1. just lay out the pcb with the normal microcontroller cktry on one side and the h-bridge
cktry on the other side, and only connect them together at a single ground point. Some
people will use an inductor [choke] or ferrite to make the connection.
Will do, but using star traces to that single ground point on the power side. Your mention of a ferrite, do you mean the small ferrite beads such as the following SMD bead type? http://tinyurl.com/aygg7hg
...and when you say make the connection, I assume I only need one? So I connect the control side of the PCB to that ground via, with a small SMD ferrite bead
2. once this is in place, simply run the h-bridge control signals from the microcontroller
across to the h-bridge. Always good to put series-Rs in these lines to help protect the
uC from motor noise feeding back from the h-bridge.
Okay - I haven't tested with series resistance yet, I'll whack some 1k resistors onto the control lines tomorrow and check my waveforms. No problem with feeding these as visa through the board, won't long traces have quite some inductance?
3. locate the main terminals for motor power on the h-bridge side of the board, so the
motor currents never cross to the other side, and never flow through the single point
ground connection of item #1. Add bypass [0.1 uF] and reservoir [100 uf or more] caps right at the same terminals.
My h-bridge is a dedicated IC, an LMD18200T - I have the 12V supply for this IC decoupled with .1uF and a 220uF reservoir as per application note, and the outputs have bootstrap caps back into the IC, so do I need these caps on the output? Or do you mean two caps across the two motor terminals? I was planning to do this already to help reduce noise.
4. it's best to use 2 separate external power supplies, one for the logic part and another for
the motors. If you can't do this, then the next best thing to do is connect a single supply
to the motor side, and jumper "1" wire over externally to supply logic power. This way,
the motor currents are always confined to the h-bridge side, and not flowing between
the 2 sides. Don't make an external ground connection between the 2 sides, as then you will have a ground loop.
I've only got one external power supply, but have a 60 way connector on to my PCB and have a couple of spare pins - perhaps providing 12v to two pins from the same source, so each side of the PCB has its own 12V pin would serve a similar purpose?
When you say external ground connection, as in - the ONLY ground between each side should be the one ground via discussed under heading 1?
5. if you have analog cktry on the pcb, confine it to a separate area from the digital logic
cktry, and also on the opposite side of the pcb from the h-bridge cktry.
My only analog signals are analog voltages going in to the ADC of the MCU, so I can't really avoid running these from the pcb connector over to the MCU....
Thanks for your help so far, by the way :-)
]Can you you arrange it so that insted of all components being one layer of the board and allk the rest the other layer, have one group be one end of the board and the other group be the other end?
Then any traces & vias needed to make the routing work can be with the similar signal levels, and the high voltage/high current traces & vias can be seperated away, not overlapping the more sensitive lower voltage/current parts.
I'm afraid not, space constraints dictate otherwise - both sides of an 74 x 100 pcb are packed tight with components, it's going to be a challenge to keep components on the signal side not underneath high voltage / noisy components on the other side.