Well, I have been reading a bit about grid current and tubes and diodes, and the most shocking thing I learned is that this 'threshold' where the diode starts conducting doesn't exist! actually, it's a smooth exponential curve with a knee. Beyond the knee the diode starts to conduct 'a lot'. Before, it conducts little, but it does. Now,the grid current is in the order of microamps, so this all takes place way before the knee. And different diodes can behave differently in this region.
When I played with this in SPICE it appeared that two bav70 diodes in parallel behave very similar to a single 1N4148, wich was used in the first version of the circuit. So maybe the answer is quite simple: our KMG ran out of 1N4148 diodes and started using the parallel bav70's as equivalent circuit..
Anyway, although I have no definite answer to my question, I learned a lot by investigating. Thank you guys for your kind replies.
When I played with this in SPICE it appeared that two bav70 diodes in parallel behave very similar to a single 1N4148, wich was used in the first version of the circuit. So maybe the answer is quite simple: our KMG ran out of 1N4148 diodes and started using the parallel bav70's as equivalent circuit..
Anyway, although I have no definite answer to my question, I learned a lot by investigating. Thank you guys for your kind replies.