Hello, I have a basic knowledge of electronics and have purchased this inexpensive (as in, a few dollars) temperature control board … my aim is to either improve it or replace with a similar but more accurate off-the-shelf item.



It comes fairly close to meeting my need: switching a thermoelectric cooling system on/off at a preset temperature. The chip used (comparator?) is LM2903.

However the behaviour of the "hysteresis" band is much too wide. The cooler switches off when the desired low temperature is reached. But it requires a roughly 10C rise in temperature before switching back on. I am aiming for 1-2C hysteresis band (hopefully my terminology is reasonably correct).

I am measuring these temperatures with a reliable remote digital thermometer - the probe for the thermometer placed alongside the probe for the controller. The temperature measured is the bottom one - minimum (relay switches off) shown in one picture, maximum (relay switches on) in the other.



Here is a close-up of the board which I have left quite large here for legibility:



The target temperature is selected via the trimpot, which from its markings is 10k. It is single-turn or less which makes finessing the adjustment difficult but after a few tries I have it in an acceptable setting. There are two LEDs, one of which shows the circuit is switched on, the other shows when the NO relay contacts are closed (i.e. when cooling is taking place).

Having examined a few simple circuit diagrams (which nonetheless push the boundaries of my knowledge!) I am theorising that one or both of the 10K and 47K surface mount resistors situated between the trimpot and the comparator chip might be involved in determining the width of the hysteresis band. Does this seem correct? The other resistors etc are mainly clustered around the LEDs and relay and presumably connected to their function.

Would it be possible to refine the hysteresis by replacing one or both of these resistors with a different value – or even with a trimpot so that I can narrow the field down to a value that would get me where I want to be?

Or could the existing trimpot be to blame, possibly a lack of sensitivity.

I have seen for sale plenty of temperature controller with LED displays and push-button settings but they are not suitable for my purpose. It needs to be adjustable by trimpot (I would probably replace existing in the circuit shown here with a detent type if available), no LED readout necessary or desired, hysteresis within a 1-2C range, and similar size to circuit shown here. If anyone knows of a cheap (few bucks) off-the-shelf circuit that does the job I would be glad to hear about it.

Otherwise all friendly help/advice appreciated! Thank you!

 

As you can see from the "simple circuit diagrams", the hysteresis is set by the amount of positive feedback from the op-amp output to the + input. Increasing the overall value of VR2 will reduce the hysteresis. Decide what difference in temperatures from on to off is best for your requirements and use the formuli in the last example to calculate the feedback resistor's value in the circuit you are using.

 

Hard to say what's controlling the hysteresis without a schematic.
You could try paralleling a resistor, say 1/4 of the value across the 10k or the 47k resistor to see if that changes the hysteresis.
Once you determine which affects the hysteresis, you likely will have to substitute a higher value to reduce the hysteresis.

Generally I would expect the resistor that controls the hysteresis to have a higher value than either of those, but it shouldn't hurt to experiment.

 

You could try paralleling a resistor, say 1/4 of the value across the 10k or the 47k resistor to see if that changes the hysteresis.
Once you determine which affects the hysteresis, you likely will have to substitute a higher value to reduce the hysteresis.

Thank you. I was thinking along similar lines. Adding a resistor in parallel would DECREASE the resistance would it not? This is obviously the easiest approach for experimentation, as opposed to placing another or greater-value resistor inline to INCREASE the resistance.

Presumably this would allow me to determine whether the relationship is direct or inverse? If I reduce the resistance as you propose, and hysteresis decreases (say, I get 5C hysteresis rather than 10C), then the relationship is direct. If hysteresis increases (to say 15C), the relationship is inverse.

Is this your thinking? Many thanks for sharing your knowledge!

 

Is this your thinking?

Yes, on all counts.

 

The hysteresis of a comparator circuit is set by the feedback resistor from the output to the POSITIVE input. To reduce this hysteresis that resistance value needs to be greater. So you need to know what type of comparator IC you have, and then discover which resistor is used for the positive feedback.
In the photo I see no identification on the IC, which would make the task more challenging.
If you have a digital meter you may be able to measure the value of each of the resistors on the board. The positive feedback resistor will probably be the highest value resistance one.

 

The hysteresis of a comparator circuit is set by the feedback resistor from the output to the POSITIVE input. To reduce this hysteresis that resistance value needs to be greater.

Bingo! Firstly I tried @crutschow excellent guidance by paralleling a lower resistance (10K) to the 47k resistor as a simple first step. Hysteresis increased. So I bit the bullet: desoldered and removed the surface-mount resistor (fiddly but desoldering wick is my friend); soldered hook-up wire to the contacts; and started trying 100K resistors one at a time.

I am now up to 300K (3 x 100K resistors) and the hysteresis has settled at around 1.5C-2C which is satisfactory, I think.

I am a bit bewildered as to why someone would build a temperature control circuit that requires a 10C change to trigger, by only using a 47K resistor. Wondering whether I am missing something that might point to the overall suitability or unsuitability of the board for what I'm doing. Maybe the band is that wide to prolong relay life by not having it opening and closing so often? Or just symptomatic of buck-apiece design?

Next step is to play with replacing the trimpot – which is really sketchy and difficult to set precisely – and replace it with something that allows control over a desirable temperature range.

I really appreciate the advice and the interest shown in helping out!

 

Bingo! Firstly I tried @crutschow excellent guidance by paralleling a lower resistance (10K) to the 47k resistor as a simple first step. Hysteresis increased. So I bit the bullet: desoldered and removed the surface-mount resistor (fiddly but desoldering wick is my friend); soldered hook-up wire to the contacts; and started trying 100K resistors one at a time.

I am now up to 300K (3 x 100K resistors) and the hysteresis has settled at around 1.5C-2C which is satisfactory, I think.

I am a bit bewildered as to why someone would build a temperature control circuit that requires a 10C change to trigger, by only using a 47K resistor. Wondering whether I am missing something that might point to the overall suitability or unsuitability of the board for what I'm doing. Maybe the band is that wide to prolong relay life by not having it opening and closing so often? Or just symptomatic of buck-apiece design?

Next step is to play with replacing the trimpot – which is really sketchy and difficult to set precisely – and replace it with something that allows control over a desirable temperature range.

I really appreciate the advice and the interest shown in helping out!

On many comparator circuits a one megohm resistor is used for hysteresis That should be astandard value for a chip resistor. For the temperature setpoint resistor there are 20 and 25 turn trimpots not much bigger that the single turn device. Is that one used as a variable resistance or as a potentiometer, with all 3 terminals connected to different points? That would mean chose the same value as the present single turn variable resistor.

 

Is that one used as a variable resistance or as a potentiometer, with all 3 terminals connected to different points?

I am not sure of the answer. All 3 terminals are connected to the board. When I adjust the pot, the resistance between terminals 1-2 rises as the resistance between 2-3 falls. And vice versa. I am assuming that 2 is the wiper.

I _think_ my results mean it is being used as a true potentiometer.

BUT I understand you can also set up a pot as a variable resistor with all 3 terminals connected, as described here: "Sometimes it is appropriate to make an electrical connection between the unused end of the resistive track and the wiper to prevent open-circuit conditions."

But if that were the case with my board, there would be zero resistance at all times between pins 1-2 or pins 2-3.

Am I correct in my assumption that it is being used as a true potentiometer?

 

Am I correct in my assumption that it is being used as a true potentiometer?

Yes.

 

Yes.

And since it is connected as a true potentiometer, a replacement should be one with the same nominal resistance. That was my reason for asking. Of course, often there is a bit of variation between the actual resistance and the marked value. A "50K' pot may actually measure 49.2 K between terminals 1 and 3, as an example.