Homebrew EFI -- Ignition!

The XLVs ignition system is dodgy at best, so it seems like a good idea to replace it.

The question being: how to make an ECU controlled CDI system?

I'm getting a lot of inspiration from this article: A high-energy capacitor discharge ignition system [alt] by John Clarke (Silicon Chip September 1997). There's now a Dick Smith kit for this, I think! K3307.

Here's a page on Transalp CDIs, including a schematic. The Transalp ones include their own high-voltage generators. And some notes on a PIC-based programmable CDI

High Voltage Supply

The first step is to generate ~300VDC which is used to charge the capacitors in the CDI units. The SC article uses a regulated stepup power supply from 9-15VDC, running at 20kHz.

This could also be used to power the XLV's stock CDIs, and this is probably where I'll start.

Microcontroller Interface

Some parts of the circuit become unnecessary due to the microcontroller's interface ... and the CDI section of the circuit will have to be doubled up for the XLVs two sets of coils. Maybe even quadrupled up to give the coils one driver each. *sigh*.


I've put it together, and already smoked out the veroboard! Oh dear ...

... I was worried that it was a fundamental design flaw, but it turns out there are limits on how much current the poor veroboard tracks will stand, and running at 100% duty cycle exceeds those limits! Why 100% duty cycle? Well, it turns out that the XLV CDIs use an SCR to short the capacitor, and these don't 'turn off' until no current flows ... whereas this circuit never stops! It also turns out that the 'kill switch' stops the engine by shorting the output of the CDIs to ground. Suboptimal.

So, I'm putting together a new version of the circuit to get around this, and rerouting the 'kill switch' lead to turn the inverter off rather than shorting it out. Additionally, the 1uF output capacitor will get a lot smaller and the regulator circuit will change to compensate for that, and a 555 circuit will be used to switch the inverter on and off with a duty cycle of ~95% at ~1kHz, imitating the voltage characteristics of the coil generator. An extra diode or two need to be added as well ...

Well, that didn't work at all. I could make it run by dropping the switching frequency to a couple of hundred hertz, but the voltage output was disappointing. I also tried gating the voltage on and off with a pair of transistors, but that did the same thing ... locked up and strangled itself, and another MTP3055V hit the bin.

As a really dumb test, I tried wiring a mains transformer backwards to feed the CDIs, but with a couple of rectifiers to knock out the negative-going sides ... which idled very nicely, but wouldn't rev above 3000rpm. I've got this horrible feeling that some part of the advance circuit depends on the generator input frequency: either that or the transformer just isn't producing enough something.

Sod it!

I've given up for the moment, and bought a new set of stator coils ... at least that way I can put the 'scope on them for volts and amps and see what they're really doing ... and my work on the AVR CDI controller continues ... since it'll be a bit smarter than the stock CDI units, it'll put less load on the coils, and maybe they won't burn out again. Here's hoping.
The very latest high-tech equipment is essential!

homepage / motorcycles / Homebrew EFI