DIY 2002 Daimler Super V8 (Jaguar XJR, X308 type)

[rossko]

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.

Good to find that.

This isn't the same as my experience mapping a bike engine on Megasquirt.

Nope, keep telling you that The petrol ECU holds the fuelling map - the slave gas ECU holds a "map" of fiddle-factor that it needs to apply to make incoming petrol pulses suit a different delivery system. An "ideal" gas system would be 1:1 all over.
On that basis a big multiplier difference between N and P is probably a clue, guessing its getting into the region of too-short times at idle.

I'd be keeping a close eye on gas pressure - any drooping under load? And gas injector times - getting too large under load, e.g. over 20mS? What is the lambda sensor telling you? RPM consistent, no glitches? Looking at the last picture ... no RPM or Inj but a reasonable MAP for a running engine - not tapped into the brake servo vacuum?

Check the gas and reducer temp display at cold one morning, see that they are consistent with ambient, so as to confirm the correct sensor values are selected. This seems to trip everyone up.

[markocosic]

Good to find that.

Even better that it didn't toast the injector drivers in the Jaguar ECU.

Bloody annoying that it'll complain about there being only a 100 ohm resistor but not that it's driving two injectors!

Nope, keep telling you that The petrol ECU holds the fuelling map - the slave gas ECU holds a "map" of fiddle-factor that it needs to apply to make incoming petrol pulses suit a different delivery system. An "ideal" gas system would be 1:1 all over.
On that basis a big multiplier difference between N and P is probably a clue, guessing its getting into the region of too-short times at idle.

Aye - it should be easier just mapping injectors rather than mapping the engine!

Idealised multipliers would look like this?
http://www.cosic.org.uk/MarkoStuff/lpg/ ... iplier.xls

(multiplier correction for flowrate, then an ever decreasing allowance for the longer open/close time)

2.5 ms = 4.0 ms; 3.0 ms = 4.5 ms. 2.36 ms to open, so the net flow of 1.5 ms (plus 1.2ms close) and 2.0 ms (plus 1.2ms close) seemed reasonably long and it certainly has a stable idle. I'm headscratching about the big multiplier difference too though - will look into this more.

I'd be keeping a close eye on gas pressure - any drooping under load? And gas injector times - getting too large under load, e.g. over 20mS? What is the lambda sensor telling you? RPM consistent, no glitches? Looking at the last picture ... no RPM or Inj but a reasonable MAP for a running engine - not tapped into the brake servo vacuum?

No droop. Overshoots when the liquid solenoids are first energised (goes up to 1.6 bar then drops back to 1.3 and holds it as you use some gas). I've not been past 9 ms on petrol (12 ms on gas) or 5,000 rpm so definitely not hitting the headroom limits.

Lamba I don't know how to read to be honest as it's a wierdo wideband sensor. It hovers around 2.0 V on petrol. Crank the mixture rich and it'll rail at "2.5" and crank the mixture lean and it'll drop to almost nothing. Plus a caveat that a misfire, lean or rich, sticks it right back at 2.5 though - as it's missing I'm not sure which it is.

Rpm is consistent with no glitches. 4-pin coil on plug with a power earth/12V and a signal earth/logic switched signal from petrol ECU. I've tapped into the ECU signal on coil 1 and it's a very stable reading.

In the last picture I'd *just* shut the engine off and hit Fn-PrintScreen so I think it's just holding the vacuum in the intake. Good spot though - I'll double check that I've not pinched a hose or swallowed something that holds vacuum etc.

It's taken from the supercharger outlet; same place as the petrol fuel pressure regulator takes its pressure reference. This won't be the same as the pressure in the intake/pressure across the LPG injectors at higher loads, because you've got the chargecooler cores to flow through. I figured that it made sense to take it from there though, as the petrol injectors also suffer from that not-quite-constant pressure difference and the petrol ECU will have been calibrated to suit.

Check the gas and reducer temp display at cold one morning, see that they are consistent with ambient, so as to confirm the correct sensor values are selected. This seems to trip everyone up.

Will do that now; haven't started the car since yesterday.


Thanks for the tips Ross!

[rossko]

(multiplier correction for flowrate, then an ever decreasing allowance for the longer open/close time)

Dunno, I work on empirical basis, and don't know the internal algorithms of the LPG systems. I start to worry if idle times are sub-2.5mS with Keihin injectors.
You shouldn't need to account for opening/closing times in practical setting up, the LPG firmware deals with that when you tell it what injector type is fitted. That falls down at the lower limits of course, which is where manual tweaks may be needed for an edge case like this car. You can't calculate just how much gas might flow during opening and closing phases.

Lamba I don't know how to read to be honest as it's a wierdo wideband sensor.

Personally i think you're wasting time trying to. So far as I can tell from poking a V6, the Denso ECU maintains a constant voltage within limits, metering the variable current needed; so monitoring the sensor voltage tells you nothing. I do not believe the LPG ECU can possibly monitor current unless you have cut and intercepted the wire.
http://p38arover.com/rover/p38a/Engine_ ... ystems.pdf
Use your OBD reader to study trims instead for setting up. Post-cat sensors might be able to be used for confirm "still rich under full throttle", which is the tricky part.

[MAP] It's taken from the supercharger outlet;

But that is after the throttle and idle air, isn't it? That does probably harmlessly explain the lingering vacuum.

[markocosic]

Ok - some progress.

Temperatures are all plausible. 28C at 2pm under the bonnet of a black car is about right for vap, gas, coolant, and IAT (via OBD)

MAP reads atmospheric at atmospheric and follows throttle openings sensibly/quickly. When you shut down it holds vacuum until you next turn the ignition on again and the throttle body cycles to the startup position.

Nominal gas pressure is 1.3 bar. It'll drop to 1.23 bar in transient high demand, spike to 1.6 bar when first switching on gas, and can spike to up to 1.5 bar if you snap the throttle shut.


I upgraded the AC Synchro software and took cold start screenshot:




I then reflashed the LPG ECU back to manufacturer defaults and the latest firmware, set it up for the Jaguar engine again, and went for a drive on petrol to warm it up. You can see what the wideband does on petrol tickover here:




Then I ran another Autcal. Where the firmware previously had "HANA" as an injector type, the new firmware has "HANA Blue" "HANA Green" and "HANA Red" as injector types. There is only one type of blue injector (1.2 ohm, 2.2 ms opening) but there are two types of green injector (1.9 ohm, 2.36 ms, or 3 ohm 2.8 ms) so I chose blue.



That's more "normal" for a multiplier. They've done something with the injector settings in the firmware since the previous revision.


Then I ran at idle in "P", at idle in "P" with headlamps, heated F&R screens, heated F&R seats, and A/C switched on, then the above in "D", then against the brakes with some throttle applied. Result:




Did some driving on 1234petrol, 5678gas to log two sets of pulsewidths in the oscilliscope files.


Recliabrated again. Once HANA Blue, Once HANA Green. Results:




Pretty darn similar.


Then I drove on gas. Big miss at higher load/rpm, restricted performance, problems. Scratched backside. Switched to half and half as 1234petrol5678gas. to collect some wideband data etc. Ran ok. Switched to half and half as 1234gas5678petrol. Big miss at higher load/rpm, restircted performance. Bingo. Problem now 50% smaller. Drove home on 1234petrol5678gas. Here's the result:



(the wideband is reading the gas)

That's a pretty close tune for "autocal and go" with no user intervention. Firmware update is good. Now, what's wrong with bank 1...



Oscilliscope files:
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 20Blue.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... rottle.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 678gas.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 0ERROR.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 0ERROR.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... ionary.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 20off).osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 678gas.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 0ERROR.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 678gas.osc
http://www.cosic.org.uk/MarkoStuff/lpg/ ... 0codes.osc


I have cut/intercepted the signal for the wideband Tubbs, so the ECU does read it. It can only read one bank at a time though, and sod's law it isn't the one you want! The wires come to a 4-way plug on the bulkhead so you can "bridge" or "read" as required.

Stag ECU logs all the OBD trims; no seperate reader required. You've got to get close (and without a misfire) before they're useful though.

[markocosic]

I'll post this and let you guess the problem. There's a clue in my next question.




So spark plugs and LPGed Jaguars.


TOPIx says:

Early (1998; 2-pin coil) nat asp cars ran NGK BKR5EVxA-13
Early (1998; 2-pin coil) supercharged cars ran NGK PFR6G-13E
They were all gapped at 1.2 to 1.3 mm

Later (1999-2003.5; 4-pin coils) nat asp cars ran NGK PFR5G-11E
Later (1999-2003.5; 4-pin coils) supercharged cars ran NGK PFR6G-13E
Nat asp cars were happed at 1.0 to 1.1 mm, but supercharged cars were still gapped at 1.2 to 1.3 mm.

Later, service bulletin 303-28 says "gap all plugs at 1.3 mm"


Jaguar says:

http://jagrepair.com/images/AutoRepairP ... 9_2003.pdf

As above, except that Iridium plugs (NGK IFR5N10) have superceded the old platinum plugs and should be gapped at 0.9 to 1.0 mm in the supercharged cars. (and all later X350s)


This Daimler has the IFR5N10s, gapped at 1.3 mm, fitted right now.


What would you run on the blown car, and on LPG?

[Gilbertd]

As above, except that Iridium plugs (NGK IFR5N10) have superceded the old platinum plugs and should be gapped at 0.9 to 1.0 mm in the supercharged cars. (and all later X350s)

These, Iridium with small gaps.

This Daimler has the IFR5N10s, gapped at 1.3 mm, fitted right now.

Right plugs, wrong gaps for every permutation (the 10 on the end of the NGK number signifies that the gaps were originally 1.0mm so they have either eroded nicely or somebody opened them up).

I appreciate that you've gone a bit overboard on this but my mates 4.2 Supercharged XJ8 runs very nicely on a standard, pro fitted BRC system and I very much doubt anything like the effort you've put in went into that.

[markocosic]

They weren't looking eroded, so I'm guessing somebody has opened them up. I've asked NGK UK's technical people if there's a more appropriate plug for LPG use (there is one listed against a PFR5G-11), else will pick up a set:

http://www.ebay.co.uk/itm/JAGUAR-XKR-4- ... 162wt_1396


A mate with BRC installed by somebody else on an X350 XJR is great but doesn't help me really. BRC injector rails (or Keihin/HANA rails for that matter) won't fit an X308 XJR (or X100 XKR) unless you violate at least one of the rules I've been told not to:

Short <300 mm hoses form injector to runner
You could mount the injectors on the inner wing, pointing downwards on a slope, but with 500 mm hoses to the intake runners

Equal length hoses from injector to runner
You could mount the injectors behind the engine, with variable length hoses to each runner. Or with equal length hoses but the longest being about 500 mm and the others a coiled mess.

Injectors not mounted upside down
You could mount the injectors upside down on the ignition coil covers, with short, equal length hoses.

If you ran the solenoid type injectors they would also need to be mounted vertically. (Ok to mount Keihin/HANA horizontally)


I've got individual (hose type/inline) HANAs mounted on short, equal length hoses and at an angle above horizontal. Best I could do with these. In hindsight - Valtek/Emer Type 34 rail. MagicJet rail. Matrix block. These would all fit next to the chargecoolers without violating the bonnet height restriction.


You can swing a cat in an X350 engine bay. Perhaps even a malnourished child. Lots of room because it was designed for a supercharged V8 and the 4.2 was designed as a supercharged engine. The cat would need to be malnourished in an X308 XJR as the engine bay is meant for an inline engine and the supercharger was an afterthought. In an X100 XKR you'd need to dice/mince the cat first.


Back to the topic at hand though: There is something up with Bank 1 on gas.


I've tried disconnecting the petrol injectors to that bank, whilst on petrol and gas, to see if the petrol ECU was capable of doing anything daft if it "saw" no injectors. Other than immediate P0201/0202/0203/0204 codes, the observed pulsewidths ran up to 12 ms, with the odd dip to zero. I'm unsure if this is what the petrol ECU actually does, or is an artefact of the gas ECU having no "pull up" inductance any more, but the important one was rpm stayed very stable so the coils are firing regardless.


I've switched one gas injector at a time in every shape and form. On 5678 the engine is smoother with all four switched (so there's still some correcting of multiplier to be done) but they switch individually ok. On 1234 it matters not which one you switch or which order you switch them in - the engine misses. Fluffs at tickover, completely misfires under load.

I've swapped all four gas injectors over to Bank 2. Problem stays with Bank 1, trims on Bank 2 unchanged. The vehicle drives fine on petrol and with Bank 1 on petrol, Bank 2 on gas.

I've verified that gas exists on both banks, and pssssssshses out for about the same length of time if you remove a hose from the distribution manifold. (I can't see the 12 mm hose having a blockage myself, but worth checking) I let the engine run on gas on all cylinders for what I thought was plenty long enough to purge the line afterwards (2 mintues).

There is a common failure point for all four gas injectors - the common electrical connection. I've verified there being power there at the gas injectors. (13.6 Volts on the commons for both banks, whilst running) Loop resistances are ~2.1 ohms; 1.9 ohms for injector plus wiring run. I would like to look at this with an oscilloscope rather than a multimeter but don't have one at present.

I'd also like to look at the injector traces and verify that these are the same across the two banks, but again don't have a scope at present.


I've removed the coils and plugs; dried the oil out of the spark plug wells (damn; needs fixing); cleaned the wells and the plug bodies with panel wipe; refitted. No change.

I've verified there being power at the ignition coil relay (from before) but not at the ignition coils themselves. Again would like to look at this (and the feed from the ECU) with an oscilliscope to see if there's any change from petrol to gas.


I can (by swapping pins in my two "breakout" connectors towards the back of the engine bay) switch banks on the gas ECU.

Petrol ECU feeds from 1234 go to Gas ECU 1234 right now. I could change these to 5678. I could then change the feeds to the petrol and gas injectors to 5678 and vice versa. Run Bank 2 on the "Bank 1" electronics inside the gas ECU and vice versa. I'll try this tomorrow.

I'll try sort of a set of loaner ignition coils too. Perhaps one (or more) are bad and "backfiring" through the low voltage side of the system when switching to LPG? (LPG being harder to fire?) The car does appear to have suffered a meltdown of coil ~2 in the past as there are scorch marks on the camcover. No receipts/history for it though, and none of the coils appear any different to each other. They'll have been shuffled around during the LPG work too.


Any other ideas?


Any experience with these pocket digital scope beasties?
http://www.ebay.com/itm/SainSmart-ARM-D ... 304wt_1163

[hardcorepit80]

If it's only on bank1 , just check again connection numbers to lpg inj. or maybe there's blockage somewhere in pipe work on that side.
Also, it might be faulty lpg ECU... if you say no matter which inj. you switch on and engine misfiring.
It can't be, that all ignition coils on bank1 are faulty.
I would get another Stag ECU just to check if problem stays with new controller.

In settings tab tick "pressure release" and set on 1.5bar that will release pressure that builds up in reducer after revs drop.
Also with lambda issue, try "UEGO voltage" instead of "full range"
Mine is set to "voltage", when I had it on "Full" it was showing some very weird values.

[rossko]

Shudda bought a V12 ...

On 1234 it matters not which one you switch or which order you switch them in - the engine misses

This symptom really is a good match for injector sequence jumbling - petrol A fires gas A, but due to physical or wiring crossover gas A feeds cyl B. This makes a terrible hash of sequential or individual changeover, but works quite well thereafter.

EDIT : a reminder that cylinder numbering has changed during the V8 life, you can get into terrible trouble using wiring diagrams or fault codes of a different vintage ...
http://jagrepair.com/images/AutoRepairP ... bering.pdf

[markocosic]

Naah - a nice XK with two SUs and clockwork ignition.

Then again, no thanks. Ever balanced the carbs on a Mk2/S-Type/420, or worse still had to adjust the points? Makes the Super V8 look like a walk in the park!


Is is the gas ECU? I swapped 1234 and 5678 this afternoon. (petrol 5678 is on the 1234 gas ECU lines, gas injectors 5678 also on the 1234 lines, switched 1234 onto "Bank 2" to tally with the OBD) No change.

Numbering is definitely correct (switched it all over and checked again this morning). Even if off wouldn't cause what I'm seeing. At worst it fluffs the changeover, emissions are a sniff off, and it might fluff accel enrichment on the first couple of cycles. I'm seeing the miss all the time, and worse with load.

The numbering is a good call too, but I'm good on that front. X308 to the X350 (4.0 to 4.2) is when that changed. I'm working to the old Bank 1/A is 1234 Bank 2/B is 5678.


Went to see Autotronix today. I actually saw two Autotronix by accident - Steve@Autronix Sheffield, and Clive@Autotronix Developments. Steve, as it turns out, is a loudmouth asserts everything knows absolutely nothing type chap. Some might call him arrogant. Others wouldn't use pleasantries. "If the check engine light isn't on then there aren't any faults" and "well if it only does it on gas then it must be the gas kit." I went to see Clive.

Clive actually knows how cars work - he does mapping and whatnot, and has a rolling road which was the primary reason I wanted to see him. He explained how those widebands work. They pump oxygen out of the exhaust, and measure how much current it takes to do this. MORE current is more oxygen in the exhaust is LEAN. LESS is RICH. However... ...if you misfire at all you're getting a dirty great slug of oxygen through the exhaust.

What you *should* see on petrol is a very high reading on the overrun fuel cut. This is when the ECU is calibrating the sensor @ 20% oxygen. You should then see a fairly high reading a tickover and light cruise (the "2" that I was seeing), that drops under higher load (the "1" that I was seeing), and is basically zero when the engine is in "rich as sin" open loop mode when you put your foot down.

I see this in the datalogs for 5678 - except for the last part. It reads "lean" again at full chat. It's missing on petrol, though not badly enough to feel it and nowhere near badly enough throw the OBD code.

(throwing an OBD code on the Jaguar needs a miss count of 10,000 in two and a half minutes; or a consistent miss at 2,000 to 2,500 rpm)

This happens sooner on LPG. It isn't going lean, it's missing, again on 5678.

It can't be, that all ignition coils on bank1 are faulty.

It can be, but only if the Gods really hate you. Perhaps I should buy that lottery ticket this week too

"When you have eliminated the impossible, whatever remains, however improbable, must be the truth."

2&3 are "Really bad", 1&4 are "Not great", but even on the 5678 bank there are some slightly iffy ones. £912.82 inc Delivery and VAT for a set of 8 from British Parts.


Other little titbits staring me in the face/pieces of the jigsaw puzzle:

Main ignition relay was iffy previously. What does this do to coils?

Receipts for "spark plugs changed" at the previous two services, even though they should last many many miles.


NGK's Tech guy says:

We would suggest using one of our specialist plugs from our LPG range.
Part number LPG1 (Stock code 1496, £20.33+VAT each), available through
your local NGK stockist, would be the most suitable for your vehicle. This
item is manufactured with a 0.8mm electrode gap setting which would not
require any adjustment before fitting to your car.

Most LPG conversion specialists reduce the electrode gap setting on plugs
when converting a vehicle to gas by 0.2mm. This is to give the ignition
system an easier time when running on gas.

You could use the existing IFR5N10 but with the gap carefully reduced to
0.8mm. The 1.3mm gap that you have been running would probably put a bit
too much strain on the ignition system for a vehicle running on LPG.

I asked:

Functionally, any difference between an LPG1 and a (new) IFR5N-10 with
the gap reset to 0.8 mm?

(other than than the time take to re-gap the plug)

NGK said:

The LPG range of plugs also use a specialist plating to prolong the onset
of corrosion that occurs with the use of LPG, which differs from the
standard plating applied to the IFR5N10. This difference in plating is
designed to prolong the life span of the plug when used with LPG with the
intention of less frequent plug changes.

LPG1 also is slightly higher spec, with a taper cut ground electrode which
also contains a copper core to help heat dissipation.

£120 for 8 LPG1s. £80 for 8 IFR5Ns.


Off to change me some coils for a borrowed set.

[markocosic]

Oh, and to add:

If you have an exhaust gas analyser (measures HC) then it is easy to tell whether you're lean or misfiring when combined with the wideband.

Dyno time (rental of the dyno and an operator, but you doing the mapping) is 40 quid an hour. Not worth the danger of trying to map the higher reaches of the map on a public road - much easier to hold it in steady state on a dyno for a couple of seconds.

[rossko]

re "missing on petrol too"
I'd look very hard for signs of tracking. Both LPG and the previous big gaps would help contribute to that ; if you jumbled the coils on reassembly the existing disease could have been spread further at that time. Once its happened, the cure is new boots and plugs at the same time. Take care bonking 'good' coils onto 'bad' plugs or you'll ruin those as well.

I've lost track whether you think you've moved your "lpg" issue or not?

[markocosic]

I've lost track whether you think you've moved your "lpg" issue or not?

So had I earlier this evening.

Feels like a game of this at times:

http://en.wikipedia.org/wiki/Whac-A-Mole


Borrowed coilpacks kill the higher load misfire and improve idle behaviour. Improve, not cure. With the reliable spark, you can recalibrate and it highlights other things - I'm hoping for the last time.

Cylinders 2 and 3 would not behave. Pronounced miss when they were activated, though less bad when active with other cylinders on that bank than on their own.

Cylinder 1 behaved perfectly. So I swapped the injector 1 onto it, with the feed from distribution manifold-injector left intact, and the pipe from the injector-manifold straightened as far as was practical. I fed this via the wiring for injector 1. (fed the car's injector 2 through the gas ECU's injector 1 wiring) I used the spark plug from injector 1 and the coil pack from injector 1.

Not one bit of difference.

I then went old school with the setup. Stuff the auto-cal and the trims and adjust multipliers one cylinder at a time.

Cylinder 1 wanted 1.4
Cylinder 2 wanted 2.3

Why hello.


Firing order of the 4.0 Jaguar V8 is:

15426378

Or:

(Bank) . Cylinder
A1, B1, A4, A2, B2, A3, B3, B4.


To cut a long story short, the gas is disappearing back up into the plenum/chargecooler at idle type flows. It is least bad on 1 and 5, due to the design of the manifold/chargecooler. A2/A3 are worst. B3/B4 suffer a little.

Crank the injector corrections (labelled with "last resort, do not use" in the AC software) to +25% on A2/A3 and +15% on B3/B4 and it idles just as on petrol. This all goes to pot when you load it up though, and the neat airflow is such that farting in the doorway results in it going outside. (pishing in the wind is easier to direct)


So it needs to come apart again. The drawing board:




Very short intake runners due to packaging constraints




Heads




Heads with "intake manifold" and petrol fuel rail




"Intake manifolds"




There is *just* space for a 6 mm OD copper pipe between this fuel rail and the head in this area.




This was my original gas inlet location, just to the side of the petrol injector. You can't get a 5 mm hose off this and between the fuel rail/head for love nor money though. Gap is too small and angles for the hose are too tight.




The fuel rail has stiffening ribs where the hose would foul. These could be removed and not leak fuel, and you'd then get a 5 mm rubber hose past, but I'm assuming that there was a flippin' good reason that they were there.




Plan B that I went with was putting the feeds here. This isn't working.



I think that Plan A was better, but I need to use something other than 5 mm ID / 11 mm OD rubber hose on an M6 nipple.

The HANA injector outlets are 4 mm ID. No point going any larger than this.


How small can one go on hose bore or stub/connector bore before flowrate becomes an issue?


Are "directional nozzles" any use, and how do they affect performance/aiflow? (they stick out a fair way)


I do have some of these, but their ID was only 2.5 mm which concerned me with regards flow rate:

http://translate.googleusercontent.com/ ... vMzQmo0eqQ

Are these available anywhere with a 4 mm bore?


Poking about I see these devices for sale too:

http://translate.googleusercontent.com/ ... YkZr-NFD8g

PTFE tube and a brass bulkhead fitting (I think compression type with an olive, then to a nipple) that then takes normal tube. I almost want the opposite of this - a bulkhead fitting/nozzle with the thinwall PTFE tube on the outside, leading to a nipple for a sort length of hose to connect it to the injector once is is out of the complicated/inaccessible part of the intake system.


Tinley Tech sell a variation of the theme:

http://www.tinleytech.co.uk/acatalog/Bi ... _2dINJNOZD

The screw-in nipple for the outside end of it looks good, but the quick-release pneumatic hose fitting for the intake manifold side has the potential to leak. (these are designed for pressure seal, not vacuum, and rely on internal pressure to seal themselves)


The kit that the Nitrous boys use looks well thought out, but I don't see that exact stuff in their standard catalogue:

http://www.noswizard.com/index.php/revo-system


Suggestions as to how you would get to that injection location welcome. It wants to be << 10 mm OD, and capable of a tight turn without kinking. 6 mm copper pressure line won't do it/can't be bent tightly enough. Thickwall Kunifer hydraulic pipe might do the trick (clutch type size rather than brake pipe size) but I don't know how you'd terminate it into the manifold easily/compactly.

[rossko]

To cut a long story short, the gas is disappearing back up into the plenum/chargecooler at idle type flows.

I don't understand the mechanism at work here, nor why moving injection point would make any odds.
Presumably you're thinking the charger rotors are creating a "back suck" at certain times? Would not the plenum even that out across all cylinders - though not necessarily affect them all equally if it is timing dependent pulses? Might be an interesting experiment to hook an old fashioned vac gauge to some of your gas injector hoses, idling on petrol.

If it is hoovering the inlet ports, moving the injection isn't going to change much - gas dolloped in under pressure will get hoovered out wherever you put it in.
Presumably the plenum will re-distribute a weak gas/air mix to all cyls, so the effect will be to richen up the unaffected cyls a tad, exaggerating the mismatch even more.

Problem doesn't seem to arise on other folks cars ... could it be supercharger wear or an assembly problem - rotor synch or summat? Or something like leaky inlet valves blowing back?

[markocosic]

I don't understand the mechanism at work here, nor why moving injection point would make any odds. Presumably you're thinking the charger rotors are creating a "back suck" at certain times?

Not the supercharger. Those chargecoolers are a plenum/resonance chamber and the intake runners are tuned resonance chambers too. You'll get pressure waves inside the runners and the plenum, and they're set up to try make sure that the high pressure point arrives when the intake valve opens. (all these weird and wonderful flaps and butterflies inside intake manifolds try to expand the rpm range over which the effect works) This is what's sucking the gas back.

X100/X308 supercharged are pretty poor in this respect. The system was added as an afterthought and there wasn't enough room to put the runners that the engine needs in there. (or even to make them all the same length; 1 and 5 are longer)

There's about 100 mm between the plenum and the intake valves. I'm dropping the gas in 20 mm from the plenum/80 mm from the intake valve at the minute. The petrol goes in about 60 mm from the plenum/40 mm from the intake valve.

The gas is being drawn back into the plenum by the resonances (and heading for other cylinders) rather than staying put in the runners. I think I need to move the injection point much closer to the valve to avoid this.

Might be an interesting experiment to hook an old fashioned vac gauge to some of your gas injector hoses, idling on petrol.

It'd flutter if the response was quick enough.

If it is hoovering the inlet ports, moving the injection isn't going to change much - gas dolloped in under pressure will get hoovered out wherever you put it in.

Depends how deep it is inside the intake port - the effect will only "hoover out" a fraction of the intake.

Problem doesn't seem to arise on other folks cars ... could it be supercharger wear or an assembly problem - rotor synch or summat? Or something like leaky inlet valves blowing back?

That's why I was so keen to find somebody that has converted an X100 XKR or X308 XJR and interesting in sharing details - I now suspect that they'll have used straws as well as the 5 mm ID hose.

(for Gilbertd - X350 and newer were better thought out in the first place, and have much much longer and more readily accessed intake runners)

Supercharger is largely irrelevant as it's so far upstream, and at idle you're working on the vacuum bypass anyway. (if you look closely at the photos you'll see that the belt is off too at the moment, as the vibration-absorbing coupler inside it is broken/on the fix-it list when the charger is next off)

There are no other mechanical issues with the engine itself; clean as a whistle for a 145k motor on petrol.


Plan A was better, but I couldn't get the 5 mm ID pipe onto the nipples and past the fuel rails so put it in the only place you could get a 5 mm hose. Didn't think about straws at the time though - assumed that it'd work without.


--
M

[rossko]

Don't buy the theory
Counter theory: gas under pressure injected into very short ports is expanding in all directions, as gas does, and spilling into the plenum. That wouldn't matter much except the general movement of air is stealing some of the spill of the middle cylinders for the end cyls.

Cures would be much the same to minimise the spill. It might well be worth trying those Bigas directional nozzles first, on the basis they'll offer less restriction at full tilt than long straws.

I didn't realise the later superchargers (a la Range Rover version) were different in this area

[markocosic]

Counter theory: gas under pressure injected into very short ports is expanding in all directions, as gas does, and spilling into the plenum. That wouldn't matter much except the general movement of air is stealing some of the spill of the middle cylinders for the end cyls.

I explained it to the mother thus:

Petrol: peeing in a doorway into the wind and hoping it all stays outside...
Gas: farting in a doorway into the wind and hoping it all stays outside...

Whatever the physics behind it, I'm seeing spill from 2/3 into 1/4 at tickover. Ditto 7/8, though less so. Not the 'charger.

I didn't realise the later superchargers (a la Range Rover version) were different in this area

http://www.gaspoweruk.co.uk/v3/lpgconve ... y_001.html

Larger dia, longer, symmetrical. (and stacks of room to work!)

I think it's for sale too:

http://www.carandclassic.co.uk/car/C301953

It might well be worth trying those Bigas directional nozzles first, on the basis they'll offer less restriction at full tilt than long straws.

Don't really want to be "trying" anything, as gaining access is at least a full day, and at least the same again to drop it all back. I was hoping for the "straw" on the outside, with a directional type nozzle into that lower intake manifold (or the head) rather than a "straw" all the down inside the port from the plenum. I'll knock another option up in MS paint later. Gas all/biased down one valve port - problematic or indifferent?

[rossko]

Don't really want to be "trying" anything, as gaining access is at least a full day

Appreciate that, but you have a throughput counter-requirement for not introducing e.g. restrictive complications, straws and elbows ... or possible new problems like ..

Gas all/biased down one valve port - problematic or indifferent?

Dunno, but it feels wrong? The impact would depend on how turbulent the particular combustion chamber design is. The only guide you have is how they handled the petrol - the end petrol injectors don't look very symmetric?

[Tubbs]

Straws im afraid. Thats the only way to mimick the petrol delivery. As you can set them down just above the two valve ports. So are you thinking its a turbulance issue ? I preume you have swapped out injectors you can get to to discount a dodgy injector ? Mind you a few dodgy injectors, hugely unlikely.
So if you want to prove the 'robbing peter to pay paul' scenario can you bang it on part or full group ?
That way you should always have enough gas at each cylinder at the oppurtune moment.

[markocosic]

Going back to this:




I think I can get nozzles in here:




Sneak them between fuel rail/head and the camcover/chargecooler in these locations:






It would mean drilling and tapping the head castings, which I'm ok with, but it will mean injecting predominantly into one valve port.

I think you'd end up in much the same place with a straw: I can't see how you'd centre a length of teflon pipe let alone keep it there so would rather prefer to inject into a known location. A chance of mucking up the airflow past the petrol injector too. (they're a 2-hole type, so even with the in cylinder head location you'd want to try not to block the spray from those)


The directionals certainly have their attractions but I'm worried that they're not long/directional enough.


Anybody know who makes these metal straws/have experience with them?

http://autogas-lpg.co.uk/gas-inlets/449-aa-GZF1001.html

http://www.autogasteile.com/product_inf ... 2bb1526a0d

http://gbo.ua/ru_podvodka-dlya-gaza.html

Looks like a metal pipe with a threaded end and taper seat that you can screw the gas pipe inlet onto from outside the runner. I could fit these to the existing holes in the plenum, and direct them to the centre of the port, but on the opposite side (valley not cam cover) to the petrol injector.

Can they be bent further without kinking?

Do they have fatigue-fracture-the-destroy-valve tendencies inside engines?