AC Stag 300 ISA2 on 3.5l Pajero/Montero/Shogun-calibration advice

[disco stu]

Hi all. New member here but I've been perusing the forums for a few days checking out things.

I've got a 1999 Mitsubishi Pajero/Montero/Shogun (or whatever else they're called in your part of the world) on which I've installed AC Stag 300 ISA2 that I purchased unused second hand.

Recently the consumption has increased a fair bit. Car has only been on the road since Christmas after I bought it cheap and did a heap of work including the vapor injection and new engine etc.

Consumption when first going was 14.5L/100km on petrol and 16.5l/100km on LPG. That was on mixed driving. I'm now getting 18.5-19l/100km on mostly highway driving, and its almost costing the same as running petrol.

While the install was checked out by the LPG guy when I got registration check, all of the install and calibration etc was done by myself. I've read the manual for calibrating multiple times, and I still think I'm missing some understanding of how it all works and how it is adjusted. This was highlighted reading some things on here. Maybe it is the translation from Polish to English in the manual. I've left it on autocalibration, not sure if that was the right thing to do, but things have gotten more out of whack between petrol and LPG curves rather than better.

One thing to note with my screen shots is that the lambda threshold was set at 0.2v for some reason (maybe some great idea of mine to try and trick it to run a little leaner, which I can't remember right now). I raised it to what it is currently, but thinking I really need to redo the autoadaptation for that to make an effect.

One thing I'm wondering is with reading the LPG vs Petrol curve. Take 13ms on my screenshot above, LPG is above the petrol curve, does that mean that the LPG injectors are open for shorter than the petrol injectors, and is sitting to the left of petrol (shorter open time) for that engine load? Going by that, am I meant to raise the multiplier at that point to get the LPG injectors open for longer time?

Looking at the open times for petrol and gas injectors on the upper right hand side of images. This is taken at idle, and petrol is at around 3.4ms and LPG is at 5.9. I thought LPG open time was the petrol time multiplied the orange multiplier line. Which in this case would be 3.4 x 1.25 = 4.25ms, except the actual time is well over that. So what is the multiplier actually doing if it isn't multiplying the LPG injector time? I think that is what keeps confusing me about adjusting that line-if I knew exactly what it was doing I would understand what to do with it.

I'm also wondering about reading the deflection map with the red line. It doesn't seem to make much sense to me, despite reading that section repeatedly in the manual and googling for info.

Would really appreciate it if anyone could give me some advice and pointers on the right way to go about setting this up properly and understanding what I'm doing properly.

Thanks in advance

Stuart

[disco stu]

Forgot to mention, I'm running Magic reducer if that makes any difference.

[LPGC]

The lambda readings in all your screenshots read close to 0.16V, did you wire the LPG ECU to the lambda wires? If not don't worry, it isn't entirely necessary but if you don't have other means (an OBD scan tool that shows live data readings) of seeing lambda voltages it is a good idea to connect those wires to lambda so you can see mixture on screen (especially for calibrating high load open loop mixture when you put your foot down).

Do you have an OBD live data scan tool to be able to read fuel trims and lambda? Any OBD error codes? We really need to make sure it's running closed loop with correct mixture on petrol before trying to find any problem with it running on gas. If it has been running with incorrect calibration or mixture the engine could be running in open loop / limp home mode and this would make it difficult to check / adjust LPG calibration.

It shouldn't make any difference where you set 'lambda threshold' because you haven't enabled ISA2 M2 adaption, M1 adaption doesn't take account of lambda readings for AutoAdaption purposes.

In your first pic the blue dots plotted are manifold pressure versus petrol injector pulse duration when running on petrol, the green dots are manifold pressure versus (would-be) petrol injector pulse duration when running on gas. The blue and green lines are kind of a smoothed average line drawn between blue dots and green dots. The idea is that if calibration is correct the blue line and green line should overlay each other, I call this the join the dots method of calibration but this isn't often the best way to calibrate a system because even if the lines overlay the fuel trims on different fuels (petrol or gas) can be different - comparing fuel trims is the best way of calibrating a system.

The join the dots method isn't much use at higher engine loads because at higher engine loads (like your quoted 13ms) when the engine switches to open loop mode the plotted points will be in the same position regardless of mixture. Unless you have wired the system to lambda and activated M2 autoadaption the only way to check high engine load mixture is to check lambda readings (you can do that on screen if you have wired the system to lambda, or you can do that using a scan tool to read lambda voltage from live data, or you could just temporarily wire a multimeter to lambda voltage).

The deviation chart is effectively supposed to show where the LPG system is calibrated lean/rich compared to the petrol system. I expect the flick in the deviation for loads higher than 11ms simply corresponds to the load at which the engine switches to open loop fuelling. In reality the deviation line doesn't give any useful info.

You have left ISA2 M1 autoadaption enabled. M1 autoadaption compares points plotted on petrol to points plotted on LPG and where points on LPG are higher than points on petrol it assumes the calibration is too lean so increases the multiplier at that point (and vice/versa). Any type of autoadaption can make a mess of calibration (even if calibration was all good as initially set up by the installer) and is something I always leave disabled. You can imagine that if the engine ever has some sort of problem such as a lambda sensor fault then if autoadaption is left enabled it can continually make changes to calibration trying to 'join the dots' but dots won't move because the engine is running in limp mode.

Can you post a pic of the 3d map (click on map, maps, 3d view) and a pic of rpm correction (click on map, rpm correction)? This may reveal any changes that autoadaption has made that could be to the detriment of correct fuelling.

Even if everything seems OK (no OBD error codes, lambda flicks OK in normal driving on petrol or on LPG) there may still be a problem with the LPG system being unable to supply enough gas to the engine for rich enough mixture when you put your foot down, but we should ignore this for now and concentrate on seeing if the engine gets correct mixture at low and middle loads. I might expect such problem (for addressing later) though, judging by ginj being 5.7ms versus pinj of 3.4ms... This tells me that the multiplier at low engine loads is quite high, which can imply that the multiplier at higher engine loads will also have to be quite a bit above 1 (multiplier is effectively gas injector duration ginj divided by petrol injector duration pinj). Imagine at 6000rpm the engine completes 2 rotations (one four-stroke cycle) in 20ms, the 20ms is the available window of opportunity for injectors to open, flow gas, and close again in time for the next cycle. If gas injectors need more than 20ms at 6000rpm to open, flow, and close again then the system can be unable to provide enough fuel for correct mixture (at high rpm and high loads). There are 4 cures for such situation if it exists - if the reducer won't flow enough gas at high enough pressure then fit a bigger reducer (but enough gas won't be the problem on this engine with magic reducer), fit bigger nozzles to the injectors (the metering is done by the size of hole in injector nozzle jets a bigger hole will flow more gas for the same pulse length, fit different injectors that take less time to open and close if the fitted injectors are slow in these respects, increase reducer pressure (but some injectors become slower with higher pressure, some injectors will perform very badly with higher pressures, some reducers can't be adjusted to high pressures). The V30 injectors have internal restrictions of 2.5mm (so even if bigger nozzles are fitted they won't flow more than they would with 2.5mm nozzles), if the reducer is a Magic Compact it will probably only be adjustable up to 1.4 bar in which case you already have it adjusted to near it's maximum pressure output (though Magic Compacts made within the last 4 months or so seem to be pressure adjustable up to higher pressures like 1.7bar and Magic Extra reducers can all be adjusted up to 1.7bar).

If you come back with the pics and answers to above questions we'll take it from there.

Simon

[disco stu]

Thanks Simon. Really appreciate your reply. I'll address some of the things you asked while I'm siting around with time to burn but will have to post other screenshots later when I get back from work.

The lambda readings always confused me on this. Its got new sensor, and I was so confused that I put an old one back in that swung back and forth. It didn't run real well with that sensor and I eventually put the new one back in. Yes, I'm wired in to the lambda sensor. This car only runs one sensor-pre cat converter.

I don't see the lambda rising above lean very often, running on either petrol or LPG, or at idle at least. It will while under power, but even then its sitting on the lean side most of the time.

This car is pre OBD2 in Australia. Only way I can do it at home is grounding one of the pins and it will flash basic codes. No codes stored, and it seems to be running fine on petrol. Though I haven't driven it a lot on petrol since consumption has increased so I'm not overly sure if its using excessive petrol also.

That is very interesting on the M2 adaptation. I don't recall seeing that mentioned anywhere in the documentation for the system at all. That would also likely explain why its not showing my short term fuel trims, which is something that I've wondered a lot about.

From memory its the magic 3 reducer that I have, so more than enough flow for my engine. I can up pressure from what it is currently if its needed, though when running auto calibration it told me that pressure was too high so I dropped it down. Can't remember what size I went to when drilling out the nozzles. Me being me I would have erred on the side of caution and gone on the smaller side of applicable-can always drill it larger, can't drill it smaller type thinking. I really should have written that down as my memory is not at its A game lately. But all that is academic, because I gather that we haven't even established that it is struggling for gas supply.

Something I should add. This is a manual 4x4. It seems to have developed a weird trait at lower speeds, when just holding at say 1500rpm in 2nd gear if you remove or add power at all it really jumps around with a decent thud when changing the power added. Not sure I can describe it better than it jumps, but make it awkward to drive it between 1100-2000rpm at lower speeds, so I generally just go up a gear and make it struggle more. It seems to only do this on LPG.

Thanks for the help so far, greatly appreciated. I'll post those screenshots later this afternoon after work

[disco stu]

3D map and RPM correction images attached.

I was thinking that there wouldn't be anything in the RPM correction, as that was blank last time I checked about a month ago. Obviously not the case now.

While I was checking things this afternoon (my time) I tried changing it from M1 to M2 autoadaptation and couldn't get it to change. How do you do that?

With the lambda readings, early on I thought I might have tapped into the wrong wire at the ECU. But changing the sensor gave a very different readout that swung wildly from rich to lean, and checking things by spraying hydrocarbons into the intake, and creating an air leak caused the lambda reading on the Stag software to change as appropriate, so it seems I'm tapped into the correct wire

Thanks again for your willingness to offer advice, greatly appreciated

[LPGC]

If last time you checked was when you calibrated there wouldn't have been any figures in RPM correction and the 3d map model would have looked flat across the rpm range, the figures were entered automatically by the ISA2 autoadaption system.

You may not be able to switch to M2 autoadaption without switching autoadaption to OBD rather than ISA2, or may be due to your version of firmware. But the best bet is probably to turn off autoadaption completely and clear (change all back to zeros) the rpm correction figures.

We can thenmanually enter guestimated rpm correction figures that will be far more likely to give good results than the automatically entered autoadaption rpm correction figures. After disabling autoadaption and clearing rpm correction figures (and making sure clearing rpm figures is also reflected in the 3d map, i.e. made the 3d map flat across rpms again) I would enter figures across all rows (0,0ms across to 25,0ms) like this: 0 rpm 3%, 300rpm 1%, 600rpm -2%, 800 0, 1000 1, 1400 2, 1900 3, 2400 3, 2900 4, 3500 5, 4100 5, 4700 6, 6000 7. These are conservative (erring on the side of not to great enough extent) rpm correction factors. The running engine will normally never see lower than 500rpm but sometimes after long cut-off (long over-run with no fuel injected) the read rpm may dip down to (say) 400rpm (even if the low rpm figure isn't shown on screen or scope) and after long cut-off the pipes between injectors and spuds (brass manifold nozzles) will be purged of gas, adding a bit of enrichment for these conditions can help bleed up the pipes (replace air that was sucked out during high vacuum no gas injector pulse conditions with gas again) and sometimes prevent stalls or at worst improve drive-ability/smoothness in conditions like return to idle.

Difficult to advise map shapes etc without being hands on but looking at your 2d multiplier (and if previously plotted points are correct) I might expect it to be too rich below 3ms.

Did you confirm the engine is running proper closed loop mode? If you have connected correctly to the lambda probes and the engine is running properly in closed loop mode the voltage readings should constantly flick rich/lean during driving on petrol or on LPG. First make sure they flick on petrol.

If it seems to be running with correct mixture because lambda flicks on both fuels then: drive on petrol with pinj at 8ms and keep constant throttle position as you switch to LPG, set the multiplier for 8ms on the 2d chart so that the pinj figure comes back to the same 8ms reading when you switch to LPG (if Pinj increases when you switch to gas and decreases when you switch back to petrol then move the point at 8ms up a bit, or vice versa). Now whenever you're driving with pinj at 8ms and you switch fuels the pinj stays at 8ms but what is ginj (gas injector pulse duration)? This will help check if nozzles and pressure are OK for this engine.

Another trick in the book for getting good results is to assume temperature correction is excessive (it usually is) and reduce it's extent like this..



What I've done in the pic above is dial out some of the default temp correction which tends to over-lean at low temps and over-richen at high temps. The default straight line isn't 0% temperature compensation it is 0% above or below the default compensation, since I assume the default gives excessive leaning at lower temps I increase the position of the line at lower temps and vice/versa for higher temps. But first I drive the car in normal conditions to see what the normal gas temp is and if I calibrated the car at that temperature I add a point at that temperature (red highlighted point at 45c in this example) so that point's compensation won't be changed. You can also see that I've added quite an increase between -25c and -30c, I do that on a lot of installs (on which the ECU wouldn't prevent driving on gas if it notices a gas temp sensor is open circuited) because if the gas temp sensor is ever open circuited it will normally lean the mixture right off, this prevents the mixture being leaned right off in the event of a broken gas vapour temp sensor (and the ECU continuing to allow running on gas).

I should have asked which model/engine Shogun is this... not GDI engine?

In parameters>gas controller settings, set 'extra injection switch off threshold' to 0.9ms.

Check 'routing' to make sure each gas injector is feeding the correct cylinder (output wires to gas injectors on each cylinder match input wires from petrol injectors on each cylinder). With the engine running smoothly on LPG you can do this by switching each cylinder in turn to run on petrol (5 cyls running on LPG, 1 cyl running on petrol). If routing is correct the engine will continue to run quite smoothly after switching a cylinder to run on petrol, if routing is incorrect you'll get a misfire on 2 cylinders because gas is turned off to the cylinder you selected but petrol is not turned on to that cylinder, while another cylinder gets both gas and petrol so misfires because it's mixture is too rich. To switch a cylinder (LPG channel) to run on petrol just click on it's number at the right of the screen under Active LPG/CNG injectors, do this for all 6 cylinders in turn.

[disco stu]

Thank you again.

I've turned off autoadaptation and have input the RPM correction figures that you gave before driving home from work. Seems to be a little better at the slower speeds, though the drive is mostly highway all the way home.

I feel I should look at the gas temperature correction once things are better sorted out.

All cylinders are on the correct injector etc. Funnily enough, while looking at the settings I noticed you can change what LPG injector is linked to whichever petrol injector in parameters>gas controller settings>Gas injector settings

This car isn't the GDI model. We didn't get that engine in australia,

If my lambda readings aren't doing that swing from rich to lean, would you say the sensor is stuffed? Like I mentioned, O2 sensor is new and I thought it wasn't giving the result I was expecting as it was constantly sitting lean. I threw an old one in I had laying around, and it gave the swinging rich and lean signal on the oscilloscope but the car didn't run as well with that sensor so I started thinking that what I was seeing with original new sensor was correct. I also checked by injecting hydrocarbons to make it rich and pulling vacuum lines to force lean and it gave the result I was expecting. Just all sitting a bit lower than what I would expect at stochiometric ratio, and not bouncing rich and lean. I might have images of the lambda readings with the other sensor somewhere, so I'll see if I can find them.

What you are saying about switching at a certain load and adjusting until petrol injector time is the same when running on LPG makes sense. Is this done on the 2D maps screen just by adjusting the multiplier line? Do you do this for a range of injector times and just interpolate in between those points? I'm picturing that this can be hard to do in practice, holding it a particular injector pulse times, especially for the longer times. Shorter should be fairly straightforward I think.

Thanks again for your help and advice. Haven't been able to find this sort of info with a lot of time looking through various places

[disco stu]

Found image on another forum of the output when I put the old O2 sensor back into the car. Can see that the reading is going from rich to lean, but I thought it would shift a bit quicker. That sensor was old though (I think). I also tried one of the old sensors out of the camry that was giving similar response on the oscilloscope. I noted the following on the forum when asking about the lambda readings (didn't much advice on it from there incidentally, seems most people don't know what the output should look like): "On lpg the car seems to surging a bit, like the mixture is bouncing around until it finally settles if the car stays at that rev range, which isn't often. Doesn't seem to do it on petrol. It didn't do this before with the original replacement sensor that was only giving 0.18 at idle"

[LPGC]

You're welcome.

The lambda reading should flick faster when rpm is faster, flick slower at idle. Particularly if LPG calibration is incorrect it may take a while to start flicking if you've just switched between fuels and may take a while before lambda starts to flick regularly. Was the screenshot at idle or while driving around? If driving around were you keeping a constant engine load or changing throttle position?

Correct about setting the multiplier points on the 2d map, yes do this for a range of injector pulse times, it is possible to set them during vehicle acceleration (keeping same throttle position in high gear to keep the engine under the same load) so you don't have to reach crazy speeds to set middle and middle high multiplier points. At the highest load points the engine will be running open loop during which pinj won't change regardless of where you set the multiplier, so in open loop mode it's not possible to set multiplier points using the compare pinj method, instead you set these load points by checking lambda reading... the engine should run rich in high load open loop conditions but not too rich, aim for a lambda reading of 0.8v for these conditions.

If the lambda sensor you removed shows flicking voltage while the new sensor doesn't it could be that the new sensor isn't compatible with the vehicle? If an incorrect sensor is fitted it could make the engine run rich all the time or make it run in open loop mode all the time... If LPG calibration is wrong it could run better on LPG with the longer petrol injector input pulse durations that running on petrol with rich mixture would give, or could run better on LPG if the petrol system is running open loop mode (longer or shorter than usual petrol injector input pulses). You need to make sure the lambda readings flick properly with the engine running on petrol before moving on to LPG calibration.

Surging on LPG can be due to incorrect shape of the multiplier line... E.g. running on petrol if the petrol ECU notices a lean mixture it might expect to correct that mixture by increasing pinj by a certain amount to give a corresponding increase in fuelling, if the shape of the multiplier line is wrong the increase in pinj might mean a very different increase in fuelling (of LPG) to what the petrol system designers expected. Another reason for surging on LPG particularly at low engine loads like idle, fast idle and over-run can be due to the ging (gas injectors pulse duration) going below the minimum duration at which the type of injectors become linear and can accurately meter gas. The V30 injectors you have fitted won't be linear at much below 4ms pulse duration in most conditions (warm injectors , warm gas and average pressure) - Manifold pressure is at it's lowest under idle, fast idle and over-run conditions, correct pinj is related to manifold pressure (engine load) and the engine's breathing characteristics (volumetric efficiency) at that rpm, so pinj is at it's lowest under these conditions, so ginj will also be at it's lowest under these conditions... if these conditions see the gas injectors pulsing for a duration below the minimum duration they can accurately meter gas this can cause surging because the engine might see much less fuelling than it should when the gas injectors are not working properly, so rpm begins to fall, but then the idle valve opens to allow more air in to increase rpm, which increases manifold pressure, which sees the injectors work properly again, which lifts rpm until the vacuum and engine load fall again and the cycle of surging repeats. Some types of surging are more likely to occur in cold (engine / gas / injector) conditions because cold gas is denser than warm gas, the system doesn't have to pulse gas injectors for as long a duration to deliver the same amount of fuel with cold gas as it does with warm gas, so gas injectors can be pushed to run below the minimum pulse duration at which they can work correctly in cold conditions.

[disco stu]

I see a common theme in your replies-lambda readings. Given that I was unsure (at best) about the plot from O2 sensor myself, I'm thinking that I need to change that before moving on. Just spewing that I bought a new one for this very reason, but it took me so long to get the car on the road that warranty has run out on that.
This could also explain the higher than expected petrol consumption. It's not ridiculously high, and others have said they get similar consumption, but I drive more sedately than they do.

My understanding is that the only real difference between oxygen sensors for different cars is the wire length and plug, assuming both zirconia and 4 wire. Is this correct? The sensor I bought was for the Montero from USA, along with most other parts. Parts are just so much cheaper there than here in Australia

[Brian_H]

Before you replace the sensor, disconnect your wire from the lpg ecu to it, and check what it shows then in the software (just tape the bare end up). I've had an ecu that failed in that way and would constantly put .32v down that wire.

Also you probabbly already know this but didn't mention it, but the actual thread size can differ, though there are universal lambda sensors that cover many models, but you typically have to reuse your old plug for those, they should come with a connector block to use, as you can't solder to the wires (it won't stick usually due to the metal used).

If you've got it disconnected, can you monitor whats happening via obd instead? If so see what the sensor is doing at that point. I'd wonder if its logging any fault codes for the sensor as I have seen on the Disco that it will log heater faults some of the time, and this seems to be a fault on the vehicle rather than the sensors as they have both been changed. Both sensors show intermittent faults on the heater circuit, but one sensor shows .44v until its been driven for around a minute or so, suggesting there is a definite issue there for me.

[disco stu]

Thanks Brian

Does the fact that different oxygen sensor gave the different readings mean there is no need to disconnect that wire?

Yes, aware of the different thread, but it seems m18x1.5 is the most common, or from what I've seen. But I take from what you say, plus the universal sensors, that if thread is right and it's zirconium then it will work. I did once try soldering those wires, for about 5 sec! I'm guessing they're stainless, with teflon insulation. Whoever had this car before me just cut the old wires and soldered new ones on, as the plug is in the most stupid spot up in transmission tunnel. Took me over an hour to disconnect it using surgical procedures.

This car was before universal OBD2 in Australia unfortunately, so can only do proper scan using the Mitsubishi tool. It will flash engine light with basic codes when you short one of the connector pins to earth. It's not showing any codes currently with that method.

Appreciate your advice, thanks

[Brian_H]

I'd be concerned that if it is sending voltage down there (which mine was) that it might kill the sensor (which it did in my case!).

I think they are stainless generally yes, due to the heat involved where it is.

That does make it slightly more problematic on the obd front, but still can be worked around usually.

[disco stu]

Thanks. I'm with you now on the voltage on that wire. Couldn't I just unplug at the sensor end with ignition on and check voltage on the ecu side plug?

[Brian_H]

That should tell you I think yes. Or pull the lpg system fuse from the battery feed and check the joint itself with the multimeter would be another option to try and narrow it down. Though if it was flicking with the old sensor, that does suggest your new sensor isn't right somehow and its not the ecu at fault. But worth checking before you risk a new sensor either way.

[disco stu]

Cheers Brian.

Just checked then. Got 12mv with petrol ecu alone (fuse for lpg pulled) and 17mv with lph fuse installed.

[disco stu]

Sorry, forgot to say that the swinging lambda readings above were taken at idle in the driveway at home. Car was warmed up and I had revved it etc beforehand

[disco stu]

So I bench tested a few oxygen sensors I've got laying around, just using butane torch to get them up to temp and checking voltage with and without flame over sensor end. All of them shows voltages as I would expect. Pulled out the one in the pajero and got basically nothing out of it.

Looks like that 16mv reading on oscilloscope is actually the voltage going down the line, rather than the sensor reading.

You guys are spot on-thank you! I guess this explains my higher than expected fuel consumption also

I'm pinching the cable off the duff sensor to put onto the one that was in the car previously. I'll reset the petrol ecu to hopefully get the fuel trims back into line. Would it be worth starting the LPG settings again from scratch with auto calibration?

[disco stu]

So, with the new oxygen sensor I thought I would just turn auto adaptation back on for my drive to and from work today. Can adjust manually over the weekend if needed. Drove to work on petrol and home on LPG. Gas curve slowly adjusted over the drive home from lean over the whole range to what you see in the image attached.

Took the screenshot at about 1600 rpm so lambda output is clear

[disco stu]

Forgot to add the RPM correction map after autoadaptation. You can see what it has adjusted after I changed to what LPGC suggested the other day

The jumpiness that I had felt previously in around 1200-2200 when cruising in a low gear has returned also. It was great in that rev range when I first put it on LPG after clearing the previous maps after changing O2 sensor yesterday.

What exactly is the RPM correction meant to achieve? I thought the corrections were based on the multiplier and based only on the injector opening time. Is RPM correction a further correction over the multiplier based only on RPM and not on engine load at all? Still trying to get my head around how all these parts/maps work together and getting things running the best. One of my big aims is fuel economy, if that makes any difference