rusefi.com

Schematics and pinout are at https://github.com/rusefi/rusefi/wiki/Hellen-128-Mercedes

At the moment we are developing two Hellen boards with on-board igniters, we have to spell out our requirements for coil output traces width and gaps.

JLCPCB with assembly limits us to 1oz.
Do we assume 400 spikes? Do we have any magic lacer/solution/else we can apply on the traces to reduce the gaps requirment?

I strongly suggest making the IGBTs replaceable in some way, either solder through with heat sinks or spaced out in a way that allows a human to replace them as the IGBTs are the number 1 point of failure on the ignition system if things are not well set up.

Primary points of failure on the megasquirts were people accidently setting the IGBTs inverted and cooking them.

andreika wrote: ↑

Tue May 25, 2021 8:34 pm

The primary ignition coil resistance may vary greatly: from 0.2 to 3 ohms. Usually it's about 2 ohms, i.e. the average current is 7A. The dwell duration in running mode is usually 2-4 ms, and for cranking it may be increased up to 10 ms or more.

This is not the complete story - coil current is determined mostly by inductance, and very little to do with the DC resistance of the coil.

See this video demo of multispark, with a high bandwidth (25MHz) current clamp around the supply wire for all 8 coils:

Note how it's a relatively constant slope (current change per time is relatively constant), which means the operating regime is nowhere near hitting the maximum DC current.

A safe reasonable assumption is probably a peak of 8 amps, which means an average of 4 amps over the dwell time (starts at 0, linear ramp to 8). Upper end of possible required dwell is maybe 50% while running wasted spark, which means an actual average of only 2ms. A 2-5ms pulse to charge the coil is not enough to cause overload in any single event (for any reasonable trace width we're considering), so the only concern is a thermal limit of the trace over time (which means design for the steady state DC case).

Plugging 3 amps (1.5 safety factor) in to a PCB calculator with 10C rise gives a required trace of 1.36mm. A Bosch Motronic 3.3 (BMW 88 pin) has ignition traces of about 1.75mm, so this is the right ballpark.

We can then round up a little bit, and go for 2mm as our standard, as that provides some margin for manufacturing variability (China...), people doing dumb things to their cars, reduction of heat, etc.

IPC-2221A says that we should keep clearance for 0.8mm for external layers and 0.25mm for internal layers. Also, we can use a trace of 1mm for average current up to 4A. Temperature rise 45° is not critical overheating. For internal layers, we should use 4x width of trace because JLC internal layer only 1/2oz thickness. Of course, wider tracks should be used if possible.

Remember the heating is worse with "peaky" loads than the average current would suggest, so it's somewhat worse than 2-4A average or whatever. We have plenty of space on these boards, so 2mm trace and 1mm clearance is totally fine.

JLC has promised to update the page which has confused me about 2oz for assembly. They do support 2oz for 2 layer 1.6mm boards.

Some additional info off a OEM ecu for onboard IGBT. this info was pulled off a Ford Copperhead ECU. 2011-14 F150/Mustang V6 and V8s. IGBT used is https://media.digikey.com/pdf/Data%20Sheets/ON%20Semiconductor%20PDFs/NGD8201N_AN.pdf trace width is 1.3mm and length was 61.7mm. 1amp 13.1mv 2amp 26.3mv 3amp 39.5mv with these measurements its 2oZ copper.

Sent to fab!

It runs! https://rusefi.com/forum/viewtopic.php?f=2&t=2095

Now with starter control, will have those hard-wired in next revision.