I actually beta tested this board, and I've tried it in a couple of different scenarios.
The issue here is that it doesn't vary the AC voltage as the PS/1 board expects to change the primer mover speed (or chuff for steamers). It simply passes the AC through and imparts the DC +/- bias based on the command inputs of horn or bell. This works great for many common conventional horn/bell boards or something like the Polar Express announcement coach. That was one of the target projects when this board was originally conceived. The output section of the schematic is below, the red arrows denote the triggers for whistle and bell outputs from the decoder. As you can see, the sound card is powered by the AC coming through the two regulators, that's how the DC offset is inserted into the AC stream. From my tests, these are good for several hundred ma of power without any heatsink, and you should be able to do quite a bit more if you bolt them to a heatsink or the chassis.
![ERR Whistle-Bell Board Outputs ERR Whistle-Bell Board Outputs](https://ogrforum.ogaugerr.com/fileSendAction/fcType/0/fcOid/48030224949526885/filePointer/48030224953956841/fodoid/48030224953956837/imageType/MEDIUM/inlineImage/true/ERR%2520Whistle-Bell%2520Board%2520Outputs.jpg)