(Thermoelectric) refrigerator woes..(mosfet switching)

[dslaski]

Me and a buddy needed to go across a couple of states this spring. He bought a box van and converted it to an RV. We got to the refrigeration and he said that we would use a cooler and buy bags of ice along the way. I suggested, with my technical background and relative bravery in the subject, that we could build a real refrigerator with some thermoelectric coolers. He liked the idea and I began.
I researched all the circuits that would be needed and started to amass the parts. Aside from the peripheral circuits, I decided on using a MOSFET (35A rated)to modulate the thermolectric coolers. (off a dep cycle 12V battery) The TE's were rated at about 15V and 18A all together. I would have them run at about 8-10A along with a fan and a few LED's to provide monitoring information. The refrigerator temperature would be managed by thermistor control (btw, if you need one...I found a Radio Shack that actually caries them....try Brookings OR)
The operation and problem is as follows: The thermistor varies in resistance with temperature, high, giving a low resistance and vice-versa. I also included a trim pot. to simulate and also trim the temprature threshold for the MOSFET. Thus, as I reduced resistance on the trim pot. (in series with the thermistor) the output from the thermistor to the GATE of the MOSFET would switch on the TE's and ancillaries (fan, LED's ect.).
The problem arose when I first tried to operate it. The fan and LED's worked like a charm, however, when I hooked up the TE's, the MOSFET let out a sliver of smoke when I varied the trim pot. I found out later that the stage between OFF and ON (about 8K ohms at the pot. out of 110K) at the MOSFET would cause a need to dissipate quite alot of heat. The current that was being gated from the TE's, I assume. It works quite well, with minimal heat dissipation, with the pot. trim to to the lowest resistance.
My first question is, I have a power resistor in series with my TE's...such as about 1.5ohms to maintain my current desire of about 8-10A(such as a current limiting resistor to my LED's). What power (watts) rating should they be? And my second question is if I can decrease the range bewteen OFF and ON such that it switches quicker, nearly like a digital signal...how can I do that? I have considered and tried a Darlignton pair. Using a transistor for the small side and the MOSFET as the large side. (it didn't make any difference, this way ) Should I have used two transistors to make it a true Darlingon pair? (Sorry for my sort of poor and undetailed wording...my first draft was erased by a slip of the finger) I hope you can see my dillema. I want to have the MOSFET drive the thermoelectric coolers without undue heat disspation in short order. It works well fully on like I said. Just the bit in the middle seems to be the problem. If you need more detail of my circuit or past articulations of (such as possible mistakes), please ask. Thankyou.

 

[NewBie/Jarhead]

You might want to go with a switching power supply approach.

Say at half power, that poor MOSFET is trying to handle 60W of heat. A decent synchronous switching supply will drop this down to 3W of heat.

An integrated part can be found from Linear Tech, but you don't really need all the fancy stuff it provides:
https://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1010,C1095,P1893,D4968

 

[Sceadwian]

If you use PWM to control the current going to the TE elements you won't need a power resistor going to each element, just keep the PWM duty cycle bellow an average current that is okay for your devices, a simple micro controller could be used to read the thermistor as a voltage and adjust the duty cycle of the PWM to control the temperature. The mosfet will be used as a switch and will hardly heat up.

 

[dslaski]

Thanx, but.....

Thanx, but these items of help are not in the 'cards', so to speak. I am trying to do this with minimum added parts, considering the effort put into the system already. I would like to make the moment of switching as short as possible. When the unit is the MOSFET is in the full ON position it works quite well. Only dissipating a small amount of heat. Enough for a small heat-sink to take care of. Transistors are 69 cents a pop... I would like some one to tell me that I can devise a circuit with these kind of basic parts (not PWM chips, microcontrollers or a whole new power supply) that will take a signal difference of .75V, this is the point of at the threshold voltage to the full ON state. I want some transistorized filtered action where the this voltage difference is compressed to a useable range such that it never hangs around this quasi state. Thankyou again.

 

[Nigel Goodwin]

Probably not what you want to hear?, but why not just buy a fridge?, they are freely available for campers and run off 12V or bottled gas. Assuming you are trying to use Peltier devices?, they aren't really very good - and only manage a fairly small temperature differential, assuming you can remove the heat from the outside fast enough?.

 

[justDIY]

or use an op amp as a comparator to drive your fets ... below a certain temp, fet off, above a certain temp, fet on. trying to modulate the TECs in some manner is pointless, due to the very slow thermal response time of the mass (a large cavity mostly filled with air)

better yet, use the op amp to drive a fet/bjt and use a relay to switch the high current. then you don't need to worry about driving the fet hard enough to ensure clean switching at high currents.

 

[chemelec]

Me and a buddy needed to go across a couple of states this spring. He bought a box van and converted it to an RV. We got to the refrigeration and he said that we would use a cooler and buy bags of ice along the way. I suggested, with my technical background and relative bravery in the subject, that we could build a real refrigerator with some thermoelectric coolers. He liked the idea and I began.
I researched all the circuits that would be needed and started to amass the parts. Aside from the peripheral circuits, I decided on using a MOSFET (35A rated)to modulate the thermolectric coolers. (off a dep cycle 12V battery) The TE's were rated at about 15V and 18A all together. I would have them run at about 8-10A along with a fan and a few LED's to provide monitoring information. The refrigerator temperature would be managed by thermistor control (btw, if you need one...I found a Radio Shack that actually caries them....try Brookings OR)
The operation and problem is as follows: The thermistor varies in resistance with temperature, high, giving a low resistance and vice-versa. I also included a trim pot. to simulate and also trim the temprature threshold for the MOSFET. Thus, as I reduced resistance on the trim pot. (in series with the thermistor) the output from the thermistor to the GATE of the MOSFET would switch on the TE's and ancillaries (fan, LED's ect.).
The problem arose when I first tried to operate it. The fan and LED's worked like a charm, however, when I hooked up the TE's, the MOSFET let out a sliver of smoke when I varied the trim pot. I found out later that the stage between OFF and ON (about 8K ohms at the pot. out of 110K) at the MOSFET would cause a need to dissipate quite alot of heat. The current that was being gated from the TE's, I assume. It works quite well, with minimal heat dissipation, with the pot. trim to to the lowest resistance.
My first question is, I have a power resistor in series with my TE's...such as about 1.5ohms to maintain my current desire of about 8-10A(such as a current limiting resistor to my LED's). What power (watts) rating should they be? And my second question is if I can decrease the range bewteen OFF and ON such that it switches quicker, nearly like a digital signal...how can I do that? I have considered and tried a Darlignton pair. Using a transistor for the small side and the MOSFET as the large side. (it didn't make any difference, this way ) Should I have used two transistors to make it a true Darlingon pair? (Sorry for my sort of poor and undetailed wording...my first draft was erased by a slip of the finger) I hope you can see my dillema. I want to have the MOSFET drive the thermoelectric coolers without undue heat disspation in short order. It works well fully on like I said. Just the bit in the middle seems to be the problem. If you need more detail of my circuit or past articulations of (such as possible mistakes), please ask. Thankyou.

I Really Don't understand Why the Resistor in Series with the Peltier Device?
Current Limiting like that Really Reduces the Effectiveness of the Cooling.
You should have just bought an 8 amp Module.

And Other than PWM, the easiest way is Full Current, Turning the devices on and off as the temperature rises and falls, Thus maintaing a cool temperature.

But you won't really get down to Refigerator Temps.
Especially if it Hot in the Vehicle.

I have built Many of these coolers.

 

[dslaski]

thankyou. Don't have time to entertain the 'slope' of the replies, however, I have found my answer in the op-amp/comparator, rather...., will explore.

 

[dslaski]

Thanx again, to the informative contributors. As it turned out....the op-amp worked to the degree, that, it would give a relatively clean signal, however, there still was the potential for a partial state of ON or OFF. I relegated to the use of a relay. This, in all its glory of being a more 'hard' touter to the operation of ON or OFF, also displayed with careful modulation an inbetween ('fluttering' operation) state... of running. The trick...that I considered and implemented, considering, states of current ripple....were/was to put a 'buffering' capacitor, in, with. It seems to do the trick. So in hard-earned consideration, the best way to hook up thermoelectrics is to either pulse width modulate them with an appropriate MOSFET and heat sink ,or, to aquire the desired cooling capacity within a thermoelectric unit and apply the above driving circuit. Idealy, as with the proffesional implementations.......PWM.

Thanx

over and out

 

[Nigel Goodwin]

Thanx again, to the informative contributors. As it turned out....the op-amp worked to the degree, that, it would give a relatively clean signal, however, there still was the potential for a partial state of ON or OFF.

You use a small degree of positive feedback to prevent that.

 

[Oznog]

Actually you shouldn't drive a TEC with PWM (pulsed DC). Depending on your voltage, the current can be uncomfortably high! The problem is that resistive losses, which are sigificant, go up with the square of current. So instead of running 5A continuous to maintain the cold, you find your 13.8V battery voltage produces 20A so you try to run it at a 25% duty cycle.

The problem is say the device had a 0.1 ohm resistance. At 5 amps continuous, that's 2.5W of heat generated. At 20A, that's 40W of heat generated while it's on. OK it's only on 25% of the time, but that's still 10W average heat! This is an efficiency killer. I have seen mfgs specifically state not to use PWM for this reason. Actually they recommend you don't simply turn it on/off either because the thermal stresses of turning it on and off every few minutes can actually cause mechanical damage, though it's unlikely you'll run into that for a single device that you're probably not going to use for a really long time.

You need a DC/DC converter. Honestly, simplest, most reliable thing you can do is SuperTex's HV9910 with an inductor, transistor, sense resistor, and diode. That will generate a constant current with little loss and with none of the stability problems of many DC/DC topologies.

You should probably be aware that Peltiers- (thermoelectric coolers) are difficult to make much effective cooling with, much less efficient- and there are hard limits to the efficiency no matter how good your heatsinking/coldsinking is.

You need a really, really good heatsink and coldsink and it requires special mounting methods to avoid mechanically breaking the device, keep the heat from bleeding onto the cold sink, and provide excellent thermal conductivity between the Peltier faces and the sinks.

Peltiers get really inefficient when you need large temp differentials. For example, if it's even 80F inside when you're parked and you need a 45F interior, that's a 35F temp difference. But with crude heatsinking, the hot face of the Peltier could be 110F and the cold side 30F, an 80F differential- which would be beyond the capabilities of most common devices.

Even for well designed commercial models, even a small one can pull like 5A-10A and thus can drain a deep cycle battery in hours once you park if you don't have plug-in power. They often can't keep the interior all that cold either if it gets to be 80F, 90F, 120F inside a car.