ACE ESC: ACE-NX (inline wiring version) 1.0

Progress on the ACE-NX project has involved mostly ditching the bs_nfet_noninverting pinout and rethinking all the MCU pin assignments.

• Gate, enable and nFault signals for the driver are all moved together and in order with the 6EDL04 driver pinout.
• Throttle input uses ICP like Afro, not INT0 like bs_nfet. This kills some jitter and half the interrupt overhead associated with INT0. Also makes the throttle routing much better.
• Phase voltage sense lines moved together.

However:

 

I also designed another board that was not what I was and am eventually intending (the mountable one), rather it's another ESC style unit. It's a combination of there probably being market demand for inline-wired boards like this, and me needing them for the T19 which I am not going to simply be replacing entirely anytime soon, and the fact that regardless of anything, since I DO want to start selling ESCs, I am either going to sell a modern driver-equipped board OR I am going to sell LC2s. So, it was time for the discrete drive to go and the LC2's beef shortages to be beefed, with the future of bigger motors and higher voltage DC busses in mind.

 

Which they have. The NX Inline is 1mm wider than the LC2 (26mm, same as Afro FS20), has beefier source connections on the lowside than the LC2, is busbar ready (use solid wire) and has significantly more highside device copper area/heatsinking even before adding a busbar.

Other improvements on the LC2 include revised FET footprints, all ISP pads in one place, plus the usual SimonK ready and warning LEDs, because blinkenlights. Also, the sense network may have finally received some due consideration. I tried to shove it to the opposite side of the board from all the noise sources like the DC bus path to the caps, and the SMPS. Something about sense networks is that the phase node, since it is directly connected to the motor, is mega low impedance, that's likely not what is going to receive noise. It's the node after the first resistor in the divider and of course the neutral. Those in my past boards had some long convoluted traces mixed in with the rest of the MCU-related traces and that is not good. With this one the resistors are right next to the associated MCU pin, the other ground resistor is directly underneath on the other side, neutral routing is very direct and the trace lengths are minimal. Will be seen whether these are cleaner starter-uppers than LC2s, it may just turn out to not matter.

Gate drive supply is from an AP3012. This uses a linear logic supply due to area and the lack of more good places away from sensitive stuff like the MCU oscillator and sense network to put a switching noise source without this board either getting bigger or being a pain in the ass. It is minimum 0805, as this is going to be the new general-purpose "not mini" ESC board and it is important that it not be a pain in the ass for either me or other people to make. Bootstrap caps, LDO input cap, boost input and output caps and driver 12V decoupling cap are all 1206. Also, it uses 40x20mil vias (mainly, for power stage reasons) which should eliminate fab issues as well with drill size.

Voltage rating depends on LDO and FETs. This can take a few different 40V devices and there are several 42 and 45V SOT-223 LDOs (the one I normally use is 40V) to correspond.

A bit of an erratum is that I post-schematic'd 3 0805 resistor footprints into this for pulldowns to ground on the lowside gates. Granted, every single FET/IGBT driver datasheet, board design and application note I have ever seen says nothing of the need for them, but to me, it just stands to reason that since the driver datasheets often don't give any indication that the output stage is NOT high-impedance with NO power to the chip, OR that there are any internal pulldowns on those gates, something bad could potentially happen at startup where the DC bus comes up a tad before the power rails to the chip do. So, I figure putting 100K or so (so as not to load the driver with much extra current) on the low sides to source will be a good safety measure against the chance that somehow, some gates are charged up when power is applied and cause a brief shoot-through. Probably just total design overkill.