i have some capacitors to give 10000joules per second to cover starting, although most commercial units to do this job have a slow start to build up the voltage and frequency to avoid storage banks and get away with a reasonably matched pump/solar panel system. when my friend does the long drive out here with another 8kw solar (glass cracked so for free
)
there will be no worries except the pumps overheating due to the lift distance. they do cut out after 30 mins or so.
so getting back to the isolated supplies for the igbt:
i have these 4x15v and 4x9v 1w psu arriving in a couple of weeks or sooner if im lucky. 2 of them will directly feed each driver hybrid module and 2 driver per igbt half bridge brick. i need to keep short leads and low inductive paths to the gates.
am i missing any more important issues or will that section be ready to box up with a lump of box section ally as a water cooling pipe for the igbts ( i think they are isolated case) and that big lump can be fed from whatever i want to output my 4 optoisolated signals for switching.
how sure are we that igbt bricks:
https://www.littelfuse.com/~/media/electronics/datasheets/power_semiconductors/littelfuse_power_semiconductor_igbt_module_mg06100s_bn4mm_datasheet.pdf.pdf can work at 1mhz? most info i read states 20khz as a practical limit to keep gates fully saturated, and thats a lot of heavy charge/discharge cycles of the gates that the little 1w psus have to supply (yeah i wish i read about the 3w) but there are 2 per gate so they share the load up and down.
80 ns delay time plus 20 ns rise time, plus 290 ns turn off delay, plus 70 ns fall time =460ns per smallest pulse completed but thats dependent on the driver system being able to shove enough drive though 3.3 ohm gate resister into a 6.2nf gate.
besides those figures really are the limit because that speed of switching gives NO saturated working pulse as all the time is going up or down and in the wasteful and heat generating linear zone. i assume i need at least a 0.5 μs gap from my pulse generator between switching one igbt off and the opposite igbt on or they will short out the supply for 460ns.
my spwm pic is giving 20μs minimum pulse width. at mark/space 50% its 320μs per cycle so at about 3khz i am not really pushing into undefined realms. (i say about because im measuring a spwm wave on an analogue scope: cue suggestions to buy a new HP dso) theres some 20mhz breakthrough from the clock but nothing bad.
i have to see to the geese first, then back to crunching some more numbers to see how much power i will need to drive the gate, how much power i will lose, and if i can stick with the pic, or change it or look for another hex file.,