I have a question regarding this mailbag video of the eevblog:
At this time he has a text at the bottom saying that the PCB ground looks inadequate and that there is to much “ground bouncy bounce”. https://youtu.be/eBzixNYF5K4?t=1956
What does he mean by that?
He is reffering to the situation when you have too much impedance (resistance / inductance) in the ground connection, and he is right, ODrive v3 has had quite a few issues with this.
Specifically, the term ground bounce is usually in relation to logic gates and drivers, but the concept is the same in general.
would you mind to explain how you went about this problem? I am still new to electronics and in my mind is this garden hose network analogy. So I suppose the problem is that the diameter of the hose that leads away from the drivers/mosfets is too small. I guess you could add a “bucket”(a capacitor) in parallel to it. Also you could make the lane bigger (increase the diameter of the hose). Or you could limit the flow of the current in front of the “knot” (gate,driver,mosfet) with a resistor. But how do you weigh these options against each other? also, is this more art or more science? Is there a feasible way to calculate the necessary width of the lines all over the circuit and the capacitance/inductance so that everything is nice and balanced?
I would say the best hydraulic analogy to inductance spikes would be a water hammer:
The standard way to prevent this in the GND track in particular is to use a GND plane. This is like instead of having the gnd in a pipe, you have a huge water-bed instead, that is so wide it can’t really build up too much momentum from the current.
For other tracks you will still have “pipes”, and yes you can calculate the inductance of those lines, and it is indeed standard to add capacitance to buffer the voltages that you need to be stable.