Thread: using narrowband sensors for lpg on disco

Bee utey – oh well, I was hoping that wasn’t the case. Seems like the lpg software for the injection type are all off the same page ie can’t do anything smarter than be one dimensionally manipulated in relation to only petrol injector opening time. Would be a great tool if once a car was at running temperature the lpg fuelling could be driven by an oxy sensor. I guess this is what the mixer type setups which you install are able to do. Interestingly, when in closed loop I can still adjust the lpg fuelling. It won’t go richer beyond the lambda which the sensors are measuring no matter how much I increase it over petrol opening time it won’t increase lpg injector opening time but it will allow me to make it leaner and decrease injector opening time.

philipA – I probably have not given an accurate description, the leds are fluctuating on lpg and petrol. I have read they should have a cycle time of around 300 mS and there is not doubt they are constantly moving up and down the led display and reading 3 times or more a second unless I lean the lpg mixture off and they might only just creep onto the start of the display. I can’t recall whether I knew at the time the leds were for zirconia only however given the output of the titania was 0-1 volt I thought I would give them a try. In practice they perform exactly as expected on petrol and lpg. Whilst the titania sensor worked off a different principle to the zirconia I read in a few spots from the manufacturers that their signal output is Vres ie voltage after it has been subjected to the resistance. So when I wired in the leds I connected them directly onto the injector plug on the 14cux so as not to impact the signal coming through from the sensor. I wasn’t 100% sure whether the 14cux was reading the volts or the resistance but once I started thinking about it it made sense it was reading the volts otherwise what is it reading the resistance of? You would have the resistance of the cable to the sensor, the resistance of the sensor itself and then further resistance of the cabling from the sensor back to its earth point(s).


I also found this in the rave cd and it is quite enlightening in how the land rover ECM manages fuelling. I don’t know whether it is for the 14cux or not but given it is written by land rover you would expect it to be how it is configured or close enough for the closed loop maps.

European On Board Diagnostics (E-OBD)
Air : fuel ratio
The theoretical ideal air:fuel ratio to ensure complete combustion and minimise emissions in a spark-ignition engine is 14.7:1 and is referred to as the stoichiometric ratio. The excess air factor is denoted by the Lambda symbol l, and is used to indicate how far the air:fuel mixture ratio deviates from the theoretical optimum during any particular operating condition.
• When l = 1, the air to fuel ratio corresponds to the theoretical optimum of 14.7:1 and is the desired condition for minimising emissions.
• When l > 1, (i.e. l = 1.05 to l = 1.3) there is excess air available (lean mixture) and lower fuel consumption can be attained at the cost of reduced performance. For mixtures above l = 1.3, the mixture ceases to be ignitable.
• When l < 1, (i.e. l = 0.85 to l = 0.95) there is an air deficiency (rich mixture) and maximum output is available, but fuel economy is impaired.

The engine management system used with V8 engines operates in a narrower control range about the stoichiometric ideal between l = 0.97 to 1.03 using closed-loop control techniques. When the engine is warmed up and operating under normal conditions, it is essential to maintain l close to the ideal (l = 1) to ensure the effective treatment of exhaust gases by the three-way catalytic converters installed in the downpipes from each exhaust manifold.

OK perhaps its time for brute force. Make a 0-1V variable supply, controlled by a knob on the dash. Toggle switch this to your 14cux inputs, with optional sensor operation on the other side of the toggle. Vary the voltage and watch your LED display change. Simple to do with a zener diode and a couple of parts.

Yes, l like where you’re coming from. I have a similar setup I’m already testing with a trimpot and diode for each bank. I have done a quick test so far on one bank but I need to get out of Hobart onto a flat quiet road to benchmark. what I saw on my test already was the sensor voltage topping out around 1 volt even when I had added approx 300mV so there was an immediate result there though I need to do this many more times and have known repeatable results to find what it is doing to the injector opening times. However, the leds showed a kind of unexpected consequence in that they never dropped back down to near 100mV, they would stay at whatever voltage I was inducing. All good fun I guess but it was a pita installing 3 wires coming off each pin 23 and 24 of the 14cux, starting to get very fiddly. There would be a nice easy way of getting those injector connectors out but I don’t seem to know it! Appreciate your thinking through this and hopefully I get some time over the weekend to test. It’s a pity your injector software only does what mine does, I am trying to get some things finished off here in Hobart and was/am to meet some friends up at the gammon ranges (north of the flinders) to go bushwalking during june and thought instead of taking my tdi over I’d take the v8 and go via your workshop.

Sadly software design isn't one of my skills, I'd love to see how well one could make lpg injection work on its own. As it is the software I have only works with one brand of injection, even though the serial adaptor fits another brand! I went to a seminar by Impco BRC on their Sequent injection systems, different again. The engineer presenting the seminar talked about all the ways that lpg injectors differ from petrol ones, including such things as the wetting of intake ports on rapid acceleration on petrol requires completely different fuelling on lpg to get the performance required.

considering the lpg software isnt that great for total control, i will eventually be using megasquirt to control my lpg injectors, i have a slight problem though, i need to know the 'opening time' for them, that is, the amount of time at 12v that the injectors take from when triggered to actually squirting fuel, this is measured in milliseconds (ms), my petrol injectors are somewhere in the range of 1.0-1.3ms, i have read some lpg injectors are much slower and somewhere in the range of 3-4ms, does anyone have any solid information?

Originally Posted by jazzaD1

considering the lpg software isnt that great for total control, i will eventually be using megasquirt to control my lpg injectors, i have a slight problem though, i need to know the 'opening time' for them, that is, the amount of time at 12v that the injectors take from when triggered to actually squirting fuel, this is measured in milliseconds (ms), my petrol injectors are somewhere in the range of 1.0-1.3ms, i have read some lpg injectors are much slower and somewhere in the range of 3-4ms, does anyone have any solid information?

Can't remember offhand, will be able to tell you actual times when I plug in to a D2 next Saturday. From memory, petrol is typically between 2.5ms and 14ms from hot idle and full load. I spend more time watching petrol times because that is how you check the mapping, change fuels and check what the petrol ecu does while on gas.

Originally Posted by bee utey

Can't remember offhand, will be able to tell you actual times when I plug in to a D2 next Saturday. From memory, petrol is typically between 2.5ms and 14ms from hot idle and full load. I spend more time watching petrol times because that is how you check the mapping, change fuels and check what the petrol ecu does while on gas.

sorry, i know what you are saying, but i'm not after the pulse width, more the lag time before actual fuel is injected, its hard to describe