Homebrew spark ECU in 1990
Posted: Mon May 05, 2014 5:27 pm
Dug this old thing out from the archives.
Built it in May 1990. It's a spark-only ECU (no micro) that you could set the advance curve in 1000rpm increments by the pots along the bottom edge of the board. The leds would show which pot was active at any given moment for when you are tweaking it real-time.
The way it works is, starting from the left edge of the board, two cascaded HC4040 counters measured the time between trigger pulses and this value is captured by two HC574 byte wide D flip flops and presented to the MC145151 programmable divider after every trigger pulse. The idea is that the slower the engine goes the longer time between input pulses so the greater the count on the 4040s. The system clock gets gets divided by this HC4040 count and so what comes out of the MC145151 divider is a clock signal that is an exact multiple of the trigger pulse rate. X64 IIRC.
This X64 clock rate is then fed to a single HC4040 counter that drives a HC540 then a R2R D/A. This produced a sawtooth waveform that had a slope time of 1/4 the time between trigger pulses, i.e. the analogue slope would run from max to zero in 30 crank deg on a 6 cyl. This sawtooth fed one side of a comparator, the other side being fed from a voltage from the pots.
Just above the leds are two 4051 1 of 8 transmission gates, one for the leds and one for the pots. When you were exactly on say 3000rpm only 1 led and one pot would be selected. 3500rpm meant the 3000 and the 4000 pots were both selected at 50/50 duty cycle and same with the leds; they would both be lit at half brilliance. I think there was 32 steps between pots. The idea was 1/ that you could see when you were exactly on the pot rpm to set it, and 2/ it would properly interpolate values between the pots for a true join-the-dots effect.
I can't really remember how I generated the variable duty cycle signal to drive the transmission gates, there's a 4008 full adder over there on the right so probably something about adding some value to a counter to get a variable rollover figure... dunno.
The switch down the bottom is to make the spark fire right at the trigger point, and the junk near it is stuff for a points input. Anyway, it worked great on the bench, but in the car (tatty old HQ with 179 (not 173) donk) it was a dismal failure. Noise problems meant it would barely run. D'oh!
Drew the board with Protel Easytrax on my 10MHz 286 with DOS 3.3 and some noname XY pen plotter. Board was done by Printed Electronics in Mt Waverley. Dunno about now, but they were "budget" in those days
Lotta water under the bridge since then. No internet forums then so I didn't know a single person who had any idea about this kind of stuff.
Built it in May 1990. It's a spark-only ECU (no micro) that you could set the advance curve in 1000rpm increments by the pots along the bottom edge of the board. The leds would show which pot was active at any given moment for when you are tweaking it real-time.
The way it works is, starting from the left edge of the board, two cascaded HC4040 counters measured the time between trigger pulses and this value is captured by two HC574 byte wide D flip flops and presented to the MC145151 programmable divider after every trigger pulse. The idea is that the slower the engine goes the longer time between input pulses so the greater the count on the 4040s. The system clock gets gets divided by this HC4040 count and so what comes out of the MC145151 divider is a clock signal that is an exact multiple of the trigger pulse rate. X64 IIRC.
This X64 clock rate is then fed to a single HC4040 counter that drives a HC540 then a R2R D/A. This produced a sawtooth waveform that had a slope time of 1/4 the time between trigger pulses, i.e. the analogue slope would run from max to zero in 30 crank deg on a 6 cyl. This sawtooth fed one side of a comparator, the other side being fed from a voltage from the pots.
Just above the leds are two 4051 1 of 8 transmission gates, one for the leds and one for the pots. When you were exactly on say 3000rpm only 1 led and one pot would be selected. 3500rpm meant the 3000 and the 4000 pots were both selected at 50/50 duty cycle and same with the leds; they would both be lit at half brilliance. I think there was 32 steps between pots. The idea was 1/ that you could see when you were exactly on the pot rpm to set it, and 2/ it would properly interpolate values between the pots for a true join-the-dots effect.
I can't really remember how I generated the variable duty cycle signal to drive the transmission gates, there's a 4008 full adder over there on the right so probably something about adding some value to a counter to get a variable rollover figure... dunno.
The switch down the bottom is to make the spark fire right at the trigger point, and the junk near it is stuff for a points input. Anyway, it worked great on the bench, but in the car (tatty old HQ with 179 (not 173) donk) it was a dismal failure. Noise problems meant it would barely run. D'oh!
Drew the board with Protel Easytrax on my 10MHz 286 with DOS 3.3 and some noname XY pen plotter. Board was done by Printed Electronics in Mt Waverley. Dunno about now, but they were "budget" in those days
Lotta water under the bridge since then. No internet forums then so I didn't know a single person who had any idea about this kind of stuff.
That is very cool. Really shows how much micros have made things easier, a bit of code and almost a single chip solution now days.
