Some general comments, and some tips to encourage a self-solution. (IMHO: the best questions are those that didn't need to be asked -- because one is equipped to figure them out themselves!)
- Why are Q8 and Q3 operated at hFE(sat) = 1? This greatly stretches out storage time, and greatly increases the current consumption (an important parameter for a bootstrap gate drive).
- Note that bootstrap only works if M4 is fired first, and regularly thereafter (or there's enough load current to forward-bias its body diode). The choice of isolated gate signal is a bit peculiar, without also isolating the DC supply; consider changing to a floating source instead. (Or, if bootstrap is adequate given inverter strays, don't bother with cooking up your own driver: just use a IR2110 style IC, perhaps boosted with BJTs if needed.)
- Is M1 a typo? You're driving a huge power transistor with itself? (M2 likewise, almost certainly doesn't have to be a 10A device. Certainly not if it's driving through 100Ω.)
- I don't know what voltage A1 is set for, but beware 6N137 specifies 5mA min. IF for operation, so it won't turn on (or reliably so) if this is a 5V gate. Simulation logic may not model supply current (correctly, or at all!), and use an assumed voltage (usually a global
VCC or VDD net).
- U8
VE is floating. This should be acceptable per the datasheet, but leaving a logic input floating does create a noise hazard in practice, and it may not be modeled correctly in SPICE. Try tying it with VCC.
- Try things, in general -- when confronted with a potentially erroneous SPICE model, it is worthwhile to take a step back from the immediate problem (the inverter), and build a test fixture featuring just the model in question, and some sources and sinks to make it go. Measure the output voltage and current range, response time, everything on the datasheet you feel like checking. (Bonus: this reinforces researching what datasheet parameters are and how they're measured, and reproducing those tests virtually, or with the real thing for that matter -- sometimes you get parts that aren't quite right, and setting up a real test can be informative.)
From this, you might now imagine some things to look into; like the VO pin, if it has the specified current and voltage range; that VCC is adequate (or the bootstrap supply in general); that R13 is suitable; or generally that the model is behaving at all.
Key insight: SPICE models don't always work right (in any given simulator). For that matter, they aren't always right in general! Always have a feel for how your circuit should be working, given everything as specified, and spot-check that things are behaving accordingly. If you find something suspicious, consider taking a deeper look. Never assume that things are as they should be, always leave open the possibility that something is just wrong in some respect. Particularly for simulation models, only certain aspects may be modeled -- they are only models after all, not the real thing!
This advice, by the way, generalizes to all of life -- never take something for granted, at face value; don't take things too seriously, too literally; always keep a critical eye out, and cross-check that things make sense, and when they don't, look deeper into it, research what's going on.
Best of luck!
