NTC thermal shutdown

[lazzer408]

As i can see in the data sheet, Softstart, Comp, Shutdown, EA input all may be used.

When use Shutdown pin, which seems easiest to me, i think, you will not need a buffer. Just design comparator as to give when T=High>> OUTPUT=1; when T=LOW>> Output= 0 (less than 0.6V) and connect its output to Pin:10 directly. There is a 5 K inside IC at Pin10.

The reference divider (to the comparator input) will have a 3rd resistor to the output of the comparator so that the reference voltage shifts in order to add hysteresis. When the comparator output is open, there will be a voltage present. I think that voltage can be used to drive the shutdown pin. Output high would be 6v with two 10k resistors making up the reference divider and supply at 12v. This does seem to be the easiest method.

 

[Ketje]

Putting all suggestions together, something like this ?
The collector of the trasistor pulls the 1k in the supply positive if at least one of the NTC resistances drops under a preset level.
Mona

 

[nec3]

The reference divider (to the comparator input) will have a 3rd resistor to the output of the comparator so that the reference voltage shifts in order to add hysteresis. When the comparator output is open, there will be a voltage present. I think that voltage can be used to drive the shutdown pin. Output high would be 6v with two 10k resistors making up the reference divider and supply at 12v. This does seem to be the easiest method.

Is 12V supply stable enough to be reference ?

 

[lazzer408]

Here's the circuit experiment. I don't think using diodes is going to work. There is too much interaction between the two sensors. Maybe a zener could work but I can't visualize how to incorporate the hysteresis with zeners unless both comparator inputs were level shifted.

 

[nec3]

Here's the circuit experiment. I don't think using diodes is going to work. There is too much interaction between the two sensors. Maybe a zener could work but I can't visualize how to incorporate the hysteresis with zeners unless both comparator inputs were level shifted.

R2 and R9 represent NTC ?

 

[lazzer408]

yes

I'm trying Ketje's design now but trying to put the NTC's on the other input so they can share a ground. His schematic is a little upside down so it's taking a little time to correct it.

 

[nec3]

yes

I'm trying Ketje's design now but trying to put the NTC's on the other input so they can share a ground. His schematic is a little upside down so it's taking a little time to correct it.

They should have been upper side of divider.

And output of diodes should go to non-inv input.

LM311 does not need pull-up ?

 

[nec3]

Ah, just saw, LM311 both has both collector and emmitter output.

Emitter out = Pin1

 

[lazzer408]

The 311 would need a pull-up but the feedback resistor should provide enough current to drive the SG3525's shutdown pin.

 

[lazzer408]

This works. But it works backwards.

 

[tauro0221]

Hi,
I agreed totally with lazzer408 suggestion to used the TMP709 temperature switch. You can use them by installed a 1K resistor on each output as a pullup then tied each output with a diode with both cathodes tie together like an OR gate and then connect the two diode end to a transistor base. Normally the TMP709 output it is high until reached the temperature setting. This will set the base to high and the transistor collector will be low. When the TMP709 reached the tripping setting the output will goes to zero.That will turn the collector high. The transistor collector need also a pullup resistor. I tried my best to explained it. I am out of town and do not have a way to draw the schematic.

 

[tauro0221]

Hi,
Attached it is the schematic. Finally I was able to draw it.

 

[tauro0221]

Hi,
Ooooppss The diodes are in the wrong direction. Here it is the right one. Sorry.

 

[lazzer408]

That was someone else's suggestion.

Mounting the TMP709 to various surfaces (some not flat) becomes an issue. It's much easier to mount a thermistor to the thermal sources. An RM12 transformer (or a toroid) and a heatsink need to be monitored. I have a small PCB, containing the SG3525, that has headers soldered to another board that contains the rest of the powersupply. I have about 3/4 x 1" left for the thermal management. An 8-pin DIP and supporting components should fit.

 

[tauro0221]

Hi,
Sorry for the confusion. You didn't mention that the surfaces were not flat. That will be an issue.

 

[lazzer408]

Not a problem. I greatly appreciate any input. Every day I learn of new devices that I may need in the future.

 

[nec3]

I think this is what you need.

Deviation because of Interaction between NTC networks is under 1 C if the temperature difference of test points is close. The maximum deviation is 3 degrees when the temperature difference of NTCs is higher than 70 degrees which is practically rare but still not worse than many commercial sensors.

After calibration at the working temperature values, accuracy is around +-1C .

R6 is variable resistance to adjust reference voltage which may also be used for setpoint.
NTC networks has to be fed by a low voltage to keep self-heating low. Ref voltage of 3525 is is good for this.

You can adjust temperature hysteresis by R3.

I saw in TI application note that LM311 can be used emmitter or collector output. This can give you chance to use Shutdown pin. Now, otput is going to SS pin as a pull-down.