Test equipment design and build: simulators, automation and measurement
We design and build test equipment, from electronics and firmware to automation and reports, for our own chargers, load balancers and firmware and for your products: simulators that imitate what surrounds a device (chargers on Modbus, a smart meter on P1, the car on the control pilot), automation that runs test sessions day and night, and measurement that logs and analyses the pilot signal, the voltage and current going to the car, and the charge, discharge and capacity of batteries, on the bench or remotely. As equipment, as a test service, or built for your product.
Chargers fail in the field on things a bench never showed: a car that falls asleep during a pause, a meter that sends its telegram a little differently, a bus that answers 10 ms too late. We build the equipment to provoke those cases on the bench, because our own boards, load balancers and firmware have to survive them. The same equipment can test yours, and we design and build new equipment for your product.
We design and build test equipment
We started making test equipment for our own use, because nothing on the market could imitate our sites, our meters and the cars we see in the field. It worked well enough that we now build it for others too. Test equipment is a product in its own right, and we make it the way we make our chargers: designed, built and documented in house, by the same engineers.
- Electronics. Our own circuit boards and interfaces: RS485, P1, control pilot, isolated inputs, power stages to charge and discharge.
- Firmware. Embedded software that imitates a device down to its timing, because a simulator that is a few milliseconds off tests the wrong thing.
- Automated instruments. Every instrument on our bench runs under software control: an automated oscilloscope, an automated programmable power supply, networked meters and an automated vector network analyser (VNA) for antenna tests. Scripts set them up, trigger them and read them out, so a measurement is repeatable and needs nobody at the bench.
- Automation and data. Test runs on a schedule, results over MQTT to a server, live dashboards and logs that survive a restart, and reports with the traces and curves.
- Mechanics and finish. Compact units on a DIN rail or in a case, with a touchscreen or without, labelled and documented so someone else can use them.
- Small series. One bench for your lab, or a series of identical stations for a production line or for partners, and, if you want, the firmware and software maintained after delivery.
Three building blocks
Every test setup we build combines the same three things, and all three are automated: simulators, instruments, logging and analysis run under software control, day and night, without anyone at the bench. The examples are ours; the method works for any device that talks to a bus, a meter or a car.
- Simulate what surrounds the device. The device under test sees a real-looking world: chargers on the bus, a smart meter on P1, a car on the control pilot. The simulator has to be as strict as the real thing, or the test passes on the bench and fails in the field.
- Automate the sessions. Cars plug in, charge, pause and leave on a schedule or at random, day and night, without anyone at the bench. The result is checked against what the device itself reports, the same way it reports in the field.
- Measure what reaches the car. The control-pilot signal (maximum and minimum voltage, duty cycle, frequency) and the voltage and current to the car, monitored and logged over hours or days, and analysed remotely while the test runs or afterwards, so a fault that happens once a night is still caught. The same logging measures batteries through charge and discharge, and runs on networked meters, so the bench does not have to be ours.
What we have
| Tool | What it does | Status |
|---|---|---|
| Site simulator (ETSS) | Imitates a complete charging site on one small unit: up to five charge controllers on the RS485 Modbus bus, with the full register map, states A, B and C, 1- or 3-phase charging, a charge current that follows the limit the controller sets, and energy counters; and a DSMR 5 smart meter on P1 whose load follows the simulated cars. A load balancer can be tested end to end, with no car and no charger | In daily use for our own development |
| Automated session runner | Drives the site simulator through idle, one-car and two-car sessions, unattended, every 15 minutes since August 2026. Each step is acknowledged by the simulator and checked against the controller’s own telemetry | In daily use |
| EV simulator | The car side of IEC 61851: states A to F, PP cable codes, and the reactions of different cars to a charging pause, scripted so firmware tests run without a car | In use for automated tests |
| Pilot and power monitoring | Monitors and logs the control-pilot signal (maximum and minimum voltage, duty cycle, frequency) and the voltage and current to the car, unattended over long runs; analysed remotely while the test runs, or afterwards | In use |
| Battery cell test | Charges and discharges a battery cell under software control, with an automated power supply, logs voltage, current and power throughout, adds up the energy (Wh) and charge (Ah) it delivers, and stops by itself when the cell is empty; every run ends with a report and a discharge curve | In use |
| Battery system test | For complete storage systems such as a home battery: round-trip efficiency (energy out against energy in), reaction time from a new power setpoint to the battery delivering it, and setpoint resolution (the smallest power step the system really follows) | In use |
| Remote voltage and current logging | Small networked meters that measure voltage, current and power on several channels and send them every few seconds, so a test bench, a battery or an installation is logged and followed from a distance | In use |
| Antenna test | An automated vector network analyser sweeps an antenna from 50 kHz to 6 GHz, calibrated per band (868 MHz, 2.4 GHz, 5 GHz), and stores the return loss as plots and Touchstone files, to compare antennas and their position in a housing or a meter cupboard | In use |
| Modbus tool | Reads and writes every register of the charge controller from a PC, over RS485 or a TCP gateway | Shipped, open source (GitHub) |
| Firmware flasher | Updates the charge controller’s firmware over its Modbus connection | Shipped (release 2.10.19) |
| EV wake-up test | Three ways to wake a sleeping car (CP interruption, “charger absent”, unplug/replug sequence) with adjustable timing, to find what a given car needs | Shipped, in the charge controller firmware |
| GridShield interoperability | Our GridShield controller and receiver tested against ElaadNL’s implementation in both directions | Tested |
What the bench has caught
A meter telegram that was too slow, by a few milliseconds
Seen. On the site simulator, the controller counted a CRC failure on the smart-meter telegram about every five minutes. On a real meter the same firmware counted none.
Found. A real meter sends its whole telegram of about 850 bytes in one burst of roughly 74 ms. The controller decides a telegram is complete when the line has been quiet for 25 ms. Our simulator wrote the telegram in pieces, and when it was briefly busy with something else, the gap was long enough to cut the telegram in two.
Lesson. The simulator now sends the telegram in one burst, as a meter does. A simulator that is kinder or harsher than the real device tests the wrong thing.
A controller that went quiet, and a simulator that forgave it
Seen. In the field, charging sometimes stopped while the site controller was reconnecting to its network. On the bench it never did.
Found. The real charge controller keeps a current limit from the site controller for 2 seconds and then falls back to zero, so a site controller that stops refreshing stops the charging. The simulator kept the last limit for ever, so a stalled site controller looked healthy.
Lesson. The simulator now lets the limit lapse after 2 seconds, like the board, and counts every lapse on its screen. The field fault is now reproducible on the bench, where it can be tested again after every change.
Supply dips nobody was there to see
Seen. A site controller powered from the smart meter’s P1 port restarted now and then, too rarely to catch by hand.
Done. An automated oscilloscope on the 5 V at the P1 connector, re-armed every minute by software, records every dip below a threshold with its minimum, average and the waveform of the deepest one. Days of logging answer whether the supply sags, how often and how deep, without anyone watching.
Proving what the car actually saw
Question. When a car does not wake up after a pause, did the charger send the wake-up signal, with the right levels and timing, or did the car ignore it?
Done. Measuring the control pilot during the wake-up sequence shows the levels and durations the car really received. Together with the EV simulator, the wake-up variants can be checked after a firmware change, before it reaches a car in the field.
What you can use it for
- Load-balancing algorithms. Simulated cars that start, stop and change phase on command, against your controller’s limit, without a car park full of cars.
- Firmware regression. Run the same sessions after every firmware change and compare the register traces and the pilot recordings.
- Protocol timing. The charge controller expects an answer within 40 ms; we measure where a gateway or a slow master breaks that.
- Smart-meter variants. P1 telegrams with import, export, 1 or 3 phases and edge cases, for a controller that reads them.
- Car behaviour. How your charger handles a car that sleeps, a slow state change or a cable with a lower rating, with the pilot signal recorded.
- Long-run reliability. Days of unattended sessions with power and pilot logging, to catch the fault that happens once a night.
- Battery cells. Capacity measured, not taken from the datasheet: charge, discharge and capacity tests, repeated to see how a cell ages, and the real consumption of a device that runs on one.
- Battery systems. How efficient a storage system really is, how fast it follows a new setpoint and how fine its steps are, the figures an energy controller depends on.
- Remote monitoring. Voltage and current logged at a site or on a bench elsewhere, followed live and analysed afterwards.
- Interoperability. GridShield both ways, Modbus against our register map, Matter devices commissioned into a controller we run.
- End-of-line test. A test station for chargers coming off your assembly line. On request.
Beyond chargers
The same method fits other devices on the grid edge. A heat pump, home battery or inverter controller also lives between a bus, a meter and a load. We can simulate the meter and the devices around it, run the scenarios automatically, and measure what the device really does, so your controller is tested before it meets a real installation.
Three ways to work with us
- Equipment. A simulator or test bench, built to order and configured for your site sizes and protocols.
- Test service. You send your charger, load balancer or energy controller; we run it on our bench and report what we found, with the traces.
- Engineering. We build a simulator, test automation or measurement setup for your protocol or product, as part of an engagement.
Prices and lead times on request. Contact us with what you want to test.
Can EdgeTech test my charger or load balancer without real cars?
Yes. The site simulator imitates the chargers and the smart meter, the EV simulator imitates the car on the control pilot, and the session runner drives both unattended. We record the pilot signal and the voltage and current to the car, and report what we found with the traces.