, could be of use? https://www.youtube.com/watch?v=ORBgemwMLS4&ab_channel=N...


Well, that is where I got the idea from, but the procedure is entirely different. His hotplate uses a PT1000 sensor to detect temperature, mine uses a type K themocouple. They work in very different ways.

Metallophile - 23-11-2020 at 13:17

Quote: Originally posted by p4rtridg3  
In diode check mode, a voltage drop of 2.150V is read only with the positive lead on pin 6, not the other way around - current flows in one direction. When I put the probe in a cold place, the voltage drop increases - it went to 2.200V after being in my freezer for 30 minutes. Heating it with a heat gun makes it drop.


I would expect a much bigger change for being put in the freezer. Are you sure that's actually the cold junction sensor? The voltage looks about right for a red LED.

p4rtridg3 - 23-11-2020 at 13:25

Quote: Originally posted by Metallophile  

I would expect a much bigger change for being put in the freezer. Are you sure that's actually the cold junction sensor? The voltage looks about right for a red LED.


It looks about right to me, about 2mV/degC. I'll try it again to be sure.

Twospoons - 23-11-2020 at 13:54

Check to see if theres any bias voltage coming out of the hot plate. The cold junction sensor may not be a diode, but an analog sensing chip. There are quite a few different types eg AD590 produce a current proportional to temperature, TMP36 produces a voltage proportional to temperature. Your measurements so far would suggest its something like that.
These types of chip are used due to being reasonably accurate without calibration (typ 1C or better)


[Edited on 23-11-2020 by Twospoons]

p4rtridg3 - 23-11-2020 at 16:35

Quote: Originally posted by Twospoons  
Check to see if theres any bias voltage coming out of the hot plate. The cold junction sensor may not be a diode, but an analog sensing chip. There are quite a few different types eg AD590 produce a current proportional to temperature, TMP36 produces a voltage proportional to temperature. Your measurements so far would suggest its something like that.
These types of chip are used due to being reasonably accurate without calibration (typ 1C or better)


[Edited on 23-11-2020 by Twospoons]


Ooh, interesting theory. Is there any way I could test this with a multimeter, or do I need something else?

Edit: I'm not sure, but some quick searching suggests that these things function only when powered. The only thing attached to the leads is my multimeter.

[Edited on 24-11-2020 by p4rtridg3]

Metallophile - 23-11-2020 at 17:11

Quote: Originally posted by p4rtridg3  

Ooh, interesting theory. Is there any way I could test this with a multimeter, or do I need something else?


I agree this sounds interesting. I would look for voltage on the connector pins on the hot plate side. You might need to jumper pin 1 and 2 to make it think a probe is connected. The AD590 that was mentioned runs from 4-30v. So some of the pins would need to supply voltage in that range, to power the sensor.

p4rtridg3 - 23-11-2020 at 17:35

Quote: Originally posted by Metallophile  
Quote: Originally posted by p4rtridg3  

Ooh, interesting theory. Is there any way I could test this with a multimeter, or do I need something else?


I agree this sounds interesting. I would look for voltage on the connector pins on the hot plate side. You might need to jumper pin 1 and 2 to make it think a probe is connected. The AD590 that was mentioned runs from 4-30v. So some of the pins would need to supply voltage in that range, to power the sensor.


Alright. Interestingly, by measuring the voltage between pin 6 and the shorted pins, I got just about 5V exactly. (EDIT: I originally said 1V, not sure what was going on with my multimeter, but anyway it's now holding quite steady at 5V) Therefore it's in the range of those sensors.

[Edited on 24-11-2020 by p4rtridg3]

[Edited on 24-11-2020 by p4rtridg3]

p4rtridg3 - 23-11-2020 at 20:19

One more thing: I found an image of the hotplate's PCB online. You can see the female end of the PS/2 port in the top left. Maybe this picture will help someone.

IMG_7889.jpg - 3.1MB

Metallophile - 23-11-2020 at 20:20

Quote: Originally posted by p4rtridg3  

Alright. Interestingly, by measuring the voltage between pin 6 and the shorted pins, I got just about 5V exactly. (EDIT: I originally said 1V, not sure what was going on with my multimeter, but anyway it's now holding quite steady at 5V) Therefore it's in the range of those sensors.


If you dare, you could supply 5V to your "test" sensor (maybe using a USB charger or battery pack). Put your meter in series with one of the legs and set it to measure the current. Maybe test that at a few different temperatures, and see if you can deduce what it is. The AD590 is 1uA/K, so room temperature would be 298uA, and it should change linearly with temperature. It might also just be an NTC thermistor, which would I think not be linear. A sensor chip would give the same current regardless of voltage, while a thermistor's current would change ohmicly.

densest - 23-11-2020 at 21:21

Making an -accurate- replacement might not be easy to fabricate. Even if you figure out what the sensor IC is the construction of the probe must match as well. You should weigh the cost of the probe against the time & tears to reverse-engineer it and the probability that it won't be accurate because you couldn't match the physical layout and construction of the original.

The placement of the sensor is interesting. The cold-junction compensation for a thermocouple has to sense the temperature where the copper wire and the thermocouple wire join. Putting the sensor in the probe handle allows the manufacturer to use copper wires from that point to the A/D in the base. Cute. That also makes it impossible (as you've noticed) to just plug in a bare type K. You would have to put the IC against and at thermal equilibrium to the copper-thermocouple wire join for your replacement probe to be consistent and accurate.

I'd agree with Metallophile. I'd vote for the IC sensor:
not a thermistor - resistance would vary far more than you've measured
not a diode - too high voltage drop and too little change with temperature

You've accounted for pins 1-2, 3-4. and 6. If 5 is connected to anything then it's a 3-pin IC.
You could put a 1K resistor in series with your meter set to a 10mA current range and measure the current between 6 and 1. If it's 5mA, then there really ought to be a connection to pin 5 inside the unit for the output from a 3-pin IC which outputs a voltage dependent on temperature. If it's much less, then it's a two-pin IC which draws current dependent on temperature. I believe that would be the commonest case.

By measuring the voltage at pin 5 (3-pin IC) or the current at pin 6 (2-pin IC) you can try to use that information to pick a suitable IC for your probe. 1uA-1mV/C are common. It would require making a little test fixture with a 6-pin plug and socket. Bus 1, 2, 3, 4 and 5 across. Put your meter & resistor in series with 6 on each side. You can also bus 6 across and measure voltage between 1 and 5. If you see 0V you definitely have a 2-pin IC.

However.... all bets are off if the sensing IC output is digital. You'd need an oscilloscope to figure that one out.

Twospoons - 23-11-2020 at 22:02

Quote: Originally posted by p4rtridg3  
Hello all,

Testing the official external probe ...



Can you open it up to see what's inside or is it all glued shut?

If not, maybe you can find another PS/2 plug and socket, wire them as an extender, so that you can test voltages on the various pins with the external probe connected and running.

[Edited on 24-11-2020 by Twospoons]

p4rtridg3 - 24-11-2020 at 06:19

Quote: Originally posted by densest  
Making an -accurate- replacement might not be easy to fabricate. Even if you figure out what the sensor IC is the construction of the probe must match as well. You should weigh the cost of the probe against the time & tears to reverse-engineer it and the probability that it won't be accurate because you couldn't match the physical layout and construction of the original.

The placement of the sensor is interesting. The cold-junction compensation for a thermocouple has to sense the temperature where the copper wire and the thermocouple wire join. Putting the sensor in the probe handle allows the manufacturer to use copper wires from that point to the A/D in the base. Cute. That also makes it impossible (as you've noticed) to just plug in a bare type K. You would have to put the IC against and at thermal equilibrium to the copper-thermocouple wire join for your replacement probe to be consistent and accurate.

I'd agree with Metallophile. I'd vote for the IC sensor:
not a thermistor - resistance would vary far more than you've measured
not a diode - too high voltage drop and too little change with temperature

You've accounted for pins 1-2, 3-4. and 6. If 5 is connected to anything then it's a 3-pin IC.
You could put a 1K resistor in series with your meter set to a 10mA current range and measure the current between 6 and 1. If it's 5mA, then there really ought to be a connection to pin 5 inside the unit for the output from a 3-pin IC which outputs a voltage dependent on temperature. If it's much less, then it's a two-pin IC which draws current dependent on temperature. I believe that would be the commonest case.

By measuring the voltage at pin 5 (3-pin IC) or the current at pin 6 (2-pin IC) you can try to use that information to pick a suitable IC for your probe. 1uA-1mV/C are common. It would require making a little test fixture with a 6-pin plug and socket. Bus 1, 2, 3, 4 and 5 across. Put your meter & resistor in series with 6 on each side. You can also bus 6 across and measure voltage between 1 and 5. If you see 0V you definitely have a 2-pin IC.

However.... all bets are off if the sensing IC output is digital. You'd need an oscilloscope to figure that one out.


Thank you so much for the detailed response! I will be testing all of these as soon as possible. Especially the test fixture you suggested - ingenious! So long as it is analog, figuring this out should be a lot less labor intensive than I thought.

p4rtridg3 - 24-11-2020 at 06:31

Quote: Originally posted by Twospoons  


Can you open it up to see what's inside or is it all glued shut?

If not, maybe you can find another PS/2 plug and socket, wire them as an extender, so that you can test voltages on the various pins with the external probe connected and running.

[Edited on 24-11-2020 by Twospoons]


Unfortunately it's pretty darn well sealed up. I tried gently prying it open along a little gap, but all I ended up doing was warping the plastic a tiny bit - I had to quit before I did any obvious damage. There is a small gap between the wire and and the edge of the port (it's hard to explain, but looking down the wire you can see a little bit around the edge of the ring leading into the back of the PS/2) and I have tried to peer in there with a flashlight, but it's just not quite enough to get a clear look at the IC.

If I were willing to destroy the end of the port, I could know definitively. But if I did that, I'd be charged at least 50% of the item price for sending it back damaged, even if I spliced it back together with a new port and restored it to full functionality. I bought it through amazon, but the seller was Thomas Scientific - they're not likely to be fooled by such a repair, even if it works 100%.

monolithic - 24-11-2020 at 11:37

Quote: Originally posted by densest  
Making an -accurate- replacement might not be easy to fabricate. Even if you figure out what the sensor IC is the construction of the probe must match as well. You should weigh the cost of the probe against the time & tears to reverse-engineer it and the probability that it won't be accurate because you couldn't match the physical layout and construction of the original.


After tearing apart an IKA hot plate and trying to reverse engineer certain aspects of its operation and non-operation, I've thought about the utility of an ugly but simple/robust open source hot plate.

[Edited on 11-24-2020 by monolithic]

paulll - 24-11-2020 at 12:25

Any chance you could take a look inside yours and see what the 8-pin chip in the top left and the big chip are? Seems quite possible that U3 is listening to whatever's in the probe, so knowing what it is might shed some light.

p4rtridg3 - 24-11-2020 at 17:35

Quote: Originally posted by paulll  
Any chance you could take a look inside yours and see what the 8-pin chip in the top left and the big chip are? Seems quite possible that U3 is listening to whatever's in the probe, so knowing what it is might shed some light.


I just opened it and took a better photo. The model number seems to be:
02333A 1CK CL16
The little blob in the middle looks to my eyes like a tiny, badly printed Texas Instruments logo, but I could be wrong.


IMG_7125.JPG - 1.9MB

Metallophile - 24-11-2020 at 17:53

Could be this precision op amp:

https://www.ti.com/product/OPA2333

paulll - 24-11-2020 at 18:16

Yeah, that's what it is. Probably being used as a buffering amp for the thermocouple so, not helpful.

Twospoons - 24-11-2020 at 19:27

J2 looks to be a connection for a K-type thermocouple, which means the black device next to it (TO92 package - circled in red) is certain to be the cold junction sensor for that thermocouple.
I would put money on the same device being used in the external probe. You should be able to see the part number if you pull off J2.
being a three terminal device it should have pins going to ground and supply with the third pin being the sensed temperature. This will either go to the Ti opamp or to the micro - I would guess it goes to the opamp, and is used to directly provide cold junction compensation. Which makes it most likely something very similar to a TMP36.

Capture.PNG - 448kB

[Edited on 25-11-2020 by Twospoons]

monolithic - 25-11-2020 at 06:01

https://www.ti.com/lit/an/sbaa274/sbaa274.pdf

For anyone else learning or reading along, section 2.8.3 may be relevant. :)

p4rtridg3 - 25-11-2020 at 14:51

Quote: Originally posted by Twospoons  
J2 looks to be a connection for a K-type thermocouple, which means the black device next to it (TO92 package - circled in red) is certain to be the cold junction sensor for that thermocouple.
I would put money on the same device being used in the external probe. You should be able to see the part number if you pull off J2.
being a three terminal device it should have pins going to ground and supply with the third pin being the sensed temperature. This will either go to the Ti opamp or to the micro - I would guess it goes to the opamp, and is used to directly provide cold junction compensation. Which makes it most likely something very similar to a TMP36.



[Edited on 25-11-2020 by Twospoons]


I will open the hotplate again as soon as I can, but I ran into a problem today. Turns out one of the screws that holds the base on is pretty stripped. Worse, it has a lock washer under it. It might take me a little while to get the damn thing out. :mad:

p4rtridg3 - 27-11-2020 at 00:24

Quote: Originally posted by Twospoons  
J2 looks to be a connection for a K-type thermocouple, which means the black device next to it (TO92 package - circled in red) is certain to be the cold junction sensor for that thermocouple.
I would put money on the same device being used in the external probe. You should be able to see the part number if you pull off J2.
being a three terminal device it should have pins going to ground and supply with the third pin being the sensed temperature. This will either go to the Ti opamp or to the micro - I would guess it goes to the opamp, and is used to directly provide cold junction compensation. Which makes it most likely something very similar to a TMP36.


[Edited on 25-11-2020 by Twospoons]


Good news! I got the hotplate open again, and it seems you were quite right about the device - it's an LM35. It would be logical to assume that Corning would use the same model in the external probe. I'm going to order a few and test it out.

p4rtridg3 - 5-12-2020 at 13:55

After testing some LM35s I got online, I can confirm that it is in fact the cold junction compensator for the probe. Now all I need to do is wire it up.

Huge thanks to everyone who helped me with this. I couldn't have done it without you!

I am now working on a writeup on the procedure so that anyone with a Corning digital hotplate can avoid the Corning Tax and make their own for cheap.

Download - 6-12-2020 at 19:54

I haven't bothered to read everyone else's posts but I'm going to suggest forgetting making a "proper" interface.

Instead, get yourself a solid state relay, hook it up inside the hotplate so it can disconnect the hotplate when power is applied, and then have an external connection which goes to a digital thermostat. On ebay and Aliexpress and those sorts of places you can pick up a thermostat that takes K-type thermocouples for less than US$5 with shipping. The thermostat can turn the hotplate on and off for you.

Doing it the proper way might look neater, but you'll spend loads of time and money figuring it out.

monolithic - 6-12-2020 at 20:03

Quote: Originally posted by Download  
I haven't bothered to read everyone else's posts but I'm going to suggest forgetting making a "proper" interface.

Instead, get yourself a solid state relay, hook it up inside the hotplate so it can disconnect the hotplate when power is applied, and then have an external connection which goes to a digital thermostat. On ebay and Aliexpress and those sorts of places you can pick up a thermostat that takes K-type thermocouples for less than US$5 with shipping. The thermostat can turn the hotplate on and off for you.

Doing it the proper way might look neater, but you'll spend loads of time and money figuring it out.


If you do this you will have to break the circuit for the heating element, otherwise you'll lose stirring as it switches the PID controlled relay on and off.

Twospoons - 7-12-2020 at 14:23

Quote: Originally posted by p4rtridg3  
After testing some LM35s I got online, I can confirm that it is in fact the cold junction compensator for the probe. Now all I need to do is wire it up.


Its kind of surprising just how much reverse engineering can be done from a photo of a PCB.

Always makes me laugh inside when I visit a tech company and have to leave my phone at reception. When a glance inside their product will tell me most of what I need to know...