Car had been running badly since I got it, but better once on FreeEMS and with a better decoder angle and so forth, but still not 100%. Today I got stranded twice and nearly stranded a third time. FreeEMS checked out fine (proven via ignition out LED flashing in harmony, no laptop handy), so it had to be fueling related as the tacho also worked, and that's coil driven.

The first time, I just happened to have a philips and the fuel hose just happened to have a hose clamp, so I pulled it off, pumped fuel into the grass, and it ran fine.

The second time the same events unfolded.

The float in the bowl must be sticky or half full of fuel, I guess. Need to put some time into that tonight and see what I can see.

I have a few tweaks to make to the ignition/ECU setup to make it nicer, too. All in good time. Reliability first.

Lots of pieces of carb all over the engine bay. Off the road yet again. Not FreeEMS' fault, though. Thank goodness I fixed the LED for ignition out, that was worth its weight in gold today. I either lost a bit of carby, or it dissolved, or Presto lost it, or something, as the float setup seems to be incomplete or incapable of restricting flow at all. Something ain't right. Grrrr.

ITBs + engine bay mounted fuel pump = near future? Fucking carbs suck.

What would it take?

Find ITBs and injectors for them (have these)
Obtain a low-power inline fuel pump (don't have this)
Obtain a stump inlet manifold to use (so as to keep the carby one available for returning to stock later) (don't have this)
4 pieces of hose to join stumps to ITBs (don't have this)
8 hose clamps to join stumps to ITBs (don't have this)
5 FETs to control the injectors and pump relay (have these)
1 day to do the required work (don't have this)
3 more days to be realistic (definitely don't have this)

On the other hand, if the carby is missing a necessary part, I could screw around with it indefinitely and not make progress. Ugh.

Hmmm.

For the record:

• Fix decoder angle to match reality established in above screen shots.
• Verify that signal from opto to CPU is robust (touching scope probe to this CPU pin caused loss of sync and should not)
• Change base timing to be 170 (-10) and setup dizzy for 10BTDC cranking.
• Lower hardware based dwell threshold and/or add hysteresis to its use (Use up to 600, start using again below 400, for example)
• Lower full load timing from 2500 to 4k where it currently rattles on 91RON

And throw carby out the window.

Took apart carby, nearly lost parts from it (some into engine, others into wrong places), put it all back together, warmed it up, revved it up, poured half a litre of water down its throat while revving, took it for a "spirited" drive, returned to base satisfied that next time it fails on me, I'll have sufficient tools handy to fix the issue. Drove nicely, though, so that's something. Unsure why, so I've almost certainly NOT fixed anything.

Manifold is a yucky shape for ITBs, though with a bit of TIG love from someone friendly, it could happen.

Currently running 83/97 degree split for the decoder, which is a LOT closer than the 69/111, but from the data above (udpated), we can see it's still not right, here are the numbers:

http://stuff.fredcooke.com/hotel.fine.s ... MAG005.png
CPU high 20ms exactly
CPU low 19.3333ms roughly
762.711864407 RPM
180 * (20 / 39.33333333) = 91.52542 degrees high
180 * (19.33333333 / 39.33333333) = 88.47457 degrees low

http://stuff.fredcooke.com/hotel.fine.s ... MAG008.png
1757.8125 RPM
CPU high 8.73333ms
CPU low 8.33333ms
180 * (8.73333333 / 17.06666666) = 92.109 degrees high
180 * (8.33333333 / 17.06666666) = 87.891 degrees low

http://stuff.fredcooke.com/hotel.fine.s ... MAG009.png
1757.8125 RPM
CPU high 8.6666666
CPU low 8.4
180 * (8.6* / 17.06666666) = 91.4062 degrees high
180 * (8.4 / 17.06666666) = 88.5938 degrees low

http://stuff.fredcooke.com/hotel.fine.s ... MAG010.png
1717.55790746 RPM
CPU high 8.33333ms
CPU low 8.73333ms
180 * (8.33333333 / 17.06666666) = 87.891 degrees high
180 * (8.73333333 / 17.06666666) = 92.109 degrees low

http://stuff.fredcooke.com/hotel.fine.s ... MAG011.png
3351.95530726 RPM
CPU high 4.25ms
CPU low 4.7ms
180 * (4.25 / 8.95) = 85.4749 degrees high
180 * (4.7 / 8.95) = 94.5251 degrees low

http://stuff.fredcooke.com/hotel.fine.s ... MAG012.png
3488.37209302 RPM
CPU high 4.1ms
CPU low 4.5ms
180 * (4.1 / 8.6) = 85.8140 degrees high
180 * (4.5 / 8.6) = 94.1860 degrees low

This is the natural variation of the engine with its sensor config. Slop in the dizzy drive, asymmetrical lag in the hall sensor setup, slop in the shaft causing clearance variation, etc.

Still, though, in no currently measured case, is 83 degrees correct. I just setup the car with 10BTDC cranking, 170 offset, and used hard timing to 1500 to set the timing of the engine. Thus any error in timing is purely down to the above variation and mismatch with the decoder angle.

To match the 3500 RPM section, we're looking at 86 degrees decoder angle, but to match idle we're looking at 91, and to match 1700 odd, we're looking at about 90. What I'll do is go down and warm up the car, do a sweep up to rev limiter slowly, record the log, update the firmware to 86, repeat, update the firmware to 90, repeat, and review the time tolerance figures, then choose one.

Fred.

83 bad
86 bad but better
90 much better but not good, will live with this

Drives OK now, logs OK now, comfy commuter project complete. Anything further is lowering the timing in the high load regions. Any JAFAs keen to volunteer to drive or tune for me?

Proof!

Ranges are 800 to 5800 for RPM and 500 to 1500 for tolerance.



http://stuff.fredcooke.com/Hotel.decode ... 3.fail.png :



http://stuff.fredcooke.com/Hotel.decode ... 6.fail.png :



http://stuff.fredcooke.com/Hotel.decode ... uccess.png :

EFI manifold ordered! If it's not very good, I'll butcher it and add ITBs. What can I do for a 50 - 100lph fuel pump? Any ideas?

get a walbro or bosch, even a bosch clone or something cheap enough IMO

probably go in-line external though, as I doubt there is an intank one there already due to the shitty card configuration, as it will make installation relatively easy

Bosch 044 or walbro 255 are three times too big. You don't want to recirculate that much fuel unnecessarily heating all the fuel in the tank (bad). Even if you plan to make a "returnless" system where the regulator is near or in the tank, I think a pump that size itself would do a lot of heating.

I recently acquired an Audi A6 in tank unit from a pick-a-part yard here in the States for $10. It is a nice small bosch unit. It looks to have a commonly available banjo fitting on the high pressure side. Something like that might work for your situation. You can have this one if you pay for shipping.

Can I take this opportunity to make a feature request on the topic of pumps that are too big? PID calculated PWM output for the fuel pump, perhaps with TPS value as the input or a fuel pressure sensor.