High-Load DC/DC Conversion
Robotics | Lighting | Electric vehicles
Most of the underlying technologies, design methodologies, and even sometimes parts are the same across several disciplines of energy engineering, from lighting to windmills to electric vehicles.
One of Luxonis’ core competencies is Energy Engineering. Several examples are given below, including lighting, power regulation/conversion, robot control, and vehicle drive, but photo above got the ‘hero’ spot on this page because of the unique nature of its deployment and subsequent challenges (overcome).
It’s a motor controller that we designed to be standard-issue for the University of Colorado mechanical-engineering course MCEN-5115; the ‘mechatronics’ class at CU Boulder. This board allows high-load robotics application under unpredictable conditions.
Why is it highlighted? Read on.
We found one smoking while supporting the first class-wide deployment of them. The 15A protection fuse had been removed, its mount desoldered, a 1/2” copper slug put in it’s place. Upon interviewing the students about this choice (after having stopped the smoking), they let us know they have removed it because ‘it kept blowing‘. The board, before we quickly disconnected it, was connected backwards (inverted polarity) to a 12V, 700A cold-cranking battery (700Amps short circuit). Since the fuse had been bypassed, the battery’ entire 700A capacity was shorting through the protection/flyback diodes on the board. Hence the smoke.
So, the students had been plugging the board in backwards the whole time, popping the fuse. So they removed it.
Curiosity being a virtue in science/engineering (see here), we plugged the board back in with the proper polarity. It still worked.
Our design goal for the board was for it to be bullet-proof, so that effectively whatever the students did with the boards, they would withstand it. Upon this discovery, though, we were worried that ‘bulletproof’ was, well, maybe a bit foolhardy.
So of course we were very happy to see that despite the first protection being physically removed by the students the secondary protection was sufficient to protect the board.
Luxonis also has experience designing products for automotive, specifically on military (24V) vehicle applications (which are similar to the requirements of standard automotive, for the most part, but with double the operating and transient voltage ratings). One design was a power conversion, protection, and filtering system for a system of radio components, while the other was the power system for the infrared lights discussed in RF Engineering.
Both designs had in common that they needed to handle without damage both voltage spike and current surge conditions, from MIL-STD-1275E, which can effectively be summarized by the following two plots:
Our design tactic was effectively identical between the two applications, broken into two parts:
Use input filtering to lower the transient voltage spikes shorter than 500 microseconds, to under |+/-180V|. So this passive input filtering would handle all spikes at +/-250V. This first stage made sure that the second never saw a voltage above |+/-150V|.
Leverage a high-voltage enable/disable MOSFET circuit, which would disconnect the following components (which could handle up to 40V) if the input voltage was 32V or above. This circuit could operate indefinitely up to +/-180V.
The way the circuits differed greatly, is how the MOSFET circuit was implemented. Because of the efficiency of the lighting circuit in the lighting application, it was relatively low power compared to the total power used by the radio system (which itself had a plethora of voltage rails, many of which were high load). So for the lighting system it was easy/simple to keep within the safe operating area (SoA) of the MOSFET disconnect circuit.
However, due to the high-load of the radio systems, and complexity (and high input capacitance) of the power conversion/regulation following this protection circuit, careful design and validation attention had to be paid to ensure the SoA of the disconnect MOSFET was not violated, and as such an additional timing chip was utilized in this design to keep the combination of MOSFET drain current, drain voltage, and transient time within the safe operating area. In short, the timing chip made sure the transition between the system being on (safe input voltage) and off (from an overload/voltage condition) occurred in such a way that the MOSFET doing the ‘turning off‘ was never exposed to a combination of voltage, current, and duration that could hurt it. Below is the system being run over a week-long period at 187V, to verify we didn’t miss anything.
It’s worth explicitly pointing out that according to the specification, 187V only needs to be withstood for say ~350 microseconds at a stretch, and usually would only occur on a vehicle if the battery somehow came completely disconnected when the alternator was simultaneously, at peak power. So a week is a long time with respect to the specification; we were testing much much further than the specification would dictate.
At Luxonis, some of our favorite projects involve Power Electronics, as there’s something specific about this work that makes it particularly satisfying in contrast to other forms of Electrical Engineering. And that something seems to be:
If you do it right, it works forever and its impressive how resilient it is.
If you do it wrong, it usually literally explodes. We’ve seen engineers vaporize solid copper before (luckily no-one was hurt).
And the second point there is why we tend to 0ver-test power electronics.