Passer au contenu

Nous améliorons nos opérations pour mieux vous servir. Les commandes sont expédiées normalement depuis Laval, QC. Questions? Contactez-nous

Bitcoin accepté au paiement  |  Expédié depuis Laval, QC, Canada  |  Soutien expert depuis 2016

NERDAXE_AP_FAIL Warning

NerdAxe – Web UI at 192.168.4.1 Unreachable

NerdAxe broadcasts the `NerdAxe-XXXX` provisioning hotspot on power-up but the captive portal at `http://192.168.4.1` will not load. Without the portal, WiFi credentials cannot be entered and the miner cannot be configured. Common causes: stale NVS WiFi credentials carrying over so the firmware skips AP mode, browser HSTS-upgrading the URL to HTTPS, iOS captive-portal redirect not firing, VPN / iCloud Private Relay / private DNS intercepting the local-IP request, phone auto-switching off the no-internet AP, or partition corruption after a botched OTA / Web Flasher session.

Warning — Should be addressed soon

Affected Models: NerdAxe (BM1366 + ESP32-S3). Applies to all NerdAxe board revisions running the BitMaker-hub ESP-Miner-NerdAxe firmware fork (AxeOS-based). Captive-portal mechanics shared with Bitaxe-family AxeOS, so most root causes are also relevant to Bitaxe Supra / Ultra / Gamma / Hex.

Symptoms

  • NerdAxe broadcasts `NerdAxe-XXXX` (or `CONFIG-NERDAXE-XXXX`) SSID and a phone or laptop joins it cleanly
  • Browser shows `ERR_CONNECTION_REFUSED`, `ERR_CONNECTION_TIMED_OUT`, `ERR_EMPTY_RESPONSE`, or `ERR_SSL_PROTOCOL_ERROR` when loading `192.168.4.1`
  • Typed `192.168.4.1` automatically becomes `https://192.168.4.1` in the URL bar - browser is HSTS-upgrading
  • Browser bounces to a Google search page for `192.168.4.1` instead of loading the host
  • Captive portal banner (`Sign in to network`) never appears on iOS or Android even after 60 s of being on `NerdAxe-XXXX`
  • iPhone shows the spinning `Sign in to NerdAxe-XXXX` sheet but the sheet itself is blank or hangs
  • Cloudflare-style HSTS warning appears: `This site can't provide a secure connection`, `HSTS_FAILED`, or a `STRICT-TRANSPORT-SECURITY` mention
  • OLED on the NerdAxe shows `AP MODE` or `SETUP` continuously - device is in provisioning state
  • OLED shows the IP `192.168.4.1` but `ping 192.168.4.1` from the joined client times out
  • Provisioning portal worked on a laptop but not on a phone, or vice versa
  • Phone reports `No Internet` on `NerdAxe-XXXX` and silently switches back to home `5 GHz` SSID or to cellular
  • Recent OS / browser update on the phone within the last 7 days correlates with the failure
  • Web Flasher session crashed mid-flash or USB-C was yanked mid-OTA - partition layout possibly half-written

Step-by-Step Fix

1

Type `http://192.168.4.1` with the explicit `http://` scheme directly into the browser address bar. Do not type bare `192.168.4.1` and trust autocomplete; do not paste from a guide that may have stripped the scheme. The explicit prefix prevents Chrome / Edge / Safari from upgrading to HTTPS via the HSTS preload list, prevents the address bar from interpreting the IP as a search query, and prevents iOS from substituting Spotlight Suggestions. Hit Enter exactly once - the AxeOS / NerdAxe captive portal should render within 5 seconds. This single keystroke discipline resolves the largest single bucket of `192.168.4.1` complaints in the D-Central queue.

2

Disable VPN, iCloud Private Relay, custom DNS, and ad-blockers on the provisioning device. iOS: Settings > Apple ID > iCloud > Private Relay = OFF; Settings > General > VPN & Device Management > disconnect any VPN profile; Settings > WiFi > tap (i) next to `NerdAxe-XXXX` > Limit IP Address Tracking = OFF. Android: Settings > Network > VPN > disconnect; Settings > Network > Private DNS > Off; pause any DNS-changer / ad-blocker apps. Browsers: try an incognito / private window with all extensions disabled. Re-enable everything after provisioning.

3

Try a different browser, ideally one that has never visited `192.168.4.1` on HTTPS. Firefox is usually cleanest - it is more conservative about HSTS preload than Chrome and Safari. Brave private window also works. Edge InPrivate is fine. Confirm the new browser actually loads `http://example.com` first to rule out a broader internet outage on the joined SoftAP. Then retry `http://192.168.4.1` with the explicit scheme.

4

Hold `BOOT` for 10+ seconds during power-up to wipe NVS. Disconnect USB-C. Press and hold the `BOOT` button on the NerdAxe PCB (small tactile near the USB-C port - check your board silkscreen). Reconnect USB-C while still holding. Continue holding a full 10 seconds after power is applied. Release. The device now emits a fresh `NerdAxe-XXXX` AP with no stored credentials, no stale STA mode, no half-written config. Reconnect, retry `http://192.168.4.1`. Mandatory if the NerdAxe was ever provisioned previously - leftover NVS WiFi creds make the firmware silently try STA mode first instead of starting the AP cleanly.

5

Disable Auto-Join and Wi-Fi Assistant on the provisioning device so it stops bouncing off the no-internet NerdAxe SSID. iOS: Settings > WiFi > tap (i) next to your home SSID > Auto-Join = OFF; Settings > Cellular > Wi-Fi Calling = OFF temporarily; Settings > Mobile Data > Wi-Fi Assist = OFF. Android: Settings > Network > WiFi > Advanced > Auto-switch to mobile data = OFF; tap your home SSID > Forget temporarily; disable `Adaptive Connectivity` and `Smart Network Switch` on Pixel / Samsung. After provisioning, restore.

6

Clear HSTS cache for `192.168.4.1` in your primary browser. Chrome / Edge / Brave: navigate to `chrome://net-internals/#hsts`, scroll to `Delete domain security policies`, type `192.168.4.1`, click Delete. Safari: HSTS records are stored in `~/Library/Cookies/HSTS.plist` on macOS - quit Safari and delete the file (it rebuilds on next launch). Firefox: Settings > Privacy > Cookies and Site Data > Manage Data. Then retype `http://192.168.4.1` from scratch. Use this if you have legitimately visited a different `192.168.4.1` device on HTTPS in the past and your browser has cached HSTS for the host.

7

Manually set DNS on the phone for the NerdAxe SSID. iOS: Settings > WiFi > tap (i) next to `NerdAxe-XXXX` > Configure DNS > Manual > add `192.168.4.1` as the DNS server, remove any other entries. Android: long-press SSID > Modify network > Show advanced > IP settings = Static > DNS 1 = `192.168.4.1`. Forces the captive-portal DNS responder on the NerdAxe to be used for any auto-detection check (Apple's `captive.apple.com`, Android's `connectivitycheck.gstatic.com`), which often kicks the captive-portal sheet open immediately. Revert DNS to automatic after provisioning.

8

Set the phone or laptop to a static IP within the NerdAxe AP subnet. iOS WiFi configure pane > Configure IP > Manual > IP `192.168.4.10`, Subnet `255.255.255.0`, Router `192.168.4.1`. Android: Modify network > advanced > IP = Static, same fields. Removes any DHCP race or hand-out failure on the SoftAP side as a variable. Mining Hacker move that bypasses an entire class of `DHCP didn't grant in time` failures the user side never sees diagnostically. Revert to DHCP / Automatic after provisioning.

9

Try provisioning from a laptop instead of a phone. Phones are the worst environment for captive portals because of all the OS-level smart features fighting you. A laptop on `NerdAxe-XXXX` with no VPN, no extensions, no Private Relay, loads `http://192.168.4.1` cleanly nine times out of ten. macOS, Windows 10/11, Linux all work. Disable any corporate / IT-managed VPN clients first (Cisco AnyConnect, Zscaler, Palo Alto GlobalProtect) - those route LAN-local IPs through the corporate tunnel and drop the response. After successful provisioning the NerdAxe joins your home WiFi and the laptop is no longer needed.

10

Hardware reset attempt: power off the NerdAxe completely, count to 30, power on. Some `esp_http_server` socket-state failures clear only with a true cold boot. A `RESET` button press does not always release every TCP socket state on the SoftAP side, particularly if you previously connected from a phone that has now wandered off. A 30-second cold power-off lets the ESP32-S3 fully de-energise, NVS state flushes, and the next boot starts with a clean networking stack. After power-up wait 60 s, rejoin `NerdAxe-XXXX`, retry the portal.

11

Capture the serial-console boot log over USB-C at `115200 baud`. USB-C the NerdAxe to a laptop with a data-capable cable. Open PuTTY on Windows, Tera Term, the Arduino IDE serial monitor, `screen /dev/tty.usbmodem* 115200` on macOS / Linux, or `minicom`. Power-cycle the NerdAxe and watch the boot log. Look for `wifi:mode: AP`, `httpd: starting`, `httpd: bound to port 80`, `nvs: read result`, `partition table OK`. If you see `httpd: bind failed`, `nvs: corrupt`, or `partition table CRC mismatch`, you are looking at firmware partition damage - go to Web Flasher reflash. If you see clean SoftAP startup and HTTP bind, the issue is on the client side - return to Tier 1-2 with the boot-log evidence in hand.

12

Web Flasher reflash with factory NVS erase. Hold `BOOT`, tap `RESET`, release `BOOT` to enter ESP32-S3 bootloader mode. Open a WebSerial-capable browser (Chrome, Edge, Opera - not Firefox, not Safari). Go to `https://github.com/BitMaker-hub/ESP-Miner-NerdAxe/releases` and pick the firmware for your NerdAxe board revision. In the Web Flasher, check `Erase NVS` / `Factory reset` before flashing. CRITICAL: do not flash a stock Bitaxe Supra / Ultra / Gamma firmware on the NerdAxe - the NerdAxe uses BM1366 and standard Bitaxe Gamma firmware targets BM1368, which will brick chip communication. After flashing, hard power-cycle (unplug 30 s) and retry `http://192.168.4.1` from scratch. If chip enumeration fails after the reflash, see sibling `nerdaxe-bm1366-not-bm1368-firmware-mismatch`.

13

Confirm or replace the USB-C cable and PSU. Marginal USB-C cables and underpowered chargers cause the ESP32-S3 to brown out during the WiFi TX bursts that drive SoftAP beacons - the AP appears to start but is unstable, phones see the SSID, fail to associate cleanly, get a partial DHCP, then drop. Swap to a known-good data-capable USB-C cable and a wall brick rated for at least `5 V / 3 A` sustained. Avoid laptop / hub USB-C ports for first-boot provisioning. After confirming clean power, retry the portal. If the serial log showed `Brownout detector was triggered` in Step 11, this is your fix.

14

Search and / or file an `ESP-Miner-NerdAxe` GitHub issue with your boot log attached. With the serial-console capture from Step 11, search open and closed issues at `https://github.com/BitMaker-hub/ESP-Miner-NerdAxe/issues` for matching boot-log lines. If you find a prior reproducer with a known fix, follow it. If not, file a new issue with the boot log, board revision, firmware version, browser / phone OS version, and the Tier 1-2 steps you have already walked. Maintainers respond, and D-Central contributes reproducers to the upstream Bitaxe `ESP-Miner` tracker on the same class of failures - the next pleb who hits this gets fixed faster.

15

When to stop DIY. If you have walked Tier 1-3 (explicit `http://` prefix, two clean browsers, VPN / Private Relay / DNS off, both phone and laptop tried, NVS wiped via 10-second `BOOT` hold, Web Flasher reflash with factory NVS erase, known-good USB-C PSU and cable, serial-console boot log captured) and the captive portal still will not load, open a D-Central support ticket with your serial log attached, or drop into the D-Central Discord `#nerd-family` / `#bitaxe-help` channel. You will get a human who has seen the exact failure on the same firmware version before.

16

When it is hardware. Hardware AP-portal failure on a NerdAxe is uncommon but real: ESP32-S3 radio damage from ESD, USB-C power-input damage from a bad PSU, partition / flash-IC damage from a botched Web Flasher session over a flaky cable. Signature: SoftAP fails to broadcast, or broadcasts but `ping 192.168.4.1` from a joined client times out and `RESET` does not change behaviour after a clean Web Flasher reflash with factory NVS erase. D-Central stocks and supports the full Nerd family - units purchased through our store get warranty honoured, units from elsewhere we still diagnose for a modest bench fee. Stop further reflashes if you see PCB damage.

17

Ship safely if the NerdAxe is heading to the bench. Anti-static bag, 5 cm of foam on every side of the outer box. Include a note with: board revision (printed on the PCB and often on the outer box), current ESP-Miner-NerdAxe firmware version, the phone / laptop / browser combo you tried, the captive-portal symptom you saw (browser error string, OLED state, ping result), the Tier 1-3 steps you have already walked, and your contact info. Context saves triage time and triage time saves your bench dollar. Canada-wide turnaround on NerdAxe-class work is typically 3-7 business days. US and international are welcomed - D-Central ships worldwide from Canada.

When to Seek Professional Repair

If the steps above do not resolve the issue, or if you are not comfortable performing these repairs yourself, professional service is recommended. Attempting advanced repairs without proper equipment can cause further damage.

Related Error Codes

Still Having Issues?

Our team of Bitcoin Mining Hackers has been repairing ASIC miners since 2016. We have seen it all and fixed it all. Get a professional diagnosis.