UPDATE: I connected the 24 volt regulated DC output of the Velleman switching mode power supply into the PIKO 35002 Electronic Throttle and I have very smooth operation. The LGB outboard digital European steam sound module's threshold voltage is higher than the starting voltage of the LGB 2085D loco, but once it starts, it is pretty cool. The engine air pump sounds, as well as the coal shoveling sounds, continue when the track power is off by means of the added capacitance circuit. The advantage of the PIKO 35002 is the reversing capability of the throttle and the throttle has a larger dial than the small pot on the Velleman voltage output. The Velleman unit by itself runs the trains fine, but would require a DPDT reverse switch.
"HONGZ" stands for HO scale, N scale, G scale, and Z scale.
Post your non-O scale stuff here!
To better match the sound to motor movement, each pair of opposite facing genral purpose diodes with legs twisted added in series to one motor wire will lower the start-up voltage by eating .75v. That loss effectively raises the motor threshold closer to the voltage where the sound board kicks in.
6v+ or (-) ___ _, --->|----,___,---->|----,___ 4.5v to Motor
. ' -----|<--' '------|<--'
That or you have a tach issue
Adriatic: Definitely no tach issue as the 2085D engine is fully analog. The sound module is a Marklin/LGB not-in-the-loco unit and relies on the variable voltage output for the supposedly synchronized chuffing sounds which are initially preceeded by a few whistle blasts. Is there a way to lower the threshold voltage on the sound unit?
BTW , the cost of the 5 amp PIKO electronic control was $160 while the Velleman DC switched mode power supply was $92 for total of $252 plus about $25 total shipping. My rational is the Velleman unit has a lower ripple output to feed into the Piko 35002 throttle.
It would be more complicated to get a voltage higher than the feed voltage to feed the sounds with more voltage to make them come in sooner. But voltage boosting can be done. If x-volt X-amp you'd get a larger X-volts and lower x-amps to use at the booster output (upper lower case x vs X notes change in value high or low)
It's far easier to use a few diodes to lower the feed voltage just to the motor (or to the sound to make it come in later if I've misread the issue).
Offhand I'd say the diode drop is also less likely to introduce noise. It's also dirt cheap. So say it does cause noise in combo with that motor; it's a cheap option to remove too. A noise filtering capacitor might cure it though. And useful as they are, you could likely use them on led lights or lowering voltage on something else.
A reminder a bridge rectifier simply holds four diodes within it and can be set up to do the same voltage drop.
The more oversized the diode/BrdgRect. the cooler the diode/BR tends to run. Going with 6-8adc means the chance of burn out should be nil as you only put out 5a.
I've read many boards have an 18v-20v max.. You have 24v on tap?(What's your operating voltage range?) If you don't have known sound board specs this voltage drop might come into use on track feed to prevent overvoltage by accidentally spinning the dial too far.
ADRIATIC: THANKS for your suggestions; you have given me some ideas to ponder! ☺
I recently added a reed switch so I can wave an LGB magnet over it and blow the whistle and ring the bell whenever I like! GREAT FUN! NOTE: My LGB Euro steam sound module is mounted near my DC controls, and not in a car or engine.
You could run an (other) remotely mounted reed(s) parrallel too if you wanted to do a trigger at the rails. An on/off too if you want to skip it, a latching relay for a "set and wait", a one time automatic trigger.
A few years back Dale H set up patterns with relays and capacitors to get varied results in length of the blasts as well. It might be searchable here or might be at J&C Studio's O Gauge Archive too.
The circuit won't care too much what style of, or how many switches are involved as long as the board contacts get closed. You could jump from the reeds legs to a regular button as well, having two triggers.
I'd guess the right reed could get picked up by a motor field. I KNOW an open frame motor can trigger some, and magnetraction; but maybe not a can motor, especially vertical. But magnets, hot glue, and reeds are effective if you wanted some "primitive" automatic sound activation to mess with 
