[dakta]

I could never understand his obsession with having the controller before fitting the turbo.

Running a car without a controller is inherently more safe that throwing a controller on 'to tune later', it did, even though i've tried not to make anything of it, put a lot of pressure on in the 'this must go out perfect' sense.

i found load of bugs when i made charles, and i remember literally being sat on his garden wall laptop in hand writing firmware updates in C# and assembly, bit rough and ready as far as development environments go but took a couple of days to go from prototype to finished in his case. Some bugs wanted a design change (as his car had quite a dirty power supply) but i worked around them using watchdog timers and registers in the processor to do a reset if the power went dodgy, which could be acheived and back to controlling the turbo in less then <5 milliseconds etc, not perfect but worked - i improved on this for yours by adding even more industrial voltage regulation components meaning the workarounds aren't necessary. I also designed the core algorithm to be quite safe, in that if the signals were dirty then it would always 'fail' vanes open - the control equation is quite long with a lot of variables, but if you model it then most inputs if they go to extremes will output a zero duty cycle, so if something did go wrong you generally lose boost but not hardware

in all fairness the only problem ive been having with yours is that the rpm signal isnt stable, but i cant get access to one long enough to examine it properly. ive had to go away, develop something, come back test and over and over and over again... and yeah it takes a bloody age