As the title says... I have a few oddities, which I wouldn't mind a couple of other opinions on!

My Zavoli LPG ECU has decided to take a dive to the point where it wasn't putting out 5V properly (I made mention of that in my engine rebuild thread). I traced a bit in the ECU, and found that a couple of the capacitors were pretty dead - bulging on the side and things like that. I've got replacements to fit, but found that even with no load on it, the 5V supply to the sensors etc was pulsing 0V-5V-0V..... found this was also the case on the 12V input side of the 5V reg, so think I have a bit more digging to do to see if I can fix it..

However - I decided that the best way forward was to get a brand new King LPG ECU, it's the later type 'D' which has OBD connection if I want (not wired in yet). Bought said ECU, and it came with a wiring loom which I've kept in the box, as it's identical to what's on the vehicle already - other than the OBD wiring.

Got it plugged in, connected up and on the laptop - configured everything (set injectors to Hana 1.9 ohm! :) ) and all good.. or so I thought.
Got it to autocalibrate, but have now noticed that the temperature sensors for both the gas rail, and the reducer are showing 110 degrees C permanently, no matter what setting I set them to in the ECU (Mine are 10K sensors from the old Zavoli ECU).

Any thoughts as to why this is showing like this? Also noticed that a couple of times the LPG switch hasn't lit up when the ignition has been in position II (same as the Zavoli ECU was starting to do) but comes on as soon as the engine fires up...

My next step is going to be checking the wiring to the temp sensors to see if there is a break/short in one of them that's causing the erroneous reading.
The RR runs OK on gas - good pick up and happy to cruise, but it is pulling the fuel trims a bit (needs fine tuning) but then I also wonder if the temp sensor values being out has affected the calibration when I ran that (I don't recall them being that far out when I calibrated it - but could be wrong - I had taken the RR out for a good spin to get everything nice and warm to calibrate it with).

Any thoughts or pearls of wisdom?
The ECU is stamped as 02/2017, so a new build, has firmware 11.52 on it, and I'm connecting to it with King software version 6.2.1.0 IC

Thanks as always...
Marty

110 degrees would suggest a short. See what it reports with the sensors disconnected, it should read -40 (or at least very cold).

That's one of the thoughts I've had (but no time to check yet) as -40 is what the HEVAC will read if there's no connection on a sensor (from memory they are 10K too),.. the sensors are soldered/heatshrunk into the loom, so I'll either attack the loom with a pair of side cutters, or I'll try popping the pins from the ECU connector to see what happens.

If the readings change, then I'll put one wire back in at a time to see if there's any change.

The lambda sensors aren't hooked up at the moment, and in guessing the are no real benefits of reconnecting them?

Cheers

Not exactly a pearl of wisdom to offer, but have you checked that the pinouts on your old fitted loom are exactly the same as the one still in the box for the new one? I found a couple of bridging links at the ECU connector (loom end plug) on mine that weren't the same as the schematics.

Ok, finally had some time to do some poking about...

The ONLY difference I can find in the loom is that the ground for the sensors ends up on a different ground pin in the new loom. I thought 'aha' this might be it, until I checked those 2 pins on the ECU and they are internally connected together - as they are both ground pins - the only difference appears to be that in my installation, all the wires are joined in a splice.

Both of the temp sensors read 9.8 ish K ohms at the moment - so there's no short to ground..

I think I'm going to try powering up the new ECU on the bench with it connected to the new loom, and the laptop and then try a 10K pot on the temp sensor wire, so I can see if it is actually reading the temp change properly.

If it doesn't, then I guess I will have to get in touch with the supplier, and see if they can swap it.

I'm surprised Simon hasn't chipped in on this thread. It may well be something he's come across in the past.

Well, the update from this afternoon's state of play:

I got it hooked up on the bench and it was doing the same thing. I could not get it to give a sensor reading at all - I even tried downgrading the firmware to the previous version, but no dice.. Decided it was either something different between the Zavoli ECU (which is a 'C') and the King 'D' ECU, or it was a faulty ECU.

So hooked up the old Zavoli one again in the vehicle, and connected straight up to it - I got no sensor readings from it either with just the laptop connected, but with the ignition on, I got readings.

Decided that whilst it was in there and working that I would re-calibrate it, now I've opened the injector nozzles up - so did that. Saved config, shut down, and swapped the ECU's over to the King one again.

Turned ignition on, no sensor reading still... Started the engine... and hey presto... sensor readings.
So there has to be something different between the ECU's or their revisions... I had wondered a bit about it as when I had it on the bench I put a multimeter on the sensor wires, and would get 5V on the sensor wire (obviously a reference voltage) when I connected the ignition wire up. However, as soon as I connected to diagnostics, the 5V would disappear... which is what made me decide to try it with the engine running (wasn't able to replicate that on the bench as I guess it uses the rpm trigger wire to sense that, and I don't have an emulator for that at the moment!)

So with that figured out, I re-calibrated the King ECU as I previously had forgotten to set the temp sensors to 10K rather than the default 4K7 (which is what kicked all this off in the first place - when I looked at the config on the laptop after first setting it up).

So fingers crossed it's all sorted and will behave itself now.. I've got a run to London to do on the weekend, so will see how she runs on gas on a decent motorway run. I'm still in 2 minds as to whether to add the OBD wire.

But for now... mystery solved... AEB version 'C' ECU with Zavoli firmware will connect to the diagnostics with the ignition off, and will show sensor data with ignition on... AEB version 'D' ECU with King firmware will only connect to diagnostic with the ignition on, and only show sensor data on diagnostics when the engine is running...

I've only just noticed this thread...

Most AEB system temp sensors are the same spec except Tartarini.

King ECU with latest software and that firmware allows you to select between a couple of temp sensor specs for both reducer and vapour...

Yes the later ECUs won't show current data with ignition off. Some won't show current data if the interior switch is broke either.

If temps were incorrectly reading 110c it wouldn't throw mixture off much during most conditions but would throw mixture off a lot if temp readings were -40c. Default temp correction goes wildly negative for -40c but is maybe only +8% for 110c.

I wouldn't connect to OBD.

SImon

Thanks for the input Simon.

The default setting in the King ECU was 4K7 for the sensors, but my Zavoli ones were 10K, and had forgotten to change that setting (as the Zavoli firmware on my old ECU has no options for sensors).

I've swapped the changeover switch aswell just to be sure, (did it when I swapped the ECU - partly because the old one won't stick back down, and this one has nice fresh double sided tape on it!) and the temp readings now look correct with the engine running.

Is it normal for the changeover switch to not light up until the engine is fired up too on the later ECU's - again, the Zavoli one would power up and light up the changeover switch with the ignition on, whereas the King one doesn't.. I'm getting used to it - but still look at it sometimes and think it should be on like it used to be!

Seems to run OK on gas at the moment, which is good. The map readings are now a lot lower with 2.5mm nozzles and the pressure set at 1.4bar, which is about what it stabilises at when running on gas.

It's got a run to London to do over the weekend, so will see how it runs on LPG, and I'll keep an eye on the fuel trims to see how far it pulls them over a long journey, which will give me an indication of how much tweaking I have to do to the map.

Noted on the OBD..

Cheers,
Marty

What temp readings are you getting now? I think the default 4.7K setting should work.

It is normal for the changeover switch not to light the first few times you turn ignition on if the ECU hasn't been calibrated, it usually stays lit while ignition is on after then if firmware is new enough to have the rpm box unticked by default, yours should be like this. Another strange aspect of AEB ECUs is that there can be a bit of misfiring on petrol the first time the engine is started with a new ECU fitted that hasn't been powered up before, can only put this down to ECU not making the connection across the petrol injector loom break at that point for some reason.

What nozzles did you fit in the Hana's? Hana's selected in firmware? Using a Zavoli reducer at 1.4 bar?

Simon

I'll have a look the next time I hook the laptop up.

The switch still isn't lighting up until the engine is running, and it's all configured and calibrated.. did the run to London today OK, but going to have a look at the petrol fuel trims to see what they look like after the 80 odd miles on gas to see how they look.

I know my sensors are 10K - so set the ECU to that at the moment, will read the temps with it running when I next have the laptop out there.

Config has LPG injectors set up as Hana 1.9 ohm, with Zavoli Zeta/S reducer at 1.49 bar. Nozzles are Hana '1 stripe' 2.4mm, but I ran a 2.5mm drill bit through them before fitting as you suggested in my other thread on the engine rebuild.

The 'revolution signal type' box is checked, and set to Standard, 2 coils. This is how it was set up on the Zavoli system - so I have just copied these settings over, as the brown RPM wire is obviously connected somewhere!

I'll post up what the fuel trims look like when I've read them.. they were sitting pretty flat on petrol before I did the run today on LPG, so will be interesting to see what they've done.

Regards,
Martin

Won't hurt to try the 4k7 setting mate..

All coming back to me now about your ex Zavoli P38 install on the other thread - I sometimes forget due to seeing so many cars and talking LPG so much!

Next question following post above might have been how does ginj compare to pinj when you put your foot down and mixture is suitably rich?

Simon

OK, I'll give it a shot the next time I hook the laptop up.

I've hooked up my Nanocom diagnostics and read the fuel trims from the engine ECU, after the run down to London today, and they look as follows:
Trim A: 1.00
Trim B: 1.00
Idle Trim A: 1.50
Idle Trim B: 1.49
Multi Trim A: 1.07
Multi Trim B: 1.08

The Multi-Trim (I presume long term trim in normal speak!) is OK to me - as on petrol it drifts a bit too..
The Idle Trim is the bit that will need looking at - 1.5 is about maxed out - and I had rich running codes stored in the ECU, so something to look at when I'm back home next week.

I'll have to set the laptop up and run it to record some data when actually driving when I'm back home - but from a rudimentary feeling of right foot and accelerator away from traffic lights, or to speed up to overtake on the motorway, it doesn't feel lacking in power on gas.

I've seen that the King software has a 'scope' option to record the data - so will put the USB extension on, and get it recording pinj and ginj times under varying conditions. I'll see if I can find a route near home where I can do 2 runs - one on petrol for a base reading and one on gas to compare... or should I just do some switches between petrol/gas on one journey?

I'll post up what I find...

Cheers,
Marty

I'm not at all familiar with Nanocom but going on what you've said it's fuel trim readings seem similar in concept to Jap import stuff where 100% Alpha would equate to Euro ltft=stft = 0, one of your trims at 1.50 might equate to Alpha 150% or ltft=sftf = 50%. I'm converting a lot of Jap import stuff these days!

Using the scope in software would have to be done at the same time as monitoring trims, because during proper closed loop operation ginj will remain the same when you adjust calibration while pinj figures change.. Suppose If trims were highly negative, then for a given pressure and nozzle size ginj might be suitably higher than pinj but only because of the negative trims, if calibration were then adjusted to bring trims close to zero ginj might be below pinj... Got to compare ginj with pinj when trims are about right to know if pressure and nozzle combination are about right, though tbh it won't make a great deal of difference on a P38 if ginj is a bit below pinj, all it will do is momentarily slightly richen mixture during positive throttle movements. The switching method can be used instead of the trims method, I use a combination of trims monitoring and switching method during the same run.

Simon

Looking in the instructions for the Nanocom, to remind myself about the trims, it says they are a ratio of between 0:1 and 2:1 with the ideal being at 1:1 obviously.

The idle trims are actually the long term fuel trims according to the documentation, which in my understanding the trim of 1.5 means the petrol ECU is detecting lean running, so is increasing fuel injection times to compensate.

Effectively the map that it's autocalibrated to for the LPG is a bit lean at the moment, so the petrol ECU has been increasing its trims.

I'm going to tweak the map a bit before my journey today (another 80+ miles) and reset the adaptive values to flatten the petrol trims again, and see what effect it has. I'm saving the configuration every time I change something, so I can revert if required.

So, I'm still getting strange readings on the LTFT's and after a 30 odd mile drive last night (all on LPG), it took 3 or 4 turns of the key to get it to start.

Did the run back home on LPG aswell, and then went to look at it this morning, so hooked up the Nanocom, and found this:

So the 'Idle trim' which the Nanocom manual shows as being the LTFT is pretty much maxed out, and the injector time is stupid (which is probably why it struggled to start). I tried starting it and nothing - just turned over (no faults in the engine ECU list either).. reset the adaptive values, and then it started up - albeit a bit lumpy (probably stinking rich if it had just been pumping fuel in before with a 56ms injector time!

I hooked up my ELM bluetooth unit and got torque on the sat nav unit (android!) so I could watch the fuel trims on each bank in real time. Fired it up on petrol, with adaptive values reset and got it so it warmed up and idling fairly happily. Flicked over to gas, and both fuel trims went vastly positive (pinj times on the LPG software also shortened) - so gave that a bit of a tweak, until the fuel trims stabilised more, and the pinj times came back to where they were on petrol.

Did that a couple of times at different rpms (whilst stationary at the moment) and at the 3 rev ranges I tried, it was all fairly rich.

I guess the next thing is to try and find a piece of road I can cruise on a bit, and get the laptop set up in the car so I can monitor the trims and the inj times and make adjustments with the engine under load.

Runs happily on petrol, and the journey last night on LPG was great when it was running - but it seems to be messing with the fuel trims something crazy!

If you're happy it runs properly on petrol, lambda flicking well and going rich during high load open loop operation, all you've got to do is match pinj while running on gas to pinj when running on petrol - then fuel trims will be the same under all circumstances as they are on petrol. P38s are particularly easy to achieve this on even without looking at fuel trims, but fuel trims can be used to confirm.

There are a few schools of though on calibration - 1 calibrate so trims are close to zero on LPG, but I'd only do that if trims were close to zero on petrol or on specific models of vehicle such as a supercharged L322's / 2 aim for trims to be the same as on petrol (which is what I just suggested above, and I'd only do that usually if petrol trims were reasonable i.e. not sailing close to being minned/maxed out), / 3 aim for trims somewhere between zero and petrol trims.

Many older vehicles only have one ltft, as vehicles get newer they tend to have more and more ltft;s each covering a specific range of rpm versus airflow (or map or less commonly TPS). Some have a different set of trims for different coolant temperatures. On some there is an underlying trim for idle before ltft is even applied which is learned on the basis of something like average trim for off-idle conditions (e.g. BMW). BMW's even have a trim for each individual cylinder called 'smooth running.value'. Out of all the Rover V8 engine'd vehicles I've worked on regards LPG, as memory serves only the late model Rover engine'd D|isco's with EGR have fuel trims which are much like the norm but they are all a low demand engine and very forgiving regards fuel trims and LPG calibration.

It is quite usual for installers such as myself, Dai and Andy to calibrate a system to be close to petrol trims, steering slightly to the side of zero, until the engine goes open loop and then our maps go a bit leaner.if we want more power on an engine that runs enough ignition advance because in this case there is no advantage to high load enrichment when running on LPG which would only start to impart on volumetric efficiency (especially on turbo vehicles with chipped woefully rich maps), or the same mixture if not quite enough advance (for LPG) means running a lean mixture would degrade power due to slower burn and overheat the valves, or a bit richer mixture if it is an engine with soft valves to protect the valves.

The default temp compensation values in AEB are exaggerated, you only really want a range of about 12% from 0c to 60c with zero comp at around 40c. Hardly any systems except AEB compensate for reducer temperature by default and on vehicles with AEB where under bonnet heat doesn't affect gas vapour temp readings I zero all reducer temp related compensation... but it can be handy on installs where under bonnet heat does incorrectly affect vapour temp readings.

It can be best not to obsess about trims on P38s, far better to get ginj suitably above pinj by selecting suitable pressure and nozzle size. Smaller nozzle with higher pressure, and the opposite, can do much the same thing when you get away from idle but at idle things work a bit differently, and Boyle's law works perfectly for an ideal gas but LPG apparently isn't an ideal gas, and you don't need as much relative pressure when manifold pressure is higher.. .

Got a P38 coming in for at least a new ECU on Thursday, and since it currently has a Piro fitted it will also at least need a new loom... but I'll have it calibrated about as good as it could within about 10 mins of changing components without using a Nanocom ;-) I'v e just converted about 10 Nissan Elgrands almost back to back, mostly 3.5L E51's but some 3.3L E50's, my first guess on pressure, shape of map and map levels was almost spot on. Nearly finished conversion of an Elgrand E51 2.5L, and dare say my first guess on pressure and map will be just about spot on for that too! ;-)

Simon

It is always easier when you've done it many times...

It's the same for me when it comes to unlocking a BECM, it takes me about 15-20 mins to get one out of the vehicle, but some people are scared to even attempt it, because it's the big box of voodoo magic... or I can get a door latch changed in about half an hour, including testing and putting the door trim back on.

I understand the theory of LPG, the fact it has lower calorific value to petrol, so needs the 'fiddle factor' to adjust the injection timings to suit.

I did the first 1000 miles on the new engine solely on petrol a) to bed it in, and b) to see what the fuel trims looked like after that distance, and also varying conditions or road, traffic, engine loads, and of course, my right foot!

I noticed than when running on petrol the fuel trims seemed to bounce around a bit +ve/-ve either side of 0 anyway, so my aim when I was tweaking the LPG just parked up was to get the pinj times looking close to what they were when running on petrol for the same given rpm, when I did this, I had the trims bouncing about the same as they were on petrol - so was good enough for me. I'm a perfectionist - I want it all to be as close to perfect as possible - but I'm also a realist, and I know I'm never going to acheive a bang on 0% fuel trim, so in my mind, as long as the LPG is following the petrol close enough and not making the petrol ECU over compensate (which then obviously seems to be causing my random starting issues at the moment as the petrol ECU trims are just being shoved to one extreme of the other).

I'll go through tomorrow and flatten out the temperature corrections - the old Zavoli ECU doesn't have them, and the underbonnet heat on P38's seems to be pretty consistent once it's up to running temp.

I don't mind it being a bit richer under acceleration, as it gives it a bit more poke (obviously don't want to drown it in fuel though) - but there's obviously something going on at the moment which is causing one of the fuel trims to be pulled way off of normal, and I can only think that this is my cause for it now not starting properly sometimes. It's only started since trying to get the LPG installed and working again, and when I run it on petrol for any length of time, it seems to sort itself out again, and will start/run happily.

Something more to look at tomorrow...

Marty

Got to be easier when you've done it many times, can also pick up a lot of insight into what's going on. I don't much get involved in fixing any others aspects of cars other than LPG installs and engine management / fuel injection problems but concentrating on those aspects will have concentrated my skills on those aspects. I'm not bad at general mechanical stuff but my mates who run garages in the area are far more likely to know all the ins and outs of most usual type mechanical jobs for a random model of vehicle than I am... I recently ummed and ahhed whether to change my son's Mondeo clutch myself (son helping) or to stick to what I do best and let mates garage change his clutch. We bought all the bits, or so we thought, then decided on letting mates garage fit them, mostly because I was too busy with LPG work. The garage said 'Ooh I'd really rather change the hydraulic release arm at the same time, when the clutch is nearly worn out the arm can in effect get stuck at the full extent of it's movement, or even if it isn't stuck it may fail soon due to being worked at the full extent of movement. Mondeo's subframes have to be lower to change the clutch and the bolts are often seized, it may involve some cutting and welding of the subframe'. We bought the arm and had him fit it at same time as the clutch, turned out he did have to cut and weld the subframe - I could still have done it all but was pleased we let him do it. If my mate at the garage were to fit his own LPG system I doubt he'd second guess me but if he did I would liken the situation to role reversal but where I decided to crack on and not change the hydraulic clutch arm and not anticipate the problems with subframe... and I could be oblivious to the pending failure of the arm until it failed.

Fuel trims will always bounce about just + and - of 0, they're supposed to do that as management never reaches a steady state, it is designed to flick lambda slightly rich and slightly lean constantly, on engines that meet usual OBD2 standards with narrow band probes the stft will always flick at least around +-0.8%. Calorific value of fuel doesn't effect calibration, if the engine doesn't make as much power due to lower cv the driver simply puts his foot down more so pinj increases to match... stochiometric ratio does effect calibration but would result in lower ginj than pinj, though the difference between 14.7:1 and 15.4:1 is only 5% and petrol trims could cope with this easily. The difference between petrol injector and LPG injector properties is the biggest reason we need an LPG ECU at all, followed by need to more accurately account for fluctuating density of fuel (pressure and temperature, due to LPG being a vapour not a liquid, which again is why we can't use petrol injectors for vapour LPG). In some respects it could all be easier for direct injection engines - supply the di injectors with LPG at the same pressure as petrol and an LPG ECU isn't strictly necessary because no calibration is necessary.

Engines running on petrol are far more likely to deliver more power with a rich mixture than engines running on LPG, you don't get the charge cooling (charge density increasing) effect from injecting vapour LPG and the LPG vapour takes up more of the volume of airlfow, so you lose volumetric efficiency. It is safer for the engine to go too rich rather than too lean but max power will usually occur a bit leaner than on petrol, implications of which will effect fuel consumption while booting it. Closed loop single point LPG mixer systems generally aim for about 0.75v lambda 4000rpm+ with wide throttle where petrol lambda reading would max out at 0.9v because petrol mixture would be a lot richer than the probe could measure. Some of the NecamKoltec engines continue to aim for stochiometric (0.5v) lambda even during flat out conditions and f they make any less power on LPG because of this it is negligible.

All AEB ECU's have temperature correction which can be adjusted in software settings if software allows it. To access those settings you need the correct dongle, dongles were originally parallel port types but later on USB types became available. I have a range of dongles for different AEB systems, they also allow you to make changes to the full map in pre V6 versions of software, set the ECU to start on gas everytime even with a cold engine (in case of monofuel i.e. no petrol tank even fitted), adjust lambda emulation waveform on the grey lambda wires (aka AEB175s), etc etc. You can probably do more with V5 software than with V6 if you have the correct dongle. I have been known to bring an old desktop PC outside with a parallel port dongle attached to access those settings on an older ECU, even run the desktop and monitor from an invertor to calibrate lol.

If LPG calibration is rich petrol trims will of course go negative, and then if engine management applies trims during cranking it can lead to problematic starting.

Simon

Above I mentioned the P38 I'm working on at the moment. No problem regards what I'm doing with the LPG... but sat in it in the yard with my foot on the brakes there was a sudden noise from under the bonnet which then continued, sounded like a PAS pump low on fluid. Got out to check and there's a pool of brake fluid near osr wheel! Noise would be ABS unit? I was late picking the missus up after tea from her mums so haven't even been under it yet to see where the leak is from, hopefully the flex pipe. Topped up the brake fluid reservoir and the noise went as expected. I've previously read comments from I think Gilbert about just turning the ignition on and ABS unit will prime the rear brakes, comments welcome in case it lands on me to sort this aspect out! How soon will ABS unit fail with lack of fluid to that corner, say should the owner decide to risk driving it away to repair elsewhere? How much of a bugger is routing new copper pipe in case it's the main pipe that's failed? Sods law this is gonna land on me over yet another weekend!

I realise I should maybe have started a new thread or posted this on a more relevant thread, might do that myself or won't mind if mods do it.

Simon