Due to a storm, my running track is not working at present. There are downed metal wires lying across the track! So I haven't been able to try running the engine. However, I did make some progress today. I have a Massoth Navigator and a Massoth 1200Z. So I connected the latter to my programming track and was then able to read the CVs one at a time, on the Navigator. A bit tedious, but I was able to get a value from all the CVs for which the loco manual gives default values. When I compared what I read with what the values were supposed to be, a pattern emerged. It looks as if every CV has had the rightmost 3 bits set to 1. So, if the value should have been 0-7, it shows as 7. If the value should have been higher than 7 then it shows as 0-7 higher than the default [depending on whether the 1st 3 bits should have been 0 or 1]. So, for example, a correct value of 64 shows up as 71. A correct value of 128 shows up as 135. However a correct value of 17 would show as 23. In particular CV 50 was supposed to be 10, but shows up as 15 - ie should have had bits 2 & 4 set to 1, but actually has bits 1-4 set to 1.
I have no idea why the 1st 3 bits have been overwritten in this way, but I shall be communicating with LGB/Marklin to ask them!
Unfortunately the Navigator doesn't seem able to write the CVs, it just reads them. It is supposed to, it just doesn't. I also tried the read/write on another loco. The read produced the correct values of the CVs for that loco - so the 3 bits set to 1 is not an artefact of the Navigator, it is actually what is on the faulty decoder. However, I was not able to write from the Navigator to the other engine either - even though I can write to it using my PC interface. I will need to find out why the Navigator doesn't write CVs. Maybe it doesn't do this on the prog track - the manual says it is a POM function. However, this is a problem for another day.
I have also downloaded and installed the Massoth Service Tool - which is essentially a substitute for the Stellwerks programming tools. This is also supposed to write CVs, so I will try that. The only problem is that it requires a template for the decoder, if you want to update more than one CV at a time, and I doubt there is one available for the Marklin decoder. Actually, I found, from CV8 in the loco doco, that the product code for this decoder is 131 - which is TRIX. I thought they made HO stuff, which I guess is where Marklin come from.
There are over 120 CVs for this loco with documented default values, and many others besides - almost all of which are wrong. I will be discussing with LGB the best way to fix this. Updating them one at a time is not my preferred option! However, I don't particularly want to send the thing back to Germany to be fixed either - unless they pay for the shipping. It is under warranty.