I've had good luck with plain old 3% peroxide from walgreens. i used a fiberglass dough rising pan and bent up some brass brazing rods(stainless welding wire works too) to bolt to the pieces(use only stainless or brass screws-plated steel will rust and contaminate the peroxide).
I keep the pieces 100% submerged (chip clips to hold the wire to the pan)in the peroxide and leave it in the hot texas sun for a day. I brought back 3 1200 cases/keyboards to white and one of the 1200's was orange when i started-literally orange. I did many A4000 pieces and some A3000 bezels also. They all came out great. be sure not to ever let the parts float up into the air or you will get stain lines(ask me how i know).
I don't use oxyclean or the paste some suggest.
I have even added distilled water to the peroxide to make up for evaporation and it still worked ok,just a bit slower. The secret i think is the light source and warmth-cold liquid doesn't work well.Texas summers can get to 100f+. UV lighting with the proper wavelength is probably best. To date none of the stuff i whitened has returned to yellow. You can boil peroxide down to make it stronger(DO IT OUTSIDE in a well ventilated place!). I have never needed to do this however.
Stronger peroxides can possibly be found at beauty supply.
Your experience is much like mine - a weak solution and naturally brutal sunlight does the trick
.
I've been retrobriting my Amigas (some) and Macs (mostly) for a bit over 10 years, though I didn't know it as that at the start. I'd had OK results *cleaning* gunk off cases with an oxy bleach - Napisan in Australia - and figured warmth made it work better so I'd soak parts in an old fish tank in the sun.
I've found re-yellowing does occur, but it's on the order of years later. I have an A600 I did in 2009, and it's not noticeably re-yellowed. An early Quadra 605 I did before that has re-yellowed to be more yellow than original mac platinum plastics, but less so than most machines of the era. Another sunlit bath in napisan+water works just fine to fix it.
I never kept accurate records of what all my pieces looked like beforehand, but I'm pretty sure if it was excessively yellowed in the first place then it re-yellows more quickly post-briting. It's like the lifetime of UV exposure that caused the breakdown of brominated flame retardents has left plenty of broken-down bromine within the plastic that's still leaching to the surface. I once had a Mac IIci with a lid that had yellowing displaying the shadow of the monitor that once sat on it. It retrobrited back to an even platinum, but eventually re-yellowed a little to show a lighter version of the same shadow again, despite being in dark storage most of the time post-briting.
In my experience, clear sealing is the *worst* thing I've tried. I gave it a go on the same Mac IIci panel above, after reading about protecting from re-yellowing. I only sprayed the rear part as a test to see if the texture would match, and it wasn't remotely close with the clearcoat I used (Mac plastics seem more matte than others, and the coat was a bit glossier). Unfortunately since the yellowing comes from within the plastic, all I got after retrobriting another time years later was clean platinum plastic where I hadn't clearcoated, but the plastic under the coat stayed yellow.
In the end like mechy, I use a weak solution and strong sun. Just a capfull (maybe 1/4 a coke can of granules) of oxy bleach, dissolved in enough warm water to cover the pieces and left in a shallow bath outside for a day. Southern hemisphere summer sun does the trick. I've tried a more traditional mix of retrobrite as a paste following a european friend's instructions. What worked in the netherlands over a day nuked originally-grey plastic almost white in under 20 minutes, before I'd noticed it was happening.
The nuked Mac. The front panel should be the same colour as the CD tray:
And a Mac keyboard, partly done with a weak solution for contrast:
