Hi Mike
Have you tried the self learning feature of the recent Märklin decoders?
This is about automatic calibration using CV7
- Install an oval / Circle with minimum R2 radius. Alternately, install the loco upside down on a foam bed and connect both poles of digital source (B and 0)
- make sure the motor type matches the reality by CV 52= 1 (SDS) =2 (No feed-back control); = 3 (C90 HLA motor); = 4 (Bell armature); =5 (DC soft motor); =6 (DC hard motor); =7 (1 Gauge motor). The real important point is not to mix SDS (Sinus) motors with regular DC motors. In your case the CV52 should be 6 or 5)
- start the calibration procedure (see page 30 of attached document). Using a CS2 or CS3, overwrite in the field "Firmware" the value 77. Alternately (MM or DCC) write 77 in CV7
- just start your loco using the speed control knob and the decoder will take over (accelerating braking stopping). Let it finish its cycle and you are all set
- finish by adjusting the acceleration delay and brake delay (Respectively CV3 and CV4)
- beware to to reduce too much the minimum speed CV2 Factory value =4 don't go below 2 or you risk burning the motor when it stalls with an heavy train)

The mSD3 manual is here mSD3 UsersManual.pdf (9,785kb) downloaded 52 time(s).

Cheers
Jean

Hi Mike
The present Märklin decoders (mSD3) have a current limit to protect both the decoder and the motor. The current limit is 1.1A permanent

I never found clear definition of DC Hard and DC Soft motors. All the recent Märklin (with CAN motors) locos I have inspected seem to be configured wit DC Hard

What do you call surging? Is this a real acceleration or a change in noise?

One possibility is that adherence is changed at the end of your curved ramp. Some locos lack flexibility in vertical curvature of tracks which can lead to adherence changes (motor running faster because the traction effort is reduced). Another aspect is the non-helicoïdal shape of tracks (a long 2 axles car will not stand on only 3 wheels. Almost nowhere helicoidal track shape is allowed (derailing and loss of traction risks)

Another possibility is the elasticity of couplings on a very long train that may cause oscillations (accordion-like behavior). Using Märklin close couplings, there is little elasticity to be expected.

Cheers
Jean