Optimizing power supplies for test applications
Most electrical engineers believe they have a good understanding of power supplies because they are relatively simple, single-function DC devices designed to output controlled voltages. However, there is much more to them than this description would suggest. Although a review of a power supply’s specifications should always be a part of the selection process, other characteristics should also be considered.
Investigate the power envelope
The most significant decision is ensuring that sufficient power is available to energize the device under test (DUT). Different types of power supplies have different power envelopes. A power supply with a rectangular power envelope as shown in figure 1a, the most versatile type, allows supplying any current to the load at any voltage level. A supply with multiple rectangular envelopes for multiple ranges (such as the two-rectangular envelope shown in figure 1b), permits higher values of one parameter at the expense of the other parameter, so it can output a higher level of current but at a lower maximum voltage. Supplies that output a hyperbolic envelope offer a more continuous transition than a multi-range power supply, with one parameter inversely proportional to the other – see figure 1c. High power output supplies tend to have multi-range or hyperbolic envelopes.
Determine the noise performance
Noise from external sources may cause problems when powering a circuit that operates at a very low voltage or a circuit that uses or measures very low currents. The supply itself is one source of noise, which can be broken into two components: normal-mode noise and common-mode noise. Normal-mode noise is generated across the supply’s output terminals due to the supply’s internal circuitry. Common-mode noise is earth-referenced noise originating from the power line and stray capacitance across the main transformer. For sensitive circuits, linear power supplies provide much lower normal-mode output noise than supplies designed using switching technology but have lower power-conversion efficiency and can be bulkier and heavier. Switching supplies typically offer more output power in a smaller enclosure. For noise-sensitive circuits, a linear supply can have just one-fifth to one-tenth of the noise (5mVp-p vs. >50mVp-p) of a switching supply. Whenever normal-mode noise is a crucial consideration, use a linear supply, such as Keithley’s Series 2200 single- and multi-channel power supplies, if possible.
Assess common-mode noise current
Linear power supplies generally have lower common-mode noise than switching supplies. Common-mode noise is generated whenever changing voltages, such as AC voltages and transients (dv/dt) on either the primary or the secondary windings of an isolation transformer, couple current across the barrier. Whenever this current flows through an impedance, the noise voltage generated can degrade load (or DUT) performance or cause load-monitoring measurement inaccuracies. Sources of common-mode noise include voltage transients from rectifier diodes (on the secondary) turning on and off and either the 60Hz line movement or the abrupt voltage transient common with a switching power supply’s primary circuit.
Figure 2 shows a simplified block diagram of a power supply. The quality of the transformer’s construction, including sufficient shielding between the primary and secondary windings, can minimize the stray capacitance between primary and secondary. With minimal coupling capacitance, the noise current flowing through the load won’t generally affect the load’s operation or impact measurements on the load. If the transformer’s primary and secondary aren’t sufficiently shielded from each other, then the coupling capacitance can be large and milliamps of current can flow into the load, creating performance problems and load current measurement errors. For low power and sensitive components, modules, or end products, evaluate the power supply for low common-mode performance. Keithley’s Series 2200 power supplies have common-mode currents of less than 10µA.
Check isolation from Earth ground
One further indication of the quality of a power supply is the isolation of its output is from the power line. A power supply with high isolation further minimizes noise on the supply’s output. A good level of isolation impedance includes parameters greater than 1GΩ in parallel with less than 1nF and shielded well enough to support less than 5µA of common-mode current. Unfortunately, few instruments meet or exceed these guidelines.
Low frequency 60Hz designs may meet the common-mode current specification but fall short of the DC resistance and capacitance figures; switching designs may have low capacitance and higher DC isolation but excessive common-mode current. In some applications, the DC isolation resistance and capacitance are more important than common-mode current. One case in which the high impedance is important is when a supply is powering a circuit driven by a linear amplifier. In this situation, the power supply is part of the load on the linear amplifier and a large power supply capacitance can create stability problems for the amplifier.
Alternatively, a supply being used to power a low voltage resistive divider or a very low current measurement circuit may need low common-mode current, regardless of the isolation impedance. Generally, the higher the isolation, the lower the noise coupled through the supply from the AC power line. The problem becomes more complex when the application involves other instruments. In this case, insufficient DC isolation in the power supply can provide a conduction path for a high common-mode current from one of the other instruments. For any particular power supply application, it’s crucial to understand the effect of the power supply isolation resistance and capacitance on the DUT, and the path or loop where the primary and secondary common-mode currents flow in order to determine if a noise voltage (common-mode current × impedance) will be developed and whether the noise will be excessive.
Ensure sufficient isolation between channels of a multi-channel power supply
If a DUT requires individual isolated power supply sections, then either a number of individual isolated supplies or a single multichannel output power supply will be required. If using a multichannel power supply, always ensure that the isolation between the power supply channels is greater than the isolation required between the DUT circuits. However, that’s not always easily determined just by reading a multi-channel power supply’s data sheet. Some power supplies don’t actually provide isolation between channels. However, Keithley’s Model 2220-30-1 dual channel programmable DC power supply and the 2230-30-1 triple-channel programmable DC power supply shown in figure 3 have two and three fully isolated channels respectively. When the isolation between circuits in a DUT is critical, consider actually measuring the power supply’s isolation between its channels.
Maximize output accuracy
If tight control of voltage at the load is essential for research experimentation, device characterization or production testing, then a careful review of the power supply’s output accuracy and read-back specifications are important. However, that accuracy can be compromised if the supply is controlling the voltage at its output terminals. What’s needed is feedback control right at the DUT, which means the supply should include sense connections (remote sensing) that can be connected to the DUT where the power leads are connected. The sensing circuits measure the voltage at the DUT so that the supply can compensate for any voltage drop in the test leads – see figure 4.
No matter how accurate the power supply output is, there’s no way to guarantee that the programmed output voltage is the same as the voltage at the DUT’s load. This is because a power supply with two source terminals regulates its voltage only at its output terminals. However, the voltage that is important to regulate is at the DUT load, not at the power supply’s output terminals. The power supply and the load are separated by lead wires that have a resistance (RLead) determined by the length of the lead, the conductivity of the conductor material, and the geometry of the conductor. The voltage at the load, without remote sensing, is:
If the load requires high current, then ILoad is high and VLead can easily be a few tenths of a volt, especially if the power supply leads are long, as can be the case in an automated test rack. A voltage at the load could easily be 80mV to 160mV lower than the desired voltage (with 2A to 4A flowing through a five-foot length of 0.004O/foot, 16-gauge wire). The remote sensing technique solves the problem of voltage drop in the leads by extending the power supply’s feedback loop to the input of the load. Two sense lines from the power supply are connected to the DUT power inputs.
These sense leads are voltage measuring lines that connect to a high impedance voltage measuring circuit in the power supply. Given that the voltage measuring circuit is a high input impedance circuit, the voltage drop in the sense leads is negligible.
The sense lead voltage measurement circuit becomes the feedback control loop for the power supply. The voltage at the load is fed back to the power supply by the sense leads. The power supply raises its output to overcome the voltage drop in the source leads and VLoad = VProgrammed. Thus, only with remote sensing can the accuracy of the power supply be applied to the load. Although power supplies can be considered simple from the standpoint that they are single-function instruments, their power output envelope, their design topology (linear vs. switching), their isolation performance, and their ability to perform remote sensing are all considerations that are important for sophisticated and sensitive design and test applications. Because some of these parameters aren’t specified, the power supply may have to be evaluated. As this article outlines, a few simple measurements can help ensure that a power supply meets the intended application.
About the authors
Robert Green is a Senior Market Development Manager focusing on low-level measurement applications.
James Niemann is a staff engineer responsible for designing instrumentation used in low-level measurements.
Qing D. Starks is a staff applications engineer.
All three work at Keithley Instruments. www.keithley.com
- Near Field Communication (NFC) technology and measurements – Part 1: Overview
- Near Field Communication (NFC) technology and measurements – Part 2: NFC RF measurements
- A measurement approach for IQ offset and imbalance of LTE mobile devices
- Mobile measurement supports onboard EHV high-voltage testing
- Perfect timing: performing clock division with jitter and phase noise measurements
- Measurement accuracy/signal integrity prove king of banner specs
- Measure surface texture and lead angle of shafts for rotary dynamic sealing—An alternative
- Understanding the impact of digitizer noise on oscilloscope measurement