You could try increasing the timeout from 500ms to 5000ms. It's also worth noting that Mango Modbus offsets are 0 indexed, so if the device only responds to reads on valid registers and there are invalid registers around where you are reading, you may want to try the offset +/- 1.
Another possibility is that there is a register that needs to be written periodically for the device to respond. I've seen that before, but not on serial devices. You could investigate that route by either searching the documentation for anything like that or monitoring what sort of output the proprietary program is doing. It's also possible their protocol begins with a preamble that isn't Modbus at all. Presumably that would be in the manual.
One thing that strikes me is the "1 bit start." I don't see any way to configure the start bits in the underlying serial port implementation we use (and I don't see anyone complaining about that in my googling, either). I plugged a USB to serial converter into my Windows machine and checked its settings with the 'mode' command. Oddly nothing there about start bits either. You could launch the proprietary program, begin communication and collect
mode output for the serial port from the command prompt, then do the same thing with Mango and compare differences. On Linux that would be with the
stty commands (and in the stty help I also don't see a way to set start bits). That should let you see if the issue is caused by serial port settings.
Also I wonder if you have LEDs on the USB to Serial converter, and you can visually tell that the device is not responding to Mango.