It's nothing new. It's something I have talked about on the forum many times over the years. If you search through my posts for words such as linear or 94Nm, you'll find it. Initially we tested using the OWUT across all weights to see what would happen. But early on we realised you need a certain amount of pressure applied during the test for the results to be comparative and reliable/accurate. The graph I posted above is from my own vcu results posted on ere years ago. There's a link in me sig below.
Are you not mistaking reliable for accurate - and is the 5kg test not accurate enough? We have seen many results with 5kg, and they appear reliable.
Your graph shows 8kg at 16 seconds and 5kg at 35 seconds. So the 5kg test is 218% of the 8kg test.
The tests by
@dfossil in post #112 come out at ...
8kg 13s and 5Kg 30s - so 5kg is 231% of 8kg
8kg 12s 5kg 28s, so 5kg is 233% of 8kg
8kg 12s 5Kg 24s, so 5kg is 200% of 8kg
So his findings are roughly the same as your graph.
@kernowsvenski tests 16 months ago (post #1171) shows 8kg at 23 seconds and 5kg test of 41 seconds (both first thing cold and frosty). So the 5kg test is 178% of the 8kg test. They are nearer, but not earth shatteringly so.
@kernowsvenski latest tests shows 8kg at 32 seconds and 5kg test at 44 seconds. So the 5kg test is now only 137% of the 8kg test.
I agree, something is wrong here.
However - what is wrong? The 2 5kg tests both gave very similar times. They were a couple of seconds longer than the previous test, consistent with a bit of aging on the VCU, but well within "the time to replace" that gets used of 60 seconds. Only 1 8kg test is shown and it is much slower than the previous one.
Would the sensible advice therefore be that the 5kg tests indicate that the VCU is past its best but not dangerously so - however the 1 8kg test is alarming, so test again at 8kg, as it is possible that this latest test gave an erroneous result?
If the test comes back again at 32 seconds, then one has to consider whether that is OK or the 5kg "time to replace" of 60 seconds needs revising.