Injectors Fully Open

[jacckk]

At 100% load and high RPM on LPG I'm getting a rev limiter type feeling. The car is a Range Rover P38 4.6 with an LPGTech system (Tomasetto Antartic reducer, Hana Injectors). The lpgtech software reports "Injectors fully open" for all 8 cylinders. It happens around 4700-5000rpm.

I've logged some data using the software which is attached. Is just a case of the map being too lean at high rpm/load or something else?

Thanks

[Brian_H]

It might help to confirm what jets (Nozzles) your using?

[Gilbertd]

Nozzles too small or gas pressure too low. Petrol injector times are 24.7mS while gas times are 45.9mS giving a multiplier of 1.85 when it wants to be between 1.2 and 1.5.

[jacckk]

Nozzles are 1.9mm I believe. I also have the 2.1mm version nozzles that I could swap to.

Is the gas and petrol not showing different times due to the app not sampling at the same time? If I look at one of the G1 peaks I get 49.9ms. Here's another picture:

[Gilbertd]

That's showing the length of time the injectors are open for. Petrol injectors open for as long as is needed for complete combustion and the LPG ECU adds a few milliseconds to that pulse length to get the correct duration for complete combustion on gas. That adding needs to be 1.2 1.5 times the petrol injector time. If too low then the clunky LPG injectors will need to be open for a shorter time than they are physically capable of achieving so you'll get lumpy running at idle. If too high then the time between 2 revolutions (and one firing cycle) is less than the injectors are being told they need to be open for. At 5,000 rpm and 50mS opening, they are never going to have time to close before they need to open again so are open permanently, hence the error. Fitting larger nozzles or increasing the gas pressure will reduce the length of time the injector needs to be open to get the required amount of gas in so they won't be maxed out as they are at the moment. But, it may need the ECU calibration to be done again once changed have been made unless it is connected to OBD to adjust itself from that.

According to this chart http://www.lpgtrade.com/valtek-injector ... -diameter/ at 1.2 bar as your graphs show, even 2.1mm are a touch on the small side. Depending on whether you have GEMS or Thor, you are looking at 26 or 28 bhp per cylinder so 2.3mm would be nearer the mark.

[LPGC]

Since you say you can swap nozzles on your Hana injectors you must have Hana Gold injectors (that have interchangeable nozzles). You don't need to buy new pre-sized Hana nozzles, you could remove yours and drill them out.

You have an Antarctic reducer, these are adjustable to unusually high pressures but even at it's near max pressure your 1.9 / 2.1 nozzles will be too small. The antarctic reducers don't seem to live long if set to pressure much above 1.6bar and your injectors will work better at more average pressures anyway. Would advise drilling nozzles to 2.5 and in increasing pressure to 1.5bar for now, or if you leave pressure at 1.1 bar drill nozzles to 2.7mm (or 3mm if you don't have a 2.7mm bit).

Following nozzle swap and/or pressure change a complete re-calibration will be necessary. If it were me I'd reset the ECU and switch it to 'standard mode' rather than the 'tech mode' default setting, this way you get a lot more control over calibration and calibration won't mess itself up over time if a fault such as a duff lambda probe occurs.

Simon

[jacckk]

I've left this for a bit and just lived with it as I've been busy with other things but now I'd like to fix it.

So now I've got 2.4mm nozzles, 1.2bar gas pressure and I've got the normal closed loop range pretty close on LPG. I'm still getting injectors fully open.

Here's my LPG map:



I haven't increased the gas pressure any further as in the normal closed loop range my multiplier according to the map is already 0.86. That means that the pressure is already too high/nozzles too big?

Here's another LPG software scope trace of what's happening which I have a few questions about:

• At the start of the acceleration in the middle of the trace why is the g.inj time oscillating all over the place? The p.inj time doesn't show the same pattern.
• Gas pressure is increasing up to around 1.54bar and then oscillating around between 1.15 and 1.54. As per my last thread I think the reducer is getting on a bit. Is that normal or could it be part of the issues?
• The lambda reading is fully rich (reducing the map slightly leads to it leaning off in places). That suggests to me my current map is roughly in the right place?

I can send the trace file to load into the LPG tech software if anyone wants to be able to look around in it and see the numbers with a bit more detail.

Thanks.

[LPGC]

Seems you already had/have 1.5 bar pressure anyway, and reducer pressure is stable regardless of load - both points are good.

For a system on a normally aspirated engine with a reducer that is pressure stable over all engine loads a multiplier of 0.86 at highest load would be a bit low even with fairly poor injectors and reducer that isn't very pressure stable. But you say reference pressure is 1.2 bar and we can see from the chart that actual pressure is fairly constant at 1.5 bar, so you have a constant over-pressure condition that the ECU will compensate for, meaning that in practice your multiplier of 0.86 will be even lower (and worse) at around 0.69 which is too low. If we look at the figures to the right of the chart, pinj for bank1 (P1) = 22.81 while ginj for bank1 (G1) = 16.5, If we divide 16.5 by 22.81 we get 0.72 which is very close the to 0.69 I just reckoned. Here's a problem with 2.4 nozzles, 1.5 bar and multiplier of 0.69... I previously suggested 2.5 nozzles and 1.5 bar assuming pressure stability and a multiplier of near 1 but you're going with 1.5 bar and multiplier of 0.69, so I expect too lean. Hopefully not too lean as to cause *misfires*.

The problem and reason for the error will occur at the point in your chart just before where P1 peaks (the red line labelled P1...). At this point P1 shoots very high almost off the scale, this will mean that the LPG Tech ECU looks at what your multiplier is set to for the 'very high off the scale pinj reading', multiplies the P1 by your multiplier (map) for such pinj and calculates that for such a high pinj reading it should pulse LPG injectors for a duration that would't allow enough time for them to close and be ready for the next injection/pulse cycle - this is what triggers the error. However, it is very unlikely that Pinj is really as high as the readings/chart at the peak suggest. It is far more likely that the LPGTech ECU is reading Pinj incorrectly where the chart suggest Pinj peaks. The incorrect readings are your main issue because an ECU that does not even meter pinj correctly doesn't stand much chance of multiplying pinj (according to installers map) to arrive at the correct ginj. We can make hardware changes such as adjusting nozzles/pressure so that if the condition/fault arises at full load fuelling will remain somewhere in the ball park but we will never be able to set the system up to provide very accurate fuelling under such conditions. If we read between the lines of points I just wrote here, you won't have set a mutliplier for a pinj of (say) 40ms, more likely the multiplier for 40ms wouldn't be far different from your multiplier for the (usual) 16ms.

A point you should concentrate on is to try to get the LPGTech ECU to meter pinj properly under all conditions no super high 'spikes' in the pinj line on the chart. Extra injection filtering doesn't work properly with LPGTech and P38s don't do extra injection anyway so disable extra injection filtering. How did you setup rpm detection (and did you connect the rpm wire to anything - Rpm detection can be hit and miss with LPGTech)? Swapping channels (engine cylinder numbers versus LPG input/output channels) shouldn't make any difference but can with LPG Tech (and a few other ECUs). If the engine has a *misfire* at all under any conditions, this can lead to erratic pinj readings on LPG Tech and a few other system ECU's (I have my suspicions why this happens but not need to go into those reasons now).

If despite trying all points above you cannot get around the issue of the ECU reading pinj incorrectly under all conditions, and if you don't want to swap to a different make/type ECU, then your best bet is to ensure that (and this only works because the issue only occurs at high/full engine load at medium/high rpm) when gas injectors are open constant duty cycle mixture remains near correct (bit rich since issue occurs at high/full engine load). I this respect - On the positive side I expect 1.5 bar and 2.4 nozzles will give near correct mixture at WOT high rpm anyway, on the negative side the incorrect pinj reading is present at medium rpm too so I expect mixture will be too rich with the fault condition at medium rpms and verge toward lean at higher rpm.

Further to the last bit, with other ECUs you might get a bit more leeway for combinations of nozzles and pressure that flow too much gas (multipliers much lower than 1) without inflicting drive-ability issues but with LPGTech with mutlipliers less than 1 on the first occasion where an LPG injector should pulse for less duration than the petrol injector it seems to always pulse for at least same duration as the petrol injector, which can cause fuelling and driveability problems. But if you have any drive-ability problems I expect they'll be mostly near flat out at medium rpms?

When we're talking multiplier, some ECU's (firmware and software) give you just the straight multiplier, some give you the multiplier when gas injector latency has been factored in, some give you both. For someone who really knows what they're doing the straight multiplier gives the best representation of the real story, what's really happening. Injector latency figures are provided by injector manufacturers but these are usually incorrect. As memory serves, LPG Tech attempts to factor in latency - the shape of your multiplier map would look very different if it didn't . I could do the maths but even without doing the maths I can say the shape of your map looks wrong now and would look still look wrong after doing the maths.

One other important point... We're talking about multiplier of 0.86 at high engine load and you;'re saying it's too low. But who/what set multiplier to 0.86 at high load? No autocal can correctly (unless due to luck really) set multiplier/mixture for open loop operation and flat out the engine will be running in open loop. Have you checked mixture under these conditions? Lack of power is usually (not always) due to lean mixture, misfires usually (not always) due to to too rich, hesitation can be either and other issues, other issues are always about momentary lean/rich mixture (perhaps just one cylinder bang) but other issues are not always due to anything as simple as the multiplier or pressure/nozzle size. Other issues can involve timing (start of ginj pulse compared to pinj pulse), extra injection filtering, acceleration enrichment, miss-match of peak and hold electrical current and spec of injectors, any aspect leading an ECU to use incorrect data when doing it's sums, all points in this 'other issues ' list boil down to the ECU or lack of installers knowledge of workarounds for the ECU.

After writing the above I just fired up LPGTech software, I was interested to see if I could suggest a multiplier of maybe 0.5 for a pinj of 40ms but it seems LPGTech is only configurable for pinj's up to 20ms... which is usually more than enough for anything much less than a big block V8 with low rpm ceiling. But nevertheless rules out
what might have been a get-around for too rich at high load medium rpms.

Anyway.. By 'anyone' I think you might mean me? Think I've already got enough info to diagnose your problem and advise from here mate.

Simon.

[jacckk]

Gas pressure is 1.2bar during normal driving conditions, it's only during full throttle that it seems to jump up to 1.5bar. It's hard to tell from that graph as the scale is so small.

I'm not sure if it helps/adds anything but I went back out and took some traces again of the p.inj times with full throttle acceleration on petrol.



I believe the ECU is connected to one of the signal wires to the ignition coils for the RPM. I would have to double check if you think it's something to do with that. The speed signal looks like it's reading correctly on the trace at all times.

Settings for reference:


Any suggestions on what else to look at to fix the p.inj readings? I don't really have any ideas.

[LPGC]

It shouldn't be anything to do with rpm... but even though rpm reads correctly shifting the source of rpm input could effect a cure. Intuitively it shouldn't I know, but it sometimes does. ECUs do a lot more in the background besides the info they present on any 'trace graph', if you want the real story you've sometimes got to use a remote oscilloscope but at least be open minded to the fact an ECU's software doesn't necessarily report what's happening at signal level very accurately.. Don't entirely trust any ECU generated 'trace graph'.

Also try adjusting the rpm drop out time and rpm signal voltage.

Pretty much saying here what I said above, swapping input/output channels would be along the same lines.

But look at your last trace for P1... 49.9ms at 5250 rpm is not possible! Hang on, Gems or Bosch P38? You have injection type set as sequential which might not be correct if Gems. If this is the case, either swapping injection type to semi-sequential (or group injection or similar) or disconnecting the rpm wire and setting rpm source as 'none or injectors' perhaps with a corresponding change to 'cylinders per coil' could prevent the silly pinj reading and thus prevent the attempted over-long ginj pulses and thus prevent the error. But if this is the case you may need a full recal and perhaps with bigger nozzle size.

in fact its worth saying that regardless of what's going on at the deeper end of tech, some aspects of which I've touched on, the most likely reason you're getting the 'holding back' symptom is lean mixture, which I would might expect with your nozzle size and calibration regardless of any 'injectors full open' errors that might have been triggered on a momentary basis where in that moment mixture might actually have been rich anyway. In other words, the injectors full open error and holding back may not be at the same time, in which case injectors full open would be reported due to the ECU sometimes incorrectly measuring 49ms as pinj but the holding back would be entirely due to your map/multiplier being too low for the pressure and nozzle size combination. Those two situations could be distinct and you probably wouldn't notice any issue when the 'injectors full open error' was triggered because this might only cause the engine to run a little rich but you would notice when your map caused the engine to run very lean at high rpm high loads when rather lean (holding back). None of this disagrees with anything I've said above.

Bed time ... Worked today, finishing conversion of an old XJ8 tomorrow, another car coming for conversion on Monday and another on Wednesday, a couple of cars converted elsewhere to put right on Friday, they'll all run perfect when they leave me. Could sort yours out in 20 mins lol

Simon

[jacckk]

It's a GEMS P38. I chose sequential because that is what the LPGTech software detected it as. Is there an easy way to tell if it is sequential or semi-sequential?

Yes I agree about the holding back and fully open not being the same issue. I was hoping to get rid of the error then I can properly map it again.

Thank you for your help.

[LPGC]

Disregarding the very long 'spike' pinj durations measured for the moment, a pinj pulse that really does go as long as 16ms will usually point to sequential because injectors that are batch/group fired need to pulse twice as often as with sequential injection (every rotation of the crank with batch firing, every second rotation of the crank with sequential). 20Ms is the available window for a sequential pulse at 6000rpm and/so 10ms is the available window for a group pulse at 6000rpm. We might see 16ms on some full group engines but they'll have a very low red line of about 3000rpm so would likely be something like an old-skool big block torque monster V8 not the V8 in your P38 which will rev to 6000rpm.

If the engine is batch injection and the LPG ECU detects rpm from frequency of injection pulses it may show double the correct rpm.
If the LPG ECU is setup to detect rpm from a cam sensor or ignition the rpm reading will be correct regardless of whether the engine is sequential or group injection (as long as the rpm divider / number of cylinders per coil are properly set up) and regardless of sequential or group injection set in software. But in this case if we set sequential injection type in software when really the engine is group injection some ECUs will add together the total petrol injector pulse period of two petrol injection pulses for the same cylinder and the longer (than batch pulse length) pinj duration reading on screen can give the impression the engine is sequential even if it is batch. The engine is likely to run much better on LPG than it would on petrol with the same type of issue (not that you'd have any easy way of changing the petrol ECU to swap from batch to sequential or vice/versa) because unlike petrol LPG can sit in the manifold waiting for the relevant cylinder's intake valve to open without wetting the manifold etc. but in certain conditions the 'adding up two pinj pulses' aspect can go wrong in all sorts of ways such as adding up total pulse length of 4 pulses instead of 2 or by not noticing the off time between injection pulses so including the off time as injector on time -- If an engine has 1 ignition coil per cylinder and runs sequential injection it will expect 1 injection pulse per ignition coil pulse, or 1 coil per 2 cylinders with sequential and it will expect 2 ignition coil pulses per injector pulse, or 1 coil per 2 cylinders with batch and it will expect one ignition coil pulse per injector pulse but now the injector pulse happens at the same time as other injector injector pulse(s). From this can see that if an LPG ECU uses rpm to define a window of time in which to measure length of an injection pulse, getting the sequential/semi/group setting wrong can lead to all sorts of problems even if the engine seems to run right on LPG under most conditions.

Gems may be sequential but the earlier Lucas stuff was group. Even if it is sequential and you made (or ECU itself made) all the right settings in software, if you're getting a reading of 49ms for pinj something is going wrong (similar to last points above) leading to the LPG ECU not measuring petrol injector duration properly. Already mentioned some points that can lead to incorrect petrol injector duration readings (misfires etc) and sometimes shifting rpm source can prevent incorrect petrol injector duration readings - If you force the LPG ECU to detect rpm from injection pulses it will use the frequency of injection pulses to define available injection pulse window as perhaps opposed to frequency of rpm pulses to define available pinj window. There are ad/disadvantages to both but they are only relevant on some types of engine, shouldn't make any difference on a P38 but just might.

Simon

[jacckk]

I tried the semi-sequential setting and changing the RPM to be read from injection impulses but that didn't make a difference.

I then swapped the GEMS ECU for a spare one I had and this one no longer jumps up to 50ms. I did some quick calibration on it and did a few full throttle runs on petrol and LPG and everything seemed good. Very strange behaviour...

[LPGC]

Well that's good news but leaves a few questions!
Maybe the first Gems ECU had been chipped to provide richer mixture at high loads, in which case would hold injectors open for longer, perhaps even constantly under some conditions at high rpm. But could be quite a few reasons, could be something as simple as a poor earth/live feed or something else that wouldn't usually be an issue such as failing none essential components (like capacitors / diodes that prevent inducted voltage spikes when injectors are powered off). But I've seen impossible pinj readings with LPGTech (and other ECUs) many times and there's usually a workaround without swapping the petrol ECU.

Simon

[jacckk]

Unfortunately the same thing happens with the other ECU now so it seems like the old one isn't at fault. Looks like I'm going to be bringing it to you Simon...

[LPGC]

Unfortunately the same thing happens with the other ECU now so it seems like the old one isn't at fault. Looks like I'm going to be bringing it to you Simon...

Yep, I'll see you on Monday and will sort it Jacckk.

Simon.