Over temperature protection circuit

[joniengr081]

Hello, I am working with a new PCB design in which I just came to know that I need to add an over temperature circuit that will send a shut-down signal to the microcontroller in case the measured temperature is out of range. I am familiar with reading RTD circuit. I actually have made once temperature monitoring board with OpAmp, ADC and a microcontroller on the board. But how to get a over temperature protection analog circuit that send shut down signal to microcontroller in case the temperature is out of range. 

[waymond91]

There might be something simpler than doing and op-amp / mosfet design. If I understand correctly they make auto-resetting fuses for DC circuits that essential "blow" when a threshold temperature is reached...

For example:

https://www.digikey.com/en/products/detail/cantherm/SDF-DF128S/1014762

If all you need to do is read a thermocouple and and switch a relay/mosfet I believe there are a lot of solutions for that too.

[bostonman]

How about a thermistor and an op-amp?

You can set the limit with resistors (cheap enough to do), a unity gain buffer op-amp, and the output into the micro. If your circuit already uses op-amps, maybe you'll have an extra one that can be used for this purpose and it will only cost you the price of the resistors.

[PCB.Wiz]

Hello, I am working with a new PCB design in which I just came to know that I need to add an over temperature circuit…

Temperature of what ? What trip temperature?
We need to know proper details, before replies can be useful.

[bostonman]

I assumed it was a PCB going into a box; maybe my suggestion wasn't practical and I should have asked the same questions.

[HwAoRrDk]

What microcontroller are you using? Perhaps it has an in-built analog comparator or analog watchdog that you can use to trigger an interrupt when a temperature sensor voltage input exceeds a threshold.

[jwet]

Maxim now ADI makes little temp switches that switch at a fixed temp- SOT-23.  The MAX6501 is the first part in the series, I'm sure they've been proliferated, I defined them about 20 years ago at Maxim.  They're use in lots of stuff.   LM-75's have gotten pretty cheap as well as DS1820's.

[joniengr081]

The first step in over temperature protection circuit is to measure the temperature on the PCB. The range of temperature on the PCB to be measured can be between 0 degC to +200 degC.

To measure the temperature on the PCB using PT RTD, I found MAX31865 which is RTD to digital.

https://www.analog.com/media/en/technical-documentation/data-sheets/max31865.pdf

I have some questions.

If I use four wire routing on the PCB then Rcable is the resistance of the copper traces, right ? I do not need to mount any SMD physical resistor for Rcable. Is that true ?

According to datasheet.

The capacitor C1 is 0.1 uF. The de-coupling capacitors for Vdd are 0.1 uF. These are the physical SMD capacitors on the PCB.

Datasheet also says that PT100 uses a 400 Ohm reference resistor (Rref), and a PT1000 uses a 4 kOhm reference resistor (Rref).

In my case I am planning to use PT100. Therefore, I will put 400 Ohm reference resistor (Rref).

I am wondering about the reading or converting the "temperature" from ADC code which I assume read by the I2C bus. The microcontroller will read ADC code and convert using the following formula into degC. The formula is mentioned on page 11 of 26.

Temperature (C) ≈ (ADC code/32) – 256.

There is a Table 9 on page 20 of 26. This table describe the values for Temperature (C), RTD Resistance (Ohm), RTD DATA Reg (01H-02H) (hex), ADC Code (dec), and the last column is (ADC code/32) – 256.

I was calculating if I read the ADC code 15901 (decimal) or 7C3Ah (hex), than

Temperature (C) ≈ (15901/32) – 256

Temperature (C) ≈ 240.91 degC

This value is written in the last column in Table 9 with the ADC code 15901 in the same line.

To calculate the RTD Resistance (Ohm). There is a formula on page 19 of 26.

Resistance_RTD = (ADC Code x Rref)/2^15
Resistance_RTD = (15901 x 400)/2^15
Resistance_RTD = 194.10 Ohm

This value also written in the second column of Table 9 in the same line where ADC code 15901 is written.

I am actually concerned about the correction of the measurement. The last column of Table 9 says 240.91 degC which match to my calculations. How do we get 250 degC, the first column value ? Is that corrected value ?

I tried to use the Callendar Van Dusen equation from page 10 of 26.

R(T) = R0(1 + aT + bT^2 + c(T - 100)T^3)

where,
a = 3.90830 x 10^-3
b = -5.77500 x 10^-7
c = -4.18301 x 10^-12 for -200 C < T < 0 C
c = 0 for 0 C < T < +850 C

The value of c parameter will be zero for positive temperature. The equation will be reduced to.
R(T) = R0 x (1 + aT + bT^2)

If I put the values of a and b parameters and the temperature T as 240.91 degC. Then I get 190 degC. How do I get 250 degC. ?

[mikerj]

If you want a digital shutdown signal rather than having the micro sample temperature via an ADC, then using a thermistor with a comparator (and adding hysteresis) doesn't get much simpler.

[joniengr081]

I just saw that  MAX31865 has SPI interface. We actually have I²C as communication interface. It is possible to add more wires to the microcontroller which is located on another board.

I am wondering is that possible to use I²C instead SPI using this I²C-Bus to SPI Bridge "SC18IS602B". Would microcontroller be able to read data from MAX31865 ?

[NpR]

Hello
What I fail to understand is that why would you use some digital bus or another to be used for "protection" functionality. IMHO, overtemp protection circuits should be independent of whether your micro is still running normally and whether or not it can actually still read the temp sensor through SPI or I2C or whatever. What you would want to do is to design a sort of overtemperature reset circuit (that, when flipped, shuts down anything that is causing the temperature to rise on the PCB dependless on what the MCU is doing).

The simplest I've used is an NTC with a 3-pin voltage ref such as the TL431 and a few resistors. The TL431 pulls its cathode close to ground when 2,5V is applied to its ref input. Just arrange a resistor and the NTC to the reference in such a way that the circuit trips at your desired temperature. Use the output to either tell your micro about overtemperature condition or simply reset the micro (provided that your other circuitry is designed to also power down if the micro is reset). The TL431 datasheet has a wealth of information and example circuits

If you wish to go down the digital polling approach, many MCUs these days have in-built temperature sensors which can be read quite readily by the internal ADC. No external components, no hassle, no additional costs / buses / converters. Unless of course you wish to measure a point temperature of a certain area of the PCB, but you did not specify this.

Regards
Nick

[David Hess]

I would go with a thermister, operational amplifier, and 3 precision resistors to make a bridge so it does not even require a reference.

[sparkydog]

If you wish to go down the digital polling approach, many MCUs these days have in-built temperature sensors which can be read quite readily by the internal ADC. No external components, no hassle, no additional costs / buses / converters. Unless of course you wish to measure a point temperature of a certain area of the PCB, but you did not specify this.

Be aware that these (so I've heard, at least) may require calibration in order to be acceptably accurate.

[joniengr081]

Thanks for your replies. These are very helpful to me.

A brief description of our application is that we have a MCU/FPGA board which is located a meter away from the application board. We are designing the PCB for the application board. The MCU/FPGA board is a kind of master board and it controls On/Off power supplies of the application board. There are linear regulators on the application board whose enable signal is controlled by the MCU/FPGA board. 

We need to measure the temperature on the application board and then implement the following two solutions for over temperature protection.

1- Measure the temperature with digital output for example using MAX31865 RTD to digital and send the digital measurement to the MCU/FPGA board. The MCU/FPGA will compare the temperature readings to a reference internally programmed. And if the measured temperature of the application board is above the certain value then it will cut off the power to the application board by disable the linear regulators.

2- The second option is to measure the application board temperature with analog output and then compare the measured output with a reference signal correspond to the shut down temperature. An OpAmp comparator on the application board which will disable/enable the linear regulators after comparing the measured temperature with the reference temperature. 

Please note that the temperature measuring circuits has independent power supplies on the application board.

[Jeroen3]

Overtemperature typicall is not that critical. If you want analog you can get an NTC and any comparator/opamp.
Like this: https://www.ti.com/tool/CIRCUIT060002
It is not accurate.

You can also use an LMT86, do the comparator/opamp but also feed the value to an ADC for readout, this is more accurate.

Make sure to add some hysteresis! Otherwise your board will powercycle.

[joniengr081]

For some reason we need to have both options open. The one with digital output for example using MAX31865 which is RTD to digital. And the analog output option. For this we need to find 1 mA current source that we connect to PT1000 gives the voltage across the PT1000 proportional to the change in resistance. Regarding the accuracy +/- 1 degC is totally fine. I have found some precise current sources but the circuit, how we connect to the PT1000 ? Any suggestion on circuity please ?

[coppercone2]

whatever you do if its important add a thermal fuse lol

I just had a water cooled thing I forgot about left on after the coolant was turned off. Its thin silicone hose so there is hardly any fluid, but even at low power levels, about 30 min later a hose popped off and squirted me with some warm coolant (i use non conductive just incase of things like this). Very small 10W load but after a while heat builds up

[Psi]

Yep, usual method for some fault tolerance is to implement a software solution to detect overtemp and handle it gracefully. But also have a hardware system that will kick in at a higher temp to force everything off if, for whatever reason, the software solution has failed to keep the temp safe. Since the hardware system will never kick in unless something is SERIOUSLY WRONG a one-time blow thermal fuse is usually fine. You want a technician to look at the device closely before anyone powers it on again anyway.

If you have to locate a thermal fuse 'off-board' you can route the low voltage power for a primary mains relay through the cable to the off-board location through the fuse and back again.  Note this also means the system wont even power on unless you connect the cable between the two systems. Which may/maynot be desirable in your application.

If using a resettable thermal fuse then with a bit of careful power routing and coding you can detect the primary relay has switched off due to this while there is still power stored in caps running the MCU and quickly store to EEPROM a flag to record this occurrence. Just need to make sure there is enough on board capacitor storage to keep the MCU running for the few ms it takes to complete the save with good margin.