Weird fuelling issue (manifold pressure affects t.inj gas)

[carl0s]

Here's a weird one.

I can choose whether my car idles with full-valvetronic (throttle butterfly wide open), and zero vacuum in the manifold, or not.

For a smooth set-off, I prefer to have the car using the throttle for idle, so I set the "minimum valve lift" to 0.8mm.

This means that the car has -0.5bar manifold pressure at idle (I think), then off idle it is back up to atmospheric as it is using valvetronic for the rest of the throttle control (valve lift controlling throttle).

The weird thing is this:

throttled idle (vacuum in manifold) requires 5ms t.inj gas
valvetronic idle (no vac in manifold) requires 4.2 t.inj gas

On petrol, there is no difference between throttled/vacuum and valvetronic operation.

the compensation tube / MAP is hooked up to the regulator, and I can see that the output pressure of the regular moves in line with the manifold pressure.

Any ideas what's going on?

It only seems to be an issue at 600~ rpm idle. If I rev the car in neutral, it seems to not show the same pattern. i.e. at 1800rpm in idle, the fuel requirement drops a lot.

[LPGC]

Maybe not so weird mate...
Are vacuum reference pipes connected to map sensor / reducer? What is the relative gas pressure with .5 bar manifold pressure / relative gas pressure with atmospheric manifold pressure? Included in the 5ms / 4.2ms is injector latency...
I've usually noted a difference between valvetronic and none valvetronic Pinj, but would suppose that since petrol injectors have much lower latency and working pressure is higher the difference in Pinj could be expected to be much lower than differences in Ginj.

Seems you have stumbled upon aspects I brought up in a much earlier Valvetronic related thread, where I suggested that gas pressure and nozzle sizes need to be set correctly to allow for the difference in injector latency between petrol injectors and gas injectors when switching between valvetronic and none valvetronic modes, particularly at idle. The difference in Ginj you have seen would account for nearly a 20% fuel trim drift between the two different engine operating modes, more than enough to easily turn on the check light....

Simon

[carl0s]

Hi.

Yes the MAP / compensation tube is hooked up, and I see the regulator pressure changing in relation to the manifold vacuum.

I need to double-check the Pinj to make sure there's no change.

Basically between valvetronic and throttled operation, the trims at idle are miles apart on gas, but unaffected on petrol.

I just have a hunch that there's a vac leak somewhere, but for months now I'm unable to find it. I'm trying to pin-point a problem with a gasp and stutter if I jab the throttle, both on gas and petrol while in neutral.

The guys at the garage are going to run the smoke machine on it tomorrow night so I suppose I'll find out once and for all if there's a vac leak.

[carl0s]

I had thought I would get around it with the "Pressure correction map", but I then discovered that this is just for gas pressure.
The Stag ECU doesn't allow mapping/corrections against MAP pressure

I just noticed that the KME Nevo software does though. My next project might use one of those.

[LPGC]

Yes Nevo does, it is a facility of Nevo I've never enabled but should do the trick. That was a good call mate, I thought only I had thought of that

And yes, the Stag correction is only for adjusting the relative gas pressure correction... so would be possible to use this for the same ends, but only if you'd fitted the hypothetical perfectly pressure stable reducer and not connected the reducer to vacuum reference... But then, for the best results you'd also have needed to have fitted the fastest responding injectors, and with such injectors the facility probably wouldn't be needed anyway!

Simon

[carl0s]

Well, upon further investigation, the reducer isn't responding to MAP pressure much at all!! I guess that will be the problem?

I have the reducer set to 1.6 bar at normal operation (no manifold vacuum).

Am I correct in assuming that when the manifold is under vacuum (0.5bar pressure), then the regulator should reduce its output by that amount? It's not doing.

Could this be because the vacuum tube I have used is pretty small?

but then, wouldn't this give the opposite problem to what I'm seeing? It should be overfuelling if it's over-pressured.

If the reducer is supposed to *increase* its pressure when the manifold is under vacuum, well then the problem is that I am already at max pressure for this regulator. The screw is all the way out.

It's not a massive problem anyway, just something that I want to have ironed out.

I can sort it out by using the 3D map of RPM vs. Pinj vs. Multiplier. In this way, I add ~20% extra LPG at 600rpm 2.5 - 4.5 ms Pinj., since there is only manifold vacuum at this point with the valvetronic/throttle blending.

[LPGC]

The reducer should reduce it's pressure when there is manifold vacuum, by .5 bar in your example. Obviously don't forget that the software doesn't show absolute reducer pressure, it shows relative pressure. If the relative pressure remains the same (relative to manifold vacuum) then reducer is in fact reducing the pressure when there is engine vacuum. If you want to confirm this you could (as a temporary measure) remove the vac connection to the map sensor but leave it attached to the reducer - you will then see the reducer pressure changing if you remove / re-install the reducer vac reference with the engine running (only when there is engine vacuum of course).

Again, the reducer pressure adjustment effects the relative pressure (i.e gas pressure above manifold pressure). If the reducer can be adjusted to a max of 1.5 bar when open to atmospheric pressure, then if you fitted the reducer to a turbo engine (with the reducer vacuum connection to engine manifold pressure) and the turbo engine has a manifold pressure of 1 bar of boost, the reducer absolute output pressure would be 2.5 bar... because it is the relative pressure that you adjust, the maximum relative pressure of the reducer is not the maximum reducer pressure.

The diameter of the vacuum line is unlikely to effect anything other than reducer pressure and map reading response times. This is hardly ever a concern and will be even less of a concern on a Valvetronc (the engine is either Valvetronic or none Valvetronic for long periods, usually Valvetronic so usually has no changing pressure for long periods, so vacuum signal response isn't a concern).

I can't see how you can use rpm correction to iron out the difference in fueling requirements (or at least gas injector pulse length requirements), at least not if you leave the engine in a state where it may idle in either Valvetronic or none Valvetronic modes. How would the LPG ECU know if the engine was in Valvetronic or none Valvetronic mode? Your idea of using a Nevo ECU's ability to adjust Ginj according to manifold pressure could do that, but not the Stag.

You previously mentioned adjusting the pressure correction in the Stag ECU - did you make this a flat line again? If not then you could see some strange results.. Can you confirm what the reducer relative pressure is showing when you have 5ms Ginj duration? When you have 4.2ms Ginj duration?

Errors like the ones you're seeing can be due to a mismatch between actual injector performance and ECU presumed injector performance... At 3ms Pinj idle duration you might set a multiplier of 1.3, so you might expect the Ginj to be 3.9ms, but the ECU adds the gas injector delay so the actual Ginj becomes 3.9 plus this offset to become 5ms (in this example adding 1.1ms for gas injector latency). The ECU shows the Ginj of 5ms on screen but the ECU firmware designers only expect your injectors to be actually flowing for the 3.9ms out of the total injector pulse of 5ms. When the measured relative pressure is anything above or below the working pressure you set in the ECU, the ECU compensates for the different pressure but only for the 3.9ms part with the gas injector delay part remaining the same constant...

Examples of the maths:
ECU set working / ECU reference pressure is 1 bar, ECU expected injector delay 1ms, actual injector delay 0.5ms, actual relative pressure 1bar, Pinj 3ms, multiplier at 3ms = 1.3. With the above figures the ECU does not need to change the Ginj pulse for pressure compensation as the measured reducer pressure is identical to set working pressure. Ginj will be 1 + (3x1.3) = 4.9ms.
Now lets try the same figures, only now with measured relative reducer pressure being 1.2 bar, so this time the reducer is supplying a greater pressure than the reference pressure and the ECU will compensate for the higher pressure.. so the equation goes 1 + ((3X1.3)*1/1.2) = 4.25ms Ginj.

If the ECU had correct info on injector latency in the firmware, the difference in Ginj pulse width between the top two examples would correct the Ginj properly and the fueling to the engine would remain correct for the different pressures. But we made the initial mistake of presuming the ECU did get the Ginj delay correct (because the ECU expects 1ms delay and the actual injector performance is only 0.5ms delay) and we haven't considered that we have unintentionally accounted for this error by setting the multiplier at 1.3 in the top example - We got the map wrong in the top example because we should have had a different constant and therefore a different multiplier.... Instead of the engine getting correct fueling at both pressures, because of the error that we built in, we effectively have the engine getting about the equivalent of a 4.4ms dose of gas in the top example, and a 3.75ms dose of gas in the bottom example. If we look at that part of the equation that deals with pressure correction again, (the 1/1.2 part) which equals 0.8333, and then apply that to the actual injector pulse in the top example (4.4ms), we have 4.4*0.83333 = 3.67ms. The error is the difference between the 3.75ms and the 3.67ms = .08ms, only a small amount but this would represent a 2% dufference in fueling, all from only a 0.5ms difference between anticipated ECU firmware injector delay and actual injector performance combines with only a 0.2bar over-pressure situation.

A 0.5ms firmware error regards injector performance would be a very small firmware error in my experience... I see a lot of ECU's get the injector response time by 2ms for certain injectors... And considering that the error will rise at an exponential rate compared to the rising proportion of incorrect built in injector delay to fueling dialled in on the mapping side of things, an error of only 2% with 0.5ms and multiplier of 1.3 might become an error of 15% (instead of 8%) with an error of 2ms and multiplier set at example only 1.1. (Side point - What did you say about your map, multiplier of 1.3, Pinj at about 2.9, Ginj at about 5 or something... that seems to suggest the ECU is probably getting the Ginj delay way too long, thus affecting your pressure compensation much more than 2% - You might want to try setting the injector type in the firmware to Matrix instead of Pan to see if this setting improves the situation). We don't normally need to consider such errors in firmware info regarding injector specs much, because at any given engine load we will usually have a corresponding manifold pressure (manifold pressure is almost the definition of engine load for most applications), and since the reducer will usually be supplying gas at a certain pressure according to engine load / rpm we can account for the errors in our fuel maps. But when it comes to Valvetronic engines, the engine load can remain the same but we can have a range of manifold pressures / relative reducer pressures, and this is when the flaws in ECU firmware info on injector specs is most highlighted.

Simon

[carl0s]

I actually did follow all that - thanks

Yes the gas pressure is reported as stable when MAP changes - so this is good then.

A further complication of this problem, is that when the car is idling with some degree of manifold vacuum / throttling (minimum valve lift set between 0.5 and 0.8mm), then during driving, the car will intermittently trip into vvt failsafe (i.e. none valvetronic), and the reported error is 27B8 - differential pressure sensor implausible (MAP sensor).

I am going to drive the car just on petrol tomorrow, to see if the same occurs then.

Can you see any reason why the LPG system might confuse the car's MAP sensor readings, causing it to trip into non-valvetronic? It is only happening when I have part-throttled idle - i.e. minimum valve lift increased so that throttle is used for idling, but valvetronic throughout the rest of the rev range. The reason for doing this is that it eliminates the hesitation that I sometimes get when setting off on gas, which I presume to be caused by lack of manifold vacuum and accumulating LPG gases in the manifold (and I don't want to fit directional nozzles).

Injectors are Hana golds btw and I have them set as Greens because golds are a new model. i think green is correct for them.

[LPGC]

I've converted a lot of Valvetronics and all of the recent ones have been a bit more tricky to calibrate than most cars... but none were a problem except the last one. You know the one I'm talking about from the other thread! Never fully got to the bottom of why canbus errors would occur under certain driving conditions when running on LPG, but it didn't seem to be anything to do with the fueling.

I have seen Valvetronics do exactly what you're describing.. Sometimes the switch back to none-Valvetronic is accompanied with a high idle speed? I think there are at least a couple of triggers... 1. Because they have two sets of fuel trims for idle, one for Vtronc, one for none Vtronic. And if the Ltft's aren't similar between modes or haven't settled down in either mode then this can happen. 2. I have recently said on another thread how BMWs don't allow Ltft's to adjust until a while after OBD errors have been cleared... I think sometimes when this is the case the engine reverts to none Vtronic mode too. Might be wrong! Just some of my observations.

Dunno what the offset will be for Hana Green in your ECU's firmware, I could describe how to find out, but given your apparent abilities at lateral thinking, man after my own heart etc, I reckon you'll know the way mate! Tired of typing today!

Cheers,

Simon

[carl0s]

I just did some test driving, and the "27B8 diff. pressure sensor plausibility" happens on petrol as well, if the minimum valve lift is set to 5mm or above. It looks as though manifold vacuum drops to 0.17bar, and from that point on, the car goes into valvetronic failsafe.

If I set minimum valve lift to less than 5mm, then this problem never happens, but then on gas I get an occasional hesitation when setting off after sitting at traffic lights for a moment or two.

The OBD number for this code (27b8) is P1199.

[LPGC]

Sorry I haven't checked this thread for a few days.

My guesses regarding the 'pressure sensor plausibility' error....

The volumetric efficiency of an engine is the engine' size versus how much air it can pump. The volumetric efficiency of an engine closely relates to how much torque the engine can make, and the volumetric efficiency for a given engine will be different at different engine rpms... We can probably calculate the volumetric efficiency of an engine at a certain rpm by how much torque it makes at that rpm.. We can measure and plot a 'torque curve' for the full rpm range of a given engine and from that we can calculate it's volumetric efficiency at any given point on the torque curve We could also calculate the volumetric efficiency of the engine at a specific rpm by taking readings from the air flow sensor at that rpm at wide open throttle.

Most engines need to have a cam that gives a good compromise between low engine speed volumetric efficiency and high engine speed volumetric efficiency. In the old days, people wanting to make their cars go faster might fit a cam with more duration and valve lift, but this would be at the expense of losing volumetric efficiency and hence engine torque at low rpm's. More recent engines feature variable cam timing to help extend the range where the cam provides good volumetric efficiency, this does help to extend the range at which the cam gives good volumetric efficiency but only by a small amount. Some engines feature valves that are controlled by hydraulic lifters, the lifters bleed down at low rpms so such lifters can also effectively reduce the duration of the cam at low rpm, this also helping to extend the rpms at which the cam can offer good volumetric efficiency.

The thing about variable cam timing and hydraulic lifters is that they only help reduce the compromises of the cam that is fitted. Most engine designers are far less concerned about the engine having good volumetric efficiency at very low rpms than they are about the engine having good volumetric efficiency from, say, 1500 rpm all the way up to the rpm ed line. Look at the torque curve of most even recent engines and they don't seem capable of making much torque at all at idle rpm - If an engine idles with a manifold pressure of 0.33 bar (a realistic figure for a modern engine) then it can be presumed that maximum engine torque available at the flywheel, at maximum throttle at idle, would be only twice the amount of torque that the engine needs just to keep running... this is a very small amount of torque and even without measuring we can assume that the engine's volumetric efficiency at idle is very poor and that this is because of the cam compromise favouring running conditions at higher than idle engine speeds.

The Valvetronic system somewhat does away with the compromises of the cam. It may be able to adjust valve duration and lift to allow the engine to develop closer to 100% volumetric efficiency even at idle, but under normal circumstances the designers didn't expect that the Valvetronic system would be adjusted to offer that performance. By forcing the Valvetronic mode with abnormal minimum Valvetronic valve lift you may have increased (or decreased) volumetric efficiiency at idle into the range where airflow measurements etc are outside the expected range of the ECU. Since your manifold pressure is 0.17 bar at idle this would seem to suggest that you have used settings that give increased volumetric efficiency at idle, and you may have increased idle engine speed maximum engine torque by a fair bit, but the designers never expected anyone to need maximum engine torque at engine idle speed... Drop a gear son, you're labouring the engine and we can feel the engine vibrating...

What would be a great engine - A variable compression engine with a kind of Valvetronic system. At full throttle the engine would work at normal compression, normal full throttle intake valve lift and duration. But at less than full throttle the engine would work with increased compression and the Valvetronic would work in the opposite way to usual Valvetronics (regards duration) to give longer duration than at full throttle in a Miller cycle.

Simon

[carl0s]

Interesting read, thanks.

The idle MAP with my adjustment is 0.5bar though, and the adjustment is one done routinely by main dealers to help with bad idle.

The drop to 0.17bar was noticed in the ac/stag oscilloscope monitoring. After the presumably unexpected drop to 0.17 bar, the car goes to non valvetronic mode, where idle MAP is, as you say, 0.3 bar.

I have some parts to change, and stuff, but haven't had time. I'll have more info soon. Right noew the LPG map around idle is somewhere in between so that the fuel trims can cope without going outside of their ranges.

[carl0s]

It sounds as though the engine is experiencing unexpected high vacuum. Perhaps the throttle plate is closing earlier than it should? It always occurs on decelleration when the gearbox changes down. Freeze frame shows the fault occurs at 720rpm.

[carl0s]

Strangely it hasn't done it since about Friday. It was doing it every 15 minutes before. It always used to be fine too.

All I have changed is lowered the regulator's pressure from MAX (about 1.65bar), to 1.5bar.

Is there any way that having the regulator on max, might cause something weird to go on with the MAP / compensation connection between the regulator and the manifold? Even when the car is running on petrol? Could the valve in the regulator somehow raise/increase the manifold vacuum? I can't see how myself. Surely just a coincidence.

[rossko]

(no. you're not talking to yourself!)
I can't think of any connection, short of a catastrophic gas leak that would simply stop everything.

Positive its not/wasn't something really silly, like added MAP nipple fouling internal throttle disc, or a foreign body?

[carl0s]

lol, thanks I do have a tendency to over-post anyway but it's good to note things down for the future anyway.

I'm not sure, but it's not an added nipple because I don't have one. I tee'd in to the link between the manifold and the fuel tank vent valve.

weird how it's stopped doing it now though after doing it almost every 15 minutes over a few day period. All I've changed is the gas pressure (but it did it on petrol too!), and a rear hub & driveshaft...

[carl0s]

I think I have a practical solution to the original query (car requires ~20% more fuel in non-valvetronic idle, i.e. under manifold vacuum, vs. valvetronic idle (atmosphere in manifold)).

1. Do not connect compensation pipe from manifold to regulator.
2. Do not connect MAP sensor to manifold.

Simples!

This way there will be more effective gas pressure under manifold vacuum, and the LPG ECU will not compensate for it. Actually I suppose you could just do (2), and let the ECU compensate for the regulator's compensation pressure drop. Not sure whether it makes any difference. I suppose it's easier to just leave off any & all MAP connections in the first place.

This also happens to be what the KME Nevo does when you use the PPO (MAF interface). It ignores the MAP sensor, and doesn't factor actual MAP vacuum into the gas pressure calculation.