I'm glad it was helpful. I don't know of any good references. Rincon-Mora has a book on the subject, but he seemed (to me anyway) to get hung up on circuit cleverness instead of the basics. Lack of cleverness in the circuit isn't the problem with bandgaps... the problems usually come from non-ideal effects that were not modeled correctly (or that are impossible to model). It is really hard to capture all the things that need to go into a robust design... mostly, it is just a lot of experience and attention to details. My advice is, if it doesn't work perfectly in simulation, go find out why. Take the time to understand the important parameters in the model (mostly IS, EG, NF,...) and try to build circuits that aren't sensitive to the non-ideal parameters (Rb, Re, IKF, etc...).
It doesn't capture everything, but Barrie Gilbert wrote a decent chapter in Huijsing's book:
B. Gilbert, “Monolithic voltage and current references: Theme and variations,” Analog Circuit Design, J. H. Huijsing, R. J. van de Plassche, and W. M. C. Sansen, Ed. Dordrecht, The Netherlands: Kluwer, 1996, pp. 269-352.
More than that, and I'd have send you a bill
rg