This document will report the results of your power supply characterization. It will include details on efficiency, voltage out, and noise.

Efficiency

Efficiency is defined as P_out/P_in. Switching supplies will usually have > 85% efficiency at normal loads. Linear regulators will have efficiency dependent upon the voltage drop.

Goal: Switchers should have efficiency > 85% across all expected inputs and loads. The efficiency of linear regulators varies much more and depends on the voltage drop.

Voltage out

Voltage out is simply the output voltage of the power supply. Vout should be very stable at the desired voltage over the normal range of Vin and loading, but may change significantly at extreme voltages and loads.

Goal: Vout should be within 2% of spec at all expected inputs and loads.

Noise

Switching supplies: These supplies introduce noise into the power lines. Measure the noise and report it in peak mV and in dB:
Noise in dB = 20 log₁₀ (AC noise level/DC level)     Note: This should be negative!
Example: 5V supply with a 250 mV peak-to-peak ripple on it.
Noise in dB = 20 * log₁₀(250mV/5000mV) = -26dB

Goal: Your supply should produce no more noise than the maximum spec'd ripple noise in the manufacturer's datasheet across all expected loads and inputs. Note: Please attach the datasheet with the maximum spec'd ripple noise highlighted. You should test ripple noise with the nominal voltage in and with a nominal load.

Linear regulators: Linear supplies should filter out noise. Simulate a noisy input voltage by summing a 60Hz 1.0V peak-to-peak sine wave with a DC offset equal to your nominal input voltage. Note: Hardware will be provided in the senior design lab for this purpose - see below. Measure the noise (peak-to-peak) on the output of your regulator and compute the noise rejection in dB:

Noise rejection  (dB)= 20 log₁₀(output noise/input noise)
Example: Input noise = 1000mV, output noise = 25mV.
Noise rejection = 20 log₁₀(25mV/1000mV) = -32dB

Goal: Your supply should have a noise rejection of at least -20dB. You should test noise rejection with the nominal voltage in and with a nominal load.

Hardware: To produce a DC voltage with a 1.0V ripple, you need a DC power supply and a 60Hz variable AC source (variac). You will find a variac and transformer pair in the Senior Design lab- the variac is blue-grey and has a large black knob on top of it; there is a wall-wart AC transformer plugged into it. The variac produces 0-140VAC; the transformer converts that to 0-10VAC. Wire the transformer in series between the positive side of the DC benchtop power supply and your test circuit. Adjust the DC power supply and variac to produce the correct waveform.

Range of measurements

Power supplies do not operate solely at one point, but at a range of input voltages and output loads. You should report measurements for voltage out and efficiency over an extended range of input voltages and output loads, with a minimum of five data points in each dimension (25 measurements). Report your results in tabular form. You may test noise (either ripple or rejection) at only one point - the nominal voltage in and nominal load.

Input voltage: Range from at least 1V below the expected minimum to 1V above the expected maximum. For instance, if your source is 4 AA batteries, the expected minimum is 4V and the expected maximum is 6V - your test inputs should therefore range from 3V to 7V.
WARNING - Do not exceed the maximum input voltage specified in your power supply's data sheet. If this is less than 1V above the expected maximum, modify the tests to include five points in the acceptable range.

Output load: Range from no load (open circuit) to 300% of the expected maximum load (or, 1/3 of the minimum equivalent resistance, if you like). See the notes on hardware, below. Note: Some switching supplies will not work without a load - in this case, use the minimum load specified on the data sheet instead of no load.
WARNING - Do not exceed the maximum current specified in your power supply's data sheet. If the 300% load will draw too much current, modify the tests to include five points in the acceptable range.

Hardware: For some projects, the load tests may require small-valued resistors (around 1 Ohm) that can dissipate quite a bit of power. For this purpose, use the variable power resistors supplied in the lab. These are about 12" long and brown in color - two are mounted on a white board.  Please do not adjust the resistors to the red zone, as this may result in burning them out.

Report

Your report should begin by describing your experiments and explaining how you chose the range of inputs to measure. You should then make a series of 5x5 tables for efficiency and voltage out, and report ripple noise or noise rejection for each of your power supplies.

If you have multiple output voltages, you should repeat the experiment for each output voltage.

Kevin Bolding January 31, 2012