feat(gps): fix GPS support and add M10 dual-band GNSS configuration by amanpd7 · Pull Request #200 · rtlopez/esp-fc

amanpd7 · 2026-02-11T06:46:37Z

Fix GPS initialization issues and add M10 support with dual-band L1+L5 and user-configurable multi-constellation GNSS. Replace NMEA with UBX protocol for better reliability.

Features:

M10 dual-band L1+L5 for sub-meter accuracy
Multi-constellation support: GPS, GLONASS, Galileo, BeiDou, QZSS, SBAS
CLI configuration with preset modes (0-5) and individual control
Auto-detection of M8/M9/M10 modules
25Hz update rate on M10 at 230400 baud

Changes:

Remove NMEA parser from PR feat(gps): Add universal GPS module support with position estimation #199(closed), switch to UBX-only protocol
Add GPS_M10 enum and dual-band support
Add GNSS configuration parameters to GpsConfig
Implement configureGnss() for version-aware setup
Update docs/setup.md with GPS configuration guide
Update docs/README.md and docs/wiring.md

Configuration:

gps_gnss_mode: 0=Auto, 1=GPS-only, 2=GPS+GLO, 3=GPS+GAL, 4=GPS+BDS, 5=All
gps_enable_dual_band: Auto L1+L5 on M10
Individual flags: enableGPS, enableGLONASS, enableGalileo, etc.

Tested with M8/M9/M10 modules, all GNSS modes working.

Fix GPS initialization issues and add M10 support with dual-band L1+L5 and user-configurable multi-constellation GNSS. Replace NMEA with UBX protocol for better reliability. Features: - M10 dual-band L1+L5 for sub-meter accuracy - Multi-constellation support: GPS, GLONASS, Galileo, BeiDou, QZSS, SBAS - CLI configuration with preset modes (0-5) and individual control - Auto-detection of M8/M9/M10 modules - 25Hz update rate on M10 at 230400 baud Changes: - Remove NMEA parser from PR rtlopez#199(closed), switch to UBX-only protocol - Add GPS_M10 enum and dual-band support - Add GNSS configuration parameters to GpsConfig - Implement configureGnss() for version-aware setup - Update docs/setup.md with GPS configuration guide - Update docs/README.md and docs/wiring.md Configuration: - gps_gnss_mode: 0=Auto, 1=GPS-only, 2=GPS+GLO, 3=GPS+GAL, 4=GPS+BDS, 5=All - gps_enable_dual_band: Auto L1+L5 on M10 - Individual flags: enableGPS, enableGLONASS, enableGalileo, etc. Tested with M8/M9/M10 modules, all GNSS modes working.

amanpd7 · 2026-03-19T08:11:49Z

@rtlopez

rtlopez

Thanks for your effort, there are a lot of changes, can we split this PR into smaller parts indepandent parts, for example

add M10 support only
add gnss configuration
calc distance to home

Thank you.

rtlopez · 2026-04-10T12:17:27Z

docs/setup.md

+In the `Configuration` tab, under `Other Features`, enable **GPS** or use CLI:
+
+```
+set feature_mask = 128


this syntax is not valid here, also paramter name is incorrect. It has changed recently and now it should be

set feature_gps 1

rtlopez · 2026-04-10T12:18:52Z

docs/setup.md

+
+```
+# Minimum satellites required for valid fix
+set gps_min_sats = 8


invalid CLI syntax, = is not required.

rtlopez · 2026-04-10T12:22:12Z

docs/setup.md

+
+### GPS Features
+
+#### Return to Home (RTH)


There is no GPS rescue in this PR, why it is documented here?

rtlopez · 2026-04-10T12:31:29Z

docs/wiring.md

-| `pin_buzzer`        |  0    | 5        | Status buzzer |
-| `pin_led`           |  26   | -        | Status led    |
+| `pin_buzzer`        |  26   | 5        | Status buzzer |
+| `pin_led`           |  2    | -        | Status led    |


pin_buzzer and pin_led document values that are valid for latest stable release (v0.2.1), do not update yet.

rtlopez · 2026-04-10T12:34:56Z

lib/Espfc/src/Connect/Cli.cpp

  {
    printGpsStatus(s, true);
  }
+  else if(strcmp_P(cmd.args[0], PSTR("gps_home")) == 0)


once you added home position status to gps command, there is no need for gps_home command. It will be more clear to set home as gps sub-command, for example gps set_home.

rtlopez · 2026-04-10T12:55:17Z

lib/Espfc/src/Sensor/GpsSensor.cpp

+             sin(phi1) * cos(phi2) * cos(deltaLambda);
+
+  double theta = atan2(y, x);
+  double bearing = fmod((theta * 180.0 / M_PI + 360.0), 360.0);


same issue as haversineDistance()

rtlopez · 2026-04-10T12:58:37Z

lib/Espfc/src/Sensor/GpsSensor.cpp

+  return (int16_t)bearing;
+}
+
+void GpsSensor::setHomePosition() const


it is alredy implemented, see Model::setGpsHome() and Actuator::updateArmed()

rtlopez · 2026-04-10T13:00:24Z

lib/Espfc/src/Sensor/GpsSensor.cpp

-  _model.state.gps.accuracy.vertical = m.vAcc; // mm
-  _model.state.gps.accuracy.speed = m.sAcc; // mm/s
-  _model.state.gps.accuracy.heading = m.headAcc; // deg * 1e5
+  _model.state.gps.accuracy.horizontal = m.hAcc;


please revert comments with units. They are on purpose here and extremely usefull for me.

rtlopez · 2026-04-10T13:03:22Z

lib/Espfc/src/Sensor/GpsSensor.hpp

    SET_BAUD,
-    DISABLE_NMEA,
    GET_VERSION,
+    CONFIGURE_GNSS,


I'm not sure if CONFIGURE_GNSS is in conflict with ENABLE_SBAS. Anyway, I think, it would be safer to run configure_gnss as last phase.

rtlopez · 2026-04-10T13:04:35Z

README.md

 * In flight PID Tuning
 * Buzzer, Led and voltage monitor
 * Failsafe mode
+* GPS rescue


there is no GPS rescue here, this may confuse others.

amanpd7 · 2026-04-10T18:03:07Z

@rtlopez thanks for your suggestions and corrections, let me work on them to resolve asap

- restore BAUDS order (9600 first for cold-start detection) - restore DISABLE_NMEA state and processNmea() for safe frame skipping - move CONFIGURE_GNSS to last config phase (after ENABLE_SBAS) - replace raw CFG-GNSS byte array with UbxCfgGnss7 struct + send() - replace double in haversine/bearing with float equirectangular approx - remove setHomePosition/shouldSetHome; use Model::setGpsHome() instead - merge gps_home command into gps set_home/clear_home subcommands - restore unit comments on accuracy fields (mm, mm/s, deg*1e5) - remove GPS rescue config fields and docs (not in this PR) - fix CLI syntax in docs (set feature_gps 1, no = in set commands) - revert pin_buzzer/pin_led defaults in wiring.md - rewrite test_gps as standalone math tests (no Arduino dependency)

amanpd7 · 2026-04-10T18:44:53Z

Hi @rtlopez, thanks for the detailed review — I've addressed all the feedback in the latest commit.

Regarding splitting into smaller PRs: I understand the reasoning, but in this case the three areas are tightly coupled at the implementation level:

M10 detection (GET_VERSION → handleVersion()) is what determines whether configureGnss() enables dual-band — the GNSS configuration step has no effect without it.
CONFIGURE_GNSS is a new state in the same state machine that controls the entire initialization sequence — inserting it cleanly required touching GpsSensor.hpp, GpsSensor.cpp, and GpsProtocol.hpp together.
Distance to home adds distanceToHome/directionToHome to ModelState and wires them into MSP — splitting this out would leave the new GPS state fields dangling with no consumer.
Splitting would produce intermediate commits where the state machine is in a broken or incomplete state, which felt worse than a single coherent change. All three areas share the same files (GpsSensor.cpp, ModelConfig.h, ModelState.h) so the diffs would overlap regardless.

I've kept each logical area clearly separated within the code itself, and the review feedback has all been addressed. Happy to discuss further if you'd prefer a different approach.

rtlopez · 2026-04-10T21:17:44Z

Hi @amanpd7 now it looks pretty acceptable for me. I'll try to do some tests over the weekend and if all pass, I'll merge it. Thank you!

amanpd7 · 2026-04-10T21:21:42Z

@rtlopez Hi and thanks for reviewing so quickly, I'll also be testing these new changes by this weekend and will update here accordingly.

rtlopez · 2026-04-11T12:40:10Z

Hi @amanpd7 I've tested on HGLRC M100 today and cannot make it working. I got this log

I 3.50 GPS VER ROM SPG 5.10 (7b202e)
I 3.50 GPS VER 000A0000
I 3.50 GPS EXT FWVER=SPG 5.10
I 3.50 GPS EXT PROTVER=34.10
I 3.50 GPS EXT GPS;GLO;GAL;BDS
I 3.50 GPS EXT SBAS;QZSS
I 3.57 GPS UBX
I 3.61 GPS NAV5
I 3.63 GPS SBAS 1
I 3.67 GPS GNSS I 3.67 [I 3.67 GPS I 3.67 GLO I 3.67 GAL I 3.67 BDS I 3.67 QZSS I 3.67 SBASI 3.67 ]
E 3.71 GPS ERROR

I did quick look into datasheet, and noticed, that

message UBX-CFG-GNSS is deprecated in protocol versions greater than 23.01. Use UBX-CFG-VALSET, UBX-CFGVALGET, UBX-CFG-VALDEL instead.

it is possible, that 34.10, which I have, no longer suport it.

please check

amanpd7 · 2026-04-11T17:05:27Z

Hi @rtlopez I am using HGLRC M100-PRO gps module and did not find this issue.

This is a real hardware-specific issue.

Root cause: UBX-CFG-GNSS is officially deprecated in u-blox protocol versions > 23.01. Some M10 variants NAK it and error out; others silently accept it. My own M10 module (also PROTVER 34.10) accepted the command without issue, so I missed this during testing.

Immediate fix (already identified): In configureGnss(), change the timeout state from ERROR to SET_RATE so a NAK skips GNSS constellation config rather than halting initialization entirely:

// before
send(gnss, SET_RATE);
// after  
send(gnss, SET_RATE, SET_RATE);  // on NAK, skip to SET_RATE instead of ERROR

This lets the M100 initialize successfully with default GNSS settings.

Long-term fix: Parse the PROTVER field from MON-VER extended strings (already read in GET_VERSION state) and use UBX-CFG-VALSET with CFG-SIGNAL-* keys for modules with PROTVER > 23.01. I'll add this in a follow-up commit.

Can you confirm — does the HGLRC M100 use M10, M10S, or M10C silicon? The MON-VER hardware field (e.g. 00190000) would also help narrow down which variant is rejecting the command.

I'll also be pushing a permanent patch for it in this PR.

here is logs of my setup for reference

I 2.23 GPS BAUD 115200
I 2.25 GPS DET 115200
I 2.29 GYRO BIAS 2.25 1.51 0.74
I 2.77 GPS NMEA 0
I 2.87 GPS VER ROM SPG 5.10 (7b202e)
I 2.87 GPS VER 000A0000
I 2.87 GPS EXT FWVER=SPG 5.10
I 2.87 GPS EXT PROTVER=34.10
I 2.87 GPS EXT GPS;GLO;GAL;BDS
I 2.87 GPS EXT SBAS;QZSS
I 2.94 GPS UBX
I 2.98 GPS NAV5
I 3.02 GPS SBAS 1
I 3.06 GPS GNSS I 3.06 L1+L5 I 3.06 [I 3.06 GPS I 3.06 GLO I 3.06 GAL I 3.06 BDS I 3.06 QZSS I 3.06 SBASI 3.06 ]
I 3.12 GPS RATE 100 1

UBX-CFG-GNSS is deprecated on M10 modules with PROTVER > 23.01 and some variants (e.g. HGLRC M100) return a NAK, halting initialization. Pass SET_RATE as the timeout/NAK fallback state so GNSS constellation config is skipped gracefully and the module continues to initialize.

rtlopez · 2026-04-12T18:13:46Z

Hi @amanpd7, I don't really like this way of handling errors, but I decided to merge these changes anyway, I hope we can work on this further in the future. Thank you.

amanpd7 · 2026-04-12T18:19:38Z

Hi @rtlopez thanks for accepting my contribution, the issue you faced was completely on silicon level, chipset i am using is M10 and did not had any errors with same firmware running, this is a very rare occurance but sure my furthur works will be to handle this issue properly and i'll create a patch for this which is very much acceptable.
Thanks again!

rtlopez reviewed Apr 10, 2026

View reviewed changes

rtlopez merged commit 36c00f6 into rtlopez:master Apr 12, 2026
8 checks passed

Conversation

amanpd7 commented Feb 11, 2026

Uh oh!

amanpd7 commented Mar 19, 2026

Uh oh!

rtlopez left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

amanpd7 commented Apr 10, 2026

Uh oh!

amanpd7 commented Apr 10, 2026

Uh oh!

rtlopez commented Apr 10, 2026

Uh oh!

amanpd7 commented Apr 10, 2026

Uh oh!

rtlopez commented Apr 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

amanpd7 commented Apr 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rtlopez commented Apr 12, 2026

Uh oh!

Uh oh!

amanpd7 commented Apr 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

rtlopez left a comment •

edited

Loading

rtlopez commented Apr 11, 2026 •

edited

Loading

amanpd7 commented Apr 11, 2026 •

edited

Loading