When is it best to use a battery or a power supply for testing my battery powered device?

As I do quite a bit of work with mobile battery powered devices I regularly post articles here on our “Watt’s Up?” blog about aspects on testing and optimizing battery life for these devices. As a matter of fact my posting from two weeks ago is about the webcast I will be doing this Thursday, June 18^th: “Optimizing Battery Run and Charge Times of Today’s Mobile Wireless Devices”. That’s just two days away now!

With battery powered devices there are times it makes sense to use the device’s actual battery when performing testing and evaluation work to validate and gain insights on optimizing performance. In particular you will use the battery when performing a battery run-down test, to validate run-time. Providing you have a suitable test setup you can learn quite a few useful things beyond run-time that will give insights on how to better optimize your device’s performance and run-time. I go into a number of details about this in a previous posting of mine: “Zero-burden ammeter improves battery run-down and charge management testing of battery-powered devices”. If you are performing this kind of work you should find this posting useful.

However, there are other times when it makes sense to use a power supply in place of the device’s battery, to power up the device for the purpose of performing additional types of testing and evaluation work for optimizing the device’s performance. One major factor for this is the power supply can be directly set to specific levels which remain fixed for the desired duration. It eliminates the variability and difficulties of trying to do likewise with a battery, if at all possible. In most all instances it is important that the power supply provides the correct characteristics to properly emulate the battery. This includes:

• Full two-quadrant operation for sourcing and sinking current and power
• Programmable series resistance to simulate the battery’s ESR

These characteristics are depicted in the V-I graph in figure 1.

Figure 1: Battery emulator power supply output characteristics

Note that quadrant 1 operation is emulating when the battery is providing power to the device while quadrant 2 is emulating when the battery is being charge by the device.

A colleague here very recently had an article published that goes into a number of excellent reasons why and when it is advantageous to use a power supply in place of trying to use the actual battery, “Simulating a Battery with a Power Supply Reaps Benefits”. I believe you will find this to also be a useful reference.

Why have programmable series resistance on a power supply’s output?

A feature we’ve included on our 663xxA Mobile Communications DC Sources, our N6781A 2-quadrant Source Measure Module, and most recently our N69xxA and N79xxA Advanced Power System (APS) is the ability to program in a value for a resistance that exists in series with the output voltage. So why do we offer this?

 Batteries are not ideal voltage sources. They have a significant amount of equivalent series resistance (ESR) on their output. Because of this, the battery’s output has a voltage drop that is proportional to the current drawn by the DUT that is being powered. An example of this is shown in the oscilloscope capture in Figure 1, where a GPRS mobile handset is drawing pulsed transmit current from its battery.

Figure 1: Battery voltage and current powering a GPRS handset during transmit

In comparison, due to control feedback, a conventional DC power supply has extremely low output impedance. At and near DC, for all practical purposes, the DC output resistance is zero. At the same time, during fast load current transition edges, many conventional DC power supplies can have fairly slow transient voltage response, leading to significant transient overshoots and undershoots with slow recovery during these transitions, as can be seen in the oscilloscope capture in Figure 2.

Figure 2: Example general purpose bench power supply powering a GPRS handset during transmit

It’s not hard to see that the general purpose bench power supply voltage response is nothing close to that of the battery’s voltage response and recognize that it will likely have a significant impact on the performance of the GPRS handset. Just considering the performance of the battery management, the battery voltage drop during loading and rise during charging, due to the battery’s resistance, will impact discharge and charge management performance.

We include programmable resistance in the above mentioned DC power supplies as they are battery simulators.  By being able to program a series output resistance these power supplies are able to better simulate the voltage response of a battery, as shown in Figure 3.

Figure 3: N6781A battery simulator DC source powering a GPRS handset during transmit

While the 663xxA and N6781A are fairly low power meant to simulate batteries for handheld mobile devices, The N69xxA and N79xxA APS units are 1 and 2 KW power supplies meant to simulate much larger batteries used in things like satellites, robotics, regenerative energy systems, and a number of other higher power devices. Figure 4 shows the voltage response of an N7951A 1 KW APS unit programmed to 20 milliohms output impedance, having a +/- 10 amp peak sine wave load current applied to its output.

Figure 4: N7951A 1 KW APS DC source voltage response to sine wave load

Programmable series output resistance is one more way a specialized DC source helps improve performance and test results, in this case doing a better job simulating the battery that ultimately powers the device under test.

Two-quadrant power supplies are better than one!

Back in October, I posted an explanation about what was a bipolar (four-quadrant) power supply (see post here: http://powersupplyblog.tm.agilent.com/2012/10/what-is-bipolar-four-quadrant-power.html). That post covered two-quadrant supplies as well. Last week, while in Lorton, Virginia, I had an opportunity to meet with some of our U.S. Army customers  - engineers working at Fort Belvoir. Many of the engineers worked in the Counter Measures Research Laboratory (CMRL). While they are very careful to not reveal any details about the specifics of the work they do, one of the engineers shared a story with me about two-quadrant operation that is worth repeating.

The story was told while I was providing a demonstration of one of our power supplies, the N6705B DC Power Analyzer (see Figure 1). I was explaining to a group of engineers that some of the 34 power modules that can be installed in the N6705B are two-quadrant power supplies: they can source current and also sink current at one voltage polarity. Other power modules are four-quadrant power supplies: they can source and sink current, and provide positive or negative voltage. This explanation inspired one of the engineers to tell the group that the N6705B helped him solve a problem!

A battery operated device (he did not mention what it was) came into his lab because it was not functioning properly: it had some type of intermittent problem. In an attempt to reproduce the problem, he removed the battery and connected the device’s power input terminals to a power supply on his lab bench. But even after running the device for long periods of time and through all of its operating modes, he was unable to reproduce the intermittent problem.

One of his colleagues suggested he try connecting the device to a two-quadrant power supply installed in the N6705B they owned. The original power supply he was using was a one-quadrant supply – it could source power, but could not absorb power. The battery that normally powers the device can source and sink (absorb) power, so perhaps a power supply that more closely mimicked the behavior of the battery could help uncover the problem. Well, this worked! With the device connected to the two-quadrant power supply in the N6705B, the intermittent problem showed up again proving that it was related to the battery being able to source and sink power – a power supply with similar characteristics was needed. Apparently, the device has a mode in which it momentarily forces current back out of the battery input terminals. That current is normally absorbed by the battery. And during that time, this intermittent problem must show up. During test, a single-quadrant power supply is unable to absorb the power and therefore does not reveal the problem. A two-quadrant power supply can sink the momentary current, and the problem was back, enabling the engineer to track it down and eliminate it! See Figure 2 for an example of the output characteristic of a two-quadrant power supply.

This example demonstrates the importance of choosing a power supply with the right output characteristics for your test. When testing a device or circuit with a power supply, the closer that power supply’s behavior is to the actual power used with the device or circuit, the more you will reveal about the actual performance of your device or circuit.  There are applications in which a two-quadrant power supply will better replicate a battery’s behavior than a single-quadrant power supply, even if you don’t expect the battery to absorb power during test. One CMRL engineer experienced this firsthand.