A bit further on now - the engine is in the plane I want to power and I have gone through the fuel map (throttle x rpm => injector pulsewidth) and got it running pretty nice. It's still a little hard to start - I expect that I'll need to add in a cold enrichment mode, for now I've been using a fuel trim control and starting at about 20% throttle (~3-3.5kRPM) but at the field that's a little more revs than I'd like while starting.
Running a 13x6 ASP prop on 10:1 fuel
il mix I'm idling comfortably at 2000 RPM but it will go down to 1800 without threatening to stall. Full throttle is about 8900-9000RPM static a little lower than with the MA Schimitar prop I was running. From the ASP tables that's around 1HP.
The engine is fitted about 40º clockwise from inverted (to get the muffler out the bottom of the plane) and I was curious about the oil behaviour, so popped the rocker cover off after about an hour of running (takes some time to work through the map calibration). On the positive side, there was heaps of oil, meaning that 10:1 is more than enough given no roller bearings on the crank. The inside of the crancase is damp with oil but no standing oil. On the negative side, there was heaps of oil (in the rocker cover), so the bleed into the intake valve bowl is not capturing much if any of the oil from the top end with the engine inverted. For now I'll drop the rocker cover after a day's flying, or be sure to fly inverted regularly (Thinking about the implications of that - I may not need a smoke system on this plane...) Long term I'll likely need a bigger vent in the rocker cover to get rid of the oil overboard rather than through the intake. It's a little sooty (not surprising since I was running dead rich and tuning back from there) but certainly nothing to be fussed about. The oil I'm using is 2-stroke synthetic Kart oil, I have some cool power blue oil on the way which should be a whole lot cleaner looking.
Initially on the plane I couldn't get the pump to stay primed - it's getting warm here (was over 30ºC today) and the petrol mix is evaporating in the fuel lines, along with the bubbles that are always going to be present at least initially. I added a tee after the pump and used a needle valve from an old OS 10 FP to bleed a little fuel back to the tank, mainly to let air bubbles through but also to keep the fuel in the lines a little cooler. The fuel bleed has been (so far) a complete success. If any air bubbles come through, they go up to the needle valve and back to the tank and there's no vapour lock issues at the injector pressure. An in-tank fuel pump would be the proper solution (that's how it's done normally) but I don't really want a brushed 180 motor sitting in my fuel tank sparking away (with the fuel slowing the motor and dissolving the winding insulation), so I'll go with the bleed for now.
I had to do a bit of a hardware hack, the receiver I was using during the initial tests was throwing SBUS out with 5V signal levels and working fine, I replaced it with a Futaba receiver which outputs 3.3V SBUS and my optoisolation was running a bit under spec, noise errors and no servo data was the result.. A resistor change solved that one.
The remaining issues before flying are purely mechanical. The first is that the prop driver has excessive runout - with the prop fitted as normally would be for hand starting (~1:30/7:30 at start of compression), one blade was about 1.5mm closer to the plane mid-span than the other. I've ordered a new prop driver to see if that's the issue (it's a little bruised around the bearing surface) but otherwise I'm going to have to pull the motor down and make sure I got the bearing properly seated on the crank (my first bearing swap so not going to swear I got it right). The second issue is that I'm using a GRP engine mount and it's pretty flexible. Wasn't an issue with the 2-Stroke but with the 4 stroke the engine is bouncing around like the kids at a trampoline park. Given that the fuel is mapped based on the throttle servo (and therefore throttle barrel) position, any engine movement is not going to help the cause. Plus it looks sketchy. I have an aluminium mount that I'll put it on over the coming days.
The ECU controls the pump and injector obviosly, but also has a switch for enabling the CDI, has 4 servo outputs (throttle plus 3 others) and take an SBUS signal from the receiver. I'm using a few channels at the moment for trimming the various parameters while tuning (+/- 30% on fuelling and for the fuel pump voltage), not sure yet if I'll need the others.
Using an injector calibration mode on the ECU the injector was run through a couple of pulsewidths at a couple of battery voltages (3ms and 20ms, 8 and 7v) to get an idea of the flow rate and opening delay. The opening delay was around 1.2 ms at 8v and 1.7ms at 7v (running well below design voltage in this application) and the flow rate around 10cc/minute at the pump pressure used for the testing (not measured but based on a pump voltage input of 3v). Full throttle pulsewidths are in the order of 5ms (so 3.8ms fuel + opentime), so if anything the injector could even flow less than this and be fine on this engine. OTOH, 5ms allows engine revs upwards of 15kRPM with 80% duty cycle so this could work on a 2-Stroke engine as well. The injection pulses are every 4 strokes with the cam timing inferred from the difference between the speeds of the consecutive magnet passes.
I'm a few posts short of being able to post images it seems - this is one more towards the target...