Readiness: Battery-Emulator Solar Storage Bridge

Verdict: Ready with caveats · 2026-09-03 · est. parts cost $58–$103 (one host board plus the commodity support parts; excludes the EV pack, the inverter, and the high-voltage DC cabling, which could not be priced to specification) Cybersecurity — software & network: Do not place on an open network

This is the cold check for this lab — there is no previous report, so everything below establishes the baseline for future drift measurement. The lab is buildable today. It is designed around six alternative host boards and needs only one of them; five of the six are purchasable right now between $11.86 and $39.90, and both CAN controller ICs are Active and stocked at several authorized distributors. Three things keep it off a clean "Ready". The exact CAN-FD controller ordering code the lab's materials name, MCP2518FD-E/SL, is not listed by DigiKey or by any ECIA-authorized distributor, so it has to be ordered as MCP2517FD-H/SL (which the lab's own materials also name) or as MCP2518FDT-E/SL. The high-voltage DC traction cable is the one safety-critical line, and none of the products found this run states the ≥ 600 V DC rating the lab requires. And the relay module that drives the battery contactors is sold as a 5 V board whose ability to trigger from the ESP32's 3.3 V logic was not confirmed. None of these is a dead end — each is a matter of buying the right variant.

Action required

Cybersecurity — software & network

Do not place on an open network — The device ships with a published default Wi-Fi password and serves its setup page and over-the-air firmware updater behind no authentication the sources record, on hardware that switches a high-voltage traction pack.

What this means for your lab

Treat this board as an untrusted device on your network until you have changed its settings. Out of the box it creates its own Wi-Fi network called Battery-Emulator with the password 123456789, and that password is printed in the project's public README — so anyone close enough to see the network can join it. Once joined, they reach the same web setup page you use, and the same over-the-air update page, and nothing in the project's documentation describes a login in front of either. On a bench that means someone could change how the bridge talks to the battery and the inverter, or replace its firmware outright, on a device whose job is closing contactors on a 300-450 V battery pack. Before you power it up in a shared space: change the access-point password on the setup page, keep the board on its own isolated network segment rather than the campus network, and talk to campus IT before connecting it to anything shared. Two things are genuinely unknown rather than fine: none of the firmware's libraries is pinned to a version, so no one can say which library code a given build contains, and Arduino libraries have no public vulnerability database, so 'no known issues' could not be established either way.

Not checked

• ESPAsyncWebServer, AsyncTCP, ElegantOTA, ArduinoJson, eModbus, Adafruit_NeoPixel, acan-esp32, acan2517FD: Unpinned in platformio.ini, so no version exists to check; and the Arduino/PlatformIO library ecosystem has no vulnerability database in OSV, so even a pinned version could not be looked up. Absence of an advisory here is not evidence of safety.
• pioarduino platform-espressif32 55.03.311: Version is pinned and was read from platformio.ini, but the toolchain ships as a vendor release archive with no entry in OSV and no product on endoflife.date, so there is no advisory or support-status source to check it against.
• Arduino core for ESP32 (framework-arduinoespressif32): Version not determined — it is pulled in transitively by the platform archive and no fetched file names it.
• Battery-Emulator firmware itself: Source-only project with no advisory database; no release version was determined.
• tools/identify_build.py: Local pre-build script referenced by platformio.ini; missing from the lab's uploaded materials and not fetched this run.
• MQTT client (broker, port, credentials): The sources state no broker address, no port and no credentials, so there was nothing to check.
• Web configuration server listening port: Not stated in the sources; port 80 is the snapshot's assumption, not an observation.
• BE-Web-Installer hosted page and the firmware image it serves: The page was not fetched this run and the binary it distributes was not examined.

Parts

Substitutes

Verify any substitute against its own datasheet before ordering — the checks above are only as good as the parametric data the distributors publish.

Other dependencies

• Fabrication — not required for the five commercial-off-the-shelf board paths. For the BECom target, the KiCad project, schematic PDF and footprint libraries are published under GPL-3.0 at github.com/rjsc/BECom, confirmed on the project wiki page fetched this run. This updates the lab's own materials, which state the design files are unavailable — they are missing from the uploaded materials but exist upstream. Layer count, copper weight and board outline were not checked; the KiCad project itself was not opened. For the Stark target no design files are published, because it is sold assembled.
• Project documentation wiki (dalathegreat.github.io/Battery-Emulator-Wiki) — live. Three pages were fetched successfully this run.
• Stark hardware vendor (shop.redispo.se) — live; the Stark CMR v2 product page was fetched successfully. The vendor notes there that it provides no software support for the board.
• Browser-based installation web app (dalathegreat.github.io/BE-Web-Installer) — not checked; the page was not fetched this run.
• External protocol-documentation sources — not checked. The lab's protocol translations were reverse-engineered from twelve external references: pelle8/inverter_resources, burra/byd_battery, flodorn/TeslaBMSV2, dalathegreat/leaf_can_bus_messages, rand12345/solax_can_bus, Tom-evnut/BMWI3BMS, FozzieUK/FoxESS-Canbus-Protocol, maciek16c/hyundai-santa-fe-phev-battery, ljames28/Renault-Zoe-PH2-ZE50-Canbus-LBC-Information, openvehicles/Open-Vehicle-Monitoring-System-3, an EEVBlog forum thread on the Pylontech SC0500 protocol, and a Google Docs spreadsheet. None was fetched this run.

Watch list

• The MCP2518FD ordering codes. Three of the four codes checked (MCP2518FD-E/SL, MCP2518FD-H/SL, MCP2517FD-E/SL) are carried by no ECIA-authorized distributor, while the tape-and-reel codes and MCP2517FD-H/SL are plentiful. Whichever code you standardise on, re-check it before each cohort — this family's stock is spread thin across ordering codes.
• DFRobot Edge101 (DFR0886). About 55 units across all three authorized distributors. Fine for one bench, not enough for a class ordering at once.
• ESP32-DEVKITC-32E at DigiKey. Zero available today with Mouser well stocked; a single-warehouse gap that could go either way by the next check.
• The high-voltage DC cable. It is the one line where the wrong purchase is dangerous, and generic shopping searches consistently return automotive cable that does not meet the voltage rating. It will need specifying by hand every time.
• The relay module's 3.3 V trigger. Cheap 5 V optocoupled boards vary in whether they switch reliably from 3.3 V logic; the products change constantly and none of the current listings confirms it.
• The twelve external protocol references. Community repositories, a forum thread and a Google Docs sheet, none checked this run — the most likely source of quiet link rot before the next check.