How do I control the temperature of burnt sugar?

To get the sugar caramelized to that perfect dark brown, I start the sugar on the stove top and finish in the oven.

I start by adding a bit of water and some corn syrup (the fructose makes the brown more intense, but one can skip it) to the sugar and keep it on a medium burner until it reaches the first caramel stage, at 155°C/311°F. As Michael notes, the sugar can go very fast from this stage to burnt. To avoid this, I place the pan in an oven that has been pre-heated to 180°C/356°F, just above the dark caramel stage. Checking every two minutes with an infrared thermometer (more often as it gets closer to 177°C/350°F), it's easy to get the sugar to the right temperature.

This post describes the method in more detail.

There's a bit of misinformation in the answers and comments here I'd like to clear up.

Tobiasopdenbrouw's suggestion in his comment that a thinner pan may work better is a good one. A thin aluminum pan is an excellent conductor of heat which is precisely why it would help in a situation like this. It is far more responsive to taking the pan on and off the heat. A heavy pan has a higher thermal mass and is thus slower to respond to changes in temperature, and it will have more carryover heat.

A thinner pan will require more attention from you as it gets near your target temperature, but it's a lot easier to put the brakes on. As you approach your desired temperature take the pan off the heat frequently to check it's temp As soon as you take it off it will quickly stop heating -- there is only minimal carryover heat from the pan, and no magic carryover heat from an other source.

Additionally, adding more sugar will lower the temperature. There is no "solution" that can crystallize. A solution requires a solute and a solvent. There is only one substance present in molten sugar -- sugar.

Spraying water on it is a terrible idea and I advise against it. At best it simply vaporizes instantly to steam. At worst it spatters molten sugar on you, not fun.

In short, treat it like you would a pot of water that you were trying to maintain at a specific temperature. If the water gets too hot, you'd add more water or take it off the heat. The same applies concept to molten sugar.

Ok, as my first answer started without really reading the question , I will give it another go.

1. As you want to minimize the temperature fluctuations, you will want to use a heavy saucepan.
2. As the entire thing is experimental, you will want to minimize the variables. Do all your testing and development with a fixed quantity of sugar that you do not want to vitiate at least until you figure out how to get consistent results.
3. Start making your caramel as normal.
4. As soon as the sugar starts to melt, move the sauce pan on low / very low heat. Pick a heat setting and do not vary it during the experiment. The Aim is to be able to reproduce the result at first, not to get the it right the first time.
5. Start a timer as soon as you lowered the heat.
6. It is best to sit besides it and check at regular intervals .. or even better continually, and take notes (time-stamp is very important here ). Continue the process until it actually burns and becomes unusable.
7. Now, with the help of the notes and the timer, repeat until the point where it still was as you wanted it.
8. If the time window in that your desired result is to short, try using less heat. If that is not possible, try using more sugar.

Hope that this helps.

I'm guessing here and I don't know if this will work, but could you try adding some more sugar to the nearly cooked sugar to 'thin' it out a bit and so make it less likely to burn so quickly?

The best way will require some effort and repetition.

First you need to figure out what your target temperature is.

Then you need to figure out how long the heat keeps climbing in the syrup after you remove from the heat.

Subtract one from the other, and you should be able to remove from the heat before it's done and allow carry over cooking to do the rest. Along the same lines as roasts climbing another ten degrees after removing from the oven.

How about spraying a fine mist of water from a spray bottle into the pan at the critical moment? (Think wisk in one hand, spray bottle in other hand, stove on lowish heat.) I think this would have two beneficial effects:

1) Provide a cooling effect by introducing a lower temperature component directly and uniformly into the pan.

2) Slightly increase the hydration of sugar solution, thereby lowering its boiling point and temporarily retarding the rising temperature and caramelization.

I presume that with a spritz or two, you would buy yourself a bit of time to evaluate the result and remove the heat if necessary.