Thanks for that! I've left the car at my folks tonight, Will have a look at it tomorrow.

I've got a link box for my TE20 to fetch from North Wales this week so it would be nice to take the RR for it's first proper run in my ownership. It'd be even nicer if I could do it at 52p/litre rather then £1.17/litre ha ha.

David.

52 a litre,, the bastards round here charge at least 10-15p more,, I wait till I hit London to get mine..

Indeed, the local Asda are 52.7p/litre, the Morrisons in Paisley is 53.7p/litre.

David.

I found one in London, 52p a ltr, but each visit you get a penny a ltr off upto 10p a ltr, when your saving between 15-25 quid on a fill up. Makes a difference 😊

The numbers in the 'modify carb' screen are not the map, they are numbers that are added or subtracted (can enter negative numbers) to the real (underlying) map which we can't see in this version of software. The underlying map starts with a set of numbers (depending on type of injectors selected) all the numbers in the underlying map are changed by the same percentage by autocal... So numbers in the modify carb screen are likely to only be correct for the specific install for which they were entered, because autocal is unlikely to set the same numbers in the underlying map on any 2 installs (even on the same spec vehicle, even with the same spec LPG components fitted). because of variables such as petrol and LPG pressure, injector flow rates. temperatures at the time autocal was run, engine tune/wear etc. Now having said that, if we have 2 ECU's both with the same firmware, same basic settings entered in both ECU's (type of injectors and working pressure) and autocal wasn't run on either ECU then the ECUs would be interchangeable on a specific vehicle and provide exactly the same fuelling, but only because the other aspects (pressures, temps, engine itself) wouldn't change if we're talking about a specific vehicle (otherwise would expect calibration to be close but not spot on). I don't remember if autocal was run on Blueplasticsoulman's install.

We did auticalibrate mine. First time i brought it and the second time after i bought new injectors. Dead easy to do and then just a tweak of the "Modify carb".

Perhaps i used the term "Map" incorrectly. I know that no two will be the same. I was just posting mine as a point of reference incase his was a million miles away.

Comparing mine and Dhallworth's, i see that that whilst his reducer is 78degrees, his gas is only at 21.

Also, he only has one reading for T.Gas and T.Petrol. Whereas on mine i get 2 for each as it's running as 2 banks. Is that right?

Hope you didn't think I was being arsey or pedantic mate, just thought it better to mention to dhallworth in case it seemed he could just enter same numbers as on yours for instant fix.

Could be that his gas temp sensor hadn't had chance to warm up (either gas temp or picking up under-bonnet heat). At 78c most AEB ECUs (behind the scenes) add zero reducer temp compensation (or about -2% for 60c or +2% for 90c), at 28c most AEB systems behind the scenes will add -2% temp correction up to a max of about +8% at 80c gas and min of -8% for -20c gas). In theory LPG systems don't need any reducer temp compensation because it's only the temperature of the vapour that matters... Besides AEB based systems only a few systems have reducer temp compensation even available to setup as an option but on most AEB systems you can't turn off reducer temp compensation and can't make changes to reducer or gas temp correction (and if you could turn off reducer temp correction you'd probably need to modify gas temp compensation). To me it's always seemed AEB systems compensating for reducer temperature is in a way an acknowledgement that the gas temp sensors are almost as likely to pick up under bonnet heat as they are gas temperature.

He should have a separate reading for bank 1 and bank 2 like you've said. But I wouldn't expect it to make any difference to fuelling because behind the scenes the ECUs run as two banks anyway (all cylinders on a bank get the same fuelling as the front cylinder on that bank). Usually the only way setting one bank instead of 2 banks can mess things up is if the box 'anticipate the injection sequence' is also ticked. Still he should select 2 banks and then we can see if both banks read similar values.

Dhallworth phoned me today and said the pressure reading with the new reducer is 1.4 bar, which is a big improvement to the pressure reading with the old reducer! But he also said that during autocal gas injector pulse length goes down to 3.4ms and stays there... So I advised the 3.4ms will be the minimum pulse duration the firmware allows with injectors selected as Tartarini, so he should decrease physical pressure (if the reducer allows) and set working pressure in software to 1bar. If this doesn't work he should try selecting 'Matrix' injectors in software instead, because the Matrix setting at least won't impose a 3.4ms minimum pulse duration (although this would give mixed results with the Tartarini injectors which probably won't work very well below 3.4ms or even 4ms).. But now you've pointed out that he only has one bank selected and we can't see readings from the other bank, I wonder if he has crossed banks (injector break plugs crossed), which would mean one side of the engine got richer and richer while the other side got leaner and leaner (and could have led to the 3.4ms reading on the bank that was going leaner).

Thanks for the input, gents.

Simon, I reduced the pressure to 0.9bar on the reducer with the screw and then set the reducer pressure to 1bar in the software. When I did this it still jammed at 3.4ms on the auto calibration.

I then did a factory reset on the LPG ECU and reconfigured it from scratch setting the reducer pressure to 1bar and the injector type to matrix. The rest of the settings were all mirrored. I ran the AutoCalibration again and the numbers danced around for a bit longer then they did previously but the calibration process appeared to stall about 1/3 of the way through the process with the T. Gas reading stuck on 0.64ms whereas the T.Petrol was still floating around between 4.5 and 6 roughly.

After the car had been running on the drive for about an hour the reducer temperature was steady at 83 degrees but the gas temp was only reading at 31 degrees.

At this point I had a few other things to do so I had to leave it there unfortunately. I wish this was as simple as single point!

David.

Lpgc wrote:

Hope you didn't think I was being arsey or pedantic mate, just thought it better to mention to dhallworth in case it seemed he could just enter same numbers as on yours for instant fix.

Not at all bud.

Could the firmware be the problem? I know on mine it's the latest firmware that doesnt allow turning off of individual injectors. You have a spare ECU with a previous firmware that we tried and it worked.

Another thing, are the injectors to blame? When i dismantled my old ones, the pistons and springs were caput

As you've said, it may be better to have it running as 2 seperate banks so you can get a better idea of what's going on. My injectors are set to Tartarini although they are generic replacements and they're set to sequential. The reducer is at 1 bar.

I also have a -2 second bank corrector.

Dhallworth txted me, he may be calling on Friday depending on whether I get this Yank pickup converted in time.

With 0.9 bar pressure I would expect even highest flowing Tartarini injectors to need to pulse for more than 3.4ms at idle, so something is amiss. Could be injector fault(s), or banks swapped (injector break plugs swapped, setting 2 banks in software would help to determine that), LPG ECU (not so much firmware but electrical aspects such as injector driving outputs) or pressure sensor problem... or could be that autocal cannot work properly because the engine is running open loop mode (perhaps due to previous attempts at autocal / running on LPG when mixture could have stayed lean/rich for too long / vehicle issue such as failed lambda probe) - autocal relies on the engine's closed loop fuelling system working properly and may continually lean or richen the LPG tuning if that is not the case. Also, the engine needing up to 6ms pinj at idle tells a story about the engine struggling to idle properly on gas (it implies higher manifold pressure necessary due to incorrect mixture on at least some cylinders).

Well, we can finally update this issue to SOLVED!!

I visited Simon this morning and he did some digging with his laptop and software and diagnosed my ECU was faulty.

He fitted a known good Romano ECU to it and even before setting it up properly the difference in idling was huge.

The switch for the Romano ECU is also different to the Tartarini one so we changed the switch in the cabin as well.

I drove whilst Simon did some fine tuning on his laptop and the only thing I can really say is Wow. The car is as smooth on LPG as it is on Petrol, if not smoother. Simon was making changes in the software and as we were driving the car was changing from LPG - Petrol - LPG and other then looking at the switch I had no idea what fuel it was on.

I’m over the moon!

David.

Cheers Dave, was nice to see you and have a chat too.

Just from what I saw when the Tartarini ECU was trying to autocal I strongly suspected the ECU was faulty so we changed the ECU at a very early stage. The existing ECU continually increased ginj during autocal until it was reading 60ms (lol).... 60ms would be so rich that the engine would have no chance of running if the injectors were actually pulsing for that long but the mixture didn't get richer regardless of how high ginj got, so although the ECU was trying to pulse injectors for a long duration they were still only pulsing for a very short duration, the mixture didn't change regardless of how long the ECU attempted to pulse injectors. Give that info, even without putting a scope on injector outputs it seemed and still seems pretty obvious that either the peak or hold part of gas injector signals was missing and most likely hold.

The Romano ECU fitted came from a Merc ML63AMG that never ran properly on gas... until the owner brought it to me from London last year. I told him I'd probably be able to sort it without changing many aspects of his install (keeping the same ECU etc) but he asked me to replace the full front end with whatever I advised. The ML runs well too now;-)

Dave made all the right moves previously, the reducer did need replacing (old one was pushing over 2 bar pressure and wouldn't allow pressure to be turned down), he would have got there on his own if his ECU hadn't also been broken.

It'd be interesting to get the old ECU on the bench and see if there's anything obviously wrong with it. How are the solenoids driven, "hit and hold" like the air suspension ECUs, so it gets a whack for a couple of milliseconds and then PWMed to reduce the holding current?

Happy to send it to you Gordon, it's no use as it is lol. Charged Dave £200, that's £100 for the second hand Romano ECU, £20 for the new switch and £80 time/labour. I couldn't find the Tartarini ECU I tried on Blueplasticsoulman's car though hence why we fitted the Romano ECU (which meant the new switch).

I've got a little digital oscilloscope, testing the old ECU would probably be best done when it's fitted as part of a normal install on a car with the scope used on the injector outputs... If/when I ever get around to it! It's definitely broke though. I've got quite a few ECU's here, some in perfectly good working order, some broke, so if an ECU malfunctions I'm more likely just to replace the ECU and shelve the old one (with a big X on the back of it) or just bin it rather than try to repair it. Years ago when an ECU might have cost £500 I was more inclined to try to fix them (swapping components from other ECUs with a different problem) sometimes with success but not often lol!

There are various design boards, this will be an older board. I've had the back off most types, the older ones usually have a row of little semiconductor looking components... On such boards with this type of problem I've seen some of that row burned, a bit of research on the net led me to believe they're diodes that are part of the injector peak/hold circuit(s). On very early 4 cylinder boards if one of those components is burned you lose peak/hold on 2 injector outputs.

It'd be interesting to take a look if you get the chance. I've had a bit of success replacing the output trannies in Citroen XM suspension modules which fail in a similar way.

No probs, PM me your address and I'll send it to you... probably won't get around to it this week because I'm so busy converting 2 Nissan Elgrands.

I've effected fixes for Citreon petrol ECU's in the past. Seems a common fault where the petrol ECU fails to read the engine temp sensor reading correctly, the engine starts OK but the cold temp reading that the fault creates causes the engine to run too rich when the engine is warm. I noticed that by applying voltage from 1.5v cells to the temp signal wire I could still get the ECU to read whatever engine temp I wanted even with the internal ECU fault, the ECU fault wasn't the reference voltage it supplied but was something in the ECU pulling the voltage on the signal wire... made a simple circuit to apply it's own reference voltage to the temp sensor and feed a modified voltage to the ECU, this voltage coming from an op-amp output with only a low ohm resistor on it's output to the ECU so the internal ECU fault couldn't pull the voltage read.