Re-worked layout, wider traces, segregated ground current, opened area

[DonaldBecker]

This is a re-worked board layout, based on an unmodified v0.3 schematic.

The following are the major changes Wider traces for paths carrying significant current, especially supplies (5V, 3.3V), injectors and LS1/LS2 Minimum 8.5 mils traces for other signals Segregated ground plane to avoid return current flow under the analog inputs Opened areas for additional features, especially additional outputs from the TLE8888 Only pull-down resistors and clamps to 5V on the bottom side of the board

The segregated ground plane would be more effective with a dedicated sensor ground pin, but that (like many other changes) would require a schematic modification.

[rusefi]

@jharvey would you be available to review suggested alternative PCB? I believe you can check out https://github.com/DonaldBecker/hw_microRusEfi

[mck1117]

There are semi-official sensor ground pins. The two large ground pins are intended as power ground, and the two small pins are intended as sensor ground.

[mck1117]

A few notes from a first look: 1) plated thru holes above the flat parts of the case are probably not a great idea, since they may short to the case. I'm talking specifically about OUT15/16/17/18 and PD0/PC12/PC11/PC10. 2) A few of the silkscreens near R117 overlap pins 3) R49/R36 silk is under the connector 4) Do we want exposed pads underneath the tc4427s? The tc4427 has no exposed pad on the bottom. 5) Fiducial to the northwest of the TLE8888 misaligned with its silkscreen 6) There are a few schematic mods we may want to make soon around the trigger inputs and power supply - let's discuss these on Slack 6b) Same deal with the programming header - we could save a bit of area if we go down to a 1.27mm header.

Also - holy crap, it's only two layers! That's awesome!

[mck1117]

7) There's a stray trace that goes across the middle of the board on the bottom layer

[DonaldBecker]

The connector's wide pins are on the opposite end of the board from the TLE8888. Two of those wide pins, unfortunately selected as the middle pins, are grounds.

I'm concerned about the ground current because it's by far the largest flow on the board. Without the e-throttle active, there is only a modest current draw from the 12V supply. (And even with it active, the path to the driver chip is short and wide.) The majority of the current flows through the low side drivers on the TLE8888, making its ground return the key path. That current previously went directly through the analog section. Now it's directed around the analog section, largely on the edges of the board, but still crosses through the input connector area of the board.

That "trace" going across the board isn't on a copper layer -- it's marking a physical (case?) feature on the Eco1.User layer. I don't have a sample of the case, so I'm relying on someone else to tell me what it is and if it's interfering in some way.

I'm assuming that the circuit board is held slightly above the case, and plated-through holes won't short out. (Although wires soldered into the holes may be at risk.) If there is actual contact we'll need a sheet of Mylar or even Kapton to avoid vibration from rubbing through the solder mask. If the gap is under 4mm we could use a thermal pad.

My plan is to add more of those through holes. Especially for IGN1-4 (and the 6 similar gate driver outputs) to see if the TC4427 chips could be eliminated.

I know that the standard TC4427 doesn't have a ground pad, but the vias were on the original board layout. Perhaps for thermal reasons, although I don't expect that the chips will be stressed by the 2mA-4mA drive current for an ignitor or smart coil.

[rusefi]

I believe the vias were on the original board layout with other TC4427 package in mind, not the SOIC-8 version

[rusefi]

What is the next step for this PR?

Are we aware of anyone planning to fab and assemble this PCB?