fmu-v5: Add support for ICM-42688-P #19306

mxiaogit · 2022-03-10T08:28:37Z

No description provided.

davids5

The select pin PF11 is SPI4_CS2 and a no connect on the other brands of V5. This should work fine. if it does not cause backfeeding. If it does cause a problem. Then the Sensor Set method used on V5X will be needed. Please verify with a scope that the CS pin is not backfeeding the ICM44688p while the VDD_3V3_SENSORS is off.

mxiaogit · 2022-03-11T01:13:43Z

@davids5
I observed that fmu-v5 does not have VDD_3V3_SENSORS_EN pin, and VDD_3V3_SENSORS is turned on by default on CUAV V5+ IMU, so there is no situation that VDD_3V3_SENSORS is turned off.

dagar · 2022-03-11T02:49:24Z

@davids5
I observed that fmu-v5 does not have VDD_3V3_SENSORS_EN pin, and VDD_3V3_SENSORS is turned on by default on CUAV V5+ IMU, so there is no situation that VDD_3V3_SENSORS is turned off.

It's PE3 here. https://github.com/PX4/PX4-Autopilot/blob/06181fedb9c9d3d33424651d31e25bf91a44063d/boards/px4/fmu-v5/src/spi.cpp#L45

mxiaogit · 2022-03-11T08:52:01Z

The select pin PF11 is SPI4_CS2 and a no connect on the other brands of V5. This should work fine. if it does not cause backfeeding. If it does cause a problem. Then the Sensor Set method used on V5X will be needed. Please verify with a scope that the CS pin is not backfeeding the ICM44688p while the VDD_3V3_SENSORS is off.

It does cause backfeeding.

davids5 · 2022-03-11T22:31:15Z

The select pin PF11 is SPI4_CS2 and a no connect on the other brands of V5. This should work fine. if it does not cause backfeeding. If it does cause a problem. Then the Sensor Set method used on V5X will be needed. Please verify with a scope that the CS pin is not backfeeding the ICM44688p while the VDD_3V3_SENSORS is off.

It does cause backfeeding.

That is a problem. The CS needs to be low while the power is off or it can latch up the sensor.

mxiaogit · 2022-03-12T13:46:04Z

The select pin PF11 is SPI4_CS2 and a no connect on the other brands of V5. This should work fine. if it does not cause backfeeding. If it does cause a problem. Then the Sensor Set method used on V5X will be needed. Please verify with a scope that the CS pin is not backfeeding the ICM44688p while the VDD_3V3_SENSORS is off.

It does cause backfeeding.

That is a problem. The CS needs to be low while the power is off or it can latch up the sensor.

How do we do it?

davids5 · 2022-03-14T13:41:11Z

boards/px4/fmu-v5/src/manifest.c

@@ -160,6 +160,7 @@ static px4_hw_mft_list_entry_t mft_lists[] = {
 	{0x0000, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},
 	{0x0105, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},  // Alias for CUAV V5 R:5 V:1
 	{0x0500, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},  // Alias for CUAV V5+ R:0 V:5
+	{0x0502, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},  // Alias for CUAV V5+ R:2 V:5


Suggested change

{0x0502, hw_mft_list_v0500, arraySize(hw_mft_list_v0500)}, // Alias for CUAV V5+ R:2 V:5

{0x0552, hw_mft_list_v0500, arraySize(hw_mft_list_v0500)}, // Alias for CUAV V5+ R:2 V:5 ICM42688P

{0x0562, hw_mft_list_v0500, arraySize(hw_mft_list_v0500)}, // Alias for CUAV V5+ Nano R:2 V:6 ICM42688P

davids5 · 2022-03-14T13:45:40Z

boards/px4/fmu-v5/init/rc.board_sensors

@@ -5,8 +5,14 @@

 board_adc start

-# Internal SPI bus ICM-20602
-icm20602 -s -R 2 -q start
+if ver hwtypecmp V552


if ver hwtypecmp V552 V562

This will cover the

2 5 CUAV V5+ ICM42688P

2 6 CUAV V5Nano ICM42688P

davids5 · 2022-03-14T13:49:17Z

boards/px4/fmu-v5/src/spi.cpp

-		initSPIConfigExternal(SPI::CS{GPIO::PortI, GPIO::Pin7}),
-		initSPIConfigExternal(SPI::CS{GPIO::PortI, GPIO::Pin8})
+
+	initSPIHWVersion(HW_VER_REV(0, 2), {


Suggested change

initSPIHWVersion(HW_VER_REV(0, 2), {

initSPIHWVersion(HW_VER_REV(5, 2), {

A duplicate is needed for the Nano with InitSPIHWVersion(HW_VER_REV(6, 2), {

While Appling power to the unit, please capture the PE3, VDD_3V3_SENSORS, and PC5, PB4 on a scope and post the images here.

The channels are PE3, VDD_3V3_SENSORS, PC5, PB4 in sequence. Details are here:DSLogic-la-220315-100851.csv

hmm this looks incorrect to me. Even the other CS line. What is the width if the PE3 off time? ---_---

PE3(VDD_3V3_SENSORS_EN) is always on after initialization is complete.

Our last version of the hardware is similar, when the chip is still ICM-20602.

@davids5 What do you think there is a problem?

davids5 · 2022-03-21T13:46:41Z

@mxiaogit The signals do not look correct to me. I need to review the power up sequence in the code.

Power should be off and all the node (CS, MOSI, MISO etc) to the sensors should be low until after power is applied.

davids5 · 2022-03-22T17:59:28Z

@mxiaogit - I found what I think is the root cause. It is probably systemic and came in when the SPI/I2C work was done by @bkueng. The initial condition out of reset for the sensors power needs be off, this was done in the stm32_boardinitialize and we get a built-in delay to when board_app_initialize is called.

I do not have current V5+ HW and It would take quite a bit to instrument it. Since you already have the HW set up can you please pull in e367997 and re-test and post the same signals timing here.

davids5 · 2022-03-23T17:36:28Z

@mxiaogit - I found what I think is the root cause. It is probably systemic and came in when the SPI/I2C work was done by @bkueng. The initial condition out of reset for the sensors power needs be off, this was done in the stm32_boardinitialize and we get a built-in delay to when board_app_initialize is called.

I do not have current V5+ HW and It would take quite a bit to instrument it. Since you already have the HW set up can you please pull in e367997 and re-test and post the same signals timing here.

Never mind I missed the call to board_control_spi_sensors_power_configgpio();. I need to step through and make sure it is really setting it to off.

davids5

The code change is fine.

The change is no worse then the existing code.

mxiaogit · 2022-03-25T10:15:36Z

@mxiaogit - I found what I think is the root cause. It is probably systemic and came in when the SPI/I2C work was done by @bkueng. The initial condition out of reset for the sensors power needs be off, this was done in the stm32_boardinitialize and we get a built-in delay to when board_app_initialize is called.

I do not have current V5+ HW and It would take quite a bit to instrument it. Since you already have the HW set up can you please pull in e367997 and re-test and post the same signals timing here.

@davids5 I found some problems with our hardware and after fixing the problem, VDD_3V3_SENSORS is controllable. Here are the waveforms tested by e367997, in order of PE3(VDD_3V3_SENSORS_EN), VDD_3V3_SENSORS(IMU1_3V3), VDD_3V3_SENSORS(IMU2_3V3), PC5(SPI1_DRDY4_ICM20602), PB4(SPI1_ DRDY1_ICM20689).

There exist two cases, VDD_3V3_SENSORS closing time has two cases, about 4 seconds or about 8 seconds.

davids5 · 2022-03-25T13:45:18Z

@mxiaogit - That looks more like I would expect.

Do you check the timing on this version of the code with the fixed HW?

What is the time from VDD_3V3_SENSORS_EN to VDD_3V3_SENSORS(IMU1_3V3) valid? Then from VDD_3V3_SENSORS(IMU1_3V3) to the first chip select or datardy going High (Zoom in a bit)

davids5

There needs to be a manifest entry for each rev/ver pair.

davids5 · 2022-03-25T13:47:41Z

boards/px4/fmu-v5/src/manifest.c

@@ -160,6 +160,8 @@ static px4_hw_mft_list_entry_t mft_lists[] = {
 	{0x0000, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},
 	{0x0105, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},  // Alias for CUAV V5 R:5 V:1
 	{0x0500, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},  // Alias for CUAV V5+ R:0 V:5
+	{0x0502, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},  // Alias for CUAV V5+ R:2 V:5 ICM42688P
+	{0x0602, hw_mft_list_v0500,        arraySize(hw_mft_list_v0500)},  // Alias for CUAV V5nano R:2 V:6 ICM42688P


What about the entries for V552 and V562?

According to the macro definition: define HW_VER_REV(v,r) ((uint32_t)((v) & 0xff) << 8) | ((uint32_t)(r) & 0xff), 0x0502 is V552, 0x0602 is V562.

right! Would you mind adding and using the format from

https://github.com/PX4/PX4-Autopilot/blob/master/boards/px4/fmu-v5x/src/board_config.h#L182-L199

So this will no be confused again.

I have modified. Please review.

davids5

Let's Bring this in and can make the #define change later

mxiaogit · 2022-03-29T03:11:43Z

@mxiaogit - That looks more like I would expect.

Do you check the timing on this version of the code with the fixed HW?

What is the time from VDD_3V3_SENSORS_EN to VDD_3V3_SENSORS(IMU1_3V3) valid? Then from VDD_3V3_SENSORS(IMU1_3V3) to the first chip select or datardy going High (Zoom in a bit)

@davids5
The following are the waveforms tested, followed by PE3(VDD_3V3_SENSORS_EN), VDD_3V3_SENSORS(IMU1_3V3), VDD_3V3_SENSORS(IMU2_3V3), PC5(SPI1_DRDY4_ICM20602), PB4(SPI1_DRDY1_ICM20689). There are two cases, VDD_3V3_SENSORS has two kinds of off time, about 4 seconds or about 8 seconds. Added zoom-in view.

Group 1：

Group 2：

davids5

@mxiaogit - thank you

davids5 · 2022-03-29T20:12:58Z

@dagar - based on the voltage supply rise time and that this is digital interpretation: The digital value of the VDD_3V3_SENSORS is indicating VIH threshold (of the logic analyzer) not stable 3V3 VDD. Note in image #5 that the CS is leading VDD_3V3_SENSORS the voltage change. Therefore we may need to increase the delay from activating VDD_3V3_SENSORS_EN to initializing the CS,SPI, RDY lines.

dagar previously approved these changes Mar 10, 2022

View reviewed changes

dagar requested a review from davids5 March 10, 2022 15:38

davids5 previously approved these changes Mar 10, 2022

View reviewed changes

mxiaogit closed this Mar 11, 2022

mxiaogit reopened this Mar 11, 2022

mxiaogit dismissed stale reviews from davids5 and dagar via 04454f0 March 11, 2022 08:54

mxiaogit force-pushed the cuav_v5_icm42688p branch from 06181fe to 04454f0 Compare March 11, 2022 08:54

mxiaogit requested review from dagar and davids5 March 11, 2022 08:56

davids5 requested changes Mar 14, 2022

View reviewed changes

mxiaogit force-pushed the cuav_v5_icm42688p branch from 04454f0 to 14af1aa Compare March 15, 2022 03:16

mxiaogit requested a review from davids5 March 15, 2022 03:55

davids5 previously approved these changes Mar 23, 2022

View reviewed changes

mxiaogit added 2 commits March 24, 2022 18:12

fmu-v5: Add support for ICM-42688-P

fda5e9a

spi: get the correct version revision

23ce05a

mxiaogit dismissed davids5’s stale review via 23ce05a March 24, 2022 10:13

mxiaogit force-pushed the cuav_v5_icm42688p branch from 14af1aa to 23ce05a Compare March 24, 2022 10:13

mxiaogit requested a review from davids5 March 25, 2022 10:15

davids5 requested changes Mar 25, 2022

View reviewed changes

davids5 previously approved these changes Mar 28, 2022

View reviewed changes

fmu-v5: add macro definitions for different version revisions

02bdd53

mxiaogit dismissed davids5’s stale review via 02bdd53 March 29, 2022 05:56

mxiaogit requested a review from davids5 March 29, 2022 05:58

davids5 approved these changes Mar 29, 2022

View reviewed changes

dagar merged commit 1870b92 into PX4:master Mar 29, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fmu-v5: Add support for ICM-42688-P #19306

fmu-v5: Add support for ICM-42688-P #19306

mxiaogit commented Mar 10, 2022

davids5 left a comment

mxiaogit commented Mar 11, 2022

dagar commented Mar 11, 2022

mxiaogit commented Mar 11, 2022

davids5 commented Mar 11, 2022

mxiaogit commented Mar 12, 2022

davids5 Mar 14, 2022

davids5 Mar 14, 2022

davids5 Mar 14, 2022

davids5 Mar 14, 2022

mxiaogit Mar 15, 2022 •

edited

Loading

davids5 Mar 16, 2022

mxiaogit Mar 17, 2022

mxiaogit Mar 17, 2022

mxiaogit Mar 21, 2022

davids5 commented Mar 21, 2022

davids5 commented Mar 22, 2022

davids5 commented Mar 23, 2022

davids5 left a comment

mxiaogit commented Mar 25, 2022 •

edited

Loading

davids5 commented Mar 25, 2022

davids5 left a comment

davids5 Mar 25, 2022

mxiaogit Mar 28, 2022

davids5 Mar 28, 2022

mxiaogit Mar 29, 2022

davids5 left a comment

mxiaogit commented Mar 29, 2022 •

edited

Loading

davids5 left a comment

davids5 commented Mar 29, 2022

	{0x0502, hw_mft_list_v0500, arraySize(hw_mft_list_v0500)}, // Alias for CUAV V5+ R:2 V:5
	{0x0552, hw_mft_list_v0500, arraySize(hw_mft_list_v0500)}, // Alias for CUAV V5+ R:2 V:5 ICM42688P
	{0x0562, hw_mft_list_v0500, arraySize(hw_mft_list_v0500)}, // Alias for CUAV V5+ Nano R:2 V:6 ICM42688P

	initSPIHWVersion(HW_VER_REV(0, 2), {
	initSPIHWVersion(HW_VER_REV(5, 2), {

fmu-v5: Add support for ICM-42688-P #19306

fmu-v5: Add support for ICM-42688-P #19306

Conversation

mxiaogit commented Mar 10, 2022

davids5 left a comment

Choose a reason for hiding this comment

mxiaogit commented Mar 11, 2022

dagar commented Mar 11, 2022

mxiaogit commented Mar 11, 2022

davids5 commented Mar 11, 2022

mxiaogit commented Mar 12, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mxiaogit Mar 15, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

davids5 commented Mar 21, 2022

davids5 commented Mar 22, 2022

davids5 commented Mar 23, 2022

davids5 left a comment

Choose a reason for hiding this comment

mxiaogit commented Mar 25, 2022 • edited Loading

davids5 commented Mar 25, 2022

davids5 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

davids5 left a comment

Choose a reason for hiding this comment

mxiaogit commented Mar 29, 2022 • edited Loading

davids5 left a comment

Choose a reason for hiding this comment

davids5 commented Mar 29, 2022

mxiaogit Mar 15, 2022 •

edited

Loading

mxiaogit commented Mar 25, 2022 •

edited

Loading

mxiaogit commented Mar 29, 2022 •

edited

Loading