Not had that PM from you yet Dave, just found this thread myself ;-)

I've just checked your LPG system map, I didn't set a fuel trim in it but would have set one if for any reason I thought necessary.

On your LPG system it's possible to set individual trims for individual cylinders, if any cylinder was acting strangely in the way it responded to running on gas I would have first checked that no aspect of the LPG install was causing that (and changed LPG parts if necessary) or drawn conclusions about your car and written you a bit of a report on it. I didn't set any individual cylinder trims.

Not saying any of the following insights will lead to diagnosis of the problem but any of them might..
A failing lambda probe can stop giving proper signal on gas before it stops giving proper signal on petrol. Another thing that can bugger lambda probes is contact with engine coolant.
Your LPG injectors are on a rail but they are individual units on a rail... so (remembering your misfire codes for 4 cyls on the same bank) it's extremely unlikely 4 on the same rail would go wrong. But if one went wrong so causing dodgy fuelling (especially if dodgy fuelling led to a misfire on that cyl) it would mess up fuelling for the other 3 cyls on the same bank and could lead to misfires on any cyls on that bank.
Leaky petrol injector.
Duff LPG ECU could cause more than one cylinder to misfire straight away but very unlikely with your ECU... the output (LPG injector) driving electronics of some other LPG ECUs use common electronics for more than one output channel (so outputs usually cock up in pairs) but it seems your ECU (and most others) use discrete electronics for output channels so channels go down individually and I've never known a channel go down on your type of ECU.
Since you've had the manifold off... Crimped pipe between LPG injector and manifold, Crimped / shorted petrol injector break wiring, Routing (check plugs to LPG injectors and pipes between LPG and manifold match up... you don't have to remove anything except plugs to LPG injectors for this. Another indicator of problem routing is if it sputters particularly badly at some point like it's down to running on 6 or less instead of 7/8?).
Individual spark plug / lead
Compression or valve lift/timing on an individual cylinder

I didn't send the PM as it started to look like it was a general car issue (lambda sensor) and not an issue with the LPG system :)

Useful things to know and check though, thanks! When I'm home I'll give it all a check.

When I had it stripped down I did number the LPG pipes going through the manifold but when I came to reassemble it all two of the numbers had rubbed off. I worried that I ended up with the connected wrong but since the car ran I figured they were ok. I assume if I had connected an LPG injector to the wrong cylinder I'd have all sorts of issues.

If you got routing wrong It might be a bit lumpier at idle and during acceleration but might still run fairly smoothly most of the time. You'd notice it mostly during switching from petrol to LPG because there'd be a point when at least 2 cylinders were misfiring at the same time.

With the engine running on petrol pull each pipe off each injector in turn and make sure there's vacuum on them (replace pipe each time before checking the next one)... this is to check a pipe hasn't come off a manifold spud. Then with the engine off try blowing through each pipe, if any are much harder to blow through than others the pipe might be kinked. It would take a much worse kink to cause a misfire at idle (when only a small amount of gas has to flow through the pipe) than at higher engine loads.

Looking at Lambda sensors now I'm home (at work).

Can anyone see a reason to not buy this?

Sensor on eBay.

I searched by part number (MHK100940, found on the LRCat website) and it says it's a Bosch sensor to fit a 4.6 P38 before the cat.

My hesitation is based on the price. It's £24 cheaper than at LR Direct. I also have in my head that GEMS cars had a different sensor to that fitted to a Thor. The eBay listing says it fits 1994-2002.

I do have a slight roughness at idle that I've now realised is only there on LPG. The idle speed wobbles up and down. Acceleration and general driving has been OK since October though.

Although it says 1994-2002 it shows the correct part number and says from VIN XA or later so it will be the right one. Assuming it is genuine Bosch and not Chinese knock off, it should be fine.

It did look like the right thing to me but thought it best to check in case I was missing something silly!

Cheap usually comes with strings attached.

If it's now running on all cylinders on LPG and has lost the original problem after being run on petrol for a while, it does sound like the trims got themselves a bit cocked up. If it was due to one dead lambda sensor that should sort it.

I've tried LPG again a few times and the issue is still there.

I've ordered the sensor so I'll see how that goes.

It is a tenner cheaper than ECP are doing, and their ones don't come with the plug either.

Best bet (generally with any eBay parts) is to send a message either with the order or before you order it giving the vehicle details. Given its specified as Bosch you should be ok, I'd only go with them or NTK having experienced cheap ones previously.

As long as it’s zirconia it should be ok, check with nano that it’s working, I got sent a supposedly zirconia one to find it wasn’t, and of course because it had been fitted they refused to take it back, the tosses on eBay need to sort there listings out..

The sensor arrived today while I was out. Came in a nice blue Bosch box and the security to check it's genuine checks out. All looks good then!

Hoping to get out and fit it this evening.

New sensor now fitted, checked in Nano that I have a fluctuating value for the bank and I do, reset the adaptive values, cleared the stored error codes and went for a test drive.

Still fine in petrol, no misfire on LPG but still down on power/sluggish on LPG. No error codes at the end of the test drive though. The fluctuating idle on LPG seems to have cleared up.

I'll drive it for a while on petrol to let the ECU learn some values for the adaptive and try LPG again.

Lpgc wrote:

With the engine running on petrol pull each pipe off each injector in turn and make sure there's vacuum on them (replace pipe each time before checking the next one)... this is to check a pipe hasn't come off a manifold spud. Then with the engine off try blowing through each pipe, if any are much harder to blow through than others the pipe might be kinked. It would take a much worse kink to cause a misfire at idle (when only a small amount of gas has to flow through the pipe) than at higher engine loads.

One of the pipes has come detached from the engine end. It's the only one that I can blow through very, very easily. When blowing through it I can hear a rush of air. It also has no vacuum with the engine running.

The other three I can blow through but there's resistance.

I guess the upper manifold is coming off on Friday :)

It's a good opportunity to ensure that each one is connected to the right injector and cylinder.

Good, getting somewhere now...!

Worth saying that with that pipe off it obviously must misfire on LPG because that cylinder won't have been getting any fuel when running on LPG. Power would obviously be down running on 7 cylinders. You'll have pulled the pipe off yourself during engine work (hg's etc)?

When I first fit the system I fit an overly long length of pipe to each manifold spud, being longer makes it easier to thread through the holes I make in the plenum and of course it's better to fit too long a length of pipe and cut it down to size than fit too short a length and find out the plenum has to come off again to fit a longer bit. To make it easier to thread the pipes on 2nd fit you can push a length of wire into the end of each pipe and push a short M6 bolt into the end of each pipe to hold the wire, that way you can thread the wire through the holes with the plenum higher up and pull the pipes through when the plenum is almost down, this also helps to prevent kinking (but still do the blow and vacuum checks just before and after bolting the plenum down when the pipes are fully routed to the position they'll be in when connected to injectors... it is possible to pass blow and vac checks but still have a problem if pipes are pushed/pulled later). You could use different colour wires to denote which cylinder each pipe feeds but for this purpose you only really need 2 bits of the same colour wire because the 4 pipes come through 2 sets of holes (so you only need to know which pipe feeds the most forward cylinder through each hole)... To save a bit of time ensuring routing was correct when ECUs didn't feature ability to switch cylinders back to petrol I used to do the 4 different coloured wire thing, then did the 2 same coloured wire thing, these days I'd quickly identify correct routing anyway and just swap injector plugs around to achieve it. Could use stripes of tippex on pipes to denote which cylinder they run to.

With that pipe off at the manifold end it will have been causing some upset for the petrol system too... fuel trims will have been more positive (mostly on the directly affected bank but also on the other bank), the cylinder with the missing pipe will have been running way leaner than the other 3 on the same bank, the other 3 would have been running a slightly rich mixture. All these effects would have diminished with increasing engine load as a vacuum leak has more effect when there is more vacuum but when an engine management system has a limited number of fuel trim ranges if fuel trims become high at low loads it can have the effect of initially causing too rich a mixture at higher loads and if that initial mixture is rich enough it can cause misfires which then cock up the closed loop functionality.. and then open loop at idle the mixture is very lean (without positive trims) due to the vac leak. You probably didn't need the new lambda probe.

Simon

Yeah, the misfire was back yesterday.

I think it must have been loose and leaking and when I've pulled on it to check it I think it came off completely. I figured you must have used longer sections of pipe when fitting the system - it was a PITA trying to get them back on as they are now :P

In the end I found some wooden dowel that fits in the pipes to act as an extension.

I wonder if this will also cure my occasional high idle issue. Every so often I find that my idle speed will jump up to around 2500rpm and stay there until I either switch the car off or put the car in drive and hold the brake. As the engine warms up it goes away. It only seems to happen on really cold days. I've been guessing that there's air getting into the intake somewhere but couldn't find/see/hear a leak on any of the visible bits up top.

The value for the lambda probe has been static since last October. I've checked it with the Nanocom periodically and have never seen a proper value for it, even after resetting the adaptive values. As soon as I put the new sensor in I got a value, even with the leaky LPG pipe. I was also getting a stored error code saying that the sensor was missing. That went away with the new sensor fitted.

I'm fairly happy that there was an issue with the sensor.

Agreed, given the symptoms started back in October it does seem you needed the lambda probe.

Anyway mate.. Is it all back together and running well now, oil and water leaks etc?

Not working on it until Friday.

Only leaks I know of at the moment are somewhere around the steering box, a heater pipe where the LPG system is spliced in and a rear axle oil seal. All of which are also getting looked at tomorrow.