There isn't much of a schematic regarding this experiment -- I used an AWG to generate the drive signal, amplified it by a DIY wide-band amplifier (LT1210 based, an off-the-shelf unit would have done just as well), and through a coupling capacitor (MLCC 4.7µF) to a common mode choke (CoilCraft
SBU9-2820R5L_ would be a decent choice). On the secondary, a schottky bridge rectifier was used along with some RFI reduction and voltage post processing stuff (the PSU had to power a sensitive analog front end with considerable power consumption -- some 5W IIRC).
In your case, you could well use the mentioned MX116 motor driver to power the system, but I would use it single-ended and utilize the capacitive coupling mentioned above. If your power requirements are really only 10mA@5V, you can use a single driver to feed all 16 transformers (8 on each output) and driving both inputs with a 70~100kHz square wave of opposite polarity. Use a voltage-doubling rectifier (2* small Schottky, 2* 1µF MLCC) and if the stability and load regulation isn't too important, you can adjust the supply voltage of your driver to provide the proper output voltage (loaded). If inrush/switching currents are a problem, you may want to place inductances in series with the diodes, but due to the very small required power, I'ld initially try without.
Of course, that's just a quick-and-dirty suggestion that would probably be a reasonable starting point that will provide very good high-pot isolating characteristics.