How to Build a Temperature Cutoff Circuit for High Power LED

[kandrey89]

Hi,

I'm trying to design a temperature cutoff circuit for a high powered LED that will be water cooled, if the user forgot to turn on the water cooling and the board overheats, the temperature sensor on the board will trigger the cutoff circuit to cut power to the LED.

I've briefly looked on digikey for a temperature cutoff IC but couldn't find what I wanted.

Do you know of an IC that does the following:
Senses thermistor? temperature
Uses resistor to setup cutoff temperature
Has a built in gate driver to drive an external power mosfet (that will cutoff the power to an LED)
Maybe has a resistor to setup restart delay before re-engaging the gate drive so the LED doesn't toggle on and off due to temperature hysteresis

[SeanB]

How about a thermal switch on the heatsink, 70C will be more than hot enough for the LED. Will open when too hot, so can be used to either diable the driver, or disconnect power.

https://www.digikey.com/product-detail/en/sensata-technologies/1NT01L-7937/1862-1060-ND/8558208

https://www.digikey.com/product-detail/en/cantherm/F20B070051ZA0060/317-1057-ND/306819

https://www.digikey.com/product-detail/en/sensata-airpax/67F070/723-1190-ND/1631957

All 3 at Digikey for US suppliers, so try from there.

[kandrey89]

It's a 4 channel LED head, 60W per channel. I need to place thermistors on the MCPCB for reaction time and more precise temperature sensing.
Those thermostats are huge, will not fit on MCPCB nor anywhere on the outside of the LED head, besides the outside of the LED head will be much much cooler than MCPCB near the LEDs because of flowing water. I also need to guard against low flow, which might cool the heatsink a bit but the LEDs would overheat.

Thus why I wrote those requirements, the IC would be in a separate box with mosfets and misc items from the LED head, and the thermistors would be connected by wire.

[Zero999]

The LED forward voltage has a negative temperature coefficient, so as long as the current remains the same, a drop in voltage means the temperature is rising. The down side is it varies from device to device, so some calibration will be required, but it has the advantage of being a proxy for the actual junction temperature.

[GerryR]

If you are going to design it yourself, why not feed the thermistors into window comparitors.  If you are going to put the controls into a separate box, then consider running the thermistor wires to some 1/32 DIN PID temperature controllers, which are fairly inexpensive and programmable.  Just a thought.

[beanflying]

I have used a few of these on Test gear fans to keep them off until really needed, generally I use 50C bolted to the heatsinks near the power bit and they are close in Temp and work well. Only for a source photo not a source  eBay auction: #273816889118 If you need more power than they will run then use a relay. Dead easy and minimal bits to go wrong. Call this the KISS option.



The other obvious question I assume you are not using Temperature rise above ambient to turn on the pump for a physical reason otherwise I would be doing this and take the operator out of the loop even add a second higher temp for a safety cutout.?

[kandrey89]

Thanks all, but I'm not looking to change how the temperature is sensed, a thermistor SMD on the MCPCB is the only way.

What I am looking for is a way to use surface mount thermistor as a temperature cutoff, send IC part number and application schematics my way please.

[ogden]

send IC part number and application schematics my way please.

Do you want assembled demo PCB as well?

[kandrey89]

How about IC part number if you're too lazy to reply relevantly.

[DrG]

Hi,

I'm trying to design a temperature cutoff circuit for a high powered LED that will be water cooled, if the user forgot to turn on the water cooling and the board overheats, the temperature sensor on the board will trigger the cutoff circuit to cut power to the LED.

I've briefly looked on digikey for a temperature cutoff IC but couldn't find what I wanted.

Do you know of an IC that does the following:
Senses thermistor? temperature
Uses resistor to setup cutoff temperature
Has a built in gate driver to drive an external power mosfet (that will cutoff the power to an LED)
Maybe has a resistor to setup restart delay before re-engaging the gate drive so the LED doesn't toggle on and off due to temperature hysteresis

I think you could do a better job of describing what you have and what you want.

I thought something when I read your first post, and then something else when I read this:
"Thanks all, but I'm not looking to change how the temperature is sensed, a thermistor SMD on the MCPCB is the only way."

Do you have a temperature sensing system in place (you are not looking to change how the temperature is sensed), or not? Since you know that their are plenty of SMD thermisters and you also know that you don't want any of the (very good) solutions already offered, I would call your attention to two chips that might be of some interest (please do not make me regret bothering to respond):

AT30TS750A https://www.microchip.com/wwwproducts/en/AT30TS750A
This is a standalone temp sensor that can also interface with a MC. You program it once for your application. It has an "alarm" signal (and that would go from the head to your external power control) that can easily be connected to a MOSFET and turn power off when the temperature reaches the programed value and then turns it back on when the temperature has fallen to a programmed value. It comes in an SMT package. I wrote up some experiences using using it here https://www.electro-tech-online.com/threads/minimalist-precision-thermostat-at30ts750a.158314/


MCP9701 http://ww1.microchip.com/downloads/en/devicedoc/20001942g.pdf  This smt device is a simple board mounted temperature sensor with a voltage out representing temperature. So, if you have in mind a thermistor+fixed resistor divider and the midpoint voltage being read by the MC, this would provide the voltage to be directly read.

[kandrey89]

DrG:

What I have: A remote thermistor on MCPCB, not an IC, must be passive due to extreme heat that could be generated on MCPCB.

What I need: A completely analog solution from an IC that can measure temperature from a remote thermistor and either control the gate of a MOSFET directly or through another IC to perform power cutoff (MOSFET switch) due to over temperature limit.

I've seen the ICs you pointed to and they're just ICs with sensors built-in, I cannot use them, they'd have to be mounted on MCPCB and I don't have space nor the environmental conditions on MCPCB.

I think my best bet so far is to use a constant current source through a thermistor and feed an opamp with preset resistor value that will match my desired cutoff temperature, somehow design in a hysteresis, a few more parts than I'd have liked but when there's apparently available off the shelf...

[ogden]

How about IC part number if you're too lazy to reply relevantly.

Lazy is one who demands solution on golden plate.

Fully analog solution for you would be thermistor+resistor and comparator with internal reference with MOSFET as switch.

[edit] Something like TLV3012/TLV3011. Other solution would be low cost micropower comparator with two (external) resistors as reference voltage.

[Zero999]

There is no magic IC which will do everything requested in the original post. At the very minimum a comparator IC and a gate driver will be required. If it's low speed switching, the gate driver could be omitted, but the comparator would still need quite a few external components to implement the timer and hysteresis.

[Mechatrommer]

you can google the aforementioned comparator based... the neatest i can find is this, although not really straight forward, you need to change the 6V zener with the thermistor... and make your own voltage measurement on the resulting voltage divider circuit and make V-temp chart..



the lesser clearer one but spot on temperature control is this...


from and description here... https://www.eleccircuit.com/fan-controller-by-temperature-sensor-using-lm393/

there are many more in google.. for mcu based, you can google how 3d printer senses print head / bed temperature. usually 4kohm (or 47kohm? iirc) high side, thermistor low side...

[james_s]

How about IC part number if you're too lazy to reply relevantly.

Lazy is one who demands solution on golden plate.

Fully analog solution for you would be thermistor+resistor and comparator with internal reference with MOSFET as switch.

[edit] Something like TLV3012/TLV3011. Other solution would be low cost micropower comparator with two (external) resistors as reference voltage.

Agreed 100%

I don't understand what makes people think they can just show up here and demand a turnkey solution be handed to them. People are happy to help and share knowledge but we are not paid consultants, you have to be prepared to do the legwork yourself or offer something in return. There is no incentive to help somebody who is just going to make demands and then disappear once they have what they want.

If I were trying to solve this problem for myself I'd probably use a cheap 8 pin microcontroller, the firmware to achieve the desired behavior is trivial.

[T3sl4co1l]

What kind of controller are you interfacing with?

If you switch the LEDs directly, what does the voltage rise up to when open circuit?  Does the power supply have open circuit fault detection, is there anything to be aware of there?  (It may be useful as an intended protection mechanism.)  Does it not, and it grenades when left open?  Or should it be treated as a true CC supply which is shorted to set the zero-power state?  (Real power supplies usually have output filter caps, so aren't CC on short time scales -- which is to say, a shunt switch would have to absorb the discharge current of those capacitors, and that has to be factored in the design.)

Would it be desirable to have the light output throttle down instead?  Should it flash on and off naturally (temperature + hysteresis)?  Or with a set delay?  Or should it be latched?  If latched, what resets it?  Power cycle?

The absolute minimum implementation, assuming a well behaved power source, is simply the thermistor, with another bias resistor to make a voltage divider; a fixed voltage divider; a comparator between the two dividers, with hysteresis; and a MOSFET.  That's it.  The comparator does need some supply voltage itself, which may require an additional connection, or a regulator.  Several of these components can be found together in a single chip, though I suspect you won't find a three-terminal, ready-to-go solution.  (Protected MOSFETs are painfully close, but their temperature thresholds are probably too high.)

If these components cannot be fitted onto an MCPCB, well, layout issues are your responsibility; we certainly can't help you without any design files to work from.

If you require a complete solution, you will have to provide your existing design files, and a requirements document listing the restrictions you are working under.  There are many contractors on this forum, available at reasonable rates.  If you are expecting an utterly free solution, who knows, maybe someone will volunteer one.  Eventually.  But, eh, you get what you pay for, as they say.

Tim