RoadTest: Enroll to Review the Keysight Battery Emulator and Profiler E36731A
Author: genebren
Creation date:
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?: I had not shopped around when this product was announced for a roadtest.
What were the biggest problems encountered?: The supplied software was difficult to obtain a license for and the functionality was challenging and confusing.
Detailed Review:
When I applied for this roadtest, I wrote the following as my Test Plan:
Test Plan for the Keysight Battery Emulator and Profiler - E36731A
1) Profile the 12V 9Ah battery (or both batteries together - 12V 18Ah)
2) Using the data gathered during the profiling process, code and test the new load controller module.
I currently have a schematic and board layout for my load controller. I will need to order boards and parts. Then I will need to assemble and test the board. From there I will generate code for the functionality, including the battery capacity functions (charge level).
3) Using the generated profile, use the E36731A to emulate the battery to test the new load controller as the emulated battery is drained, using an external programmable load, monitoring the charge level logic.
4) Assemble and commission the solar system with the new charge/load controller module and test the system through multiple days. (Note: the charge/load controller will log charge and discharge events and report data through a Wireless link (EnOcean protocol) to a PC using a EnOcean transceiver modules, or a handheld EnOcean terminal (also under development)).
The project that I envisioned to test this product on was a solar powered fountain/birdbath (water feature) for inside one of our native plant gardens in our yard.
This was all well and good, and I was looking forward to taking on this project and roadtest, but as often happens, life has different plans. During the time span of this project, I had to deal with injury (shoulder/pinched nerve in my neck) and a prolonged respiratory illness and some very strange (still undiagnosed) intestinal issues. But I pushed forward where I could and postponed those things that I could not yet manage. What I present here is a compressed scope and project presentation.
A large, heavy, sturdy, well-traveled package arrived at my home late in October (Thanks Randy!)
External examination of the box revealed no obvious damage, but I was still very interested to check further.
The instrument was cradled in foam inserts, protecting it from damage and vibration. The small box layered into the foam contains cables and paperwork.
Here are the contents of the accessory box. The hefty terminal block is the front power DC power connectors. And here are some views of the Instrument.
The Instrument is beautiful. With great controls, displays and connectivity options. There are both front and rear DC power connections, with sense connections, plus a host of other features. Here are some User Guide Images showing greater detail on the connections and controls.
After some back-and-forth exchanges with Keysight staff. I finally received a license file for the BenchVue Advanced Battery Test and Emulation software. I do find it a little annoying and confusing that an instrument that is called a Battery Emulator, only has that level of functionality when driven with external software (with a timed license). But this is sort of the world that we live in.
I ran several charge and discharge cycles on a single battery to get a better understanding of the operation of the profiler functions. This is a lengthy process on an 8.5AH battery that is specified at a 20-hour charge rate (~425mA rate), so I opted for a 1.0A rate to try and fit it all into a one day run.
Here is a screen capture of a discharge cycle of the 8.5AH battery:
After capturing the data for the discharge run, I imported it into Excel to extract the Voltage to %Charge data for my project. Here is that data:
For my lookup table I wanted to be able to look up the average capacity of the battery for every 50mV step in Voltage from 11.0V to 12.5V. So, I pivoted the chart and reduced the table to 31 data point and re-plotted the curve.
This data can then be imported into my code for the Solar Controller to use for determining the daily pumping schedule given the charge state of the batteries.
This was an interesting exercise, and the E36731A was up to the task and performed flawlessly.
Prior to getting the proper license for BenchVue, I decided to try out different aspects of the E36731A Power supply and DC load functions. After exploring the capabilities of the Power Supply mode of the E36731A, especially the test sequences and slew rate settings, I decided that a great test would be measuring the efficiency of the RECOM R-783.3-05 DC/DC converter. My project is using two of these convertors to convert the 12V battery voltage down to the logic level of the boards (3.3V). I decided to test the efficiency of these converters by slewing the voltage applied to the inputs from 10V to 20V, while holding the load on the outputs at a steady value of 500mA. I used a BK Precision 8500 to supply the output load.
First, I set the power supply to output 10V (the minimum of my range of interest). Then, I modified the voltage slew rate to 0.05 V/sec.
To generate the voltage ramp and to collect the data with the datalogging function, I decided to use a sequence list to specify the test steps.
The first entry just reiterates the starting voltage with a 1 second delay. The second entry with command the Power Supply to 20 Volts at a slew rate of 0.05 Volts/Second, taking roughly 200 seconds. The 'Time' of 255 is just to make sure that everything settles at the final voltage.
In addition to the above list, there is a Sequencer List Property setup screen.
This setup determines how the sequence list is processed, including how it is triggered. I choose the "List Run/Stop Key" as the trigger source.
Next, I needed to configure the data logging functions to capture the sequence run data. I wanted to capture the Voltage and Current readings, along with the computed Power. Unfortunately, the E36731A does not seem to include the power data in the log output, but it does display in on the Oscilloscope screen. Not sure if this is a firmware bug, or just some lack of understanding by me.
Here are Data Logger Properties setting that I used to capture the data.
I choose to show the Voltage at 5.0V/division, Current at 0.2A/division and the Power at 1.0W/division. Since the sequence runs for 256 seconds, I choos to set the Horizontal Time per Division at 40 seconds to capture 320 seconds to data. The operating sequence of events is to start the Data Logger first (from a menu button) and the then the Sequence (from the List/Run/Stop button), so have the Horiztontal capture time set to 320 leaves plenty of time to get all the data captured on the screen.
Next, it is necessary to select how the data is stored and where.
Here are the selected settings. Note: Again, not sure if this is an operator error or not, the "Append data and time to file name" selection does not seem to be working.
So, finally we are ready to go. The sequence is started from the Data Logger screen.
The Data Logger is started by pressing the button under the "Datalog Run Stop" menu item. Then I started the sequence list by pressing the "List Run/Stop" button. As the test ran, the oscilloscope display showed the progress. At the end of the test, a popup message informed me that the data was being written to the USB drive. I transferred the USB to my desktop system and open the resulting log file to find that it was a binary image, with a text header. That was not going to be easy to deal with. I noticed an export function under the "Data Logger - Target File Selection", so I decided to try that out.
Pressing the button below the "Export" menu item, initiated the export function. The export file was a CSV formatted list.
I imported the CSV file into Excel, adding some additional columns (Power (somehow missing from the log) and Efficiency (input power is calculated and output power is assumed to be fixed at 3.3V @ 500mA or 1.65 Watts)). I then used a scatter chart of a range of the data (10V to 20V), plotting both Input current and computed efficiency versus Input Voltage.
The measured results are a pretty good match of the published specifications.
I have always wanted a water feature for our backyard. My Wife and I have bounced around several plans throughout the years, but this past year we finally settled on a plan. We were wrapping up a design of a new garden of native plants that was once a large section of lawn. We wanted to add a centerpiece structure which would be a water feature, combination of a fountain and a birdbath. The basic functions were to be an upright tower, which would hold four glazed ceramic platters, where water would flow into the platters and then spill to a rock bed. The water would flow through the rocks into a hidden reservoir below. A submersible pump (enclosed in a filter box) would pump the water up through the fountain structure. Power for the project would come from a solar panel (so we would not have to bury AC power through the recently planted garden).
Here are some early pictures of the garden (not quite completed at the time of this picture) and some mockups of the fountain.
The next steps in the construction of the Fountain/Birdbath were to dig a hole for the foundation forms and to connect into the garden water line for the auto-fill feature.
The gray pipe is the 12V electrical run to solar power system. The blue pipe is the water line. The second picture is the water line connecting to a new shutoff valve, dug into the existing pathway (just in case of a software bug causing a overfill situation).
The first image is a loose placement of the lower reservoir tank area, after the concrete blocks were modified to fit the foundation and attaching rebar. The second image shows the decorative 'chop' block that will be mortared into place above the reservoir. Between the 'chop' and concrete blocks will be a grid of rebar and two mesh screens (1" x 2" heavy mesh or livestock panel and 1/2" x 1/2" mesh) to support the river rock mix, that will fill nearly level to the top of the 'chop'. Dirt will be back filled to just the bottom level of the 'chop'. Once finished my wife will plant some native grasses around the fountain and other native plants in order to blend with the rest of the garden space. The lower reservoir will be coated with several layers of epoxy pool paint to make in fully watertight. The third image shows the beginnings of the notches for the rebar as well as view of the plumbing. The top most block will have a removal lid (cement block), below which is a solenoid operated valve (for the auto-fill feature). Water will be routed through the, to the other side of the block to the tallest pipe that leads back down to the reservoir. A flexible pipe will also pass through that pipe (in the opposite direction) for the pump(s) to fill the upper trays. The shorter pipe is going to be used to hold the water level gauge (as it is open to the water reservoir). The Pump Controller will be placed inside the block with the second pipe, in will read the water level gauge and control the solenoid valve and the pumps (under one set of birdbath trays).
This is an early rough mockup of the Fountain/Birdbath with the copper pipes and ceramic trays in place. You can notice in the background of the first image, the screening that I intend on using on top of the rebar to hold up the river rock above the reservoir. The second image is a movie, showing a very early mockup of the fountain top section, running on my patio/deck.
The basis of the solar system is a package from Tycon Solar. The system is a RPPL 12-18-15 model, which has a 15 W solar panel, 12V 18Ah battery (two batteries) and solar charger/controller. This is a pole mounted, waterproof system.
Inside this system, I am installing a custom-built solar controller that can monitor and log the performance of the system. The solar controller will measure the battery voltage to determine approximate charge on the batteries and will supply power to the pump controller module. Both modules are equipped with EnOcean transceivers allow them to communicate wirelessly. The solar controller has a real time clock which is used to determine when to log the PV voltage/current and when to run the Fountain/Birdbath pumps. If the charge is low on the batteries, the pumps will be limited to a short period of runtime per hour, otherwise they will run according to the programed schedule (offset times from sunrise and sunset as determined by a table of sunrise/sunset for our location). The solar panel and controllers mounted atop the pole in the second image (just a few feet away of the fountain).
The Solar Controller and the Pump Controller share a common design and PCB. The two modules are populated slightly differently during assemble. Also, there a common sensor connector design, with are populated slightly differently. Here is a system overview.
The Solar Controller will be inside the solar power cabinet, while the Pump Controller will be embedded inside the Fountain/Birdbath tower. The two modules communicate with each other via a pair of EnOcean 90Mhz transceivers. They will also communicate with a base station (attached to my main Desktop) or a soon to be developed handheld remote that will come in handy for making schedule changes, etc.
Here are the schematics:
Here are the PCB layouts:
Here are some images of the bare PCBs and some of the assembled Solar controller.
The first image is of the two bare PCBs, shown in a rough alignment as if they were connected together. The second image shows the Solar Controller module loaded in its enclosure with the connector exiting the side of the case (when finished the connector outline will be sealed to keep the elements out). The final image is of the Solar Controller with its sensor board attached and cables attached for testing.
The population matrix for each of the PCBs is as follows:
Board Type | INA219 - I Sense | EEPROM - 512Kb | RTC/SuperCap/Osc | 12V Power switch |
Solar Controller | 1 | 1 | 1 | 1 |
Pump Controller | none | none | none | 2 |
Interface Type | INA219 - I Sense | High current conn. | 4 pin 200mm conn. | |
Solar Controller | 1 | 1 | none | |
Pump Controller | none | none | 1 |
The Base Station board is built from a (heavily hacked) prototype, from another project. I am using a Master module from my Kitchen Lighting System, document here: ( Kitchen Lighting System Phase 2 - PCBs, Schematics and new sensor module. ). I developed a simple software package for this board to receive commands from a Visual Basic App, via the USB to serial convertor. The data is read in by the microprocessor via a serial port and converted into an EnOcean data packet and sent via a second serial port to the EnOcean transceiver. Any reply messages sent by the transceiver and reformatted and sent back to the Visual Basic App.
To assist in tracking the EnOcean Radio traffic, I am using a copy of DolphinView and an EnOcean USB dongle to monitor packet transfers.
The feature set of the Solar Controller are:
The feature set of the Pump Controller are:
To test the voltage and current sampling, filtering and data logging, the logging feature was enabled for PV and Battery voltage and current, set to a 1 one second rate (debug mode). The system was set up with a single power supply connected to both the PV and Battery/Load ports and a DC load connected to Pump power port. While the logging was occurring, the load was turned on and off. The slope of the changes in the logged data reflect the filter rate. Here is a plot of the data collected from the logfile and sent to the control panel.
The charts showed the proper filter rates and no issues.
Now for the Emulated Battey tests, using the E36731 to power up the Solar Controller, at different levels of Battery Capacity, using profiles generated earlier in the test process. The goal here is to test the Voltage to Capacity function written into the Solar Controller firmware, using data from the same profile. I will test this function at a couple of data points.
First, I will start with the emulation with a fully charged battery.
With the Solar Controller powered up, for at least a minute to allow the filters to catch up, I requested Voltage and Capacity values from the Solar Controller. On the BenchVue screen, you will notice two small negative bursts in the current monitor section. This appears to be short bursts where the EnOcean receiver to shut down, during the packet transmission. The good news is that the BenchVue Capacity (100%) and Voltage (13.1023 V) and the Solar Controller readings (100% and 13.110V) are in agreement.
Now I will reset the emulation at 75%.
With the Solar Controller powered up, for at least a minute to allow the filters to catch up, I requested Voltage and Capacity values from the Solar Controller. The good news is that the BenchVue Capacity (75%) and Voltage (12.2677 V) and the Solar Controller readings (77% and 12.274V) are in reasonable agreement. I also requested the current reading from the Solar Controller, represented the load it is placing on the emulated battery, The measured result is 13.22mA, which is reasonable.
Now I will reset the emulation to 50%.
With the Solar Controller powered up, for at least a minute to allow the filters to catch up, I requested Voltage and Capacity values from the Solar Controller. Now the BenchVue Capacity (50%) and Voltage (11.990 V) and the Solar Controller readings (57% and 11.90V). While the voltage measurements continue to track, the capacity values are moving further apart as the emulated capacity decreases. This is a little puzzling.
Now I will reset the emulation to 25%.
With the Solar Controller powered up, for at least a minute to allow the filters to catch up, I requested Voltage and Capacity values from the Solar Controller. The bad news is that the BenchVue Capacity (25%) and Voltage (11.6421 V) and the Solar Controller readings (38% and 11.648V). While the voltage measurements continue to track, the capacity values are moving even further apart as the emulated capacity decreases. This is looking pretty bad right now.
Now I will reset the emulation to 10%.
With the Solar Controller powered up, for at least a minute to allow the filters to catch up, I requested Voltage and Capacity values from the Solar Controller. The bad news is that the BenchVue Capacity (10%) and Voltage (11.3566 V) and the Solar Controller readings (28% and 11.365V). While the voltage measurements continue to track, the capacity values are moving even further apart as the emulated capacity decreases. This is where I begin to get a sense of the problem.
The Voltage to Capacity function used data from one of my battery profiling sessions. But if I compare the data I used, to the way the Emulator is using the data, I see were the issue lies.
Apparently, the Emulator has normalized the capacity data to cover the range of 100% to 0%, where the original range was 100% to ~22%, over the same range of battery voltages. So, it would seem that if I set the Emulator at nearly empty, this would likely match my Voltage to Capacity calculation of ~22%.
So, now I will reset the emulation to 0.1% to see if this matches the projected ~22% capacity.
With the Solar Controller powered up, for at least a minute to allow the filters to catch up, I requested Voltage and Capacity values from the Solar Controller. Now the BenchVue Capacity (0.1%) and Voltage (10.9892 V) and the Solar Controller readings (21% and 10.996V), causes the Solar Controller's response to capacity at 10.996V returns a capacity of 21%, which closely matches the original (non-normalized) profile. I will have to recompute the variables for my lookup table.
The BenchVue software did make the debug process fairly easy. I will have to try this again when I continue my project.
There is still a bit of work to finish my Fountain/Birdbath project. Hopefully I can pick up on the construction of the project soon, weather and health permitting. Here is a list of steps that remain to be taken:
On the project side of things, one of the best things that I learned or discovered in the course of this project was the usefulness and accuracy of the Texas Instruments INA219, Zero-Drift, Bidirectional Current/Monitor with I2C Interface. This part has been a really high point of the project and is something that I will plan on using in other future products.
The Keysight E36731A is an absolutely amazing piece of equipment. The features built into this product far outperform any similar instruments that I have used in the past. I can see this instrument adding so much more capability to my product development and testing. This is the most accurate, and powerful combination of Power Supply and Programmable load that I have ever seen. I did not start out as a big fan of the software, but once I tried out the Battery Emulation feature, that view changed greatly. I went from skeptical as to if I would renew the software license to pretty darn sure that I will have to consider it.
Thank you element14 ( rscasny ) and Keysight for giving me the opportunity to roadtest this amazing instrument!
Top Comments
message to the sponsor : if they could supply an eternal license with road test devices / software, the tester would likely keep on using the device, and keep on showing its capabilities in blogs. Way past…
Hi Gene,
Very useful review!