Stephan's Sonex #1627 Builder's Log

116.40 hours
Task List 48 n p.png
Log #DateDescription Avionics Power Conditioning Unit Time
0022 2013/05/19

I worked on the design of the power conditioning unit today. The power conditioning unit takes input power from the onboard alternator and regulator and provides the necessary voltages for use by the MCU and sensors. A total of 3 different voltages will need to be provided: +12V, +5V and +3.3V, with a total combined power output of approximately 36W. This is just a preliminary design goal, as I have not finalized the EFIS design yet.


For the power conditioning unit, I chose a typical SEPIC converter design because of the need to be able to handle both input voltages well below the minimum desired output voltage, as well as above the desired output voltage. Although it is more likely that a failure in the regulator will cause an overvoltage condition (e.g. 16-18V), the possibility does exist that a partial regulator failure could cause a significant drop in output voltage (8-10V).


The power conditioning unit must accept this wide input voltage range and continue to produce clean +12V, +5V and +3.3V voltages to the EFIS. I used Linear Technology's online web tools to determine the best possible controller chips for the task. I could have started with National Semiconductor, but I've had good luck in the past with Linear so I started there. There were a number of chips that matched my input range and current criteria.  I then cross-referenced these results with DigiKey to ensure that the chip was stocked and readily available.


I settled on the LT3759. This is an MSOP package chip and available in the industrial temperature range (-40°C to +125°C). This is important, as I may decide to have the MCU firewall-forward, as it will be easier to connect the thermocouples and other sensors this way.

I started with the reference design for the LT3759 and then made changes to accomodate the input range as well as the line conditioning and slow start delay. I used the SPICE circuit emulator from Linear Technologies called LTSPICE to do the circuit analysis. I primarily used their tool because it had a SPICE model for the LT3759 controller.


I still have an issue with the voltage spiking to +30V before steady-state is reached, but I think this is because I'm not giving SPICE enough time to calculate the circuit's dampening factors, so it's making incorrect assumptions. (I have since determined that the issue was because I had used an ideal voltage source. Once I specified the internal resistance of approx. 20 milli Ohms and parasitic capacitance typical of a lead-acid battery, the voltage spike was greatly diminshed). I'm only interested in the steady-state, and from that point, the results look good.


In the graph below, you can see the performance of the circuit. The blue line represents the input voltage from the alternator/regulator. I purposely set this up in the simulation as a 7V p-p sinusoid with a 10V DC component. This causes an input voltage swing from 3V all the way to 17V. 


The output of the SEPIC controller is the green line. You can see it is fairly steady at 12V regardless of the input voltage. The noise is approx. 400mV p-p at a frequency of 300kHz. I'm not going to bother filtering this any further, because I'm going to have the controller feed traditional regulators for the +5V and +3.3V.

The output current is the red line. It is simply the output from the +5V regulator into a 4.8 Ohm load. 


SPICE Circuit Simulation - Blue - Input Voltage; Green - converter output; Light Blue - 5V Regulator output; Red: Load current

Preliminary Power Conditioning Unit for the EFIS

Revised Power Conditioner with +12V, +5 and +3.3V outputs

Revised Power Conditioner SPICE Simulation output