Add ADS131m0x support for load cells by garethky · Pull Request #838 · KalicoCrew/kalico

garethky · 2026-02-19T02:20:14Z

This adds support for the ADS131M02 and ADS131M04. These share a general architecture so they can have a single MCU driver.

The main difference between the variants is the frame size. Each frame is composed of 24 bit words. Each frame is 2 words (status & crc) + 1 word for each channel:

M02: [STATUS, CHANNEL0, CHANNEL1, CRC]
M04: [STATUS, CHANNEL0, CHANNEL1, CHANNEL2, CHANNEL3, CRC]

All writes to registers require sending a complete frame and reading back a complete frame. The register addresses are the same across all models in the sensor family (02, 04, 06 & 08). TI designed it so we don't need separate code for each one.

The 04 variant can simultaneously sample 4 channels. The goal with that chip is to support under-bed load cells where a reading would be taken independently from each load cell. In the commercial world there is a piece of hardware called a "Summing Box" that electrically sums multiple load cells together to make large scales (think for semi trucks).

Basic summing box support is in this PR that fuses multiple sensor channels into a single LoadCell. The MCU side code for the sensor also sums the values and passes the sum to the load_cell_probe for triggering. More advanced features, like taring of individual channels is left for another future PR.

Checklist

pr title makes sense
added a test case if possible
if new feature, added to the readme
ci is happy and green

dalegaard

Have done quick review, but not reviewed/commented on anything saying TODO.

docs/Config_Reference.md

klippy/extras/load_cell/ads131m0x.py

klippy/extras/load_cell/ads131m02.py

klippy/extras/load_cell/ads131m0x.py

src/sensor_ads131m02.c

dalegaard · 2026-02-24T08:28:22Z

src/sensor_ads131m02.c

+    adc->sb.data[adc->sb.data_count + 1] = counts >> 8;
+    adc->sb.data[adc->sb.data_count + 2] = counts >> 16;
+    adc->sb.data[adc->sb.data_count + 3] = counts >> 24;
+    adc->sb.data_count += BYTES_PER_SAMPLE;


Since the sensor data is 24 bits, can't we just send 3 bytes and do sign handling on the host? It seems like one byte would always be zero here? counts at the call site looks to only fill 3 bytes

This is actually an optimization: Python doesn't have a 24bit type that can be used with struct.unpack (because c doesn't have a 24 bit datatype).

Early on when I tried to get the sensors to go fast I discovered that manually bit shifting in python to create each int was one of the major performance bottlenecks. Switching to stuct.unpack moved this into python's c routines and is a huge speedup.

This improvement was folded into FixedFrequencyReader and now any data sent over the BulkSensor mechanism must be amenable to struct.unpack.

ADC Sensors can sample more than a single channel if multiple physical load cells are connected. The LoadCell instance will sum those discrete readings to give a single weight value. This is similar to a "summing box" in a real world phyiscal load cell setup. The data for the additional channels is re-broadcast so it can be charted in debugging tools. TODO: * Display individual channel data in commands like LOAD_CELL_READ and LOAD_CELL_DIAGNOSTIC

dalegaard

Overall looks very nice. Will get it tested on my board asap!

klippy/extras/load_cell/ads131m0x.py

klippy/extras/load_cell/load_cell.py

src/sensor_ads131m0x.c

garethky · 2026-03-19T22:21:28Z

klippy/extras/load_cell/ads131m0x.py

+        # GAIN register (0x04): PGAGAIN0[2:0] at bits 2:0, PGAGAIN1[2:0] at bits 6:4
+        self._write_and_verify_reg(GAIN1_REG, self._gain_register_value())
+        self.reactor.delay_for_ms(50.0)
+        status = self._read_reg(STATUS_REG)


Reading the status here is just done to aid in debugging. But nothing about it is verified or used. This could be removed.

While not actually used, I like having this status output. It should prefix with the sensor type and name though, like the other logging statements.

garethky · 2026-03-20T06:42:53Z

klippy/extras/load_cell/ads131m0x.py

+        super().__init__(
+            config,
+            sensor_type="ads131m02",
+            sample_rate_options={


The sample rate options are duplicated. these can be pulled back up into the base class.

Moved to the constants area

dalegaard

Seems to work well, just a few tiny changes and then we'll get this merged!

src/sensor_ads131m0x.c

klippy/extras/load_cell/ads131m0x.py

dalegaard · 2026-03-23T20:21:05Z

klippy/extras/load_cell/ads131m0x.py

+        # GAIN register (0x04): PGAGAIN0[2:0] at bits 2:0, PGAGAIN1[2:0] at bits 6:4
+        self._write_and_verify_reg(GAIN1_REG, self._gain_register_value())
+        self.reactor.delay_for_ms(50.0)
+        status = self._read_reg(STATUS_REG)


While not actually used, I like having this status output. It should prefix with the sensor type and name though, like the other logging statements.

* Support for configuring one or more channels to be summed. * Configure the sample rate and gain * Enable high perfomrance mode by default * Check for unexpected device resets and CRC data errors while sampling Signed-off-by: Gareth Farrington <gareth@waves.ky>

dalegaard reviewed Feb 24, 2026

View reviewed changes

garethky force-pushed the pr-sensor-ads131m02 branch 3 times, most recently from b70c306 to be1b3b0 Compare March 1, 2026 03:20

garethky changed the title ~~Add ads131m02 support for load cells~~ Add ADS131m0x support for load cells Mar 1, 2026

garethky force-pushed the pr-sensor-ads131m02 branch 2 times, most recently from c9a380c to f5c63aa Compare March 18, 2026 22:42

garethky marked this pull request as ready for review March 18, 2026 22:51

dalegaard requested changes Mar 19, 2026

View reviewed changes

klippy/extras/load_cell/ads131m0x.py Outdated Show resolved Hide resolved

klippy/extras/load_cell/ads131m0x.py Outdated Show resolved Hide resolved

klippy/extras/load_cell/load_cell.py Show resolved Hide resolved

src/sensor_ads131m0x.c Show resolved Hide resolved

garethky force-pushed the pr-sensor-ads131m02 branch from f5c63aa to 285b2f3 Compare March 19, 2026 20:24

garethky commented Mar 20, 2026

View reviewed changes

garethky force-pushed the pr-sensor-ads131m02 branch 3 times, most recently from 5780c9d to ed0b8c3 Compare March 20, 2026 23:22

dalegaard requested changes Mar 23, 2026

View reviewed changes

garethky force-pushed the pr-sensor-ads131m02 branch from ed0b8c3 to 6e10bcd Compare March 23, 2026 21:11

dalegaard approved these changes Mar 23, 2026

View reviewed changes

dalegaard merged commit 8f9b61f into KalicoCrew:main Mar 23, 2026
3 checks passed

Uh oh!

Conversation

garethky commented Feb 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Checklist

Uh oh!

dalegaard left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

dalegaard Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

garethky Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

dalegaard left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

garethky Mar 19, 2026

Choose a reason for hiding this comment

Uh oh!

dalegaard Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

garethky Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

garethky Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

dalegaard left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

dalegaard Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

garethky commented Feb 19, 2026 •

edited

Loading