Burning Breadboards with a Peltier TEC and N-channel MOSFET

So, I'm building a peltier cooler for a wine fridge and naturally want to get it as cool as possible, SO...

525 W Computer PSU (Amazon.com)
TEC1-12710 (Amazon.com)
Arctic Cooler Alpine 11GT r2 (http://amzn.to/1qRZu4L)
Mosfet transistor [IRF520]
Arduino Uno R3

After many a google, I ended up following this setup:
http://bit.ly/1qRZ4vk
and pulling the power from 4-pin molex from the PSU.

All began well, controlling the peltier through PWM, starting at 25. I get about two degrees F difference on each side. I let that run for a few minutes and the temps stay about the same, no problems. Then I kicked it up to 100, and it definitely started to get warm so I killed the power before I melted the thing.

Now that I confirmed it worked, I hooked up two CPU coolers on both ends, pulling power from another 4-pin molex. Tested the fans independently - all good again. Then I restarted - kicked the peltier up to 25 - again, looking good, 1 degree F temperature difference between heatsinks. Then at 50, I got about 3 degrees difference.

Feeling bullish, I went to 100. I read 6 degrees of difference on the heatsinks after about a minute. And then I smelled burning. Suspecting the transistor because of the high amperage, I hit it with an IR temp gauge and read 180 deg F and climbing. I killed the power and as I did, started to see burning edges around the collector/emitter (see pics).

Alright, so my guess I'm an idiot and the PSU is way overpowered at up to 25A through the +12V molex, the peltier is rated up to 10A, and the mosfet is up to 9.7A (optimistic). Pathetically, that's about where my smarts end, and I'd love any advice on how to make it all work.

TLDR; toasted my poor transistor because I'm an idiot.

Pictures attached for you enjoyment.

IRF520
You need a lot more than 5V to get that to turn on.

(It's not your "PSU".)

You should look for a "logic-level" MOSFET or a drive circuit for your IRF devices. Your pick.

Bread board is totally useless for high current.
For low current it is only useless.

Sort on Rds, find the lowest you think you can afford.

Example:

At .004 ohm Rds, you can run 25A thru it and it dissipate 2.5W in the transistor, so some heatsinking will be needed.

For your application, a relay might better. On or off, no PWM needed.

The IRF520 is ancient, pathetic specs compared to modern devices, and
its not a logic-level MOSFET.

If you want high current at low voltage you'd never go picking a 100V
rated device, higher voltage MOSFETs have higher on resistances, all
else being equal. For 12V a 20V or 30V device is appropriate, although
there are some good automotive rated devices around these days, they
tend to be 55V (truck with dead battery and broken regulator).

You always choose device by its Rds(on) rating, ignore the current
rating completely, you won't be near that limit if you aren't liquid-cooling
the device.

Breadboards are good to, say, 100mA, per connection without concern,
probably will handle upto 500mA without immediate damage, but several amps
is way beyond them.

Those Molex connectors aren't rated for 25A through one connector.

What is with the massive pictures everyone posts? I can see the Matrix in them.

Grumpy_Mike:
Bread board is totally useless for high current.
For low current it is only useless.

Sorry, I have to completely disagree breadboards are NOT entirely useless. Cheap breadboards CAN be useless, quality ones like 3M are usable. They have their place, especially when you are learning.

That being said, in this case, they are a FIRE HAZARD. OP, you need to remember, at most they are rated for 1A and that's a high-quality one. Even then, it should be avoided. At THESE current levels you should be looking at using protoboard or something similar. This doesn't even begin to address the choice of MOSFET or the fact that you stupidly pushed a lot of amps through a MOSFET with no heatsink.