Under the hood
A close-up of part of the switching power supply
Controller IC:
In the centre of the photo is the heart of the switching power supply: the synchronous Buck controller IC. It has soft-start, thermal protection, output current sensing, cycle-by-cycle current limiting, and a host of other features which make for an extremely robust design. The controller IC drives two external 60V MOSFETs, each avalanche rated to 25 Amps. The PCB layout has been heavily optimised to minimise radiation from the two high-current loops. The current-mode control loop is unconditionally stable under all load conditions.
Inductors and capacitors:
Although inductors and capacitors aren't particularly sexy components, their performance is critical: designers agonise over their choices here (myself included), because these are the components that make or break a power supply design. Toroidal inductors have held a favoured place for many years, largely because of their low magnetic flux leakage, but they do suffer from being impossibly large. Fortunately miniature shielded inductors have now reached a point where they are not just a viable alternative, but are offering performance advantages. We use one of these miniature shielded inductors in the ServoStation.
In the output capacitor arena, finally there are some technologies that offer high capacitance & low ESR that (a) aren't the size of a house, (b) have reasonable environmental ratings, (c) don't degrade under bias, and (d) don't fail short-circuit and then catch fire. One such technology is conductive polymer tantalum, but there are others. None of them are particularly cheap, but they are just too good to ignore. We use several.
Reverse polarity protection:
Reverse polarity protection is not implemented using a Schottky diode - that would have wasted at least 0.4V of low voltage performance and generated 1.6W of heat at full load. The circuit we use has a forward voltage drop of 40mV - and that's at 5 Amps output current.
Loss of ground failure mode:
Using a Schottky diode is not only inefficient - it can be dangerous, because the diode will continue to conduct regardless of whether a ground is present or not. Consider for a moment the situation where the ground connection has failed: voltage regulation is now impossible (there's no ground reference), and so - depending on the type of regulator - input voltage may be appearing at the output. Any devices unfortunate enough to have a return path to ground will be exposed to the full unregulated input voltage, potentially damaging or destroying them.
The reverse polarity protection circuit we use requires a ground in order to conduct, avoiding this problem altogether.
Connectors:
A lot of products that use standard servo connectors simply line them up next to each other. Spacing the connectors out was more difficult to design and is more expensive to manufacture. So why on earth do we choose to do it this way?
- Current capacity: On a 0.1" grid there is simply not enough room to route PCB traces that can carry the required 5 Amps. Spacing the connectors out to 0.2" allows much wider PCB traces and 5 Amps (and more) becomes possible. When we see claims of 10 Amp capacity with 3 Amp rated connectors arranged on a 0.1" pitch grid, we sigh deeply and look the other way.
- Spacing: Most hobby-grade 0.1" pitch connectors are a tiny bit wider than they should be. If there's only a few of them it doesn't matter, but with 8 in a row they need a bit more room to breathe.
- Vibration: Those little tabs aren't just there for decoration. They lock into the servo connectors preventing them from working loose and falling out during flight.
- Shielding: Putting 8 connectors hard up against each other would result in two 0.3" x 0.8" slots in the enclosure. This would seriously degrade the RF shielding. Breaking the slots up so that the maximum dimension is 0.3" more than doubles the high-pass cut-off frequency of the enclosure.
