I thought I might put my two bits in about temperature stability. It is not correct to assume that a bigger amount of water will give you better stability. What you will see is a slower reaction time to putting in cold food. The important feature of any of these is the software running the controller and it's proper adjustment followed by a design that has no extra material between the heating element and the water and a reasonably sized heating element. If you have a slow response when the water cools, you might expect a slow recovery when it heats back up and thus more time out of your desired temperature range. If you have your own PID controller, you can use auto tune to get pretty good parameters but even that is not going to be your very best settings all the time. It is normal for an industrial process to tweek the settings for better performance. If you build your own setup, you want to select things that do not hold a lot of heat themselves. For instance, when I tried placing the big chafing dish I use on a large hot plate, the heat up time of the plate was so long (due to the mass of the metal slab built into the unit to make the temperature stable, like the big water bath idea) that by the time the water warmed up, the hot plate was so hot it overshot by 4 or 5 degrees. Same overshoot on cooling. Not an acceptable outcome. Controllers can be designed to deal with this with precision but they will not be the PID design that you can pick up easily. The software is not written to cope with the extra thermal effects. I use a PID controller from Automation Direct at about $100 (I already had it) and a small pump. My heat source is a 150 watt lab heating pad that I put under a large stainless chafing dish and I control current with an SSR. I can hold temperatures within about .2 or .3 F with that using the autotune feature to set the PID parameters. I have not tried to tweek the PID settings to improve things as I am happy with what I got. I have a second controller and thermocouple coming out of eBay for $32 that I will use for a second setup on a turkey roaster. I expect it to do the job just the same as the $100 unit. I used the $100 controller on a turkey roaster as a test and got results that were less effective control than my chafing dish setup after using the same auto tune process. The temperature fluctuated by about 2F up and down. The difference was probably the result of the size of the heating element and some thermal inertia inside the cooker that I could not see and probably the ease of heat loss of the unit. I expect to be able to tune out the wide swings but it is more fiddling. I hope that helps put things in perspective. I am sure the Sous Vide Magic equipment is just fine and I assume they have good temperature control since they are designed for this task. Immersion heaters are probably just great but kind of spendy. If you DIY your setup, you need to put some effort into understanding the the proportional, integral and derivative settings on the controller but auto tune might do all you want. You also need to verify that the temperature read on your electronics matches the real temperature of the water. I use a good glass lab thermometer for that but you could just check 32F with an ice/water bath and 212F with boiling water (at sea level). If those look good, you are probably pretty well off.
