Dealing with fan failure

I've built several passively cooled lamps and am considering trying my hand at a fan cooled lamp. I've avoided fan cooled in the past because I didn't want to deal with fan failures and dead COBs because of it. Is there an easy way to cut the power on fan failure?

Alternatively, what's the chance of a COB surviving on a decent heat sink designed for use with a fan? How about in a vented enclosure? I know those questions are circumstantial but any general opinions would be appreciated.

Design your heat sinks to be passive at around Tc=80C, then put fans on it and run the fans at low speed.

That's the basis of most of my "semi-passive" designs.

To be honest, I'm looking at using those star sinks if I make another lamp.

Rahz said:

I've built several passively cooled lamps and am considering trying my hand at a fan cooled lamp. I've avoided fan cooled in the past because I didn't want to deal with fan failures and dead COBs because of it. Is there an easy way to cut the power on fan failure?

Alternatively, what's the chance of a COB surviving on a decent heat sink designed for use with a fan? How about in a vented enclosure? I know those questions are circumstantial but any general opinions would be appreciated.

Click to expand...

Some companies are paying attention.

http://shop.stevesleds.com/Thermal-Switch-Overheating-Protection-8794102497.htm

I put them on all of my cpu/led coolers. Simply add it inline, after the cob. The circuit will open and the led will shutoff at the rated temperature. Thermal adhesive it somewhere it will fit and screw it down so it dosen't get knocked off. I have some pictures somewhere in my thread..

Also a good idea to use insulated quick connects and a little liquid electric tape on the exposed connections to isolate everything

Interesting, so placement and the efficiency of the heatsink in passive condition would determine the Tj at which the unit turned off.

Have you done testing to determine it generally opens at or before Tj85?

Positivity said:

Some companies are paying attention.

http://shop.stevesleds.com/Thermal-Switch-Overheating-Protection-8794102497.htm

I put them on all of my cpu/led coolers. Simply add it inline, after the cob. The circuit will open and the led will shutoff at the rated temperature. Thermal adhesive it somewhere it will fit and screw it down so it dosen't get knocked off. I have some pictures somewhere in my thread..

Also a good idea to use insulated quick connects and a little liquid electric tape on the exposed connections to isolate everything

Click to expand...

Normally the base of the heatsink near the cob is very close to the Tc temperature on a small cpu cooler. It turns off right around the same time my temp display says it should. I have some at 80c and some at 60c, i prefer the earlier cutoff at 60c

That would be ideal. I could do some testing and try some nominal increases in heat sink size to see if I can keep it under 80 without too much more expense if the planned design doesn't do the job.

churchhaze said:

Design your heat sinks to be passive at around Tc=80C, then put fans on it and run the fans at low speed.

That's the basis of most of my "semi-passive" designs.

To be honest, I'm looking at using those star sinks if I make another lamp.

Click to expand...

churchhaze said:

Design your heat sinks to be passive at around Tc=80C, then put fans on it and run the fans at low speed.

That's the basis of most of my "semi-passive" designs.

To be honest, I'm looking at using those star sinks if I make another lamp.

Click to expand...

Well said. My sentiments exactly. Big sinks and just enough fan to keep air moving. Semi-passive design has the advantage of wide margin for fan failure. I actually ran my main overhead light for several days passively but my Veros were getting up to as high as 70c. Cooler is better for several reasons but the light would have survived and I know now that a fan crapping out isn't too much of a problem.

How about something that cuts power altogether in case of the big "what if", like electrical fire, etc...?

All I have come up with so far is a MOV....... and this is cheating but the Arduino has a flame sensor, I have really been pondering....lol...meant for robots but now could be part of thermal safety circuit..?

I've noticed most drivers have temp cutoffs as well. Depending on how hot the drivers got in the enclosure in the event of fan failure they might be able to save the COBs.

Arduino + TMP36 temperature sensors + SainSmart relay modules. If temperature of TMP36 which is mounted on opposite side of heat sink directly under cob goes over X degrees either dim the led or cut the power using SainSmart relays. Or you can use spark.io Core (vs arduino) and you have a wifi connected device you can send data to the cloud - sms you of high temperature situations etc.

PS: bonus points for controlling the fans PWM values with speeds of temperatures. let me know if you have any questions - I have this exact setup running ATM and its pretty sweet

beodrone said:

Arduino + TMP36 temperature sensors + SainSmart relay modules. If temperature of TMP36 which is mounted on opposite side of heat sink directly under cob goes over X degrees either dim the led or cut the power using SainSmart relays. Or you can use spark.io Core (vs arduino) and you have a wifi connected device you can send data to the cloud - sms you of high temperature situations etc.

PS: bonus points for controlling the fans PWM values with speeds of temperatures. let me know if you have any questions - I have this exact setup running ATM and its pretty sweet

Click to expand...

This is where I was at.....but just attaching a thermocouple and then writing a code to set shutdown @ 60c or so.....

But these are even easier right now.....
http://www.amazon.com/THERMAL-temperature-TF-242-60°F-Fahrenheit/dp/B00EN20T0C


My problem with the switch from Steve's.....What happens say if your fan goes out in the start of the day......Doesn't that mean that the diode will basically strobe on and off all day, till you rectify? That part seems weird!!!! When the fan dies I WANT the power to cut...Period!

@Positivity


What are your thoughts on the Steve's switch...I have looked at similar for a while now and really like the idea, but after thinking about it last night and posting ^ .....what are your thoughts on that? Have you actually tested the switches on a Fan-less cob, just to see the reaction?

Abiqua said:

This is where I was at.....but just attaching a thermocouple and then writing a code to set shutdown @ 60c or so.....

But these are even easier right now.....
http://www.amazon.com/THERMAL-temperature-TF-242-60°F-Fahrenheit/dp/B00EN20T0C

My problem with the switch from Steve's.....What happens say if your fan goes out in the start of the day......Doesn't that mean that the diode will basically strobe on and off all day, till you rectify? That part seems weird!!!! When the fan dies I WANT the power to cut...Period!

Click to expand...

Yep...it will turn on and off as it cools. I haven't really looked further as i'm pretty happy with things as is. The fuse looks like a nice option too..

I don't like having nothing, thats for sure..

Say worst case scenario, at least one of the worst, light ends up on the ground face down on something flammable. The fuse should blow before a fire starts. Very nice....

The thermal switch would go on and off. Possibly starting a fire.

I keep small auto fire extiguishers over my tents just in case. Between the thermal switch and extinguisher i can sleep well at night. I'll look into the fuses, i do prefer a stay off function....other than the having to replace them..ugh...lol

The thermal switch works smoothly. I can watch the temp on the fly with my pc temp/fan controller. It cuts off at about 58c according to my thermocouple. Takes a few minutes before it starts back up again.

Got the idea from my "upgraded" to cree china direct light that would constantly cut off...lol. Thats what happens when you rely on a fan blowing at a thin pcb....dodos..

Well,there is the "analog electronics " way also ..

But first ,let's define "Fan Failure" ..
Fan failure can occur:

-Due to fan malfunction and/or total failure .
&
-Due to Fan PSU malfunction and/or total failure .

Making a system ,analog (ICs ) or digital (MCU or PLC ) for checking upon both the fan and the fan PSU ,
ain't exactly the best idea (pricey,spacey ,wirey and more ) ..

So ,correctly you gentlemen are discussing for thermal checking ...

Forget the COBs ..
Place the heat sensor somewhere on the heat sink .
But ,first you should measure the temperature of the heat sink spot ,
when the operating COBs have reached their thermal equilibrium .
Then check also the ambient Temperature .

So ,if Ta=30°C and the COBs are operating more than 6-8 hours
-more than enough in most cases ,for the COBs to reach their thermal equilibrium-
Thermally steady state -
and that spot of the heatsink Ths= 45°C ,then
the "power cut-off " threshold should be T ambient MAX + Ths .

If the maximum ambient temperature of the example can reach up to 38°C ,then
Tthres= 38+45 =83°C .

Now..
A mechanism of some short is needed to check the heat sink spot ,by sensing the heat and cut-off the power if
the temperature goes above 83°C -remaining at the previous example- ...

So a sensor is needed and an interface with the drivers power line (the AC side always -NOT THE COB DC SIDE,NEVER! ).
The sensor can be negative thermal coefficient thermistor .

(a type of resistor that changes it's resistivity greatly ,
depending on the temperature of it's material .Negative coeff if the resistivity decreases as temps rise ,positive coeff when resistivity increases as temperature increases )

It can be a thermocouple .

(two wires of different metals welded together at one of their tip.
Because metals have different electron charge ,between them they create an electron dynamic.
-Remember the "Galvanic Corrosion " ? That's why ..-
That means that they are creating a small ..hmm..battery ,sort of speaking ...They create a very small voltage
at their free/open tips.That yes,changes dramatically ,
when the temperature of their welded together tips point ,changes too .)

or dedicated ICs lIke LM35

The "interface " that cuts-off the power to the drivers ,
of course is either a mechanical or solid state relay .
(NO=normally open => OFF when no signal to relay
NC=normally closed=> ON when no signal to relay )


So ...In case one wants to stay off from -bulky- microcontrollers or programmable logic controllers ,
then there's the analog electronics way -
at least for such simple circuits -..The good ol' way ...

But firstly ...

_Learn to etch your own PCBs ..


Very easy -Very cheap -Very neat.

All you need is a B/W cheap laser printer ,some blank single sided FR4 pcb plates ,
an etching solution and the freeware (300 pins limit ) edition of this software :
http://diptrace.com/download-diptrace/
http://diptrace.com/download-diptrace/

Then you can start entering the world of electronics and make your own thermostat controller for your COB light ,starting from here (just an example-search there are plenty more like that ) :

http://electronics-diy.com/electronic_schematic.php?id=1043

Very easy -Very cheap-Very neat
DIY to the bone !

Cheers.

Was hoping you'd chime in sds. You always catch what i miss...

So never on the DC side? It does work really well there for the function of shutting it off when its hotter than i like.

It'll be easy to move it to the AC side though since everythings in a case. That would be a problem for open setups having to run a new set of wires to the cob.

I'm actually just waiting and hoping for some next gen drivers that will have led thermal protection included as a feature. I think philips already has it but not all the models.

Thermal Foldback....thats whats its called?

Went back and checked philips and it seems it only has a driver thermal protection where it decreases current to protect the driver and not the led itself.

This article from digikey shows how it could function with a led.

http://www.digikey.com/en/articles/techzone/2013/jul/how-thermal-foldback-improves-the-reliability-of-led-lighting-fixtures

You could use the open-close on the dimmer wires too on the B's (probably).

open for full power, closed for low power.

@stardustsailor

Thank you for the DIY thermostat circuit....Do you have an opinion on using with Atmel or with an Uno and Relay shield and attaching the thermistor or temp sensor thru I2C?


Thanks for confirming the AC side, I had been wondering about this and had come to the conclusion that even after cutting DC, what about the AC line in, was it different than being at the wall.......apparently it Very Much is different and could be live even after a DC cut, thanks for confirming that.....

Abiqua said:

@stardustsailor

Thank you for the DIY thermostat circuit....Do you have an opinion on using with Atmel or with an Uno and Relay shield and attaching the thermistor or temp sensor thru I2C?

Click to expand...

I haven't played much with the I2C protocol to be honest ..

But check these experiences of mine ,please ,if you like :
https://www.rollitup.org/t/diy-cree-cxa-arduino-thermal-monitoring-protection.824070/

https://www.rollitup.org/t/thinking-of-a-new-light.822679/page-6
https://www.rollitup.org/t/thinking-of-a-new-light.822679/page-8

Abiqua said:

@stardustsailor
Thanks for confirming the AC side, I had been wondering about this and had come to the conclusion that even after cutting DC, what about the AC line in, was it different than being at the wall.......apparently it Very Much is different and could be live even after a DC cut, thanks for confirming that.....

Click to expand...

-Constant Current switching power supplies ,unlikely the constant voltage ones,they are not supposed to operate w/o a load(=> infinite resistance ,as the relay has opened the output circuit to the COBs ) .
They enter the hick-up mode and if they stay at such state for some time or get to that state oftenly ,then they might be permanently damaged .

-Some CC drivers do not have an over voltage protection on the DC output side .In such case if the relay opens the DC side in case of say of simply overheating and NOT 'cause of fan failure,there's the chance that when the relay will close again after the unit has cooled down ,some capacitors of the driver will be fully charged ,as for some time there was no load on the DC output when the relay was open .
Without over voltage output protection ,when the DC output will be closed again ,the fully charged capacitors will briefly
discharge on the DC output line .Thus it might some diodes of the COB to be fried or even worse (althought very difficult) the whole COB to fail.Usually though ,just some diodes are fried and the COB still operates ,but permanently "wounded" .

So ,any relay based automation incorporated with constant current supplies it should control their input AC line (relay just after mains input socket/plug end ,before the drivers ) and not their output DC line (to the COBs).

It is noted on the spec sheets of the most branded drivers.
(check the last page of the spec sheet of every meanwell cc driver that has the 3in1 dimming function )

Cheers

Check out TMP36 - they are analog and dirt cheap - around 2$ a piece. You can hang them off analog inputs on atmel, ardunio etc and pull it down with a resistor (0.01 if you are using a spark core). What I have done and happy to share my code if you want for spark is to read the fan RPM via the green tach cable. If the fan RPM goes under 500 then I cut the power to the lights.