Which diodes switch faster then? 1N4148 or BAT41?
I did quite a few consoles for jtag and Zephyr motherboards seem to be most sensitive to switching problems causing the hack to fail followed by E79.
On these motherboards I first remove the solder from the holes with braid (MS' solder is really hard to attach a wire to), and replace it with my own solder after applying some flux (I have this can of flux wich I can spray onto the points (smells really nice)).
Then I solder my 1N4148 diodes with one leg into the jtag holes and the other leg a kynar 28g wire (as short as possible) to the rf-board and the other point.
I dont cut the legs of the diodes since I'm paranoide the heat of my iron will f*** them up causing them to not do the job.
Instead I apply hot glue all over the legs of the diodes and on the rf-board point (since I'm also paranoide the signal for wireless controllers will interfere with the jtag hack, lots of people report more errors when powering the xbox from a wireless controller).
After this the zephyrs will work most of time.
But still one in the 10 times it will fail to boot.
I get the feeling the jtag wires/diodes are really sensitive when it comes to XBR on Zephyr.
So if BAT41 handles the jtag hack better by switching faster I'm happy to give them a try.
I used 4148 like you did, but on a falcon and it never gave me any issues whatsoever... I dont know about the bat41 but the 4148 have 4ms switch time which is pretty darn quick.
I did all my solders, and then if I had any ends that were slightly elongated I removed those ends with snips once they were soldered in and then glued the wires to the board with hot glue, at the cooled solder points of the motherboard... I soldered mine to the underside of the board, easier access for soldering and cleaner look on the top of the board.