Unexplained changes in circuit output

[ericgibbs]

Hi Eric

I can't believe my measurements, laid low today with an aggressive tummy bug. Will remeasure on monday evening.

Cheers

hi,
The position of the thermistor is important, if you are to get the control range you require.
When you feel better post the type number of the thermistor.

 

[Andrew Leigh]

Hi Eric,

a quick note before I pop off to work. When I started this project I did some research, one of the articles which seemed to make a lot of sense to me stated that it was bad to locate the temperature probe on the condensor element. I can't remember the reasons for this but due to the results I was getting this made sense. The old probe however was located on the condensor.

I did some brief research last night and found a couple of articles which stated exactly the opposite. One said that to get the fridge down to 5°C that the condensor would need to be at around -8°C and that it was the condensor temperature that needed montoring and therefore the thermistor needed to be there. It would appear as if measuring the air temperature like I am trying to do is a big no no. That's one of the problems with the internet, always check your sources, I did not. Another article I found last last night was extensive and excellent, they are the FAQ pages of Dometic Fridge Repairer. Almost like a workshop manual.

Anyways, at some point when the evaporator temperature become clearer we would need to alter the temperature profile of your circuit to suit the cooler temperature. Thinking through this last night locating the thermistor on the evaporator would make a lot of sense and would also reduce the hysteresis due the evaporator being like an inverse heat sink. Each time we opened the door the thermostat would switch on as the air temperature had altered.

Does this make any sense to you?

Regards
Andrew

So tonight will locate the thermistor at the condensor, this will completely alter the temperature profile of my existing unit and I may need to match the resistors accordingly.

The 25k that you simulated on the thermistor has to be correct, even appears in the manufacturers temperature table so either I can't measure your I damaged the unit when soldering, I have a spare to check against.

 

[Andrew Leigh]

Morning all,

what a learning curve.

Well have attached the thermistor to the evaporator plate and this appears to have made a significant difference to the hysteresis and thermal control.

Measured some temperatures this morning with the existing thermistor and controller, these were;

05:30 - 2.3°C
05:45 - 2.4°C
06:00 - 2.1°C
06:15 - 2.1°C
Fridge door opened for a little while to check things
06:30 - 2.4°C
Fridge door opened for a little while to check things
06:45 - 2.5°C

These are all in a significantly narrower band than previously recorded.

Then, taking the spare thermistor bulb in free air next to the thermometer, it measured 16.3kΩ at 17°C (yup it shoulf be summer here it's flipping freezing). Took the same bulb and positioned it directly adjacent to the remote temperature probe in the fridge it measured 24.5kΩ at 2.5°C.

I then pressed the bulb (without touching it) against the same position on the evaporator that the original thermistor is, it measured 27kΩ, immediately thereafter to check if there was error between the two bulbs I measured the resistance of the original bulb, it measure 15.8kΩ, clearly a problem.

So circuit aside for the time being a couple of issues are clearer / more apparent. The first is that the postitioning of the probe is critical as you correctly stated. When positioned on the evaporator plate the thermal control is significantly better and hysteresis is much improved.

The original thermistor is either damaged, defective or was damaged by me during soldering. This need replacing but will spend the time to find one with better spec's and one that has a screw thread mounting for more intimate cointact with the evaporator plate.

When I get there I will finally be be able to confirm the resistance of the thermistor at desired temperature so that you can tweak the circuit.

Eric, have you checked the circuit layout and schematic I produced, nothing stops me from getting the board produced now as only the resistor values will change .... or should I wait a while.

Thanks for all the help thus far
Cheers
Andrew

 

[ericgibbs]

hi Andrew,
Your results for the thermistor are very close to the expected values.

The last circuit you posted looks OK to go, dont use the earlier version with the SSR's.

The +12V should be regulated for best performance.
The hystersis resistor value should be selected to suit the high/low limits of the required tempr.

 

[Andrew Leigh]

Hi Eric,

thanks will proceed accordingly.

Just a point of clarification on the regulated 12V. When driving on 12V supply the power source will be a battery fed via the alternator, at 14.4V will there be enought headroom to use a 7812? Is there a more elegant solution using perhaps a zener?

Cheers
Andrew

 

[ericgibbs]

Hi Eric,

thanks will proceed accordingly.

Just a point of clarification on the regulated 12V. When driving on 12V supply the power source will be a battery fed via the alternator, at 14.4V will there be enought headroom to use a 7812? Is there a more elegant solution using perhaps a zener?

Cheers
Andrew

hi,
I would just use the LM2904CT low voltage drop out 12Vreg, its drop out is 0.5V.
The 7812 needs approx 3V overhead to be sure.

 

[Andrew Leigh]

Hi Eric

Been searching for a datasheet for ages, that component would not perhaps be a 2940 rather than a 2904?

Cheers
Andrew

 

[ericgibbs]

Hi Eric

Been searching for a datasheet for ages, that component would not perhaps be a 2940 rather than a 2904?

Cheers
Andrew

hi,
Woops, a typo, attached the d/s.

 

[Andrew Leigh]

Thanks mate

 

[Andrew Leigh]

Hi Eric,

feel like a right idiot, will explain later.

I understand many things but unfortunately electronics is not one of them. I really battle with electrons. Another two issues if I may?

New Circuit (yours)
Before I get the board made some clarification please. The LM 2940 calls for two capacitors to be included under certain conditions. C1 if the regulator is located far from the power supply filter and C2 for stabilty. You had already included a 22uF in your original circuit, must I add another one?

Attached is the revised schematic and board.

Is it good practice to etch only around the tracts like I have done (ratsnest)?

Old Circuit
I am using the old circuit now to experiment with to get a better understanding.

The reason the two thermistors measured differently is because they were, an oversight on my part. The circuit originally called for 15k resistors, in order to get the balance right this required a thermistor to be at ~15k when at target temperature, this translated to a thermistor value of 4k7 at 25°C. Indeed it was a 4k7 thermistor in the circuit, that why it measured 14.5k at target temperature. I am sorry I told you 10k I forgot that I had bought two lots of thermistors. Further more the tolerance on the base temperature of 4k7 is ± 10% and on the temperature measurement is 5% which at target temperaure will account for ± 740Ω. In the area of the temperature curve we are operating at, the curve is for all intents and purposes flat so (3°C - 25°C) / (14k7 - 4k7) = 450Ω per °C.

So if my thermistor was at the extreme of the spec I could have a unit of effectively that could be somewhere in the order of ± 1.6°C. Does this tolerance means that the thermistor will fluctuate between this range or that it was manufactured with a specific error of X% that it will always operate at?

Cheers
Andrew

Thermistor data https://www.electro-tech-online.com/custompdfs/2009/10/WMF11-XXX_AMPRON.pdf

 

[ericgibbs]

hi,
The two additional caps close to the Vreg should be 100nF.

Ref the etching, its hard to see on that image, but do you mean that you have only etched a copper free border around each track/pad.?

If yes, whats the width of the clear gap.?

The thermistor at 4k7 @25C should be OK, I will do a calculation.

The curves on a therm are not linear.

 

[Andrew Leigh]

hi,
The two additional caps close to the Vreg should be 100nF.
Will do

Ref the etching, its hard to see on that image, but do you mean that you have only etched a copper free border around each track/pad.?
Yes

If yes, whats the width of the clear gap.?
They were 0.20mm have increased them to 0.50mm. See new image (does not include the caps).


The thermistor at 4k7 @25C should be OK, I will do a calculation.
Thanks

The curves on a therm are not linear.
OK

Thanks for that, looks like it will be difficult to get 1°C accuracy purely due to the thermistor, am I correct in stating that?

Cheers
Andrew

 

[ericgibbs]

hi
This the thermistor plot 4.7K @ 25C, so it fits nicely with what you have found in practice.
I wondered why our two readings differed.

I'll look at pcb.

EDIT: the etch looks OK, its not the way I would do. If you are confident you will get no solder bridges try it.

Most domestic fridges dont work to a 1Cdeg tolerance, providing the tempr stays lower than 5Cdeg, the recommended temp for keeping food cool and higher than 1Cdeg , so it dosnt freeze your milk, I would settle for that.

 

[Andrew Leigh]

hi
This the thermistor plot 4.7K @ 25C, so it fits nicely with what you have found in practice.
I wondered why our two readings differed.
My head is hung in shame LOL

I'll look at pcb.
Ta

EDIT: the etch looks OK, its not the way I would do. If you are confident you will get no solder bridges try it.
No will revert to purely the tracks then, better safe than sorry


Most domestic fridges dont work to a 1Cdeg tolerance, providing the tempr stays lower than 5Cdeg, the recommended temp for keeping food cool and higher than 1Cdeg , so it dosnt freeze your milk, I would settle for that.
So i'm being a little pedantic then LOL?

What the equation to get the plot?

Thanks
Andrew

 

[ericgibbs]

hi,
Ra= Therm resistance at the desired tempr.
R0 = 4k7
T25= resistance at which the 4k7 is measured.
Tabs= 273C
Ta = Therm tempr

4100 is the Beta value for the therm material, there are a range of Beta values.

Ra= R0 *exp(-4100*(1/[Tabs+T25] - 1/[Tabs+Ta]) )

Ra= 4700 * Exp(-4100* ((1/(273+25))-(1/(273+2)))) = ~ 14.8K

OK.?

 

[Andrew Leigh]

Thanks for that

Andrew

 

[Andrew Leigh]

Hi Eric,

I have been going through Mr Paisleys tutorial on volatage comparitors. Slowly it is dawning on me, I think. In it's simplest form would this statement be correct.

"There are two inputs, these are basically like an electronic N/O or N/C state in a relay. In our configuration we have set the reference voltage on the non-inverting input, this means that as long as the reference voltage exceeds that at the inverting input that the circuit will output signal from Pin7.

We have a voltage divider at Pin 3 (Ra=Rb) so therefore the reference voltage to Pin 3 in a 12V circuit will be 6V. Connected to Pin2 is what I would call a variable voltage divider as one of the elements is a variable resistance i.e. a thermistor.

As the resistance in the thermistor decreases so the voltage at Pin2 increases, when this voltage exceeds the reference voltage at Pin3 then Pin7 stops sending a signal and the circuit is effectively in the OFF state.

However time between cycling is nominally wide as the ON or OFF state is in degrees and not a sudden ON or OFF. To reduce this effect a resistor is introduced between + and Pin 7. This resistor provides some feedback voltage to the reference voltage at + allowing the switching to improve in response time (will need to get my head around this)"

If this is a about right then I want to learn how to calulate the voltages around the circuit to really understand whats going on.

Cheers
Andrew

 

[ericgibbs]

Hi Eric,
"There are two inputs, these are basically like an electronic N/O or N/C state in a relay. In our configuration we have set the reference voltage on the non-inverting input,
Our Vref or Vset [ req tempr] is on the Inverting input.

this means that as long as the reference voltage exceeds that at the non inverting input that the circuit will output signal from Pin7.
The output of the comp will be low, so the transistor is ON.

We have a voltage divider at Pin 3 (Ra=Rb) so therefore the reference voltage... not the Vref/Vset but the Vsense from the Thermistor... to Pin 3 in a 12V circuit will be 6V.
The pin3 voltage will be dependent upon the thermistor resistance

Connected to Pin3 is what I would call a variable voltage divider as one of the elements is a variable resistance i.e. a thermistor.

As the resistance in the thermistor increases so the voltage at Pin3 increases, when this voltage exceeds the reference voltage at Pin2 then Pin7 goes high and the transistor is switched OFF

However time between cycling is nominally wide as the ON or OFF state is in degrees and not a sudden ON or OFF. To reduce this effect a resistor is introduced between the non inverting and Pin 7.

This resistor provides some positive feedback voltage to the thermistor voltage at non inverting , which adds to the thermistor voltage, allowing the switching to improve in response time.
This extra added voltage means the thermistor resistance must reduce much further before the total voltage at pin3 is less than pin2. [to switch the transistor ON state]

When pin1 goes low again,at switch over, as thermistor warms up, this extra positive feedback voltage is removed, so the 'gap' or switching points are changed by the positive feedback

Cheers
Andrew

hi Andrew
Hope you follow that.

 

[Andrew Leigh]

Thanks Eric,

I appreciate the help

Cheers
Andrew

 

[ANTlabs]

I think you should check all of the soldering and if that doesn't work, you should get a new thermosistor.