I may be a bit late to this...but kudos for getting this to work!
I don't have the spare time to do something like this (I spend my spare time gaming and golfing!)
I might be able to offer you some "thoughts"...
WAY back when (in the late 1990s) when I did a lot of heavy electronics/firmware engineering (now I'm a COO do the COO stuff - hehe) I did a lot of work with TEC cooling. It was primarily focused on cooling signaling and communication lasers, and then we moved into some DoD work (which I can't talk about - hehe).
We added TEC coolers to the laser as it gave us rock-solid temperatures for the laser (laser wavelength will shift with temperature.) We went with the TEC approach mainly due to the fact that controlling temperature profiles was only limited by your cooling capacity and thermal mass. We had to both cool and heat the laser to keep the temperature stable (if you run current in the other direction, the "cool" side of the TEC will heat up). TECs are very inefficient cooling devices. However, they do give you many options that don't exist with other cooling methods.
As we were trying to make a commercial product, running straight DC current through the TEC proved ineffective. What we ended up doing was using a PWM signal and amplifying that through driver electronics. We chose H-Bridge motor controller ICs as they hit the current target we were aiming for. A thermistor measured the temperature, and we ran that into a spare A/D channel on our main DSP board. We ran the control loop in spare cycles in the DSP, and used one of the PWM motor control outputs to drive the H-bridge amplifier. After a coding a simple PI controller, we could dial in our temperature to whatever we wanted and everything happened automatically.
The PWM with control loop approach kept us from burning up the TEC (as we could set the PWM limit to 90% or so), and gave us an efficiency gain as we only used the extra power when we needed it. (You can easily build a PI controller with a few op amps.)
This was successful, and we launched the first commercial TEC cooled signal and communication laser products.
For the higher power DoD applications, we used our "standard" TEC, built a copper cooling plate, added more TEC, new cooling plate...effectively, we "stacked" TEC coolers in an inverted pyramid to cool the higher load. Each pyramid level having the cooling capability to cool the "load" and it's own power consumption, with a bit extra for design margin. Each TEC was setup with it's own H-bridge (with the drive current capability matching the TEC) and its own control loop. The "top" or "biggest" TEC layer was connected to an external cooling device to dissipate heat. The prototype was bulky, but you should have seen the look on people's faces when this thing took the laser from ambient to 10 C in half a second...and then kept it at 10 C during full load and modulation testing!!!
Hope this makes sense!