EW10J4 (RFN) engine power differences

[Rokassileika]

I have two identical cars with same engines. One has Stag 4 eco + tommaseto achille, ac-w02 injectors, other one stag 4 plus (unknown reducer), valtek 3 ohm injectors. They have that combination, because today I exchanged the injectors with each car. (The stag4 plus version had ac-w02 and vice versa before) The car with stag 4 eco always had problems- trailing, wasnt accelerating fast etc. The Stag 4 plus version was always good at acceleration. After replacement of the injectors now both cars accelerate like "meeh" .. While calibrating the Stag 4 eco version I got a message when auto calibrating- nozzles too big but I still managed to join the dots. BUT while calibrating I noticed the gas pressure jumping a lot- 1.4- 2.2 bar felt kind of unstable (will rebuild the reducer soon). The stag 4 plus calibration went well, no messages, joined the dots easy. My questions are:

1. Is it normal for the stag 4 eco version the reducer pressure to be that high? If no then what can be causing that?

2. The stag 4 plus version with Valtek injectors is feeling weaker because maybe the injectors are not as accurate as ac-w02's?

3. I believe that this kind of jumpy pressure in the reducer could be causing all this trailing?

4. Cars and engines are identical. How the heck gasoline curve is that much different from the other? Different pressure sensors for LPG causing that?

Thanks in advance!

[Rokassileika]

Somehow these did not upload for some reason

[LPGC]

You could see pressure fluctuating between 1.4 and 2.2 bar if you keep your foot down and pressure remains at 1.4 bar above manifold pressure then you suddenly lift off the throttle... Foot down will see manifold pressure same as atmospheric pressure, LPG pressure is relative to manifold pressure so now you have LPG pressure at 1.4 bar above manifold pressure (= 2.4bar absolute) pressure. When you lift off the throttle the engine will be in over-run mode and using no fuel with a manifold pressure of around 0.2bar, which is 0.8bar below atmospheric pressure. 1.4+0.8bar = 2.2 bar, this is what it is reading.

You're thinking in over-simple terms about calibration and setup. The join the dots method isn't the best method of calibration and won't help you set high load mixture. If/when the engine switches to open loop high engine load mode the only thing that will affect plotted LPG points (versus petrol points) is how you've set the multiplier at that load point regardless of mixture. The plotted points are dependent on RPM because the engine's volumetric efficiency changes with RPM. You haven't mentioned what reference pressure was set in both ECU's. If you have set reference pressure correct for the lower picture then it seems the nozzle sizes are too big for that one... The LPG ECU bases LPG injector pulse length on petrol injector pulse length, so LPG injector pulse length shouldn't be shorter than petrol injector pulse length otherwise it will run too rich for at least one injection cycle whenever the driver switches to a new load point, also LPG injectors need to pulse for longer than petrol injectors to meter fuel as accurately as petrol injectors, most particularly at around the low load range (the multiplier needs to be at least 1). If your top picture is correct the engine won't be able to run with correct mixture at high loads and high rpm anyway because it will run out of 'window' / time available to pulse injectors between 4 stroke engine cycles, because that map has a multiplier of 1.2 around the high load area, it needs to be around 1 at the high load area... But the shape of both maps will be wrong anyway. It is usually OK to have multiplier lower than 1 around the high load area, this is because we can expect pressure to fall a little from what it is at low load conditions under high load conditions, the reference pressure should be set to what pressure is under low load conditions, the ECU will compensate for low pressure by increasing the LPG injector pulse length, increasing the pulse length is a bit like increasing the mutlipler.

[Rokassileika]

Is this the reference pressure You were talking about? I never really paid attention to this setting, but I don't really understand the concept of Gas pressure that is shown in the right upper corner. So this is absolute manifold pressure? (Manifold + LPG) That is shown here? So how do I know how to set everything right if it has an impact on how well the lpg mixture is set. And to have that bare simplicity and point of view. If joining the dots is not ok, what exactly do I need to do then? For example, in high loads. I observe ltfts and stfts while driving on petrol in high load area and at that same moment i switch to LPG- I notice that stft + ltft is lets say is -4 i Decrease the multiplier a little so it gets closer to zero? Is this the correct concept of adjusting the multiplier on these low/ mid / high load areas? Besides, how would I know what nozzle size I have to drill on ac-w02 ? (I was talking about that car with too big nozzles) Thanks!

[LPGC]

Yes the pressure setting in your screenshot above is the set reference pressure.

In most 'screens' in your LPG software you can see LPG pressure and manifold pressure. The LPG pressure reading is LPG pressure above manifold pressure, not absolute pressure. The manifold pressure reading is absolute pressure. As far as we and the LPG software / system is concerned it is LPG pressure above manifold pressure that we're interested in, not absolute LPG pressure, because the amount an LPG injector will flow is relative to the nozzle size (or LPG injector internal channels size if that is smaller than LPG injector nozzle size), the pulse duration and the LPG pressure above manifold pressure. LPG above manifold pressure affects how much gas will be injected (for a given pulse length) to the extent of the square root of this pressure.

Example - If you set reference pressure at 1.6 bar in LPG software, then whenever the LPG pressure minus manifold pressure was also 1.6 bar the LPG ECU would not have to compensate for pressure because it is expecting 1.6 bar and reading 1.6 bar. (Also keep in mind that like I said the pressure reading in software is LPG pressure minus manifold pressure, it is not showing absolute LPG pressure, it is showing what we are interested in). But if reference pressure was 1.6 bar and the pressure reading was 1.3 bar then that is an under-pressure situation and the LPG ECU will compensate by adding some percentage to the length of pulse duration to compensate for that under-pressure situation, if it didn't compensate for the lower pressure the mixture the engine would get a leaner mixture than we intended. The formula for LPG pressure correction is usually close to comparing the square roots of the pressure reading and the reference pressure. The square root of 1.6 is 1.2649, the square root of 1.3 is 1.1402. We would have to add around 11% to the square root of 1.3 to get 1.2649.... If you set reference pressure to 1.6bar and the gas pressure reading in software is 1.3bar the LPG ECU will add around 11% to gas injector pulse duration to compensate for the under-pressure sitaution.

If the gas pressure (again worth saying above manifold pressure, and above manifold pressure is what is shown in the gas pressure reading on screen not absolute pressure) is actually 1.3bar but we set a reference pressure of 1.3 bar, then the LPG ECU will (behind the scenes) be adding around 11% to the calculated ginj (calculation based on the multiplier, rpm correction and also gas temp correction). Or if we set the reference pressure at 1.3 bar and actual pressure was 1.6 bar the LPG ECU would behind the scenes subtract around 9.7% (1.402 - 9.8% = 1.2646). Note that the pressure correction percentage is different for a given extent of over pressure than it is for the same extent of under pressure. We could set reference pressure at some ridiculously high pressure such as 2.8 bar and perhaps still be able to calibrate the system even if the actual pressure is really only 1.3 bar but one problem with this is that the multiplier map has a certain resolution, let's say it has a resolution of 8 bits (which is 255 points) and for this over pressure situation (2.8 bar square root being 1.6733 versus 1.3 bar square root being 1.1402) the LPG ECU would be subtracting over 31.5% so we would need to set our multiplier numbers low for correct fuelling and with the low (small numbers) in the multiplier we start to lose resolution - The resolution could be termed accuracy, there is a 2% difference between numbers 50 and 51 so if we had a number 50 in our map the nearest next number we could set would be 50 or 51 which is like a 2% jump. But if we set numbers closer to 100 in our map the difference to the next number we could set is only 1%, this is twice the multiplier accuracy we could get with the numbers in the 50's in our multiplier map. We also gain accuracy because the LPG ECU isn't having to calculate those square roots correctly or use a pressure lookup compensation table that might only be able to compensate to the nearest 1%. And we gain accuracy if the correction values are close together because behind the scenes the LPG ECU will probably simply be adding up all the correction perecentages, so there's a correction percentage for pressure but there's also a percentage correction for temperature... Imagine an engine still warming up and the gas vapour temp is cold, so the gas vapour is denser than when it is at normal operating temperature, let's say the LPG ECU compensates for this gas temp be decreasing ginj by 10%. If it is already decreasing ginj by 30% because we have set the reference pressure at 1.3 bar when actual gas pressure is 2.8 bar the 10% temp correction pales against the 30% pressure compensation, it effectively makes our temp compensation incorrect too and effectively further decreases the resolution/accuracy of our multiplier map.

The best way to calibrate the LPG system is often to set reference pressure to actual pressure, it is always the case that it is best to set it reasonably close, but there can be some exceptions and it can depend on what ECU/system we're talking about. Most ECU's are better and more accurate at doing positive corrections than negative corrections, so it can be better to set reference pressure the same as or a little higher than actual pressure. However, on many systems the actual pressure falls when the driver gives the engine some stick, so in that case it might be better to set reference pressure somewhere between actual idle pressure and high load 'stick' pressure, in this case the ECU might apply a little negative pressure compensation at idle but not have to add as much positive compensation for the high load 'stick' conditions (of course the better setup would have a reducer that doesn't lower pressure under 'stick').

I didn't say joining the dots isn't OK, it is another tool in the box but you have to remember that the dots could be plotted at different rpms and the dots are higher at rpms where the engine volumetric efficiency is higher and lower where volumetric efficiency is lower... If dots are plotted at one rpm on petrol the dots on LPG should be in a different position at a different rpm. The ECU will probably attempt to use dots plotted at around the same rpm but that won't be possible unless you've driven it on both petrol and LPG at the same RPM. The dots lose a lot of their meaning when the engine is running in open loop mode because in open loop mode the dots will be plotted in the same position when running on LPG as they are on petrol anyway. If the engine doesn't have a fuel return then while you're driving around on LPG the petrol is sat in the petrol injector fuel rail being heated by under bonnet and engine heat, so becomes less dense, so then when you switch back to petrol the fuel trims on petrol will probably be a little higher than they would normally be on petrol, if you compare plotted dots on LPG to dots plotted when running on petrol that is less dense than normal it will make it seem the LPG system is calibrated a bit richer than the petrol system. When you run on LPG the LPG is injected as a vapour into the engine, the petrol would be injected as a liquid, the LPG vapour takes up more volume than liquid petrol in the manifold... This can have a different effect on an engine that uses a Map sensor as primary input for fuelling compared to an engine that uses a MAF sensor as primary input for fuelling. In both cases let's assume you keep your foot in exactly the same position on the accelerator pedal as we switch from petrol to LPG - In the case of the map sensor the reading increases because we are pushing some vapour into the manifold, this vapour is in addition to the air that is coming through the throttle, to the engine that uses MAP as primary input this looks like a slight increase in engine load so the petrol system reacts by slightly increasing pinj. In the case of the MAF sensor being primary input the manifold pressure increases because we pushed some vapour into the manifold, so there is less vacuum in the manifold, the less vacuum means not as much air flows through the throttle so to the petrol system it seems engine load has decreased and it may respond by decreasing pinj. In both cases the change in perceived engine load may be enough to have the ECU start applying a different long term fuel trim range, so if you were only watching STFT's (and not watching LTFT's) you could see a change in STFT that only occurred because of the switch to a different LTFT and assume the mixture changed when really it didn't and really LTFT+STFT didn't change. In the case of the MAF setup (which most engines use these days) if we switch from petrol to LPG and don't move the throttle the applied LTFT won't change, at least not at loads above idle, and the small drop in MAF reading means that our plotted LPG dots shoud point to a slightly richer mixture on LPG than on petrol even though the actual mixture is the same. The real difference is that due to the LPG vapour taking up that bit of extra space in the manifold (than liquid petrol) the power drops slightly and if we want to still have the same power that we had on petrol before switching to LPG we have to very slightly (perhaps unknown to the driver) push the accelerator pedal a little more so the the engine gets the same air intake as it was getting when running on petrol.

All the above are small effects but are real and noticeable, especially to a practiced / trained eye. And the effects can make a difference to fuel trims. You tend to find that if you manage to plot points on LPG at exactly the same rpm as points for the same load position on petrol, and you have accounted for petrol temperature changes and everything else, then for fuel trims to remain the same the plotted dots will usually imply it is calibrated marginally richer on LPG than it is on petrol.

You can sometimes also see related effects at idle if you have a particularly sensitive map sensor. One aspect than relates to engine efficiency is pumping losses, the amount of power the engine has to use to suck in air and fuel and blow out the exhaust gasses. If an engine idles with (say) 0.3 bar manifold pressure when running on petrol (remember the manifold pressure reading is absolute, the LPG pressure reading is usually relative to manifold pressure reading) then the manifold pressure reading under the same idle condition may fall slightly when running on LPG, this is in some ways counter-intuitive because weight versus weight LPG needs more air to burn at correct mixture than petrol... however, weight for weight the LPG creates more heat than the petrol so the engine doesn't need as much weight of LPG as petrol to maintain the same idle speed as petrol and by pumping some volume of gas into the manifold through the LPG injectors it also slightly lowers the induction pumping loss. There may also be an unrelated effect that slightly offsets our comparison of MAP readings, both the petrol ECU and LPG ECU may have a slightly different effective reference voltage (or ground reference voltage) that they provide to their respective MAP sensors depending on whether the petrol inejctors are being driven or the LPG inectors are being driven... But when you consistently see lower MAP and MAF readings from both the petrol ECU and LPG ECU when idling on LPG compared to idling on petrol it confirms that the engine needs less air to idle at the same rpm speed and load on LPG than it does to idle on petrol, so it is idling more efficiently (at least in that respect) on LPG than on petrol.

Anyway, your takeaway from this is that it is OK to use the join the dots method as long as you also check fuel trims. And you cannot use the join the dots method to set up high load fuelling if the engine switches to open loop mode at high loads, in that case you use lambda readings.

[Rokassileika]

I'm sorry about my brains barely "braining" so far, but did I understand it correct that operational pressure has to be set slightly higher or about the same as it would be shown in the right top corner (gas pressure) when the car is stable driven in low- mid loads and minimum has to be set just a little below from the reading I get at WOT?

[LPGC]

To keep it simple just set reference pressure to the pressure reading at idle. Then calibrate. If you change the referece pressure you need to do most of the calibration again because of that pressure compensation I discussed in detail. Also to keep it simple for now just set minimum (switch back to petrol) pressure at half the actual and reference pressure.

A minimum pressure setting in software is the pressure at which the system switches back to petrol because you have run out of LPG in the tank. A lot of old systems didn't allow you to set a figure for this, they set switch back to petrol pressure at around half the reference pressure, most current systems do allow you set the minimum pressure. When setting minimum pressure / switch back to petrol pressure the considerations are 1. What is the reference / usual pressure, remember that as pressure falls the ECU has to start compensating for pressure by adding a percentage to ginj, the pressure compensation isn't totally accurate so if actual pressure is much below reference / usual pressure it can make fuelling less accurate, 2. How pressure stable is the reducer between idle and full engine load, if you fit a big reducer on a low powered engine the actual pressure probably won't drop by much between idle and full load, so a small drop of pressure might point to the tank being empty and you could prevent a short time of less accurate fuelling by setting a relatively high minimum pressure not far below reference pressure. Or if you have a reducer that idles with say 1.6 bar pressure but pressure falls to say 1.2bar pressure but is consistent in 'only' falling to 1.2 bar pressure you wouldn't want to set switch back to petrol pressure at 1.3 bar because that would then prevent running on LPG at full engine load, the ECU would switch back to petrol as though you'd run out of LPG when you put your foot down. 3. I wrote about LPG pressure being referenced to manifold pressure, the LPG pressure reducer usually has a vacuum connection to the inlet manifold and LPG pressure rises and falls along with manifold pressure... This means that when the engine is idling and the manifold only has 0.3bar absolute pressure, if the LPG pressure reading is 0.7bar then the actual pressure in the rubber hoses between LPG reducer and injectors is the same as atmospheric pressure at 1bar absolute pressure, or if (again with manifold pressure at 0.3bar) the LPG pressure reading is only 0.4bar then there is only 0.7 bar absolute pressure in the LPG hoses which is actually a 0.3bar vacuum compared to atmospheric pressure which is 1bar. If we set the LPG pressure reducer to output a higher pressure the most likely reason for a 0.4 bar LPG pressure reading is that the LPG tank is empty (of liquid LPG) and the engine is idling using the LPG vapour that remains in the LPG tank which is now down to around 0.7bar absolute pressure. If the LPG tank is at 0.7bar absolute pressure it is obviously at a 0.3bar vacuum relative to atmospheric pressure and this means that air might flow into the tank from the filling point (the bit of vacuum in the tank might be enough to open the one way valve in the tank fill port and the one way valve in the filling point and air might flow into the tank). If the engine continues to idle on the vapour in the tank that is increasingly being diluted with air then even if the ECU pressure compensation is correct the mixture will get leaner and leaner because the LPG injectors will be injecting a mixture of LPG and air instead of just LPG. This isn't much of a problem because when we refuel with liquid LPG the reducer will once again only be fed with liquid LPG so the injectors will again only inject liquid LPG. But when I fit most tanks I purge the air out of them before sealing them with the valve(s), this has the advantage of allowing the tank to fill to thr required level (usually the standard 80% fill limit) regardless of the pump pressure. Pump pressure can change seasonally because in hot weather LPG tank pressure (the vehicle and the filling tank) are higher than in cold weather. In really cold weather pump pressure might only be (say) 6 bar, in hotter weather it might be 13bar. If air isn't purged out of a tank before it is filled with liquid LPG the air in the tank will be compressed above the liquid LPG in the tank. If we leave air in an LPG tank and then pump the tank to 80% of completely full with liquid LPG then we have compressed the air in the tank into 1/5th it's original volume, so now the air in the tank will be at 5 bar absolute pressure which is 4 bar gauge (relative to atmosphere) pressure, the pump would need to be pumping at at least 4 bar to be able to fill the tank to 80%. I often modify tank valves to allow them to fill to 90% instead of the usual 80%, at 90% we are comressing the air in the tank into 1/10th it's original volume so the air will be at 9bar gauge pressure.. The LPG pump would need to pump at 9bar to be able to fill the tank to 90%, yet in very cold weather LPG pressure might only be 3bar and if the pump gives a 'head' of 4 bar that adds up to only 7 bar, so unless we purged air out of the tank before the first fill of liquid LPG then whenever we tried to fill at this (4bar head) pump with LPG that has a static pressure of 3 bar for total 7 bar we wouldn't be able to fill to the intended 90% because the pump wouldn't have enough pressure. To get around that I do purge tanks or air before putting some liquid LPG in and to prevent air from entering the tank through the fill port when the tank is empty I never set switch back to petrol pressure lower than 0.7 bar.. The 0.7bar added to the 0.3bar idle manifold pressure is 1 bar which is atmospheric pressure, setting any lower than 0.7bar switch back to petrol pressure might allow air into the tank during engine idle when the tank is very nearly empty. But if you were running a system that uses 0.9bar actual and reference pressure the 0.7bar minimum / switch back pressure might be too high because pressure might fall to 0.7bar at full load anyway. And if you didn't purge air out of the tank before first putting LPG in the tank, or if you live in a warm climate, or if you are happy at the standard 80% fill (as opposed to 90% fill), or if you use a pump that always gives a high head pressure, then there may be little point in being concerned about air in tanks or using a switch back to petrol pressure lower than 0.7bar... But you still wouldn't want to set a switch back pressure that is lower than half actual or reference pressure really.

Also don't forget that actual and reference pressure play a bit part in the calibration. You don't want a map that has a multiplier lower that is effectively lower than 1, especially not around the idle load area. If you set actual and reference pressure at say 1.6 bar but find that at around idle load ranges the multiplier is below 1 then you might be able to 'get away' with lowering reducer and reference pressure to a much lower level, like say 0.9bar. The difference an injector will flow for a given pulse length (within the injectors more linear pulse length range) between around 0.9 bar and 1.6 bar is around 24%, so if the injectors were more or less in their linear range at the ginj they were pulsed for at idle loads with the 1.6bar pressure and multiplier of 1 and decreased actual pressure and reference pressure to 0.9bar then we would expect to increase the multiplier from 1 to around 1.24. Notice that I have been talking about 'within the injector linear range'... LPG injectors take longer to open and close than petrol injectors and while they are partway between closed and open (or open and closed) they are partly flowing. The amount of pressure can speed up or slow down either or both the opening and closing stages. The amount of gas they flow when partially open depends on many things such as how partially open they are, internal design, how far the restricting nozzles are from the internal opening/closing bits, pressure (again), etc. Nozzle size can usually make more difference to how much gas an injector flows than pressure but there are exceptions, for example one example of an exception is if the injectors has nozzles that are fitted some distance from the opening/closing component inside, if nozzles are some distance and the nozzles are small then under most conditions when the injector is even partly open the volume between the opening/closing parts and the nozzle will be pressurised with LPG vapour and that higher than manifold pressure volume of gas between the opening/closing bit is going to go through the nozzle into the engine almost regardless of how big the nozzle is. So unless the injectors have nozzles very close to the opening/closing internal bits pressure can make more of a difference to very short opening pulse times than nozzle size, then even before the injectors are within their linear range the nozzle size starts to make more difference than pressure. But pressure can also affect the opening and closing time. The best bet for accurate fuelling at short ginj's is to fit very accurate and linear injectors, and if using small nozzles make sure the injectors have nozzles that fit up close to the opening/closing bits. There's also a difference in flow between internal physical size injectors, the injectors we use on most car sized engines are fit for purpose but they would be massively oversized for (say) a 5cc model/toy car engine even if we fitted extremely tiny nozzles, on the other hand if we had a massive engibe with 3000cc individual cylinders the injectors we normally use on car engines (with 200 to 1000cc cylinders) would be massively too small even if we fitted the biggest nozzles possible so we'd probably want to fit multiple injectors to one cylinder.

When choosing components, setting up and calibrating the aim is to end up with a system where pressure remains fairly stable across the entire engine load range, peak ginj is close to the same as peak pinj at full load, the injectors are within their linear range at any engine load. Because the LPG inectors take longer to open and close than the petrol injectors the map should then have highest multiplier at around idle loads and lowest multiplier at peak engine load.

[Rokassileika]

I’m really grateful for the amount of detail and explanation you’ve provided. Honestly, I ended up making myself a small “cheat sheet” with different facts and theories from your posts, and little by little everything is starting to make much more sense now.
I still have one small question though.
As I understand it, the behaviour of the LPG system should match the behaviour of petrol as closely as possible, so that the ECU does not “feel” a large difference between the two fuels. Because of that, in the low and mid load ranges I should switch from petrol to LPG while driving in the same conditions, observe how pinj changes, and then adjust the multiplier so that the combined STFT + LTFT on LPG is similar to what it was on petrol.
But theoretically, shouldn’t the ideal situation simply be:
LTFT+STFT= 0
Why can’t I simply adjust the multiplier until STFT + LTFT becomes approximately zero on LPG, without comparing petrol vs LPG by switching between them?
Thanks again!

[LPGC]

You could calibrate it so STFT+LTFT = 0 (or close to 0) when running on LPG. But there can be problems doing that... For starters, as you already know it can make for more of a change when you switch between fuels. The bigger that change is the longer the engine runs with slightly incorrect mixture when you change fuels. But also it can be a bad idea because some vehicles base open loop high load fuelling on the LTFT for the highest load position before switching to open loop. If that LTFT were to depend on which fuel you had recently been running then when you switched to the other fuel and put your foot down before that LTFT had been relearned the high load fuelling could be wrong.

Depedning on which make/model engine I'm calibrating an LPG system on, and depending on how the petrol system handles high load fuelling (does it run wide band lambda sensors and continue to run open loop into the high load area, does it run narrow band, if narrow band is it compleltely open loop or apply high load LTFT's, what are the LTFT's on petrol, etc) I might aim for the same LTFT's on LPG, or somewhere between petrol LTFT's and zero, or zero LTFT's, but usually I'll aim for same LTFT's unless the LTFT's seem a bit too far away from zero in which case I'll usually aim for somewhere inbetween current and zero.

[Rokassileika]

I came back for the Camry today. And I noticed a few things. On petrol at idle the stft + ltft is jumpy.. and high I would say. On one bank +7 on other +13, or +12 and +14. On mid loads the sum is +20 sometimes. When I cold start the car, sometimes I can clearly smell raw gasoline. Maybe I have a leak, but in an over looked area? That fluctuation makes it really hard for me to set LPG depending on petrol. Another thing is when I believe Ive set correct multiplier on idle- switching from lpg to petrol- no misfires. But when I switch from Petrol to LPG i can clearly feel short misfires and stft + ltft falls down to 0 or even negative. And eventualy gets back to prefered values as petrol- +5 and +10 on both banks. But still kind of fluctuates. Just realy hard to catch the numbers. Any suggestions what could that be?

[LPGC]

When you start most petrol cars from cold you might be able to smell petrol, the choke effect / excess fuel factor means some petrol comes out of the exhaust.

The pipes between LPG injectors and manifold might be too long, this can explain some lumpiness when changing from petrol to LPG. Or it could be that LPG pressure rises when it's running on petrol, in which case there is a chance that there could be a problem with the reducer. Or you might have set some 'changeover overlap' in settings, this causes petrol injectors to continue workig for a very short time when the LPG injectors (or individual injectors) are first turned on, in which case the mixture usually does go a little rich for a moment because the engine receievd both LPG and petrol. The figures you're seeing in fuel trims could just be due to engine load changing due to 'short misfires'... When an engine struggles to idle and the idle speed falls the electronic idle speed control will open the throttle a little more to try to maintain idle speed, this can cause a momentary blip to open loop mode and/or a different set of LTFT's to be applied.

[Rokassileika]

I fit smaller nozzles- 2 mm. (It was ~2.6-2.7 before)The autocalibration still leaves me a message- nozzles too big. And leaves me with multiplier curve below 1. I think I don't understand the concept of LTFTs. The LTFT at idle running on petrol is +14.7. STFT remains +/- 3 . I moved multiplier points manually arround 1.2-1.4 on idle and low loads. And I get pretty much the same fuel trims if combined. Of course, the pinj is less at that point than petrol would be. And then i switch to petrol. Engine almost stalls (misfires) fuel trims become outrageous (+33 stft, +40 ltft) and eventually it evens out. Stabilizes to +14.7 combined... Switching back to LPG doesn't make the stall effect and trims don't jump that crazy. I think there is a user error. I must be not understanding something. Maybe at idle and very low loads it is best to do the "joining dots" method and later on rely on fuel trims and then lambda (open loop high loads)? Something similar is happening with my Camry. So as I said, there must be user error- I'm the problem or there are Vacuum leaks, weak fuel pump issues on both of the cars. Need help. Seriously

[LPGC]

Is your reducer leaking gas into the inlet manifold by the vacuum connection tube?

Some injector designs will output quite a lot of gas even with small nozzles even if only pulsed for a very short ginj.. the internal design might have a slow closing time and/or a wide and long output shaft before the nozzle (lots of 'pre nozzle volume'), in which case the 'shaft volume' is full of gas and once gas is in this area it will all go through any nozzle because it has enough time to all go through the nozzle (when pressure is higher in 'shaft' than manifold vacuum on the other side of the nozzle).

It isn't too unusual for LTFT's to momentarily 'flick' high/low during fuel changeover but still the changeover should be smooth, particularly changeover from LPG to petrol.

Do you have routing correct - If while running on LPG you switch each cylinder back to petrol in turn / one at a time does the engine continue to run on all cylinders or does it cause a misfire when you switch any of the cylinders back to petrol?

[Rokassileika]

Sorry for late reply. According to the website I was banned for some reason. Using VPN now. Found a few vacuum leaks, isolated it. Seems to be ok so far, just a little weird that vacuum leaks on an engine with MAP sensor would do anything bad. Anyway it seems to have stabilized at least for now.

[LPGC]

Vacuum leaks on an engine with Maf sensor are more likely to lead to problems than vac leaks on engines with map sensors. On the Maf setup the Maf reads less airflow, which can make fuel trims more positive and/or cause which LTFT is applied/used (but less of the 'or', more of the 'and'). On the Map sensor setup the idle control valve / throttle may not need to be open as far for the manifold to get the same intake air flow but the map sensor will read the same as it would without the vac leak, at least to a certain extent.

[Rokassileika]

Another update. Smoke tested the EW10J4 today. I have two small vacuum leaks near throttle body, one small leak on the oil dipstick and a huge vacuum leak on the PCV hose connection. Will isolate it and report back if it does anything.

[Rokassileika]

Isolated vacuum leaks. After isolating them found even more. The stud was untightened that lpg hose goes on it. Camshaft sensor seal was leaking too. After all, the results were immediate in fuel trims. On petrol it evens out to 0, on lpg it is a bit rich still. But noticed that sometimes when the car is idling, it randomly misfires, sometimes even stalls out of the blue sky. The trims go mad for that one moment and after that it all evens out again like nothing ever happened. Sticking throttle body flap maybe?