Update on this:
Still running with the supercharger disconnected, pending finding a replacement one. (the coupler was damaged so replaced the coupler which cured one horrible noise, only to highlight another bearing noise)
The LPG is otherwise finished though, returning 17 mpg on gas in mixed use. Interesting effect of the Valtek 74 high flow tank valve, 8 mm feed pipe, twin liquid filters and twin reducers is that it'll actually cruise the LPG tank completely dry. (it'll run on vapour
The flashlube needed setting to "1.5" on the scale rather than "6" as the instructions suggest as a starting point. Switchover is set to 3 second delay, 200 ms between cylinders, and 30C reducer temperature. Doesn't stay on petrol long at all (perhaps 1/4 mile) and unless you leave it overnight the heat-soak means the second start of the day is always over 30C so 3 secs then gas.
LPG mapping was something of an anti-climax compared with the rest of the install and the fun we had with the Rover. Here's how you map a Stag sytem:
Your basic control is the 2D multiplier. What multiple of the petrol injector time should the gas injector time be?
(the equivalent on a AEB system would be editing all the rpms at the same time, using a key on the keyboard)
There's an auto-calibration routine to get you one point at idle and guestimate the rest. You can then move points up/down and left/right indivudally or collectively, and add as many as you please.
I switched one bank to petrol and the other to gas, then drove gently and had kid brother tweak the multiplier until the petrol injection times on each bank were the same. Idle is easy. Then pop it in D and turn on all the electrical loads for another datapoint. Hold the car on the brake and add a little throttle and you get a datapoint at about 3.5 ms. We added three more (5, 8, 11 ms) and found a hill to cruise up to set the 5 ms and 8 ms multipliers.
With the 5 datapoints set the car drove pretty nicely, so you then take it for a drive on petrol to collect the petrol map: (petrol injector time when running on petrol)
Then take it for a drive on gas to collect the gas map: (petrol injector time when running on gas)
You need a dual carriageway with a steep hill to collect the upper reaches of this, as the car sets off in 2nd and 2nd will take you past 70 mph. You need the hill to slow the acceleration down enough for the gas ECU to decide that things are in "steady state" rather than "transient" otherwise it won't record the injection times. With the supercharger connected this vehicle would need a dyno for proper setup.
Now flick back to the 2D multiplier and it overlays the maps you collected:
It needs a little more at the top end - bringing the 11 ms datapoint up to 1.2 resolved this perfectly.
You can also look at this on a 3D "deflection" map:
This shows the % difference between injector times on gas and petrol. It needed less gas on overun. Bringing the 0ms and 2 ms datapoints down to 1.0 multiplier resolved this. The high-load low-rpm datapoints are unreliable (an auto never really operates there, so the ECU is collecting transient injection times not steady state ones) so ignore those.
If you need rpm based correction, you can switch to a 3D multiplier:
(this is a graphical equivalent of the table in an AEB system) There are also temperature based corrections and whatnot available.
The Jaguar and HANA injectors needed none of this - everything is nice and linear/well behaved to so that is it really. HANA Green. 1.0 multiplier until 2 ms. 1.2 multiplier from 3.5 ms onwards. A little slope in between.
You'll see a graph in the bottom of the screen. This is an "oscilloscope" of all the data on the right hand side of the screen, PLUS all the OBD data, and you can record it all whilst you're driving for later recall. Drive, then interrogate everything, incl the fuel trims and ONE wideband oxygen sensor, later. This was useful when things were going wrong earlier, but once the nozzles were in the right place I haven't looked at the OBD data other than post-mapping curiosity. (if injection times match and lambda does what it should with the check engine light off, things are dandy) I haven't turned on the auto-adaptation either - there appears to be no need for it with a good base map.
The one thing OBD *is* useful for is clearing the occasional P0430 (bank 2 cat inefficiency) code that can show after a hoon it around then snap the throttle shut, but that's a aged catalyst/oxygen sensor combo that triggers the same fault on petrol. It works well enough for MOT purposes anyhow:
(petrol, fast and natural idle)
(gas, natural idle only)