Table of contents

RoadTest: Enroll to Review the Keysight Battery Emulator and Profiler E36731A

Author: saadtiwana_int

Creation date: 28 Jan 2024

Evaluation Type: Test Equipment

Did you receive all parts the manufacturer stated would be included in the package?: True

What other parts do you consider comparable to this product?: There are several manufacturers making Electronic Loads and Power supplies, but I couldn't find one that integrates both in one unit, like the E36731A does. There are also SMU (Source Measure Unit) instruments that could theoretically perform similar functions but the wattage is usually very limited. The Keithley 2281S comes closest as battery emulator up to 20V, 6A.

What were the biggest problems encountered?: I had some issues/observations related to Benchvue software and the current firmware. The documentation needs more details on the various functions as well.

Detailed Review:

Introduction

Batteries, often perceived as a mundane and dry subject, were no exception in my eyes when I embarked on my career journey 15 years ago. Yet, as I delved into numerous projects revolving around batteries over the past decade and a half, my intrigue grew exponentially. The industrial applications I was engaged with demanded a high level of performance over extended periods (imagine 7–10 years), which meant every aspect of the batteries' performance held significance. From capacity degradation over countless charge and discharge cycles and years, to optimizing shelf/storage life, calibrating the BMS inside battery packs, devising optimal charging and discharging methods, to strategizing the use of multiple batteries in a system, and more. These pursuits not only enriched my projects but also expanded my knowledge about batteries beyond what I had initially envisioned.

As a result, when I saw the Keysight E36731A roadtest listed, I was very excited and felt extremely privileged when I made it to the final list. Over the past couple of months, I revisited many battery tests that I had previously conducted with discrete instruments (Power supply and Electronic Loads). However, this time, I had at my disposal an instrument specifically designed for battery work! Needless to say, roadtesting the E36731A has been an absolute delight.

Unboxing and First Impressions

I wrote a fairly detailed "unboxing and first impressions" blog soon after receiving the unit. In the interest of avoiding repetition, I will encourage you to read the blog I posted earlier, linked below:

(Blog Post) Keysight E36731A: First look and Impressions

For this review post, I will just summarize by saying that the Keysight E36731A is a beautiful, well-built piece of instrument. It is not the smallest instrument for the benchtop, but it's definitely not big for it's category of instruments, and certainly smaller than if you were to replace it with separate Power supply and Electronic Load instruments of similar ratings.  The form factor is such that it is deeper than it is wide or high, so from the front side it does not take a lot of space on the desk. The hardware is very capable and feature-rich, and the BenchVue application adds some very desirable functionality on top.

For the full details, please take the time to read through my full blog post.

What is the E36731A (and why you need BenchVue app)

The E36731A is essentially an instrument with 2-quadrant power supply and Electronic load hardwares combined. The power supply hardware is rated up to 200W (30V, 20A) while the Electronic Load part is rated to 250W (60V, 40A). The datasheet and user manual of the E36731A give very detailed specifications for the E36731A and I encourage you to have a look at them for more details. However, here I want to give a high-level view of what it is (and what it's not) and to clarify some things related to the name. I have made a simple diagram to try to explain this:

Basically, when you buy the E36731A, you buy the hardware "instrument". Inside the E36731A instrument, there is circuitry for the power supply and Electronic Load functions, with connections to the front and back of the device. The instrument runs its own firmware, which controls and executes the instrument's functions as well as lets the user interact with the instrument. In this standalone form, the E36731A is usable as a power supply OR an electronic load (not both simultaneously!) and you use the front panel to configure it and observe the live status values. There is also the option of controlling the device via its USB or Ethernet interface using Labview, SCPI commands and such for automated testing.

Now to go from here to having a "Battery Profiler and Emulator", you need to purchase a license to the BenchVue "Advanced Battery Test and Emulation" software. This lets you run the battery test & emulation app inside BenchVue on your PC, which is the clever piece of software that lets you profile, emulate, charge/discharge and cycle batteries. Without this application, you have an instrument that can run as a power supply or Electronic Load (but not those advanced battery functions).

Btw, if you are curious about the hardware internals, Dave from EEVblog made a very informative teardown video some time ago, linked below:

https://www.youtube.com/watch?v=fZ6JXMoievE

Software Installation, Licensing & Getting Started

The E36731A unit I received had two ports at the back to connect it to a PC: a USB port and a Ethernet port. The optional GPIB port was not populated. In my case, I have been exclusively using the USB port to connect directly to the PC I had designated for this.  USB over ethernet is more of a personal preference for me (unless I really need to put distance between PC and instrument).

I found the software installation process very simple and straightforward. I googled "Keysight E36731A" and the top search result took me to Keysight's product page:

https://www.keysight.com/my/en/product/E36731A/battery-emulator-and-profiler.html

On this page I found everything I needed including the datasheet, user manual, and link to BenchVue software. There is also a link to check the latest firmware version. I compared it with my unit to make sure my unit was on the latest version, which it was. As of this writing (Jan 2024) the latest firmware is dated April 2023.

The PC application I needed for the E36731A is called "BenchVue Advanced Battery Test and Emulation". This application actually runs inside (or on top of?) the main BenchVue application, along with other apps that you have installed. So first of all, I proceeded with installing the BenchVue software, which was as simple as installing any other PC software via installation wizard. The download was < 450GB and the installation consumed slightly below 1GB disk space which is very reasonable.

After launching the BenchVue app, I clicked "Applications" on the home page which gives you a list of all the available BenchVue applications. I installed "BenchVue Advanced Battery Test and Emulation" from the list.

With regards to licensing, there are three ways to use the Battery Test and Emulation application:

1. After installing the app, you can get a 7 day trial license. This is what I used for the first week.
2. After the 7 day license gets expired, you can apply for an additional 30 days license after registration with Keysight.
3. For longer term use, you need to get a subscription based license. This can be node-locked or floating. There is no perpetual license available.

A few weeks after receiving the instrument, I received an email from the Element14 team to generate my machine's ID via the BenchVue application and share it with them. Keysight used this to generate a node-locked license file, which I received via email. This particular license type is referred to as "BV9211B" and costs ~850$ as of this writing. Installing it was quick and simple. I'll add a few annotated screenshots to show the process.

I noticed on the internet that some people were struggling with licensing, but personally it was "smooth as butter" for me. The only part I did not like too much is the fact that my subscription will end after a year!

On a side note, since the E36731A has an electronic Load inside, I was hoping that the BenchVue "ELOAD" application would work with it too, but I saw that the E36731A was not on the list of supported instruments. I hope this gets added in a future update because, from the screenshots, the ELOAD application seems to have some very nice features too. The same is the case with the "BenchVue Power Supply" application.

The Battery test and Emulation application can be launched in two ways. You can either execute BenchVue and then launch the battery app from the list of installed applications. I found it more convenient to launch the battery application using the desktop shortcut. 

Connecting to the instrument was quite simple. With the instrument connected to the PC via USB and powered up, you need 2 simple steps:

1. Click "Search and Connect to Instrument. The E36731A appeared below in a few seconds
2. Click the connect icon in front of the found instrument.

Once the instrument successfully connects to the BenchVue application, the user is presented with the four main operating modes to choose from:

I have included this screenshot because I think it gives a good overview of the main functional modes of the device offered via the "Advanced Battery Test and Emulation" application. In the subsequent sections, I have attempted to document my journey exploring these features in more detail.

Before moving further, I want to mention that there is an "Instrument control" tab that you can open to see what's being displayed on the instrument's physical screen. However, despite the "control" word, you can only see but not control anything from this tab. Being able to control the instrument from here would have been a nice feature from a remote control perspective. Perhaps it'll come in a later software update.

Safety measures

Batteries, especially the Lithium based batteries, can store a lot of energy and release that energy in a short amount of time. Thus, if not handled with care, they can pose a very real risk of heating, smoke, fire, and explosion. In my test plan, I laid out some ground rules for my testing so I wouldn't endanger myself or anyone around me. I stayed committed to these throughout my testing. The main points are the following:

• Don't leave any experiment running unattended. No experiments are to be left running overnight.
• Setup experiments in a place away from any materials that can spread fire (wood, cloth, plastic, etc)
• Keep the test setup out of reach of the curious toddler at all times.
• Keep the cell/battery packs inside "Lipo-safe" bags at all times during testing.
• Use connectors on all batteries and cables to avoid accidental shorting of terminals while handling. (I standardized to XT60 connectors as I had plenty of them.)
• I kept a fire blanket and CO2 fire extinguisher nearby. (CO2 is not the right type of extinguisher for battery fires but that's what I had available at home)

Luckily, I found that the ESD mat on my home-lab desk had a fire-retardant composition (I tested by trying to burn a leftover piece), which gave me some peace of mind when running experiments on my desk. In some images here you may see the battery/cells without LiPo safe bags, but those images are for illustration purposes only - I always moved the battery/cell back to the LiPo safe bags when running the actual tests.

There was one instance when I had to make an exception to the "no experiments running overnight" rule. This was when I was testing the battery cycling feature. In this case, I put a smoke detector right beside the test setup and slept on the floor nearby so I could get alerted in case of an issue (I'm a light sleeper).

Battery Charge/Discharge testing

Working with batteries in past projects, one of the things I spent a lot of time doing was charging or discharging batteries for one reason or another. Reasons included testing BMS over full cycles, testing aging, measuring capacities at different temperatures, calibrating coulomb counters, etc. So naturally, the first thing I tried with the E36731A was charging and discharging batteries.

It was a few weeks between receiving the E36731A instrument and getting the BenchVue license sorted. So in duration I experimented with just the instrument in standalone mode (without BenchVue application). On one side, I was amazed by the amount of options available on the device. For example, in the battery mode it gives a lot more options than most power supplies (slew rates, on/off delays, OCP delay, etc) I have used over the years. On the other side, I found that it didn't give me the drained capacity (in mAh) when I was discharging the battery in Load mode. This is something I have come to expect from any Electronic Load by this point. I really hope this gets added to a future firmware upgrade because this is a very fundamental measurement for Electronic Loads in my opinion.

Once I had BenchVue installed and working, it was super simple to perform charge or discharge cycles. I have quite a variety of batteries available, mostly Li-Ion or Li-Polymer of different sizes and configurations. In the end, I decided to stick to some 18650 cells I had received from a reputed manufacturer for a project. This was because these cells were in brand new condition and also because the bigger battery packs were taking much longer times for each charge/discharge cycle. In a typical day I can only manage 2-3 hours of uninterrupted time for my home projects and I did not want to leave any tests running without being around.

I first sourced an 18650-type single-cell casing to avoid soldering wires directly to the battery terminals. I then prepared cables with XT60 connectors for reverse-proof connections. I standardized on XT60 connectors since that is what I have on most other battery packs of mine as well. It's always a good idea to test with non-reversible connectors instead of just wires to minimize any chances of accidental shorting of terminals. I prepared the cables such that I can use the E36731A in 4-wire mode by utilizing the sense terminals in addition to the main terminals. I have annotated the image to explain what goes where:

The reason I prepared the extension cable for power and also a longer wire for sense terminals was because I was hoping to perform some tests on batteries inside a thermal chamber to compare performance between room temperature and cold temperature. Eventually, that did not happen due to some logistical issues and at some point I cut the sense wires to shorter size to avoid the wire resistance influencing my readings, just in case. 

In the above image you can see my full setup for these tests.  The white bag on which the battery is placed is a Lipo-safe bag I mentioned earlier. The actual tests were performed with the battery inside the LiPo safe bag.

For setting up a charge cycle, the main parameters you must provide are the Charge Mode, charge current and maximum battery voltage. I used the cell datasheet as a guide to make sure I wasn't exceeding the cell maximum ratings (voltage and current) during charge or discharge. I performed the tests at various charge rates and the results were exactly as I expected. I took some screenshots to share how it all looks in BenchVue.

The above screenshots were saved after the cycles had finished. On the left side you  can see the settings used, while the graphs on the right side show the current, voltage and capacity graphs. Note that when the cycle is running, these are updated in real time. We can also see that although we selected "CC" (Constant Current) mode, in practice it means CC mode charging followed by CV (Constant Voltage) charging. This is the correct way to charge batteries, and it's good that the software follows the same. CC charging until the end wouldn't make sense anyway. If we had selected the CV mode, then the CC mode would have been skipped, which takes much longer to charge the battery.

Similarly, the discharge tests were performed at different current settings, and the results were exactly as per my expectations.

Discharging a single cell at 1A current. Note in this case that the discharging happens in constant-current mode until the end. This exactly how we expect the discharge cycle to go under Constant Current mode.

Note that apart from the graphs shown, each charge or discharge test saves the data as a "DataLog" file also, which can be loaded back into the "Battery Viewer" tab at a later time and exported out in csv format for any further processing.

One thing that I really liked about the data logs is the information about the settings used in the cycle. This way, you don't have to make manual notes and can always come back and see what the settings for different experiments were. A small detail by very useful one.

My thoughts: Comparing with my previous experience of charging and discharging batteries in the past, I found doing the same via E36731A + BenchVue battery application an order of magnitude more convenient. Previously, I had to use a power supply for charging and an Electronic Load for discharging, and I had to manually move the battery between the two every time. This meant frequent trips to the test setup, if not being stationed there for the entire time. I also had to note down the readings manually. Lastly, I had no visibility over the voltage and current trend over the full cycle, because I could only see the current reading on the instrument at any time (and no graphs). I wish I had this instrument + software combination available to me 10 years ago...it would have saved me so much time and effort!

Battery Profiling

When I am roadtesting a product, I try to integrate the roadtest into my ongoing projects as much as I can, so I can spend more time doing the testing and also understand the real-world usage. "Battery profiling" may sound like something people wearing white labcoats would do when studying batteries in a science lab. But actually has practical use beyond just getting to know your battery better. For me, the ultimate goal was "battery emulation", because I wanted to test some electronics with batteries without actually using battery packs. For one, I did not know beforehand whether I would need a 5 or 6 cell Li-Ion pack because of certain factors in my design that I was not clear about. Secondly, I did not have enough cells with me yet to build the battery pack I needed to build eventually; I only had a few cell samples from the manufacturer. Last, but not least, I was apprehensive about directly testing my electronics with a battery pack of the size I was roughly expecting due to safety reasons. Li-Ion batteries are not as forgiving as some other chemistries (Ni-Cd, Ni-MH, Lead-acid), and testing new electronics with "live" batteries can be stressful. Emulation would be the perfect alternative. But for emulation, you need to first have a model of your battery pack. 

The E36731A provides the battery pack profiling functionality via the BenchVue software. However, I did not yet have the battery pack to profile. So instead, I started with profiling the cell sample I had. The plan was to build battery packs with this type of cells later. The cell samples were 18650 cells from Molicel, which is a major reputable manufacturer. The physical test setup for these tests was exactly the same as my previous (Charge/Discharge) tests. 

To set up profiling, we need to use the BenchVue software (it can't be done on the instrument itself without the PC software). More specifically, we need to use the "Advanced Battery Test and Emulation" application inside BenchVue. So out of the 4 modes in the application, I chose the "Profile" mode. 

For any testing with batteries, since batteries hold charge, the initial state is important. For most of my battery tests, including the profiling, I standardized to the datasheet's rated conditions. For example, before doing charge profiling, I performed a "rated discharge" on the cell (using the BenchVue app). As per the cell datasheet, this meant discharging at 0.7A constant current down to 2.5V. This gave me a consistent starting point for my tests.

Cell Charge Profiling

Setting up the charge profiling was simple. I basically translated the cell datasheet values to the options in the profiling window for a "rated" charge profile at the ambient temperature (~30C).

Cell specs as per datasheet:

Rated Charge: 1.7A current charge followed by voltage-limited charge to 4.2V. Stop condition: when current drops below 100mA

Maximum Charge Current: 1.7A

Rated discharge: 0.7A constant current discharge down to 2.5V

Maximum discharge current: 10A

Rated capacity: 3.35Ah minimum, 3.45Ah Typical.

These were easy to translate to the BenchVue settings. I left most other settings at the default values.

After this, I clicked start and the process started. For my single 18650 cell, the process took a little more than 3 hours to complete. The process finished with a message stating an actual measured capacity of ~3.37Ah. This, to me, was quite good since the battery datasheet mentioned rated capacity between 3.35-3.45Ah.

The application software gave a good deal of information on the various parameters measured over the whole cycle from start to finish, as can be seen in the graphs below:

The graph on bottom plots the measured Internal resistance and Open-circuit terminal voltage versus the State of Charge. This is essentially the "model" of the battery that we needed to build, and this is all the info we will need for battery "emulation" later. 

The graph on top gives info about the charge current and battery terminal voltages during the cycle. We can see a constant current charge until the middle of the cycle, after which the current starts to drop off. Eventually, it dropped below 100mA and the cycle stopped. The voltage went from ~3V to 4.2V as expected. The vertical lines are the sample points where the charge process was paused to measure Internal resistance. WE can zoom into this part to get more details on how the internal resistance is measured:

Going from left to right, we can see that the "normal" charging is first paused for about 1 second to give a "resting period" to the battery. Then, a 10ms current charge pulse is given to the battery, which I think is when the terminal voltage is captured again to compare with the resting terminal voltage to calculate the internal resistance. After this, we see a discharge pulse of equal time and magnitude, which I think is there to compensate for the previous pulse and leave the battery in a somewhat undisturbed state.

One thing to note is that the process generates two different files:

1. The Charge/Discharge profile (csv format) -> This is the "battery model" to be used later during emulation.

Opening the file in excel, we can see the information stored in the file. I really like the fact that it stores all the settings in the files themselves. This makes it easy to see what settings/conditions were used when a particular model was created, instead of having to store this info elsewhere.

2. A log file in a special ".hdf5" format, which can be loaded into the "Battery Viewer" tab inside the BenchVue application to view the raw data (current, voltage, etc) and can be exported out to a csv file. The below image shows the log file for this run loaded into the "Battery Viewer":

Notice that, like the model file, all the run settings are stored and displayed with the log files as well. The "Export" button on the bottom right allows us to export to a more portable csv format. The export also allows for choosing things like the delimiter as well as what to include or exclude in the export. Overall nice, simple and useful.

Cell Discharge Profiling

Next, I repeated a similar process to create the cell's discharge profile. The difference being that I chose the Profiler -> Discharge option this time. I decided to discharge with a 1.7A current instead of the 0.7A rated current since that was closer to my expected use case later. Also, I had to fit the entire cycle within my evening without leaving the test setup unattended for safety reasons. 1.7A constant current discharge down to 2.5V would take about 2 hours for my battery's capacity. As per the datasheet, I chose the stop condition as the voltage threshold of 2.5 volts. As with the charge profiling, I performed this in 4-wire mode with the same cables I had prepared earlier (with sense wire connections).

The test started and finished in about 2 hours, as expected. You might notice that the discharge test took less time than the charge. The simple reason is that the discharge happens at a constant set current from the beginning till end, so you can quite accurately calculate the time by dividing capacity (in Ah) by the set drain current in Amps. In case of charge, initially the charge happens at a fixed Amperage while operating in Constant-Current mode, but later when it moves to Constant-Voltage mode, the current drops and the charge process slows down more and more. A good rule of thumb is that, for the same charge/discharge current, the charging will take ~50% more time compared to the discharge.

Once finished, we can see the final curves in the BenchVue software. If you compare this with the charge profile curves screenshot, you may notice the constant current discharge from beginning to end for the discharge profiling, which is what we expect.

The profiling showed a depleted capacity of ~3.46Ah, which conforms to the capacity as per the cell's datasheet.

Again, we can zoom into the current/voltage "spikes" in the graphs to see how the profiling is done during this process:

Here, we can clearly see how the profiling is done. First the normal discharge cycle is paused and the battery is left to rest for 1 second (as per the default setting). My guess is that the open-circuit voltage is captured at the end of this resting period. Then the battery is discharged for a short amount of time. The voltage at this time should give the loaded terminal voltage, to be used for internal resistance calculation. Then the cell is charged for the same amount of time with the same amount of current. This would be to undo the disturbance to the battery's state caused by the measurement. After this, the normal discharge cycle resumes. The main difference between the internal resistance measurement part between discharge and the charge profile is that the polarities of the measurement and compensation pulses are reversed.

Overall, the charge and discharge profiling processes are similar and produce the same output files. One thing I noticed is that you cannot use the same file to record both the charge and discharge profiles; the software will give an error that the settings don't match and not allow you to continue.

Issue Observed: I want to mention here that I did encounter some issues several times during this process whereby the BenchVue application would lose connection with the instrument while running the profiling cycle. When this happened, I lost any data after the point of disconnection and had to reset and restart all over again. This happened a few times. I will talk about this in more detail in a later section.

Battery Emulation

As I mentioned in the previous section, one of the reasons profiling the 18650 cell was so important for my roadtest was because of what I wanted to do next: Emulate a battery pack! In my initial exploration of the BenchVue application, I had noticed the option of using the profiled model to simulate a battery pack built from the same cells. This caught my eye because I had been working on a power board design that needed to both charge and draw power from a 6S/5S Li-Ion battery pack depending on whether an external power source was connected or not. The following diagram should give a good idea of what my board is supposed to do:

In simple words, if the external power is connected, charge the battery while also generating the 12V output from the external power. When the external power is removed or not present, use the battery to generate the 12V output. The board had some other functions too but they are not relevant to this testing so I am not going to mention those to keep things simple.

Here I want to take a moment to explain why testing charging circuitry is not as easy as you might think:

• A lot of battery charging ICs these days will validate a battery by giving a voltage/current stimulus and observing the resultant behavior to determine if a valid battery is connected or not before it starts charging. If you just connect a power supply output in place of the battery, it will not work in most cases as the behavior is different.
• You cannot just connect an Electronic Load in place of a battery for the same reason. It will not get recognized as a valid battery.
• Even if you manage to fool the charging IC or smart charger into starting a charge cycle, the voltage will not increase over time with charge going into the battery. You need this to test behavior over the full cycle.
• Battery's internal resistance is a real factor and is not modeled by a discrete power supply or electronic load.
• Sometimes you don't have the actual battery in hand at the time, or don't have samples of different types of batteries you want to support, or not sure of what size will suffice your requirements beforehand.
• Charging/discharging real batteries to prepare for a test with your electronics wastes so much time!
• Safety - real batteries have a lot of energy and can start smoking, catch fire, or explode in the event of certain mistakes or failures.

In short, when you think about it, emulation is a VERY useful tool to have for testing any electronics that need to work with batteries. The E36731A + BenchVue battery emulation software include all the hardware and software needed to enable emulation. I have to admit this is by far my favorite feature of the instrument and software!

By this point, the boards were built, the components had been assembled, and I was ready to test—the timing was perfect! Here is how the board looks:

To test, I connected a Tenma 7208340A power supply to the "External DC Input" port. This power supply can source up to 1.6A for voltages up to 60V. This would allow me to vary the input voltage between 0-40V range and see how the board performs.

Then, instead of connecting a 6S Li-Ion battery pack (that I did not even have at the time), I connected the E36731A and configured it to "Battery Emulator" mode. This was very easy to set up via the BenchVue application and took less than a minute:

•  Set "Operation Mode" to "Emulate."
• "Open Profile" -> "Without Temperature" -> Select the file name for the charging profile that was created earlier. This loaded the cell profile into the application. 
• Set "Emulate Mode" to "Auto" (default) because I wanted to emulate both charging and discharging of the battery.
• I left the capacity rating and current limit to the default values (these were loaded from the profile) as they were within my expected test range.
• I left the "Initial State of charge" to 50% to emulate a half drained battery.
• To emulate a 6S battery pack instead of the single cell I had profiled, I turned on the "Battery pack" option, and set No of Series to 6 and No of parallel to 1. This would emulate the 6S battery pack as I needed.
• I set the Cut-off condition type to "Cutoff Voltage" (default is cut-off SoC - State of Charge), but left the Lower and upper threshold values to default (since this is an emulated battery, I don't have to worry about the battery pack catching fire due to over-charge or discharge!)
• Lastly, I changed the "connection type" to 2-wire instead of the default 4-wire to match my actual setup.

The final settings are as following:

One observation I want to mention is that you can only select one "without temperature" profile at a time. So in my case, I could either select a charge or a discharge profile. If you select "with temperature" profiles, it allows you to select up to 10 profiles, but they all must be of the same type - Charge OR discharge. You cannot mix. Nevertheless, the emulation will still happen in both charge and discharge modes as long as you select the "Emulate Mode"  -> "Auto" option. (You can also limit it to just charge or discharge if you wish so).

Anyway, with the settings done, starting the emulation process was just a matter of clicking the "Start" button in BenchVue. 

Measuring Standby Power

Standby power, also called vampire power, phantom load, ghost load, or leaking electricity, refers to the way electric power is consumed by electronic and electrical appliances while they are switched off or in standby mode. This is the first thing I wanted to measure on my device. This measurement would tell me how long the battery will be able to retain charge while installed in my device while the switch is turned off. Measuring this was very simple. I kept the "External DC input" power supply turned off, and the power switch on my power board (refer to the diagram above) turned off as well. Looking at the Battery Status tab, I could see that the emulated battery was sourcing ~0.8mA of current at 22.5V.

Practically, this means that if I put a fully charged 8000mAh 6S battery in my device and leave it in the switched-off state, it will take 8000mAh/0.8mA = 10,000 hours (~416 days) for the battery to deplete.

I can also use this measured draw to suggest a charging schedule when the devices are stored on shelf. Normally you would want to maintain the battery charge level between 30% to 70% for long-term storage to protect batteries. This means that once the devices are recharged to 70% for storage, they can be left on the shelf for (70-30)*10000 = 4,000 hours before needing to be recharged back to 70%. This implies a recharge cycle for each device every 4000 hours, which is ~5.5 months.

Measuring No-Load Power consumption

After this, I wanted to measure how much power is used up by just the 12V Buck-boost converter (and some other circuitry on the power board) without any external load. For this, I turned on the switch on the board, and saw in BenchVue that the current draw had increased to ~33mA. The "external DC input" was still turned-off in this case. At 22.5V, this is about 740 mW of power being consumed by the power board itself. It's a bit higher than my liking and mostly getting wasted as heat and light (few LEDs). I made a note to investigate this further to see where I can save up on some power consumption since it will directly impact the battery life of the device. Once I make any changes, I can come back to the same setup and use the observed values as baseline to measure the improvement.

Testing Battery Charging circuitry

At this point, I wanted to move on to testing the battery charging circuitry. This is the part I was most apprehensive about because I had not tried the circuitry using its EVM or any other such means before integrating it into my design. The charger design is based on the BQ24600 charger IC from TI and I had designed the circuitry around it to charge 6S Li-Ion battery packs with a max current limit of <1A.

With the E36731A already in Battery Emulation mode and configured as a 6S pack, testing the charger at this point was as simple as turning on the "external DC input", which in this case was connected to the Tenma 72-8340A power supply. After a VERY long second (and skipping some heart beats in that time), I saw the battery status change from "Discharging" to "Charging" and the current value from 33mA to about 950mA. I was relieved to see that the charging circuitry was doing something right, and without any magic smoke coming out of it.

I have annotated the screenshot to show how it all appears in BenchVue's emulator window. All the values are updated in real time. A small point: the "Charging"/"Discharging" label is very useful because it gives a direct indication of the direction of current without having to think about what the sign of the current value means.

One of my absolute favourite features in the BenchVue application for the E36731A is the ability to start the Emulation cycle at ANY state of charge (0% to 100%). I found it an extremely useful feature when trying to test and find issues in the connected electronics.

I have worked quite a bit with batteries in my preceding years, and this has always been a pain point because charging/discharging batteries to bring them to the required starting SoC before a particular test wastes so much time and gets very exhausting! With the BenchVue application, I absolutely fell in love with the ability to jump between charge states quickly to test at various States of Charge without having to wait tens of minutes or hours in-between.

In this particular testing scenario, I had seen that the charging was working at 50% SoC, however I didn't know if it'll work across the full SoC (State of Charge) range.This was easy to test by changing the "Initial State of Charge" parameter in 10% steps and restarting emulation (you can't change SoC while the emulation is running). I quickly found out that the charger wouldn't charge at 70% SoC or above. To get a better picture of what was going on, I set the Initial SoC to 59% and started the emulation, and observed how the cycle progressed. After half an hour later, I had the answer. The charger was charging the emulated battery fine until about 69% SoC, and then it suddenly stopped. I reset the emulation back to 59% starting SoC, and performed the experiment again and got the same result the second time as well, which showed this was consistent behaviour.

This showed very clearly that I had an issue in my design that needed fixing, and the graphs showed me exactly how to reproduce the issue when testing again.

As of this writing, I am still working on solving this issue in my design (along with some other issues I found). What I am VERY glad about is that I have this super-power to start my testing at any State-of-Charge I desire, and see visually exactly how the charging cycle happens. 

Speeding up Emulation cycles

One unique thing about testing with batteries is that they need patience and time! This is because you are limited by the charge/discharge current values, and the batteries themselves take time to move between States-of-Charge. For example, even for the 18650 cells I was using, a single charge cycle took ~3 hours at the rated current. This was a major issue for me since one evening means no more than 2-3 hours of free time on a normal working day. I wanted to test the performance of my charging circuitry throughout the charge cycle (0% to 100%) under different conditions and with variations in the populated components to fix the charging issue and optimize performance. I calculated that each cycle would take >3 hours which would slow me down and I wished there was a way to speed up the cycles.

At one point, I had the light-bulb moment, when I realized that I was dealing with emulated battery, which meant I could simply reduce the capacity of my emulated battery to effectively speed up the cycle. So I changed the "Capacity rating" to one-tenth of the actual and started a test, with the initial SoC set to 0%. This gave me the desired speedup I was wishing for and I was able to run a full charging cycle in less than 20 minutes!!

I was very happy to see this not just because of the speed-up, but also because this gave me a good overview of how the charge cycle went at different states-of-charge. Since this was a shorter cycle, I could also just sit around and wait for it to complete. This way, I was able to hear some audible noises coming from my board mostly during the times when the current waveform is showing noise. Seeing this has given me several ideas on possible root causes, since now I know the issue isn't just happening at ~69% SoC. Best of all, I'll be able to reproduce the issue at any time, run quick cycles and compare performance whenever I want to try any component/design variation. I hope to get back on this in the coming weeks once I have solved the issue.

Issue with external-power-preferred OR'ing:

The ease and flexibility with which the E36731A allowed me to test my power board electronics helped me make observations resulting in finding a fault that otherwise would have needed me to test explicitly for. In this case, observing the currents and voltages between the instruments led me to notice an issue in my "external-power priority OR'ing" circuitry. My design intent had been that when external power is connected, the battery should not get used, even if the battery is at a higher voltage than the externally connected DC input. This was implemented as an Enable signal that would disable the ideal diode in front of the battery when external power is in valid range. However, I noticed that this was not happening, and the load was always drawing current from the higher voltage source (battery or external power). This led me to a more thorough read of the ideal diode datasheet which confirmed my oversight. In this case, the enable signal only enabled or disabled the MOSFET gate's "enhancement", but the body diode of the mosfet would still conduct in the forward direction. Such an obvious mistake in hindsight, but at least I am glad I managed to catch it quickly.

Battery Cycling

As I progressed through my roadtest, I came to the realization that testing the battery cycling feature over tens or hundreds of cycles would be a major challenge. The simple reason was that even one charge-discharge cycle on the smallest of my 18650 cells was taking 4-6 hours. I experimented with using higher charge and discharge currents, but the battery temperature would start to rise, which made me uncomfortable for a long-running test. I had noted in my test plan that I would not be leaving any battery tests unattended for safety reasons. Ultimately, the best workaround I could come up with was to source a very small rechargeable battery. I had recently helped a friend replace the battery inside her "Beats Flex" earphones and had noticed the very small 100mAh 1-cell battery inside. So I decided to source one for my battery cycling tests as well. 

When the tiny battery arrived, I first soldered a connector to it after confirming terminal polarities. This was to prevent accidental shorting during handling. Notice the small size of the battery next to the XT60 connector.

The heat-shrink tubing and kapton tape were for added insulation. I had sourced the battery online for ~10 US$ so you can imagine it didn't come with a datasheet. The only thing I knew about it was that it was a 1 cell Li-ion/polymer type, and 100mAh capacity. No specs for maximum charge or discharge current were available. Usually, tiny batteries such as this are not designed for high charge and discharge rates anyway, and "1C" is a safe bet. 1C for a 100mah capacity battery means 100mA current for charge or discharge. To confirm, before starting the cycling I used the "Charge/Discharge" mode in BenchVue to conduct some preliminary tests. First, I performed a charge cycle at 1C, which went fine without any noticeable heating and yielded a capacity of ~66mAh. Then I discharged at 2C (200mA) down to 2.5V. This time, I noticed the cell was getting noticeably warm, and yielded a capacity of 58mAh. Due to the heating, I decided to stick to, 100mA current for both charge and discharge.

Starting the cycling was very simple. You enter the "Cycler" mode in BenchVue, and then set the create a new Sequence. After this, an new popup window opens where you can create the steps for your profile. After this you can set the repeat count and start the cycler. I used the 4-wire mode since that was my configuration this time.

This is how I set the steps for one (charge & discharge) cycle:

1. Charge with 100mA current to 4.2V (CC-CV mode). I set the stop condition to C/10 = 10mA (common practice to use C/10 as stop condition)
2. Rest for 1 minute
3. Discharge with 100mA constant current down to 2.5V
4. Rest for 1 minute

Based on my earlier tests, I estimated one cycle for this small battery to take around 1.5 hours, and so I set the cycler for 10 cycles to see if I could observe any capacity drop. The test ran overnight, and I stayed besides the test setup (even slept besides it) so as not to leave it running unattended. By morning the test finished and,...I have to say I was VERY impressed. Let's look at what we have.

Below screenshot shows the BenchVue screen when the test was done. I have annotated it to point out some noteworthy details:

We can see how the cycler executed the 10 cycles as per my setup and kept meticulous record of the live parameters as well as the capacities and time taken for each step. Zooming into a single cycle reveals the full details of each cycle:

From the graphs we can see that each charge cycle is executed as a CC (Constant current) charge followed by CV (Constant voltage) mode charge until the current drops below 10mA. Then, after a 1-minute resting period, the discharge step is executed with a 100mA constant current. The cycle ends with a 1-minute rest period before moving to the next cycle. The graphs confirms that the cycler is doing exactly what we expect it to do.

Battery ageing over multiple cycles

Coming back to the previous screenshot, the most important and interesting piece of information is in the table at the bottom of the window in the "Battery Cycle" section. A crop of that is below:

This is the main information I was after. We can clearly see the capacity absorbed by the battery in each charge cycle as well as the capacity measured in each discharge cycle. The actual exact time taken by each cycle is also good to have.

What absolutely blew my mind was that literally from the second cycle, we can start to observe the drop in capacity with every cycle. Even more impressive is the fact that the drop in capacity per cycle is a very small amount on an absolute scale (less than a milli amp hour!). I had never expected I would be able to observe THIS small drop in capacity so quickly, and this just blew my mind! I am VERY impressed!!

I put this data into excel (you cannot copy from the BenchVue directly but one of the saved files has it in the form of a table if you open it in excel). And then I applied some basic calculations to get some insights:

The first three columns are the data from the cycling in BenchVue. The next two are calculate from these values. We can make some observations from the data:

• First thing to notice is that the very first step (charge step in this case) will almost always yield somewhat different results. This is not an issue with the instrument. Rather this is because batteries store charge, the observations in a particular cycle depend on what parameters were used in the previous cycle. All the cycles in this test used the same parameters for charge and discharge, but the discharge cycle preceding the first charge cycle above was done at a higher current. Thus the absorbed capacity is smaller in the first cycle, which is expected!
• The 4th column gives the cummulative capacity loss with each cycle. The reference for all capacity loss values is the highest capacity (charge or discharge) seen in the first two cycles. I got this definition from the help file in BenchVue, which makes a lot of sense. We can see an average of ~0.5% capacity loss in each cycle. The trend is very consistent across the cycles, which gives confidence that the instrument is not operating near it's noise floor (the values would jump in that case!). The loss might seem big (as percentage) for a battery but there are several contributors. Firstly, this was an "el-cheapo" battery with questionable quality. Secondly, these tiny batteries are usually designed for much lower discharge currents than the 1C rate we were using (higher rate is worse for capacity drop). Lastly, we were also going from 0% to 100% in our cycles, which causes a much faster capacity drop than staying between 20%-80% for example.

Overall, I was VERY impressed with the cycling mode, and how small a change can be reliably noticed by the instrument. My expectation before the test had been that MAYBE I will start to see some observable capacity loss by 10th cycle or maybe I will need to do few tens of cycles more. Kudos to Keysight designers for building such an accurate design!

One last observation I want to mention about the cycling mode is this: Cycling batteries can take days to weeks, and it's not always possible to leave a test running for this long in one stretch. Keysight has included the feature to load a previous cycle sequence which was stopped or interrupted, and then continuing from where you left things in the last run. This is via the "Load Cycler" option in cycler mode.

I was very relieved to see this feature when I saw it. Before that I had been thinking of how I will be able to run cycling for this long without any interruption. The Keysight engineers have been spot-on on this one!

E36731A as a calibration instrument

The Keysight E36731A is not marketed as a calibration device, but the fact that it gives such accurate instrumentation on voltage and current as Electronic Load, I found one more good use case for it in my home-lab. I have a Tenma 72-8340A power supply that I have been using to power devices on my desk requiring up to 60V DC (limited to 1.7A current). The thing I don't like about this power supply is that it always goes out of calibration after some time. Difference of one volt or more between the set point and the output is not uncommon. I think it's not an issue with my particular unit because I had the first unit replaced, but the replacement is no better. Searching on the internet, I found that these are also sold as Manson power supplies and found the calibration manual. 

To calibrate the power supply, you need a decent voltmeter for voltage calibration and an Electronic Load for current calibration. In the absence of an Electronic Load, you'll need a (variable?) power resistor and an Ammeter in series to measure current. For this reason, I was never able to calibrate the current in the past. With the E36731A, all I needed to do was setting it to "Load" mode and connect the output terminals of the two instruments together.

• Voltage calibration: In Load mode, and without the load turned on, the E36731A display shows the exact unloaded voltage output of the Tenma power supply. I used this to match the actual voltage with the displayed voltage as per calibration procedure.
• Current calibration: For this, I set the E36731A (in Constant-Current Load mode) to different current values and used the knobs on the Tenma power supply to match the values between the two instruments.

It took me no more than a few minutes to recalibrate the Tenma power supply, and with such convenience! Note that the accuracy of the E36731A is orders-of-magnitude better than that of the Tenma Power Supply. Yes, it's an overkill for this, but shows how versatile and useful an instrument like this can be. Reminds me of what a wise person once said "A good instrument... is a joy for ever!" 

Issues, Suggestions & Wishlist

The E36731A is an excellent piece of equipment and gives solid performance in all modes of operation. However, there are some aspects that can be improved, in my opinion. Interestingly, these are all related to the BenchVue application software or the device's firmware. I hope these can be taken into consideration for future firmware updates.

• The first issue I noticed with the E36731A was when I tried using it as an Electronic Load (standalone mode). Having used many Electronic Load instruments in the past, I have come to expect that any such instrument will display the cumulative discharged capacity (mAh) while it is discharging - in standalone mode. This is not the case with the E36731A even though it's purpose-built for working with batteries. I strongly urge Keysight to add this to the display in a future firmware update.
• One aspect I absolutely did not appreciate during my testing is that whenever the E36731A is being used with the BenchVue application, the instrument screen stops displaying any live information other than the mode the instrument is being used in. For example, in emulation mode it just displays "Battery Emulation". I understand the need to prevent user from altering instrument state/settings from the front panel while it is under control of the BenchVue application. But the current state (voltage, current, and other parameters) should still be displayed on the screen. I found this rather uncomfortable, especially when the BenchVue application would lose connectivity with the device in the middle of a test. It was very uncomfortable not knowing what state the battery was in and whether it was still being charged or discharged and I had to disconnect the instrument due to this. Many other Keysight (and other) instruments that I use regularly at work as part of manufacturing test systems don't show this behavior and you can still see the values on the screen. It even helps in debugging when the test sytems don't work as expected. I really hope this gets improved in a future firmware update.

• When operating in Emulation mode, using the 2-wire configuration (physically as well as in the Emulator settings), the software stopped the emulation several times prematurely, giving the error below. This was strange since I wasn't even using the 4 wire method and the sense leads were not connected anywhere. In some cases, restarting the application seemed to fix it, while in others I had to restart both the software and the instrument.

• A lot of electronics these days work in switching mode (charging circuits, power converters). As a result, when emulating a battery the current & voltage waveforms can be noisy under certain conditions. This happened in my tests too. In my observation, the E36731A managed to show this behaviour quite well in the live data graphs. However, I did find myself wishing that there was a software low-pass filter available that I could turn-on to view the live data in a more averaged form whenever I was more interested in looking at the "big picture" over a longer run instead of the micro variations/events.
• BenchVue application has the very nice feature to continue cycling tests by loading the file from previous run and resuming from where you left. However, there is no similar pause/resume functionality in the Charge/Discharge mode. This would be helpful since battery charge/discharge cycles can sometimes require very long durations, especially if testing with smaller currents, which is not always possible in one stretch. For example, at work we have a policy not to let any battery test run unattended overnight due to safety concerns so a pause/resume functionality would be very helpful. Currently you can only stop a test, after which you need to start over (or combine the data from partial runs manually later). 
• The live graphs in BenchVue battery application do not have a mode where they update with live data while also scaling the axes to display all data from start until most recent sample. So I had to press the “Auto scale all traces onto the screen” button whenever I wanted to get the most updated full picture. I also noticed the application would feel sluggish after acquiring data over a very long run. I was using a i7 PC with 16GB RAM which should  have given enough processing power and memory. Perhaps the software libraries that load and display graphs can be optimised further.
• Earlier in the testing, I made an observation that if a battery is connected to the instrument terminals and the instrument is powered-on without doing anything else, the instruments starts draining the battery with few tens of millamps. This also happens any time the instrument is switched to power supply mode with the output off. My guess is that this is because of some internal discharge circuitry that keeps the output terminals discharged for safety reasons but this means the battery will gradually get discharged if the user is not careful about this. I was caught off-guard when I had the battery connected to the instrument and I power-cycled the instrument for some reason. I immediately disconnected the battery physically so as not to alter it's state (this was in-between tests).

• Requiring an external DAQ just for temperature acquisition (profiler mode) is not so good in my opinion. An instrument in this price range should natively support connecting a thermocouple to it directly for logging battery and/or ambient temperature while profiling or during emulation/charge/discharge run.
• Several times, BenchVue lost connection by itself from the E36731A instrument while running a charge or discharge cycle. In such a case, the BenchVue application was showing disconnected while the E36731A kept running the task. However, since the instrument doesn’t show the live data (Voltage, current) when running a task from BenchVue application, there was no way to know what state the program cycle (or connected battery) was in. So I had to reconnect the BenchVue application to the instrument by pressing the “Connect” button. As soon as this is done, the instrument is automatically stopped, and you have to start over again. During this period of disconnection, no logs are saved either. I also experienced this while profiling a battery. The instrument kept running, while the software had lost connection, so I didn’t manage to profile it completely and had to start over again. Both the PC and E36731A were placed on a side, away from any movement, so the probability of a physical USB disconnection is very low. Nevertheless, I tried direct USB connection as well as through docking port and accessed the PC via VNC to make sure there is no movement but the issue still re-occured a few times. I couldn't try on a different PC due to the node-locked license.
• The "Instrument Control" tab in BenchVue shows what is currently on the screen but does not actually allow controlling the instrument from here. This would be nice-to-have for remote control, although certainly not a high-priority feature.
• Occasionally, I had connection problems when connecting the instrument to the BenchVue app (as in, the app not finding the connected instrument). What I noticed was that if the instrument was turned on and connected to the computer before launching BenchVue, there were more issues connecting compared to vice versa (i.e, launching the BenchVue app first and then turning on the instrument). Sometimes this issue would not happen for days.
• The E36731A has two very nice rotary knobs, one labeled "Voltage" and other labeled "Current". However, my view is that they are not very consistently implemented in the firmware currently. For one thing, the control is too "fine" and so it takes long time to make bigger changes in Volt/current values. For this reason, I always resorted to using the keypad instead. Secondly, the implementation is not very consistent. For example, if you are in Load function (CC mode), rotating the Voltage knob changes the current value, while the current knob does nothing. Due to such inconsistencies, I have found it hard to train my brain to use the knobs so far. I hope this is improved in a future firmware update, including a good "velocity control" algorithm on these knobs. The knobs have a very nice feel to them, by the way!
• The way Lock/Unlock functionality is implemented in the current firmware is somewhat odd. It requires a long-press to lock and a short press to unlock. It should either be the opposite, or both should require a long press.
• Although all the documentation is quite comprehensive in terms of covering the topics, I feel the amount of detail/explanation is not always enough. There were instances when I felt some more explanation would have been nice. I also saw some cases where the documentation and current firmware were not in sync. For example, Battery Emulator Users Guide.pdf page 30 says under CC mode: “Current may be programmed in one of three overlapping ranges, a Low, Medium and a High current range”. However, in the instrument menu I only saw two ranges (Low-4A, and High-40A). 

Conclusion

The Keysight E36731A is an excellent piece of instrument. It looks beautiful and performs solid. In standalone mode it has features that I haven't seen in other benchtop instruments I have used (and I have used a lot of good power supplies and Electronic Loads in the last 15 years!). And when you add the BenchVue Battery test and emulation application on top, then it takes things to a whole different level. The more I interacted with it, the more I found myself wishing I had this instrument at my disposal 15 years ago when I embarked on my journey with batteries at work. It could have saved me countless hours of manual labor. While there are some areas for improvement, it’s also important to remember that this instrument is a recent addition to the market. My prediction is that the E36731A with the Battery Test and Emulation application is going to make a lot of engineers very happy! If I had to choose one absolute favourite feature, I would definitely pick the battery emulation. Now that I have tried it for some time in a real-world testing use case, I don't think I can go back to testing without it. 

I am grateful to Element14 and Keysight for affording me the opportunity to roadtest this remarkable piece of technology. It has undeniably earned a permanent spot on my desk. While I am wrapping up my roadtest review for now, my exploration with it is far from over. I look forward to sharing more of my adventures with it on the element14 community in the coming times!