gnss: auto-detect L5 receiver with separate config arrays by leoleovich · Pull Request #153 · Orolia2s/oscillatord

leoleovich · 2026-06-15T12:00:55Z

The F9T-10B supports L1+L5 but not L2. At startup, probe the receiver's RAM for CFG-SIGNAL-GPS_L5_ENA to detect the receiver model:

F9T-00B (standard): uses default_configuration_v220 (L1+L2 signals)
F9T-10B (L5): uses default_configuration_f9t_10b (L1+L5 signals)

The correct config array is selected before comparison, so an already- configured receiver won't trigger unnecessary hardware resets.

Also adds:

Per-band satellite monitoring via UBX-RXM-RAWX (always enabled)
L1/L2/L5 signal counts in per-second log output
Proper maxKv sizing for rxGetConfig (must be >= batch + 64)
RTCM base station output with raw observations and ephemeris

L1/L5 receiver:

Jun 15 08:19:12 time12 oscillatord[81921]: 08:19:12 INFO  L5 receiver detected (F9T-10B), using L5 signal config
...
Jun 15 08:22:41 time12 oscillatord[82177]: 08:22:41 DEBUG GNSS data: Fix time (3), Fix ok: True, satellites num 29, signals L1=40 L2=0 L5=26, survey in error: 0.70, antenna status: 4, valid 1, time 1781526161, leapm_seconds 18, leap_notify 0, lsChange 0, timeToLsChange 11360240, lsSet: True, QErr(n) -3077, qErr(n-1) 3730, pos_accuracy 0 m, time_accuracy 21 ns

L1/L2 receiver (L1 antenna):

Jun 16 05:40:53 time11 oscillatord[283304]: 05:40:53 DEBUG GNSS data: Fix time (3), Fix ok: True, satellites num 32, signals L1=41 L2=0 L5=0, survey in error: 4.52, antenna status: 4, valid 1, time 1781602853, leapm_seconds 18, leap_notify 0, lsChange 0, timeToLsChange 11283547, lsSet: True, QErr(n) -824, qErr(n-1) 3517, pos_accuracy 4 m, time_accuracy 23 ns

leoleovich · 2026-06-15T12:31:29Z

@nbourdau please take a look

nbourdau

Could you split up this commit into smaller and more atomic commits?
Being a big one merging several changes make it hard to review properly and quickly. Having the changes split gives the occasion in the commit message of each to explain at least why each one is needed and make clear by looking at each one what code change is relevant for which semantic change.

A natural split is hinted in your current commit message (the "also adds" list + auto-detect L5), but another split could be chosen.

leoleovich · 2026-06-16T08:55:44Z

@nbourdau done, split in 5 different commits, but kept 1 pull request

nbourdau

Please ensure that all commits compile

nbourdau · 2026-06-16T15:18:54Z

@@ -90,18 +120,24 @@ UBLOXCFG_KEYVAL_t *get_default_value_from_config(int *nKv, int major, int minor)
 	char     current_line[128] = {};

 	{
-		size_t config_length = default_configuration_size;
-
-		// If version >= to 2.20 apply 2.20 ubx config
-		if ((major == 2 && minor >= 20) || (major >= 3))
+		/* Select config array based on firmware version and band mask */
+		const char (*cfg_array)[128];
+		size_t config_length;
+
+		if (bands & GNSS_BAND_L5) {
+			cfg_array = default_configuration_f9t_10b;
+			config_length = default_configuration_f9t_10b_size;
+		} else if ((major == 2 && minor >= 20) || (major >= 3)) {
+			cfg_array = default_configuration_v220;
 			config_length = default_configuration_v220_size;
+		} else {
+			cfg_array = default_configuration;
+			config_length = default_configuration_size;
+		}

 		for (size_t i = 0; i < config_length; i++)
 		{
-			if ((major == 2 && minor >= 20) || (major >= 3))
-				strncpy(current_line, default_configuration_v220[i], 127);
-			else
-				strncpy(current_line, default_configuration[i], 127);
+			strncpy(current_line, cfg_array[i], 127);
 			IO_LINE_t line = { .line = current_line };


This makes the commit not to compile, this part must be introduced later (with "gnss-config: add band mask parameter to config selection" commit)

nbourdau · 2026-06-16T15:20:12Z

+		if (bands & GNSS_BAND_L5) {
+			cfg_array = default_configuration_f9t_10b;
+			config_length = default_configuration_f9t_10b_size;


This must be introduced no before "f9_defvalsets: add F9T-10B (L1+L5) config array" commit

nbourdau · 2026-06-16T15:26:45Z

+
 bool check_gnss_config_in_ram(RX_t *rx, UBLOXCFG_KEYVAL_t *allKvCfg, int nAllKvCfg);
-UBLOXCFG_KEYVAL_t *get_default_value_from_config(int *nKv, int major, int minor);
+UBLOXCFG_KEYVAL_t *get_default_value_from_config(int *nKv, int major, int minor, int bands);


Missing src/gnss.c update for use of get_default_value_from_config

rxGetConfig() internally checks if there's room for another 64-item batch after receiving data: (totNumKv + 64) > maxKv. This means maxKv must always be at least received_items + 64, otherwise the call fails with "Too many config items". Fix by using a 128-element result buffer for batch reads of 64 keys at a time. Also simplify check_gnss_config_in_ram() to read only the keys we need (from the config array) rather than using the ALL wildcard which can overflow on receivers with many config items.

The ZED-F9T-10B supports L1+L5 signals instead of L1+L2. Add a dedicated config array (default_configuration_f9t_10b) with the correct L5 signal keys: GPS_L5_ENA, GAL_E5A_ENA, QZSS_L5_ENA = true BDS_B2_ENA, GLO_L2_ENA = false (no L5 equivalent)

Add GNSS_BAND_L1/L2/L5 flags and pass a band mask to get_default_value_from_config() so it selects the correct config array for the detected receiver model. At startup, probe CFG-SIGNAL-GPS_L5_ENA from receiver RAM. If L5 is enabled, load the F9T-10B config array instead of the standard L2 config. This ensures an already-configured L5 receiver won't trigger unnecessary hardware resets on restart. As fallback, if L5 wasn't detected but the L2 config is rejected by the receiver, reload with L5 config and retry.

Always enable UBX-RXM-RAWX on UART1 (moved out of RTCM gate) and parse it every second to count L1/L2/L5 signals. Classification uses both gnssId and sigId per the u-blox UBX-RXM-RAWX signal numbering table: GPS (0): sigId 0,1=L1 3,4=L2 6,7=L5 SBAS (1): sigId 0=L1 3=L5 Galileo (2): sigId 0,1=E1(L1) 3,4=E5a(L5) 5,6=E5b(L2) BeiDou (3): sigId 0,1=B1I(L1) 2,3=B2I(L2) 5,6=B1C(L1) 7=B2a(L5) QZSS (5): sigId 0,1=L1 4,5=L2 8,9=L5 GLONASS (6): sigId 0=L1OF 2=L2OF Signal counts are included in the per-second log line: signals L1=N L2=N L5=N

leoleovich · 2026-06-16T15:57:02Z

Thanks for the feedback verified each commit

leoleovich · 2026-06-17T12:02:25Z

Thank you @nbourdau please issue a release when you can we don't have pending updates.
Side node - you probably want to enable automated branch removal on merge.

leoleovich force-pushed the feat/l5-signal-detection branch 4 times, most recently from 97fb191 to 5aea3d2 Compare June 15, 2026 12:22

nbourdau requested changes Jun 16, 2026

View reviewed changes

leoleovich force-pushed the feat/l5-signal-detection branch from 5aea3d2 to c05cd90 Compare June 16, 2026 08:53

leoleovich requested a review from nbourdau June 16, 2026 09:10

nbourdau requested changes Jun 16, 2026

View reviewed changes

leoleovich added 4 commits June 16, 2026 08:46

leoleovich force-pushed the feat/l5-signal-detection branch from c05cd90 to 6c25739 Compare June 16, 2026 15:56

leoleovich requested a review from nbourdau June 16, 2026 15:57

nbourdau approved these changes Jun 17, 2026

View reviewed changes

nbourdau merged commit 6f36b9b into Orolia2s:main Jun 17, 2026
1 check passed

leoleovich mentioned this pull request Jun 18, 2026

Please create a release #154

Closed

Navigation Menu

Repository navigation

gnss: auto-detect L5 receiver with separate config arrays - #153

gnss: auto-detect L5 receiver with separate config arrays#153
nbourdau merged 4 commits into
Orolia2s:mainfrom
leoleovich:feat/l5-signal-detection

leoleovich commented Jun 15, 2026 •

edited

Loading

Uh oh!

leoleovich commented Jun 15, 2026

Uh oh!

nbourdau left a comment

Uh oh!

leoleovich commented Jun 16, 2026

Uh oh!

nbourdau left a comment

Uh oh!

nbourdau Jun 16, 2026

Uh oh!

nbourdau Jun 16, 2026

Uh oh!

nbourdau Jun 16, 2026

Uh oh!

leoleovich commented Jun 16, 2026

Uh oh!

Uh oh!

leoleovich commented Jun 17, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

leoleovich commented Jun 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

leoleovich commented Jun 15, 2026

Uh oh!

nbourdau left a comment

Choose a reason for hiding this comment

Uh oh!

leoleovich commented Jun 16, 2026

Uh oh!

nbourdau left a comment

Choose a reason for hiding this comment

Uh oh!

nbourdau Jun 16, 2026

Choose a reason for hiding this comment

Uh oh!

nbourdau Jun 16, 2026

Choose a reason for hiding this comment

Uh oh!

nbourdau Jun 16, 2026

Choose a reason for hiding this comment

Uh oh!

leoleovich commented Jun 16, 2026

Uh oh!

Uh oh!

leoleovich commented Jun 17, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

leoleovich commented Jun 15, 2026 •

edited

Loading