The engine revs on 769s are very odd, you take power in say notch 2 and revs rise but then they increase markedly again at a few mph despite not touching the power controller again. When you then open up to notch 3 there is another slight increase in engine revs but increasing from notch 3-4 seems to have no noticeable effect on engine revs.
On braking to a stand the engine revs only die down to idle when the unit reduces below around 5mph. This is consistent every time and isn't related to the compressor starting up - when the compressor runs the engine note doesn't noticeably change, whether at idle or taking traction power.
To be honest it was only speculation, hence my saying 'may be'.
On the subject of the acceleration patterns, it makes more sense, There are two nominally independant control systems here:
1) the traction electronics whose purpose is to take DC power from the bus line and according to the driver's setting of the control lever, apply the appropriate power to the DC motors - exactly as it has always done on a 319.
2) the genset control's role is to maintain the bus line voltage, - in effect replicating the 750VDC supply from the 3rd rail collector shoes or the rectified output from the transformer when fed by 25Kv 50Hz OLE.
So, when the driver selects a notch on the controller, (notch 2 in your case for starting), the traction electronics will set a current for the motors to meet that demand that must also be correct for the speed. At brakes release, the motors are stalled, so their current must be limited until their speed is such that it can be increased. This places a load on the bus line which causes the genset to increase it's output to maintain the voltage which initially only requires a moderate speed of the engines. Once the traction electronics senses the drop in the motor current due to its rotationa speed, it can increase the drive up to the normal current for notch 2. This loads the bus more and the genset responds by spinning up and increasing the excitation to produce more power. Selecting notch 3 then applies more power to the motors, causing a further demand on the bus, to which the genset responds. Now further traction demand (e.g. nootch 4 or an adverse gradient) requires more power but if the engine and generator have been optimally matched the engine will already be near the speed at which it can produce maximum power, so the excitation on the genset can then extract more power from the generator without much (if any) increase in its speed. When I was on a 769, once the speed was in the '60s the engine was effortlessly humming away at a relatively moderate speed and was almost drowned out by the whine of the traction motors 3 cars back.
Apologies for the convoluted explanation but hopefully it explains the fact that the driver controller does not directly affect the diesel engine spin speed and that the spin speed is not necessarily proportionate to the load on the engine, - the two independent control system manage that.
That is why a diesel-electric traction system is generally quieter, smoother, more efficient, and dare I say it here, more reliable*. The MAN D2876 is a proven rail traction engine, and in this application is running well within its design limits. The ABB generator is a proven commmercial 3 phase alternator and the traction electronics and motors, although 30+ years old under diesel power are running at less than 80% their class 319 maximum power levels. Indeed, the track limitations on the 769 speed even under OLE should further mitigate the reliability impact of their age.
* yes I know, this is a 769 thread, but it seems that the subsystem on the trains are in themselves reasonably reliable, - I believe that most of the issues are problems with their integration.