Pedro Rodriguez

ECE480

Application Note

Design Team 11

Power System Overview and Buck Converter Operation

A power system is needed in the Electronic Parachute Deployment (EPD) system to convert the systems power source to the needed voltage potentials required by the active components. This power system was developed after the key active components in the design were chosen to ensure complete compatibility with the rest of the system. The most significant challenge of this system design was balancing efficiency, physical size, and cost. With the final power system design the team has developed a system with a good balance of these design criteria.

The power system of the EPD must meet the voltage and current needs of the other components within the system. The system needs three main outputs to the other components of the system. It must provide a 5V output to both the analog altimeter and the signal condition amplifier. The system also must output 3.3V for the microcontroller. The last output is a short 2A burst output to power the igniters.

The power source chosen for this EDP system is a simple 9V battery. This is the standard power device for commercial rocket altimeters and provides a large amount of charge in a small physical space. Other options considered were a Lithium ion coin cell battery. The 9V battery holds roughly five times the amount of stored energy

Once the power source was determined some power conversion was needed to change the 9V input to a 5V and 3.3V output. There are a few topologies that would allow this step-down conversion. The system could use a linear voltage regulator, a charge pump or a buck (step-down) converter. These power systems each are unique and offer different benefits.

A linear voltage regulator is typically quite cheap and usually requires two decoupling capacitors and the voltage regulator itself to operate. These devices can take a voltage at a high potential and output a lower voltage. This is done by dissipating excess voltage as heat, and therefore is quite inefficient.

A charge pump system operates by using a single integrated circuit that contains a series of switches and capacitors. Within the IC the input voltage enters and then is passed between capacitors using switches. These capacitors are to be sized in such a way that the voltage can be increased or reduced at the output of the device. These devices are more efficient than linear voltage regulators. The issue with charge pumps is that all of your power must go through the IC, which has a fixed amount of power it can handle. This usually limits the output current of charge pumps to the milliamp range.

Finally there is the buck converter topology. This topology is based around using an inductor as a storage device. Figure 3.5.1 below shows the basic design of the power system.

Figure 1 Simplified Buck Converter

When the switch in the buck converter is closed, the inductor coil will charge up to Vin according to the equation:). This assumes that the system started at 0V. This inductor will create a ¡®ramp¡¯ of current that will go into the load assuming Vin is a constant. After a certain time, the switch will open. The inductor will try and dissipate its stored energy out to the capacitor and the load that would be hooked up in parallel to this capacitor. This along with the fact that the switch is open will cause the node between the diode and the inductor to be at a negative potential compared to ground. This will forward bias the diode and allow an amount of charge to flow back into the inductor. During a single switching cycle the amount of current through the load capacitor (using Kirchhoff Current Law) would be equal to the current out of the inductor minus the current going out to the load. If this were to be calculated or graphed you could notice that average current that flow through the capacitor would be 0V. Kirchhoff Current Law is used to describe current entering a node at any instantaneous point in time. This Law could be extended to show that the average current entering a node must be the average current leaving the node. When this is now considered, assuming the average current through the capacitor is 0V, and then the average current through the inductor must equal the average current in the load.

By carefully regulating when the switch is opened and closed and by choosing the inductor you will be able to store energy in the inductor and release it slowly during discharge cycles. This allows the step-down of voltages with a theoretical 100% efficiency. In real systems power loses will be seen mainly in the inductor resistance, the gate resistance of a switch transistor (what is typically used as the switch in this system), and the voltage drop over the diode. Even with these loses the system can reach high level of efficiency in the 90% range.

The team initially chose to try and use a step-down (buck) converter to do this power

conversion. This converter would be quite efficient, but could be a bit more difficult to implement. In buck converter system you can buy an integrated control that will operate the switch and it will have some switching regulation control method. Also buck-converter integrated circuits may also move a switching transistor or the diode on to the chip, so less external parts are needed.

The team initially looked at trying to use a two-output buck converter IC, for example the TPS54386 from TI. This single chip along with external components would then change our 9V input to a 5V and 3.3V output. These dual type buck converter chips from TI required many external components and were quite complex. Using a dual buck converter system to get both output voltages would require more physical area and could increase the cost of the system. Another potential hazard is having the two inductors form the different supply rails couple in some way causing instability.

It was then determined to attempt to just use the buck converter system to output the 5V rail. This conversion from 9V to 5V could then be made very efficient. This would be beneficial to the overall efficiency because the two active devices, the analog altimeter and its instrumentation amplifier, on average require more current then the microcontroller running on the 3.3V rail.

With the 9V to 5V conversion in mind we set out to find a buck-converter IC that will be efficient, require few external components, and be fairly inexpensive. Our first choice was Texas Instrument¡¯s TPS560200. This converter met all the previous requirements, however when we received the chips the team realized that a new criterion was needed. This ICs were very small and would be difficult to solder by hand. So the team started searching for another buck converter IC with a better package. A new converter that looked promising was then found. This was the TPS62160 from Texas Instruments. This part worked well initially, but when more thorough testing was done a potential hazard was found. When testing this buck converter IC the team added an increasing amount of output capacitance to the output of the system. This was done while observing the effect of the total output capacitance on the output voltage ripple. More output capacitance would result in less output ripple which is desired in the EPD application. While increasing the output capacitance, the team noticed that the IC would go unstable and into thermal shutdown with an output capacitance greater than 220nF. At the point of instability the output ripple was quite small. It was decided not to use this IC knowing it has this stability issue. A third buck converter IC, Texas Instrument¡¯s LM2574, was then tested thoroughly and no hazards were found. This was found to be very robust, it required very few external components, and was the cheapest solution we had tried, so this is the chip chosen to be in our design.

With the buck converter then 5V supply was created, but there still needed to be a 3.3V rail created. Applying the 5V rail into the linear voltage regulator completed this objective. This method is not as efficient as the buck converter, but is very cheap and requires a minimal number of parts. This method does decrease the efficiency of the system, but the 3.3V power rail only powers the MSP430. The MSP430 requires little current to operate so the power dissipation in the linear voltage regulator is minimized. Initially the part used for this voltage regulator was the LM3940 in a T220 package. This then was changed in the final circuit revision to a LP2950, due to size and power needs. The original LM3940 was quite bulky and had the ability to provide a large amount of output current, which in our application was unused. The LP2950 was nearly identical in functionality to the original voltage regulator, except that it was smaller and had a limit of 100mA of output current. For the EPD system application this new voltage regulator will work.

Figure 2: Our Final power schematic

This designed power system provides a good mix of efficiency and cost for the EPD system. With the way the system is designed the power source is also flexible to allow voltages between 7 and 40 volts to be used. This feature was never sought after in the design, but this does allow the system to provide good voltage levels even when batteries lose their voltage potential slightly before dying. More specifically the system could also be used with the 12V A23 type battery that is fairly common. This power system has been in place through violent rocket launch testing and still is holding up to the power needs of the EPD system