Tiny PWM controller for TEC.. Need some.

[NotSoCoolJ]

As the title states I need a PWM controller for a tec of around 80-100watts. I plan to use a dell Laptop AC adapter with 19.5v and 7.7 amps available for the power supply. The reason I need it small is because I plan to make some can coolers for my family for Christmas. You know like mini desk fridges. I don't want them to freeze the cans and I don't want them to be turning them on and off a lot. So, I need something that will keep the little fridge around 40 degrees F all the time and not use an excess of power doing it.

I have looked around and I have found several controllers that would do the job. The problem with them is that they are huge. I guess I'm gonna have to break out the soldering station for this one.

So far my ideas are:
555 timer based with a trim pot and a power transistor.
Pros. Cheap and easy
Cons. It wont add more power when you put a warm can in. That means it will take a long time to cool it off.

Two of the above for a high and low setting with a temp switch for switching duty.

Any ideas? If I use just a temp switch I'm afraid it will kill the module in short order.

[laKewl]

you could use a 556 as a stable clock then use a thermoresistor as one of the timing to set the width of the pulse and then a trimer cap that you set the temp with (would need to be set buy the user
if you need more info let me know ill bust out my ECAD stuff

[Uncle Jimbo]

Here's a complete design. It has temp control too. The link shows a design to run 14 TECs, but the only difference between that and what you want is the size of the MOSFET and heatsink.

This is pretty simple - 8 or 9 parts. For low current applications it can be quite small - you can bread board it and mount it on whatever you are using to cool the TEC. The whole thing would maybe take 2 inches square to mount.

http://www.xtremesystems.org/forums/...3&postcount=26

I have parts laying around to do this. You might too - the only hard part is the MIC502. There are a lot of other fan control ICs around, but I like these because the setup is very simple. You can use any n-MOSFET with good amp capacity - blown-up power supplies are full of them, and they are cheap online too. Get something with way more current than you need, since there's not much difference in cost and that will reduce chances of it blowing up.

For example, the IRF3315 is a 27A Id part, $1.50 at electronic goldmine:
http://www.goldmine-elec-products.co...?number=A10330

I assume you mean 80-100W cooling power, something like the 12709 and 12710 TECs available on eBay. Those are way bigger than you need, and you complicate the design because they use more current at 19V max than the Dell laptop supplies. A much easier design is to use a 12706, or something even smaller. Those are about $5 on eBay and are a perfect match for the rest of what you want to do. The Dell supply is also way more than you need - a 12V 4A supply is more than enough.

The power needed to keep the fridge cool once it reaches temp is very small if it has decent insulation. With PWM your power will be a few watts at that point.

When you throw in a six-pack of warm cans, the TEC will kick on full, but since it is not in direct contact with the cans, it will take the air in the fridge down to whatever your setpoint is in a few seconds. You won't need a fan, or even a big heat sink - an old CPU sink with the fan off should work fine, or really anything about that size. I don't know how you plan to use the cold side, but a wide but not deep heatsink on the back side of the interior should work fine. A bigger TEC won't work any better unless you find a way to get thermal contact between the cans and the TEC - say fill the fridge with water or something.

I did something similar with a 1.5 gallon foam cooler and a 12706 TEC. I set it at just above freezing, since I like those icy cold drinks. We used it in the car instead of an ice chest for long trips, it just plugged into the lighter socket. If I filled it full of already cold cans, the current draw was less than 1 amp. If I filled it with warm cans, it would draw about 3 amps for a minute or so, then ramp down. It would hold about 12 cans, and take them from room temperature to just above freezing in about 15 minutes.

[NotSoCoolJ]

I was actually thinking of just a single can arrangement. You know, I was thinking of making it look like an old Coke machine. Just enough room for one can. I was also planning on direct contact with the can and TEC, possibly using some water in the bottom of the fridge maybe a half inch deep. I want to be able to cool a can quickly. 15 minutes is a long time when your thirsty.

Your comments about the TEC in general make me think I've got the wrong idea totally when it comes to power requirements. I was under the impression that if I used more voltage the amps would be lower. It seems these are current driven devices though.

That means I can spend a lot less on the power supply then.

This is the module that I had in mind.
http://www.allelectronics.com/make-a...OOLER/-/1.html
Do you think a 50watt would get it done?
Ebay is nice and all but I hate buying things there. It always feels like a gamble.
I do have 3 defunct power supply units here and one that works but I don't trust it with my equipment since it is a cheap Coolmax. I have already raided some of the parts like the X2's and some capacitors, but they all still have the transistors intact. My wife hates the fact that I wont throw out old broken computer parts. Every time I recycle a part I make sure she knows

So I thinking of a parts list. Just to get something down. If you see any problems or omissions let me know.

PS. 12v-5amp. Any ideas where I could get these for cheap? I suppose I could build them but with only a radio shack available here It could get fairly costly quick.
N-MOSFET. The EG ones look good enough for 1.50 a piece. I could pull them out of the PS I have here but that would require work.
Thermistors. http://www.goldmine-elec-products.co...p?number=G2005
With these the resistance goes up when the temp goes down. Will that work?
Clock Hmm. Maybe I'm dense but I don't see what the difference in a 555 vs a 556 in this application. This is all new to me. Will I need a dual timer?
That PWM fan controller is starting to look tasty. I have a bad Motherboard here with fan control. Maybe I can recycle them.
Various resisters and capacitors I have on hand. I even have a few cermet pots lying around here somewhere. Probably the wrong range but we shall see.
I have a true RMS meter with a K-type. So temp calibration shouldn't be a problem.
The schematic you linked. looks fairly easy compared to this whole getup I'm trying to get into. It should be small too.

[NotSoCoolJ]

After inspecting the dead board here it appears it uses lm354 oscillator/comparator. Not what I'm looking for. You wouldn't know where I could get a few of the MIC502 fan controller ICs would you. You weren't kidding. Those things are hard to find! So far I have come across one source and that is Rapid in the UK. I doubt I'll be able to get them cheaply. Anyone need a few hundred of these things? LOL.
So far the alternative is the radioshack fan controller kit which is 10$ if they have them. It is 555 based though. You should have seen me trying to get a few lm386n's the other day. I needed 8 and I went to 3 radioshack locations and came up with 3. Oh yea 3.50$ a pop. Some body is making money.

[Uncle Jimbo]

I was actually thinking of just a single can arrangement. You know, I was thinking of making it look like an old Coke machine. Just enough room for one can. I was also planning on direct contact with the can and TEC, possibly using some water in the bottom of the fridge maybe a half inch deep. I want to be able to cool a can quickly. 15 minutes is a long time when your thirsty.

Your comments about the TEC in general make me think I've got the wrong idea totally when it comes to power requirements. I was under the impression that if I used more voltage the amps would be lower. It seems these are current driven devices though.

That means I can spend a lot less on the power supply then.

This is the module that I had in mind.
http://www.allelectronics.com/make-a...OOLER/-/1.html
Do you think a 50watt would get it done?
Ebay is nice and all but I hate buying things there. It always feels like a gamble.
I do have 3 defunct power supply units here and one that works but I don't trust it with my equipment since it is a cheap Coolmax. I have already raided some of the parts like the X2's and some capacitors, but they all still have the transistors intact. My wife hates the fact that I wont throw out old broken computer parts. Every time I recycle a part I make sure she knows

So I thinking of a parts list. Just to get something down. If you see any problems or omissions let me know.

PS. 12v-5amp. Any ideas where I could get these for cheap? I suppose I could build them but with only a radio shack available here It could get fairly costly quick.
N-MOSFET. The EG ones look good enough for 1.50 a piece. I could pull them out of the PS I have here but that would require work.
Thermistors. http://www.goldmine-elec-products.co...p?number=G2005
With these the resistance goes up when the temp goes down. Will that work?
Clock Hmm. Maybe I'm dense but I don't see what the difference in a 555 vs a 556 in this application. This is all new to me. Will I need a dual timer?
That PWM fan controller is starting to look tasty. I have a bad Motherboard here with fan control. Maybe I can recycle them.
Various resisters and capacitors I have on hand. I even have a few cermet pots lying around here somewhere. Probably the wrong range but we shall see.
I have a true RMS meter with a K-type. So temp calibration shouldn't be a problem.
The schematic you linked. looks fairly easy compared to this whole getup I'm trying to get into. It should be small too.

Making a circuit with all the features of the MIC502 using a 555 or 556 would be complex and use a lot more parts. I have used other fan control ICs but nothing else is as easy to use. You don't need any other clocking components, and it has other features you can use if you decide to get fancy.

It looks like Rapid Electrinics has the MIC502 in qty 1 for £1.95 - not super cheap but not bad.
http://www.rapidonline.com/Electroni...ement-IC/80431

Your AllElectronics TEC is fine, you won't need the cutoff switch with PWM control. That will draw a max of 6A at 12V.

To change the temperature of 1 gram (mL) of water 1 degree C requires 4.18 Joule (watt-second). With 220Ml in a can, if you get good contact, that TEC will cool it 1C every 12 seconds when first turned on, dropping to 1C every 15 seconds as it gets colder. Your 40F target is about 4.5C. So your can will go from 25C to 4.5C in about 4 minutes.

For power, you might be able to get away with a 5A 12V adapter. Here's one for $20 - http://www.censuspc.com/product.php?...8&cat=0&page=1
That is nice and tidy, and would cool your can in 5 minutes.

If you don't mind adding a bridge and a cap, you can use a lighting transformer which gives more power for less money. Your blown-up PSUs should have plenty of big full wave bridge chips - they usually have markings right on the chip showing diode position, or ac and dc input. You also need to get a good sized 25V cap, but those are also easy to find in almost any computer gear. A 2200uF or bigger would do the job.

Here's a 12V 12A unit for $7.50. You may be able to find it at even lower cost.
http://www.usalight.com/105_watt_ele...1_pr_1427.html
It has overload built in, but you won't overload it in this design. The output will be about 15V, so you will also get more power out of the TEC - you will probably cool your can in about 3 minutes.

Almost any pots you have on hand will work, with a little fooling around. 10K are very common in the parts that come with fans and such these days.

The key is getting the thermistor. It needs to be negative temp coefficient, the opposite of the one you have in your post. The ones that come with most temp controlled fans and other PC temp sensors are 10K negative, perfect for this design.

Or you can probably get free samples at http://www.ametherm.com/ntc_thermist...FQoRswodMSw3hg
Here is a data sheet:
http://www.ametherm.com/Data%20Sheets/NT03%2010391.pdf

[NotSoCoolJ]

Yes, Crunch I looked there first. Digikey has no stock. I could just as easily use a 555 as use anything else. I just like the fact that Uncle Jimbo's circuit is already tested and found to be satisfactory. It is low parts count and can be made small. I have familiarized my self with the data sheet already. Like I said, if I didn't care about the size, radio shack has a kit with damn near the same design that would work. It's just too big. I might still get the RS kit just to play with it. Just have to change that power transistor with a better one and add the thermistor.

[Uncle Jimbo]

have you tried digikey in usa

have looked at AMC6821

MAXIM fan controllers

have you ask for IC samples, some times they will give them free

KaptCrunch - the Maxim chips are not PWM, they are just on-off
The TI chip is nice, but it is a 5V chip so needs some voltage regulation, a little more complex than the MIC520. That's easy to do though, and is probably something I should have put in the original design. Just put a resistor between the + side of C2 and the supply line - 500 Ohm for the MIC502, 1K for the TI chip. Then put a Zener diode across the cap C2 (line points to the + side). Use a 5.1 for the TI chip, a 12V for the MIC520.
the radio shack parts are:
5.1V
http://www.radioshack.com/sm-buy-the...i-2049726.html
12V
http://www.radioshack.com/product/in...entPage=family

[NotSoCoolJ]

New list time
PS. 12v-12amp $7.50 Needs a rectifier and capacitors. http://www.usalight.com/105_watt_ele...1_pr_1427.html
Or. 12v-8.5amp $36 ready to go http://www.mini-box.com/110w-12v-8-5...-Power-Adapter
MIC502. $3.86 http://www.rapidonline.com/Electroni...ement-IC/80431
Newark Loves US! 759 available $2.29 http://www.newark.com/05C4624/semico...ICREL-MIC502BN
NTC thermistor. 10kohm. Maybe I can get some free samples. I thought the one I posted was NTC? It's not 10k though.
N-MOSFET. $1.50 http://www.goldmine-elec-products.co...?number=A10330
TEC. $14.75 http://www.allelectronics.com/make-a...OOLER/-/1.html

I thank you fellas for all your help. I'm a novice at this still. Your comments have cleared a lot up and given me some confidence.

Questions about the circuit.
I don't want to overload the power supply so I really don't want to try it with the 5amp one. I have seen what happens to those laptop supplies when they get overloaded. They let out the magic smoke. I'll have a look and see what I can find. Like I said, the lighting transformer would be fine for me. But for my family I want something UL certified and non-ugly. I will probably use the working Coolmax 600w PS I have here and check to see how much current it is drawing at max output.

Edit: Do you really think I'll need a Vreg? I was under the impression that the laptop supply is regulated already.
Never mind Your talking about a resistor and diode.

I ordered some samples from ametherm. I knew the thermistors would be the easiest to get. Thanks for the link.

[NotSoCoolJ]

I hope you don't mind but I have edited your original schematic with the additions you posted. I also had a question about the connection to VR1 from pin1. Is that a diode?

I'm sorry for all the questions and once again I thank you for all your help.

Edit: I removed the schematic I messed with so there wont be any confusion for others.

[Uncle Jimbo]

I hope you don't mind but I have edited your original schematic with the additions you posted. I also had a question about the connection to VR1 from pin1. Is that a diode?

I'm sorry for all the questions and once again I thank you for all your help.

Uncle Jimbo's TEC control.

The 'arrow' on VR1 was just my sloppy graphic - it's the 'center' or wiper on the pot. Also, the diode D1 is a snubber - when the MOSFET switches off, the toroid inductor will try to maintain the current by generating voltage. Because the other side is tied to a big cap, the voltage on the high side of the MOSFET can spike negative. Without the snubber, that can blow up the MOSFET.

The zener works by breaking down at a specific voltage, so it is 'backwards' too. After a little thought I dropped the resistor R1 to 220 ohms. A higher value will put the MIC520 at a lower operating point than we want at 12V, and even if the supply is at 20V, and a 1 watt zener can manage the .4W.

I should have done a decent schematic rather than just talk through it. So here it is, with all the components labeled, I think. Let me know if this answers your questions.

[NotSoCoolJ]

That pretty much clears it up. I didn't think I was following your directions properly. I see what you mean now about the zener diode. When it reaches break down voltage It passes through and the cap takes over. Thats a nifty little diode regulator. Now about the toroid. Is there a specific inductance figure I should look for? Will any old one off a board or out of a PS work?

Thinking about the toroid I Have found about 6 various types around the house. I would think that by your high current notation I should go with the the one that has the fewest turns of the largest wire. The toroid that I have in mind has 5 turns of large diameter wire and was located near the ATX power input on a dead main board. I don't have a circuit set up for inductance measurement. However, that 4700 micro farad cap is quite large so I would think it's not of great importance.

Tomorrow, I will order the MIC502. I will be getting 10 so I have a few extras. If anyone wants one or two I'll gladly ship them out to you. Right now, I plan on useing 4 and I'd like a few extras. So I will have 4 available. Just rock me off a PM if you want one. Uncle Jimbo I will send you one of these controllers if you want one. Just let me know. Or I can just send you the parts. Too me building these things is half the fun.

Woops.. Before I buy these things I was wondering about the pin out in your diagram. In the data sheet it appears to be opposite.

[NotSoCoolJ]

Well never mind on that. It appears that the mic5021 and MIC5020 are different chips all together.
The actual part number is MIC502BN or MIC502YN for the lead free variety.

Ok so after some more digging I found the right chip. from a US supplier.

I will edit the previous Link location for the US supplier to the correct one.

[Uncle Jimbo]

That pretty much clears it up. I didn't think I was following your directions properly. I see what you mean now about the zener diode. When it reaches break down voltage It passes through and the cap takes over. Thats a nifty little diode regulator. Now about the toroid. Is there a specific inductance figure I should look for? Will any old one off a board or out of a PS work?

Thinking about the toroid I Have found about 6 various types around the house. I would think that by your high current notation I should go with the the one that has the fewest turns of the largest wire. The toroid that I have in mind has 5 turns of large diameter wire and was located near the ATX power input on a dead main board. I don't have a circuit set up for inductance measurement. However, that 4700 micro farad cap is quite large so I would think it's not of great importance.

Tomorrow, I will order the MIC502. I will be getting 10 so I have a few extras. If anyone wants one or two I'll gladly ship them out to you. Right now, I plan on useing 4 and I'd like a few extras. So I will have 4 available. Just rock me off a PM if you want one. Uncle Jimbo I will send you one of these controllers if you want one. Just let me know. Or I can just send you the parts. Too me building these things is half the fun.

Woops.. Before I buy these things I was wondering about the pin out in your diagram. In the data sheet it appears to be opposite.

That is exactly the kind of toroid I had in mind. It doesn't matter much what value it is - we are just using it to smooth out the power.

I would love to have one of the finished units - not only because it would be cool, but also that way if something goes wrong I have the identical unit to debug.

Good luck with it.

[NotSoCoolJ]

I don't think The circuit has any problems. Outside of the resistor network, the whole circuit is pretty much as described in the data sheet notes. Right now the project is on wait until I get the thermistors. I went for the free sample option on that one. It could take a while. Newark didn't have a problem taking my visa Do they ever? The IC's should be here sometime this week. When they get here I'll take a trip down to the local ratshack and get a few more things I need. The tip on my for my station got over heated the last time I used it. I was pulling caps off of some of this new lead free crap I have around here. That stuff takes some real heat to dislodge. I had an old weller pen I was using for that duty but I loaned it out. Haven't seen it since. One of these days I'll get one of those nice hakko units but this digital ratshack station just keeps going. It's like an old Corvair.. Every time you use it you wish it would die but it just keeps going.

When I start building there will be pictures. Be warned. My camera is utter poo.
No matter, my meter is good. I dunked the temp probe in some ice water and I got between 0 and 1.

You know I was thinking. This circuit could be used just as easily on one of those 12v Danfoss compressors. Too bad those things are around 400 dollars.