LPG Discussion Forum

markocosic wrote:Length of flexi - nope! That sir, is an essential component. The flexible coupling, between a tank mounted in the rear crumple zone of the vehicle, and a hard line that runs from rear bulkhead to front bulkhead in one lenght, is essential for safety purposes. If UKPLG disagree then they can do so in writing and it'll be attributed personally, publicly and be citeable in court.

It isn't essential at all. Copper pipe is flexible and, with the S bend or loop that CoP11 says you must put in the pipe, in the event of an accident that moves the tank it will just bend. The same applies if you had used poly pipe for the whole run as you were advised in the first place. What you have done is introduce a weak point, the connector, that is far more likely to break in the event of an accident, not to mention introduce another potential leak point. However, in the event of an accident severe enough to break the pipe, there is the excess flow valve on the output of the tank that will shut off the flow anyway.

markocosic wrote:The intention of that paragraph is fine (don't be lazy and make pipes up in sections) but there are legit reasons to have a short length attached to the tank. (essential servicing every 10 years for pressure test/replacement is the other/, as well as making it easy to have 5 attempts at a decent flare and fancy bend without having to replace the entire pipe run each time)

There is no essential servicing every 10 years, there is tank replacement every 10 years. At which time you simply disconnect the pipe from the tank outlet and remove it, you still don't need a short length of pipe attached. You are trying to come up with arguments as to why your bodge is acceptable and failing.

Mark, its all well and good arguing the finer points of the cop 11 with UKLPG. Unfortunately at the end of the day if you are going to take the car in for inspection then its up to the inspector to make the decision as to wether he feels the car is up to standard and puts the car on the database.
End of story im afraid. I know of an inspector that wants all the vac lines clipped as well as the pressure lines obviously. But there is nowhere in the COP 11 that says anything like that, however he wont pass it unless they are, so if you want it to pass with him then you clip the vac lines....
Its that simple really. You can argue it till your blue in the face, i can pick the cop 11 to bits, if you paid 'to the letter' attention to it you would never be able to convert any jaguar x and s type's or Audi's or some Merc's either.
Im sure there are more that i have come across too.

markocosic wrote:If UKPLG disagree then they can do so in writing

They have, it's published as CoP 11 and there is no requirement for a flexy section and definite discouragement for extra joints. This is something they have thought about, as it does recommend stress relief curlies.

There's more than one way to skin a cat, and UKLPG's "should" instructions recognise this. (GoLPG used to issue Jaguar LPGA certs with structural underbonnet duck tape and bubble wrap, for example)

My personal preference would be hard lines (preferably kunifer, rather than copper, as its not susceptible to the work hardening and will stretch/deform far more) between bulkheads, then an overtly flexible line to any component mounted in a vehicle crumple zone. Couple that with a "breakaway" electrical connection the solenoid valve plus inertia switch and you're very unlikely to have issues.

UKLPG's suggestion is an s-bend or coil in a 6 mm copper pipe supported every 600 mm, but it's up to the individual fitter to issue a cert or not. Simplest is 2 connections using copper pipe. I'm not going to lose sleep over an extra two for the convenience given during install - but will double check that any inspector is happy with these deviations (tank at an angle but with functioning 80% fill stop is the other) before taking the vehicle there though.


Speaking of lost sleep...

...Jaguar gremlins. Lovely.

Got the car back together yesterday and fired it up. Mechanically happy but not so electrically - running like isn't on a pair of strombergs. Checked through the wiring I'd done line by line from the engine bulkhead connector to the injector. Injectors transposed on the "intercept"plug - whoops. Switched the pins over and fired up again. Absolutely no difference made; the car still has a random miss. Eventually throws P0300 then P0301 through P0308 (random miss, then each cylinder miss) and P1316 (misfire bad enough to affect emissions, not bad enough for catalyst damage). This P1316 code disables the fuelling adaptation.

Also throwing a knock sensor bank 2 code (P0332). Don't rest things against knock sensors kids, or stupider still bolt things to their retaining screws. It's like attaching a metal bracket and water hose to the diaphragm of a microphone then facepalming when it doesn't work. Obvious now, but completely didn't think of it at the time. Supercharger removal required to rectify. Groan! This code limits throttle opening and limits rpms to 3,000.

Driveable at cruise and I'm off on hols this evening, so I cut losses and reassembled the rest of the car. Boot floor supports made, a wishbone that was particularly clanky (they'll all get done later) changed, and a few other maintenance tasks done. Dropped the supercharger belt off to prevent boost. Car is fine to run like this and without boost or high rpms can't come to much serious harm - time to scratch my head some more and try work out where and why the miss is coming from. Gas ECU has been disabled (rpm pickup, oxygen sensor pickup, CAN pickup, and power/earth to no effect)

Service history shows new plugs at the previous TWO services. Suspect that I may have awoken a intermittent "genuine reason for sale whilst it's still working ok" problem. Will get some photos up this evening if I can find the lead anyhow.

Oh, and to the chap who mentioned wideband AFR sensors. Spot on. The gas ECU shows a constant 3.8 volts and some more digging says that they revised the sensors for '99 cars. My 97 one has voltage sensors and '98 ones do too, but '99 onwards is constant voltage, current-based widebands. Stag can only read a single one of these. Can the AEB based units read two?

markocosic wrote:but will double check that any inspector is happy with these deviations (tank at an angle but with functioning 80% fill stop is the other)

The difficulty you may have is demonstrating that your fillstop is still 80% despite the angle.
(I usually pack the tank level in the first place)

markocosic wrote:My 97 one has voltage sensors and '98 ones do too, but '99 onwards is constant voltage, current-based widebands. Stag can only read a single one of these. Can the AEB based units read two?

Lost track ... which ones (sensors and systems) are you interested in? It's difficult to extract a meaningful signal from a constant voltage sensor without interfering with it. I don't know of any gas system that includes Hall effect current sensors or suchlike for an interference-free method.

Demo empty tank by powering solenoid with a cracked flare nut/compreesion union. Demo 80% fill at a pump. Same as checking any other tank.

Hmmm, bloody phones - typed a long reply on the train and it's gone!


I can confirm that the later X308 models (4 pin coils, not 2-pin) use a constant voltage (varying current) wideband linear afr sensor - not the usual 4-wire voltage output O2 sensor. The Stag can intercept and read one of these, but not two. If the AEB systems can intercept and read two of these then that's another feather in their cap, though from what I understand the O2 hookup is for your convenience only.


Observations on the drive to the airport:

I'll upload the description of the fuelling adaptation cycle and the OBD cycle. Pulling the battery lead on these resets fuelling adaptations but not stored fault codes. It takes about 50-100 miles of normal driving to do an OBD cycle. (convert a P1000 to P1111)The knock sensor appears to have some adaptive behaviour, as the fault code for that disappeared. Perhaps there's a "gain adjustment" for the sensor built into the ECU. Vapourisers were easy to bleed, and they'll sit (LPG off) at 90-95C in traffic or 85C on the open road. Fully hot within 5 miles; can't say when they got to 40C threshold as they were long since warm when I first checked.

ODB scanners are honey traps. Left the car parked at Hopwood Park Services whilst I had lunch, OBD reader on the passenger seat, LPG filler visible under bumper, only to find three AA patrols crawling about it when I got back. They were playing the guess what's wrong game whilst having their lunch, heh! Current best guess is the ignition coil relay. This powers up all 8 coils and sits next to the engine ECU, in the large plastic box on top of the passenger footwell. This box is a convenient tooltray/kneeling whilst working, and if there's a dicky connection anywhere then bouncing a few knees and tools on the top of it would do the trick. We'll see as soon as I can get the bleedin' security screws out.


Random hodgepodge of photos of the install thus far are here:

http://gallery.cosic.org.uk/index.php?/category/13


A few selected ones follow:

General layout:


Gas ECU and the Flashlube system are fitted under the panel covering the brake pedalbox on the driver's side. Main liquid feed runs into the engine bay on the drive's side of the gearbox tunnel, along the front chassis leg, then to two filter units mounted inboard of the shock tower. Two liquid feeds run from them back along the chassis leg, up the bulkhead, across the top of the bulkhead, then around the false bulkhead on the passenger's side to two vapourisers mounted inboard of the passenger's side shock tower. (the OEM power steering reservoir was relocated forwards next to the ABS/ESC valve block) No liquid lines or high pressure components are fitted in crumple zones, and those on the bulkhead are fitted as high as possible so as to minimise risk of the engine hitting them.

Vapour filters mount to the shock tower, down the inside wing, to the front passenger side corner. One runs to a fuel rail mounted to the new power steering reservoir bracket, and the other runs across the slam panel to a fuel rail mounted to the false bulkhead on the driver's side. The combined MAP/temperature sensor is mounted in the longer of the lines, but there's a balance pipe located immediately after it so the pressures afterwards should be identical even at high flow. (there's another balance pipe between the two vapourisers at the vapouriser end) Injectors are mounted either side of the chargecoolers and feed into the intake runners via 270 mm long hoses.

Vacuum lines to the MAP sensor and two vapourisers are individual, so as to avoid pressure drops/aid response times compares with teeing off. These feed from the supercharger outlet, as does the petrol injector regulator. (any pressure drop across chargecoolers the OE ECU compensated for) Flashlube feeds into the throttle body, immediately downstream of the throttle plate for reasonable atomisation at lower loads. This is pre-supercharger - it'll help lube/seal the eaton and should be adequately churned around by that point.

The clearance issue:


The OE engine fits where it touches. Less than 1/2" clearance to the chassis legs, and it rubs on the underside of the bonnet as standard. This was a big problem when trying to fit the HANA injectors.

Liquid feed entering engine bay:





Two M6 rivnuts, a bracket with another M6 rivnut in it, and a P-clip. You can see the pipe behind the factory heatshields there. The bracket is because you don't want to drill the "seam" of the bodyshell closest to the feed pipe in that area, as it is VERY highly stressed. The conduit next to the liquid pipe contains the wiring for the tank.


Liquid filters:





Another bracket and more rivnuts (counting?) for the liquid filters. 8 mm tee to feed the pair of these then on to the vapourisers in 8 mm again. Waves and S-bends where practical, and another two p-clips into M6 rivnuts onto the driver's side chassis leg. (not visible in photos) where the liquid pipes bend back down towards it before heading up towards the top of the bulkhead.


Liquid pipe runs:






Backwards along the driver's side chassis leg, a p-clip each, up the false bulkhead on the driver's side and onto the main bulkhead. A pair of p-clips into an M6 rivnut on the driver's side, then re-use existing M6 fasteners along the top of the bulkhead. A pair of P-clips onto the false bulkhead on the passenger's side, then around to the vapouriser with the longer of the two pipes getting a P-clips. There's room under the plastic covers for these, and a convenient cutout on the central one near the passenger side that both pipes fit through.

Drilling/tapping/helicoiling "bosses" on those plastic false bulkheads works a treat by the way:

Oh - when you've finished cutting/fettling/harassing the pipes, deburr and blow through with an airline. (you're post liquid filter here remember)


Vapourisers:





They fit like so. Big bracket that picks up on the shock tower with the vapourisers bolted to that. Note the "injector connector" on the temperature sender - the Stag kits don't come with a plug for that. Two outlets on each vapouriser. They're set up with one outlet pointing down, and a U-bend between them for balance. Hopefully most of the sludge will accumulate there rather than inside the vapouriser. Tee pieces look ugly but are functional for the 16 mm water pipes, and made for a convenient bleed point. (but it turned out not to be needed) These run towards the back of the engine bay. Pressure adjusters are easy to get to.

Liquid unions are a pain in the proverbial on the KME Gold Reducers because they're recessed into the body of the vapouriser. Use "long" flare nuts, not short ones like I did if you like your life simple!

Gas outlets flip around into two vapour filters, mounted to a bracket that picks up on the shock tower bolts and they themselves are tied down with exhaust clamps. 45 mm nominal on the filters. 1 7/8" clamps worked fine.


Water feeds:
You need to take a feed before the heater valve. This means mucking about in the supercharger valley. I thought I'd do this at first:



It didn't work. Too much of an angle on the passenger side hose, and not enough clearance around the throttle body. So I did this:




Here's a nice mistake:



I didn't want the hoses rubbing on the knock sensor studs, so made an L-bracket and bolted it to the knock sensor. Silly. Bolt it to the intake manifold instead. (same bracket but upside down!) Rather than T-pieces and fugly (functional; ugly) domestic water fittings, a 45 degree outlet would have been useful here.

These are interesting shots of the factory "valley" cooling hoses:





They get hot and degrade. 10 years/100k is a good time to be swapping these anyway!

Routing out of the engine valley is as follows:





They take a "U shape" then runs forwards along the passenger side chassis leg. Stand on top of the engine (valley/cam covers, not the chargecoolers) and rock the heck out of it and you'll see how the OE hoses move. Cable tie to the throttle intake elbow and associated brackets and to each other. The U does the wobbling around and is clamped to the passenger side false bulkhead by the coolant expansion tank. Not my proudest install, but that's about all I could see to do. Jaguar do much the same and rely on the "cobweb effect" to strain relieve the OE hoses and wiring...

I'd like to see this area of a pro install on an XJ8 or XJR - suggestions for improvements by somebody who has worked on one of these in the flesh welcome.

Red goo visible is red rubber grease. Veggie oil based and safe on rubber parts - I like it for fitting cooling hoses in tight spaces. Note my mistake with the part load breather pipe. This wants to fit UNDER your water pipe, not on top. It'll fit but is a pig. I'll rectify when the supercharger is next out.

Injectors:

The only place you'll get nozzles is here:




There's the cylinder head, then a short "intake adapter" as Jaguar call it, then a short length of runner and the chargecooler/plenum. Fit the nozzles to the short length of runner on the plenum side, oriented as shown to clear all the plugs on the petrol injectors. It's about 40 mm further from the valve than the petrol injector, but < 150 mm total.

You can't fit the nozzles here:



The fuel rail on the later supercharged cars has lots of "reinforcing ribs" that make it enormous, and there's no room to pass a hose between the fuel rail and the head casting because of this. If/when you make this mistake, tap a tapered thread, screw a bolt in, then lop the head off and peen it over with a punch they grind the port flush again on the inside...

Tricky to see in the photos, but I've ditched the OE wiring trays for the injector wires and tied the injecor harness directly to the fuel rail. Makes disassembly slower, but gains you precious room under the charge coolers. No way would you fit "plug and play" harnesses under there, so save yourself the effort of bothering. You can see the just-slightly-too-small gap between fuel rail and head for the hoses in this shot too:



(don't drop anything into the intake when refitting these short adapter manifolds, will you...)

Injectors themselves would have been MUCH MUCH MUCH MUCH easier if I had fitted rail mounted, side entry, bottom exit types. If I did this again I would use Valtek/Emer Type 34s. MagicJets also qualify but their longevity is questionable. As it stands, I had inline style HANA injectors, so HANA it was:



You are allowed to mount these horizontal without causing wear issues. (confirmed with manufacturer) but vertical is best for reducing gunk buildup. This is impossible in an X308 type XJR. (the top of that chargecooler plenum rubs on the bottom of the bonnet as standard, remember)

For added fun, there is naff-all room between the engine and the air intake, although the intake is designed to move with the engine rather than stay attached to the car body. There's also little room between the passenger side and the vapourisers, and the bonnet slopes down more the further forward you go. You'll see that the spacing forward-backward of these injectors is a little odd to clear both the oil filler cap (funnel only now) and the engine breather tube - deliberate!

Detail of the mounting bracket:





If the soviets made cable ties, they would look like these. Bought them from MegaPol.pl and think they're for install ducting or similar. Either way, they cinch down tight, stay cinched, and don't melt. Dremel for the slots.


Vapour distribution:

Passenger side front corner with the fusebox bracket out:


Add fusebox bracket back with custom PAS reservior and fuel rail bracket:




The aircon pipe (hard line, engine side) needs tweaking to clear the vapouriser. I've also tied it to the dipstick tube.

Also visible is the temp/map sensor and the balance pipes for the vapour side. The sensor doesn't have any mounting tabs - designed for the cable tie it and run crowd - but the cover for the electronics is held on with four self tappers. If you remove these, copy the hole pattern to a piece of metal, then screw some slightly longer self-tapers back in again, then you've made yourself a MAP sensor with a mounting tab.

ID of 12 mm pipe is 12 mm
ID of the vapouriser outlets, filters stubs, and fuel rail stubs are 11 mm
ID of the brass tee pieces is 9.5 mm.
ID of the temp/map sensor is 7.5 mm.

This is why I have the tees and balance pipe after the map sensor - I suspect that there'll be a non-trivial pressure drop across it at full chat if you tried to run an entire 200 bhp bank from a 7.5 mm ID pipe.

More soviet style cable ties across the slam panel for the driver's side fuel rail supply, which can be seen here:



I later added P-clips to hold the individual pipes to the ignition coil cover panel.

More generally in the photos, you can see three nipples fitted to the supercharger outlet. One for the temp/map sensor, two for the vapourisers. The temp/map sensor is designed for 4 mm hose and has a tiny ID. The vapourisers are for 6 mm hose with large ID drillings inside them. I didn't want to feed a vapouriser from the narrow ID map/temp sensor, or tee both vapourisers from one 6 mm drilling. Probably OTT on the latter but the niplpes are cheap enough and space was not at a premium there. (those vac hoses run into the valley and follow the water hoses for the vapourisers, and follow the OEM coolant bleed hoses for the map sensor)


Misc:

Found another shot of the water feeds:

Flashlube:

Reservoir and ECU fit here on a bracket that mounts to the bonnet catch mount:






The flashlube is the shi**iest piece of design I've come across so far - absolutely no thought seems to have gone into it. What's wrong with an all-in-one unit? (the ECU can/should stand any heat that the plastic bottle full of flammable fuel oil can stand) Ever head of flying leads long enough to do something useful, or plugs? Not impressed with what you get for your 150 quid compared with 150 quid's worth of LPG ECU and harness...

Nipple on throttle body:




As before, the supplied nipple is NOT a self-tapping part and WILL shear off if you try to screw it into anything other than cheese/plastic, and is then hardened when you try to drill it out. Use a proper nipple and a hot air gun to ease the pipe over the end. I cable tied the feed pipe to the A/C flexi pipe run.


Underbody pipework:

You saw previously where this entered the engine bay. Here's the other end of the pipe when it escapes that heatshield:




It then runs along the floor stiffener, as high as possible. M6 rivnuts and P-clips kids.

Trick for fitting the P-clip that you need for the <600 mm regs between the engine bay and that crossmember? Drill from the INSIDE of the car after peeling back carpet inside footwell, about levl with the A/C condensate drain:



Be very careful here:



Factory wiring runs on top of the floorpan just next to the stiffener rib, and in the "rear crossmember" area. Have an assistant life the carpet and wiring, use rivnuts not self-tappers, and place the wiring either side of the rivnut after drilling. You will be annoyed if you self-tapper into those harnesses.

Run along the chassis rail type structure, past the carbon canister:




And up the chassis leg:



As early as is humanly possible, just after the stiffener/mounting for the rear subframe, bring the pie onto the "underside" of the rear chassis leg rather than the "inside" of it. Copy the OE fuel rails on he driver's side. You want as much room around the rear spring pan as possible, as it needs to drop vertically downwards in order to remove the rear subframe or service the rear suspension. (dampers every 50k or so on these) You can see the rear subframe mount more clearly in this shot:



You can also see a stainless heatshield and exhaust wrapping too. (picks up on the bumpstop mounts and one rivnut in the rear floorpan) Overkill perhaps, but didn't take too long and should keep inspectors happy with the <100 mm rule.




(CAD - cardboard aided design - and the front of an old dishwasher acquired from the tip)

No, the rear upright/wishbone-driveshaft won't hit it before it hits the bumpstop. It sits well inboard of the lump on the casting that does the bumping!

Inside the boot:

Pipe and wiring run through one of those 30 mm dia bulkhead things with a 2-hole rubber glove on and into the boot:



(not finished in that pic)

Remove spare wheel mount by drilling out spotwelds. Dint the side and front of the spare wheel well to fit 720 mm dia tank. You can see the vent pipe outlet on the LHS near the back of the car in the boto floor. Plenty of P-clips to the rear and side - you can re-use the OE boot floor mbracket mounts in many instances. Repaint. On active suspension models, you need to move the accelerometer down a bit so that it is still under the floor. That's two M8 rivnuts and some wiring visible just under LPG pipe.

Toolbox anti-slip mat to prevent chafing between tank and floor:



Skip a few shots and the tank looks like this:



DIY tank straps are 40x4 mm with M10 fasteners and large angle-iron brackets on the back side of them. M8 tensioners tension them against the M10 bolt heads. Most "strapped" and "bolted through the boot floor" tanks would fail a 30g decel test - if you're not confident of being able to TOW the vehicle with the tank then it isn't secure enough.




Framing is 2x1 softwood with some 1/4" ply corner stiffeners. Rests on a ledge at the front of the boot and rests on the chassis legs at the rear. Also rests on the tank in the centre. OE floor (one side cut off the piece that covers the batter/fusebox and the carpet re-glued over the edge to make a flat piece from the right-angled piece) looks like this on top:




I'm a little pissed at the Stako style multi-hole tank. "720 by 270" it says but the bloody valve box is huge, so the boot floor sticks up by one wood thickness at the back of the car - precisely the amount that the valve box sticks up past the top of the tank. Next time I'd use a 720x270 single-hole donut style tank, as then you'd get the floor looking pretty much OEM. (higher at the front, but the same height as before at the back)




The yellow sleeve over the gas pipe is just because it was there. Everythign is ubileed onto the tank box, incl wiring (heatshrink sleeved with hot glue under the sleeve to make it sealed) and the JIC filler (some 22 mm ID water hose over the outside of it brings it to the same OD as the "rubber glove" from the tank box)


Filler:

We put it here for now:



This would need a bumper cut to fill it, AND an extension, as it is too deep for the filler to reach if the filler has one of those anti-splashback type shields. Also the tangs are the wrong orientation. Either the car gets a towbar and it fits there, or it gets fitted opposite the petrol filler cap in the bodyshell. all TBC, but for the moment we've made minimal/discrete mess of the bodywork.


Wiring:

Later...

Cracking mate, its good to see it all coming together. Its also nice to hear that you would perhaps do a few things diffently on your next install. Its one of those kind of things that, if you are an engineer with a wish to develop your ideas. Sometimes you have to come to the conclusion that a few choices along the way where maybe not entirely the right ones. Dont get me wrong, im sure the system you are using, which is over specked anyhow will run the car without any problems. But there are certian things that would make life easier. Some lovely work, and pretty much very well considered though.
Its a problem for us proffesional installers, you can only charge what the market dictates. When you get into specialist conversions you can end up paying youself well below minimum wage for the hours you do on it. But if you charge a well considered ammount the customer would probarbly run a mile at the quote. Finding a happy medium, a living wage for a good standard of workmanship is the key. So how many hours do you reckon you are up to now then ?

Cheers tubbs

Far from perfect - but purpose of forum is to share benefit of hindsight. Water pipes are my least favourite part. Others could be easier but jot nasty.ut. The learning/ordering process was more rushed that I would like as car bought on 18th and wanted to fit kit starting njune 1st. with fuel prices can afford to make mistakes and still net save, so wen agead anyhow so to speak, ever.y week on petrol is cash burned, and timing coincided with some holiday. Pro install I would say now 3.5k - 1.5k bits and 2k labour to do. Not realistic here.

Would guess 25 hrs learning about Lpg. 120 hours so far on conversion I would say, but that also included other maintenance work. With rail injectors and some prior knowledge you could do in 50 hours, especially if you use the magic self tapprs, crimp clamps etc. Beware that on early cars the pedalbox area has more stuff inside.acuum cruise actuators where fkashlube fitted and ignition coil drivers where I put ECU. you may need to squeeze into passenger side ECU . Room thete but tight.

50 hrs ? Man i spent far more hours on that Bentley ! Never again. but having said that, if one walked in the door tomorrow i would consider it.
So as you havent really finished yet, what do you reckon, somewhere in the 60 hrs region in the real world ? Even at £20 an hour thats £1200 plus the bits. £2.5-3k would be about right.
But how long would that take to pay off before you where in profit so to speak ?
its a shame, you would struggle to find a customer that would be prepared to pay that. Especially as a done in a day company would quite happily balls it up for a grand !

LPG hardware (incl wiring) is all in; I've just not dug through the photos yet... Unless mapping throws up something special I'd be including that and assuming that you could do these commercially at 50 hours.


Misfire
Symptom - random misfire that's present at idle and worse as engine load decreases. This turned out to be a problem of two parts:

Remove ECU box cover security screws. You CAN remove this with a normal flat blade screwdriver, if you give them a little tighten first to break the loctite, then undo them whilst applying LOTS of pressure on the screwdriver to discourage it from sliding along the ramps on the security screw heads. Inside ECU box you'll see the brown relays, and the ignition coil relay on this car was faulty. Swap with the horn relay for now, and the random misfire under full load is completely gone.

Misfire at idle is still present, suspect vacuum leak on the part load breather that I talked about earlier. Verify by disconnecting it from the intake elbow, and having an unwilling assistant block the hole with a finger whilst you run the engine and the misfire is gone. I'll fix this when the supercharger is back out (wohoo) to do the snout drive coupler - and replace the crummy corrugated hard plastic hose that the factory use with some proper vacuum hose whilst I'm at it too.

Then we can at last begin playing with the LPG...


Economics?
Petrol @ 1.30/litre and LPG @ 0.70/litre.
Petrol mpg 22 and LPG mpg 17. (properly accounting for difference in calorific value)
This works out at 26.8 pence/mile on petrol and 18.6 pence/mile on lpg.
The most you'll save is 30%, even if the conversion is free.

@ 1,500 DIY conversion cost it's 18,000 miles to recover your money/break even. After 45,000 miles you've saved 2 grand. (about 20%) After 100,000 miles you've saved 6500. (about 25%)

@ 3,500 professional conversion cost it's 45,000 miles to just even. After 100,000 miles you've saved 4000.

Kid bro' (whose car this is) does 30k/yr or so. He'd probably run 5k in his MX5 and 25k in this so it'll be paid off by next year and should knock 20-25% off the fuel bill within 3 yrs.

In no shape or form is it the "50% saving" that the more unscrupulous advertisers claim - once you sit down and run the numbers with honest LPG mpg figures. No shame in running the maths and deciding not to convert. Most punters can't do maths though!

I think it depends how much you pay for your gas. I pay 55.3 inc vat on my account. A lot of folks are paying 60 odd from calour centres. And that makes it all slightly more worthwhile. You may be plesantly surprised about mpg figures though. I regularly get a 10% drop, sometimes with the smaller engines between 5 and 7. Older single point systems always ran 20% or more.
As long as your trims are running really well they will turn out good figures. You are never going to match petrol figures obviously. And the other problem with LPG is you are more likely to drive harder, because its cheaper ! So if you can manage to drive the same on both fuels, and use the same pump everytime then you can get a clue. As i said though, you need the trims to be good.
Im just messing about with my vito van. First tank i tested i got 24 mpg. Mostly lanes and stuff. Throttle response was a bit flat coming off tick over. So i had a fiddle with it and ritchened up a few other bits on the map. It drove really well, so over another tank it turned out 18's. More fiddling about today. I must have over egged the pudding so to speak.

Hmmmm, got impatient today...

Went to pick up a supercharger coupler from these guys:
http://www.powerhouseautomotive.co.uk/c ... rades.html

Who weren't in. So I chucked five quids worth of LPG in, had a goodly sniff about, then brimmed the tank. Took 75L from fully empty - close enough to the 80% on a 95L tank for LPG pump accuracy purposes, in spite of the tank angle.

Popped back to their unit and had a play with the LPG system. Laptop in, autocal run with one injector to energise the solenoid valves. Whoops. That union on the rear reducer would be fingertight then...

One borrowed "special tool" from a nearby industrial unit (1/4" whitworth open ender, sides ground down, head bent to 70 degrees, and handle cut short...) and that was nipped up and after 20 minutes of sniffing/autocal all was declared good. On a still day I dare say there's no need for the leak detector - that tracer they put in the gas absolutely mongs in the slightest concentration.

That's about as far as I got before the Powerhouse owner Lawson turned up. "Err, WTF are you doing on my... ...Oh! Cool!" Cue a 2 hr chat, buying the supercharger coupler, and me sat in the corner of his yard playing for the rest of the afternoon.

Job zero - "fix" the part load breather leak with enough tape to let me play with the LPG kit.

Job one - switch the wiring from voltage based lambda to current based UEGO.

Job two - set up the vapourisers. Easy enough one at a time as described. Clockwise to REDUCE the pressure is the opposite of what I thought. Initially set to 1 bar. (the readout on the Stag software is the differential pressure, not absolute, so you set it to 1 bar not 1 bar above manifold pressure)

Job 3 - autocal. Hmmm, it does 1, then 1/3/5/7, then all 8. Big regular miss, injectors too small, multiplier 2. Turn 1 bank onto gas injectors and "go manual" - adjusting the vapouriser pressure and multiplier manually until the gas injection times are little more than the petrol ones. Settled for 1.3 bar and a multiplier of 1.4 at first. Switch all to gas, and we have a wooden leg on our caterpillar...

Regular miss. Switch them individually and there's something up with #5 gas injector - regular miss. Oh well. Let's play with the other 7 for now and leave that on petrol. Actually no, let's play with the "good LPG bank" and leave the other bank on petrol as an injector time comparison. (both banks have near as dammit identical fuel trims on petrol)

Observations

Cold-start on petrol gives 7 ms pulsewidths. These rapidly drop to 3.5 ms.
Idle on petrol in Park gives 2.5 ms pulsewidths. 3 ms when idling with the autobox in "D" or "R" positions.
Fast idle on petrol (2000 rpm) in Park gives 2 ms pulsewidths.
Against the brakes in D, 90 kPa MAP at 2,500-3,000 rpm is a petrol pulsewidth of 8 ms. Not a lot?
The UEGO/wideband/current sensor reads "around about 2.1 units" in the Stag software, whatever those units are. "Very rich" is 2.5 units. "Very lean" approaches 0.5 units.
Reducers generally sit at 80-85C, gas at 60-70C at low demand.
Tickover in Park is 640 rpm; 600 rpm in D/R. Quite low, but steady as a rock.

Cold start enrichment is huge! It disappears pretty quickly though.

I think these petrol injectors are much bigger than the 440 cc suggested, or the engine is a lot more efficient than my crude spreadsheet assumed - 2 milliseconds is a low pulsewidth at fast idle, no?

If you crank the multiplier hard then the petrol ECU will adapt by up to +/- 20% without throwing an OBD code. Throwing in 50% more gas than it needs makes it a little "fluffy" at idle but it doesn't object, so it should switch over happily before the full warmup cycle is complete.

Spun back through town and took a look at the petrol map that the ECU collected thus far. It'll need a lot of road to populate that even without the supercharger.

Will take a look at that #5 tomorrow when the engine isn't at 100C. Good job it's that one as it's only one of two where you can get to both ends of the pipe, what with the spud being on the front side of the charge cooler. I'm thinking either duff from new (unlikely; seems to tick), debris inside the injector (hope not; I did try not to), or a piece of hose/debris inside the spud (hopefully). The HANA outlet stubs aren't the easiest to push the pipe over and I think I could have "cut" a piece of the rubber pipe off with the end of the injector when fitting it. Well, I hope that's it anyhow!

Not too expensive if the injector is toast/I've gunged it up, but it's more wait time:
http://translate.google.com/translate?c ... pair=pl|en

Glad that they're readily swapped without removing engine bits? Oh yes.


Oh, and 33 ohms works a treat for preventing the petrol injectors from opening, triggering the gas ECU, and not setting an injector fault code on the Jaguar. I swapped the resistors inside the ECU rather than running the separate emulators, as it saved mounting yet another box and yet more wires.

Hmmm, last post seems to have disappeared into the ether.

So that injector #5...

Removed
Hoses checked clear
Spud/nipple in the intake cchecked clear
Injector refitted backwards to "blow it through"
Injector swapped for #8 (in correct orientations)
Fault stays with #5

Time for a cuppa.

Dig out the diagnostic screwdriver
Petrol injector #5 also firing when LPG injector is enabled

Time to think about which bits I've mucked about with.

Emulator resistors inside the LPG ecu. Yep - I had a solder whisker between the petrol ECU side of #6 and the petrol injector side of #5. This will have meant that #5 and #6 were firing simultaneously on petrol (oops) and that on LPG the petrol ECU still fired #5 petrol injector from the #6 output. (oops)

That helped.

Ran autocal. Didn't like the answer is gave - too rich. Tweaked the multiplier manually until Tinj (from the car ECU) was the same on petrol and gas. This works in P and D at tickover. Multipliers 1.2 and 1.4 respectively.

It'll drive (gently) on gas. Maps overlap/don't deviate too much for a first pass, but it misses with higher load or rpm. Modifying the multiplier appears to have no effect.

I tried pulling the petrol pump fuses, just in case the petrol injectors were opening (with the 33 ohm emulator resistors) at higher pulsewidths, but this had no effect.


This isn't the same as my experience mapping a bike engine on Megasquirt. With that you just made up a fuel curve that looked something like a torque curve, scaled it against manifold pressure, and provided that it was somewhere between 10:1 and 15:1 the engine would fire it pretty reliably.

The LPG multiplier seems to have no such effect - I can stall the engine with a value of 0.7, or smother it with a value of 2.2, but inbetween it just doesn't seem to want to run. What am I missing here - this should be fairly simple to map, no?


I have acquired the petrol map (and vehicle runs fine on this)
I have an OBD connection and it'll read the trims and codes (without flagging "electrical fault like many aftermarket readers do on the XJ8)
I have a reading from the OE wideband one one bank (can be swapped; Stag only supports one current based sensor)
I can run the car nicely at two load points.
It appears unresponsive to changes at other load points.


Suggestions? I'm going to upload some screenshots and ask on the AC forum in the morning.

--
M

Software setup:





Gas layout:



Water layout original and LPG: