Actually ignore the centre pin bit, the centre pin is ignition switched 12v of course and easy enough to test...
I've just checked and that is showing 12v once the ignition is on.
So that rules out the wiring to the coil pack.

you can check the primary & secondary coil resistance, but that will not tell if they are breaking down inside. Primary is 0.8 ohm, secondary is 13k ohm.

Also, make sure the body of the coils are properly earthed. They get their earth through the mounting bolts so you do really need to have all 3 bolts on each coil. The original ones have 8mm flange heads so I would think they have an M6 thread.

As it's a Thor, the access to the coils is difficult but you can test them by swapping coils around and seeing if the problem moves to another cylinder.

So another update. I bought new coils and leads, as I was pretty confident it was the coil pack and decided I'd rather have a spare to swap straight away, than to take one out and then have to wait for a replacement to arrive. Also if I was buying 1 coil pack, I may as well buy 2 and new leads, then it's done properly, knowing that they are a bit of a pain to get to and bolt in (Plus I had no idea what brand or age the current ones were).

So swapped to the new coils on the bracket, but found they only had 2 bolts in each coil, but they are about 30mm long M5 bolts and I only had shorter M5 bolts, so had to leave them with just 2 in each coil (in opposite corners). I'll get some longer M5 bolts so I can fix them properly. I did however buy new bolts for the coil bracket, so that I could fix the bracket to the engine properly with all 4 bolts. I also put the full set of new leads on.

Rebuilt it all, cleared the faults and reset the adaptive values, then started it up, initially it felt smoother, but after 30-60 seconds it started to feel rough again (although not quite as bad as before). I checked with the Nanocom and the roughness timing was like before, high on cylinder 1, not good on 4 and 6 either, but lower values and the a little roughness on 7. Misfires were showing on 1 and a couple on 4.
I checked the faults after this and P0300 and P0301 are back.

It's so annoying, it has to be coil pack related, for it to be across both banks, but only on one coil of cylinders. However it's not the coil pack, it's not the leads and whilst I was pretty sure it wasn't the spark plugs, I bought a new set of NGK, as the current ones have been in for a while now. I've just managed to swap all the spark plugs over this evening, reset the errors and adaptive values again and started it up - it's ever so slightly less rough, but still far from right. If I rev it a little it goes a bit smoother, but it's still not right.

It's also quite clear from the roughness values that it's worse on the furthest away cylinders and therefore longer leads, than it is on the closer cylinders. I'm convinced it's electrical, but can't determine where, as both coil packs have 12v on ignition and the 2 switched ground pins per coil pack have continuity back to the ECU plug.
I'm thinking next step is to get another 3 pin connector for the problematic coil pack and bypass the main loom, connecting it straight back to the ECU pins.

Oh and I also pulled the fuse to the LPG setup and started it up again, to rule out the LPG ECU doing anything weird to the injectors. That didn't change anything, it still gave the same roughness values and poor running. Though I really don't think it's injector related anymore, as I've tested them with an LED and you can smell the fuel. It's definitely got to be an ignition/spark issue, from the passenger side coil pack.

The Nanocom roughness values are approximately:
Cyl 1 - range from 8 to 10
Cyl 4 - range from 3 to 7
Cyl 6 - range from 2 to 5
Cyl 7 - range from 0 to 2

The other cylinders very occasionally show a number other than 0, but so quickly and always 0.something - so I think that's just anomalies.

Any other ideas?

Very odd. All those cylinders are fed from the one coil pack but as you have replaced it, It is pretty doubtful that is the problem. Other than the wiring, the only other thing is the ECU but even then it would be a bit of a coincidence that 2 circuits would fail and it would be the two circuits that feed the one coil pack. I'm still inclined to think it is something you disturbed when doing the other work, but damned if I know what.....

Unless..... there is a problem with the ignition switched supply to that coil pack. You've tested it and found ignition switched 12V there, but that was presumably with the engine not running. If there is some resistance in it so the voltage drops when the engine is running, that might be an explanation. You could always try hot wiring direct from the battery to the White wire at that coil pack.

Thanks Richard.

I found a 3 pin connector the same as used on the coils, in my spares box. So I quickly soldered up a bit of multi core cable to it, then cut the 2x switched ground wires at the ECU and connected them up to the multi core. Then gave it 12v on the centre pin, just directly to the battery to test.

Again, reset the adaptive values and started it up. Exactly the same. :( So I re-soldered at the ECU end and switched back to the actual engine loom.

I've just run it again and initially when you first start it, it's not quite as bad. However as soon as the revs drop back down to idle under 1000rpm, the full roughness and misfires on cylinder 1 are more apparent. Still the occasional misfire on cylinder 4 too. However today it seems the misfires are not enough to trigger the fault codes, despite seeing them on the live inputs page of the Nanocom.

The new spark plugs have done something, as actually cylinder 1 roughness settled down to 8-9, rather than the 12 I was seeing the other day. I used to have Bosch spark plugs, only because it had them when I bought it and they seemed to work well, so when I replaced them previously, I just bought the same ones again. This time I went for the NGK (the 16mm ones, although the Bosch were 21mm and always fitted fine) as recommended by many with these engines.

I agree, it has to be related to the work I did. However I've run out of options really on what to try next. I also find it highly unlikely to be an ECU issue, that's suddenly appeared coincidentally. If it was all on one bank, I'd say something else got messed up, but the fact it's also on cylinder 4 and 6, which haven't been touched at all (well before I started replacing the leads, coils and spark plugs of course), it seems to be spark/coil related.

I've triple checked every connection that has been unplugged, tested pin outs for continuity and shorts etc multiple times now. Going to have another think what else to test.

The casting on some of the later cylinder heads have a smaller indent for the spark plugs so ones with the 21mm hex head won't fit, or at least there isn't enough clearance to get a socket onto them. Other that that, they are identical.

I've run out of ideas too. I've got a spare Thor ECU if you want to try that (as you have a Nanocom you can easily sync it to the BeCM) but I'm on a ferry to France to sort out another poorly P38 first thing in the morning so I wouldn't be able to send it to you until the weekend.

Yeah, to be honest I never really thought about it last time I replaced them, as I just went for the same again. I suspect the heads were replaced before I got it. I bought a fully refurbished set from V8 developments the other year and they are the same, enough space for a 21mm spark plug socket.
I ended up keeping the other heads, as I plan at some point to get them skimmed and maybe one day build up a fully reconditioned engine from a top hat block.

That's really kind Richard, let me have a think too this week and see if I can come up with anything else to test or progress any further in diagnosing and I'll let you know if it's time to give the ECU a test swap.

I've been thinking all week about this and searching online for any other ideas. I checked the coil packs that I had removed and the secondary resistance on the coil for 4 and 7 reads around 3.3M ohm, whilst 1 and 6 was a touch under 13k and the other 2 coils on the other pack were both a touch over 13k ohm. So there clearly was an issue with that coil, which made me think maybe some damage had happened to the ECU from such high resistance. I concluded it may well be ECU related and ended up ordering a Thor ECU on Thursday for Friday delivery.

I've just tried the other ECU, after setting the immobiliser code and clearing the adaptive values again. Sadly the same. Although, if I sit and let it run I'm seeing random roughness on the other cylinders a little more too. It's only logging misfires on cylinder 1 and occasionally cylinder 4. If I rev it to 2000rpm it stops the misfire count increasing until it drops back to idle again. Although it doesn't sound smooth when holding around 2000rpm, it feels a little better, but still sounds like it's running rough. As it warms up it runs a little worse than first cold start though. I let it warm up until it auto switched to LPG and it didn't affect how it was running when it switched.
Looking over the wiring loom again, the only other sensor near by is the crank position sensor really. I've just looked at the pin out of this to the ECU and tested for continuity or shorts and then also tested between pin 46 and pin 35 on the connector that goes into the ECU whilst the position sensor is connected and get 0.9k ohm - which from checking online appears correct for a good sensor.
However the other random cylinder roughness today is making me think it could in fact be the crank sensor going bad, as at this point I've replaced every other component for spark. A couple of times the roughness would settle down to just show on 1 and 4, before then randomly showing on other cylinders as well. Cylinder 8 seemed to run rough for a few seconds and then be fine, intermittently.

After all the test running, I checked for fault codes and there were none logged for misfires, just codes Pxxxx Permanent voltage supply fault drive cycle A, occured 1 times, signal too low, not currently present and P1000 Lambda sensor upstream catalyst bank 2, drive cycle C, occured 1 times, signal invalid, fault not currently present.

It's so frustrating, as I can't seem to get any closer to finding the issue and I just want my car back now!

Presumably you've swapped the coil pack over so they are all giving the same resistance? The fact that it makes no difference when running on petrol or LPG shows that it isn't fuel related and must be sparks. A crank sensor normally either works or it doesn't but they normally fail when they get hot and the engine just stops. Let it cool down, or give it a dose of brake cleaner or even cold water, and it starts to work again. There is also the camshaft sensor although what affect that has when it fails I have no idea, I've never known one fail.

Yeah I did, I've got both new coil packs on, new leads on all plugs and all new NGK spark plugs.
The new coil packs were all about 13k when I checked them before install last week.

The part that makes me think it's neither the crank or cam sensor is the fact that it's so predominant on the 4 cylinders from the same coil pack, even if it seems to not be 100% only these cylinders today.

I'll have a look and check the wiring to the cam sensor, as I can't remember exactly the route that takes right now, I think that is lower down, rather than by the intakes/top of the engine, so not likely been disturbed, but worth double checking.

Were the replacement coil packs new or used ? Either way check the primary & secondary resistance on them.

Yeah, both brand new, I went for Allmakes ones, as didn't really want to try Britpart. I checked the resistance on each coil before install and they were all around 13k secondary resistance. I double checked again yesterday across the coil for 1 and 6, but through the ignition leads to save trying to reach down to the actual coils. It was around 14k, which seems fine considering the leads were in the chain too.

Just spent another few hours on this. I removed the coil packs again so that I could clean up the bracket and put all the bolts in, as I decided with everything else changed, it must be an issue with the grounding.
I removed the coil packs from the bracket and thoroughly cleaned them up, using a Dremel to gently sand the rust and dirt off the surface, especially where the coil packs touch the bracket and where the bracket touches the intake manifold and bolts. They were pretty crusty and probably not providing a great ground.
Then I cleaned the threads for the coil back bolts and replaced all of them with brand new bolts. I'm not sure why it seems that when looking online, originally Land Rover only used 3 bolts per coil pack, when there are 4 threads. I put all 4 in per coil pack. Previously it only had 2 per coil and has been like that since I bought the car.
I tested all the resistance again - the primary on all coils was 0.4 ohm exactly and the secondary was around 11k on all with no leads connected.
I put everything back together again and started it up - initially it was showing roughness on 1, 4 and 8... However after a few minutes and letting it idle for a bit it went back to the usual roughness on 1, 4, 6, 7 - with the roughness being progressively worse from 7 towards 1.

Next I tried running it with the MAF disconnected, but that didn't make any difference.

So then I made up a short 3 pin male to female jumper cable to sit between the loom and the coil pack. I cut in 12v on the centre pin and used the original loom with the short little bypass cable to provide the switched ground. Exactly the same :(

Sometimes it struggled to start, other times it was okay. I'm not sure if that points to issues with the crank sensor though or rather just that some of the cylinders are filling with fuel and it's not burning.

I struggle to think it's MAF, crank sensor, cam sensor etc related, when it shows that it's always the same 4 cylinders with issues and they are all related to one of the coil packs.
I'm also doubting the lambda sensors, but then that doesn't make sense either, as it's 2 cylinders per bank, not all on one bank.

Errors today are just P1000, P0301 and P0300

I've been cycling through the different inputs on the Nanocom and of course the fuel trim is a bit of a mess and the only other one that stands out is target engine speed being 3000rpm at idle... That doesn't seem right.

I've been resetting the adaptive values between each run and also the battery was fully charged with my trickle charger before I started, to make sure it was kept topped up and low voltage would not cause any issues.

At this point I'm completely out of ideas. I'm debating having to take it to someone else.

If disconnecting the MAF makes no difference, that suggests it isn't doing anything or the output from it is so far out of what it should be, that the ECU is ignoring it and using a default fuel map.

Hmm, that is true!

It's a Bosch MAF, I can't remember when I fitted it now, but has been fine previously. I'll check the wiring to it tomorrow and take a look at the values it reads with the Nanocom.
Although I guess if it was the MAF it would be more likely to effect all cylinders, rather than the weird 2 on each bank.
I'm going to check the readings from the lambda sensors tomorrow too, although if it was these, I'd expect one bank to be running rough.

I just can't understand it still, I want to doubt the roughness readings and which cylinders this shows up on, but I can't see how it would detect roughness on the wrong cylinders all the time.