Just had a Disco with Thor engine here, modified for off road use and the owner says he's won a few competitions with it. LPG converted (Prins) and was running slightly better on LPG than on petrol but on either fuel it was idling rough (as though it had a big cam in it), though the owner said he'd rebuilt the engine and it only has a standard cam. The two OBD error codes pointed to both front lambda sensors failed and lambda readings for both banks were full lean but I found when I blipped the throttle the lambda readings momentarily went rich as they should, which pointed to maybe the the lambda sensors (at least signal) showing actual mixture so the mixture actually lean. It was looking as though 2 failed lambda sensor heaters were causing the rough idle but when I unplugged the MAF and restarted the engine it ran perfectly, correct mixture on both banks and lambda readings flicking like they should.. It runs a lot better with the MAF unplugged but I warned the owner it might refuse to start, or start then refuse to run, or if it runs OK at light engine loads it might still not run OK under heavier loads. I asked him if the ECU had been modified, he said not. He's going to fit a new MAF, see how it runs with that fitted and maybe new lambda probes if it runs better with the new MAF but codes pointing to lambda's still reoccur.

Any other thoughts on it running better with the Maf unplugged?

As usual when I see a Disco / Rangerover I pointed the owner to this forum

I hope you told him that the Thor doesn't play nicely with aftermarket MAFs, it needs to be a genuine Bosch one.

Don't forget that with it disconnected it will run a default fail safe fuelling map, whereas with it running from a duff one, it will fuel to suit the incorrect readings from the MAF. Did you check the live readings? MAF output should be 17-23 kg/hr at idle and 57-63 at 2,500 rpm.

With the Maf disconnected it was running closed loop with flicking lambda voltages and fuel trims working as normal, which seemed a bit odd.

I forget the figures but the Maf flow figures but it was under-reading airflow so causing lean running. Also with it connected short term trims didn't climb to try to correct fuelling as usual, instead if just reported duff lambdas and went to open loop fuelling which was lean.