Gumstix Raspberry Pi CM4 + Pixhawk FMUv6

@Keith_Lee I’ve had a quick look at the schematic. You’re powering the servo rail from an onboard regulator and have the barometer/s on i2c??! Why?
I haven’t looked at dma mapping, timer allocation, or the add-ons like ftdi yet. How locked down is the design?

As per https://github.com/pixhawk/Pixhawk-Standards/blob/master/DS-012%20Pixhawk%20Autopilot%20v6X%20Standard.pdf, pg7 the Bosch BMP388 pressure sensor is on I2C4.

The servo rail is actually powered by a current-limited distribution switch: TPS2041BDGN

I myself was not involved in part selection, but I’m sure this has something to do with the capacitive load of the servos and allows for over-current protection.

the distribution switch is going to be problematic: the servo’s shouldn’t be pulling board voltage at all, and there’s going to be strange behaviour when it cuts in and out. It might be fine for electronic speed controllers, but it won’t be for actual servos on the rail. I’m somewhat confused as to why it’s there. I assume the enable pin can be switched?
As for the baro: there really should be some justification for using I2C. It’s not a high speed sensor, but there are some nuances with those things and power on I2C: if it glitches on the bus it might not be able to be restarted.

I am unaware of any condition wherein a power glitch would occur that would prevent the sensor, or the bus, from restarting, but I’m very curious. What would the steps to reproduce this be?

This is really more of a concern with the v6 standard than with the CM4 + FMUv6 PCB, as all of the elements you describe above are documented in the spec.

What’s the status here? From Pixhawk FMUv6 it seems like the v6 standard itself is still a work-in-progress. Is the Gumstix board usable with ardupilot in its current state?

No, there’s been no work done to bring up ArduPilot on this board.

Curious if there have been any developments for getting ArduPilot for this board? Seems if not so far, there must be some deeper issues and no point looking forward for this board any time soon.

I’m waiting on an update re shipping. There should be boards on their way to @peterbarker and @ppoirier

Thanks @james_pattison , so it hasn’t completely grinded to a halt. Good to know!

Hello @james_pattison,

Could you please tell us if there has been any movement with this? It is a very interesting board and it would be great to know if it is an option.

Thanks,

Dylan

Gumstix sent some out.
First pair:

Second pair:

They’re marketing/selling the device but haven’t addressed the heating issue. Patrick wasn’t comfortable continuing the port, and I didn’t push the issue.

1 Like

Thanks for the response. :slight_smile:

Wow, this is really disappointing! I think it is really promising to have this level of integration with a RPi CM4. Pity the Gumstix team is not more committed to this project (or at least that is how it seems).

Yeah: I’m not sure they accept that it will cause problems.
We’ll likely get around to finishing off the hwdef for the autopilot component, but it’s it’s low priority given we don’t have confidence in the complete unit.

This makes me a little sad. Back in the day they used to make some great SWAP optimized single-board computers with ethernet and wifi.

if any one try this board can share your experience with it and what is the problem with it ?

Thanks for the feedback everyone. We are waiting for some more Coral units before we can build the next version which includes a DIP switch that will disable the PCIe->USB3 chip and the Coral module. That will eliminate the heating issue for people not interested in the Coral. For others, adding a heat sink to the PCIe->USB3 chip worked for me and kept the board at a reasonable temperature.

I’d also like to remind everyone of our current promotion (until October 15). You don’t pay the manufacturing costs ($1999 value) on Upverter boards using the FMUv6U module. You pay only for the cost of the modules placed on the board.

Go to Upverter and design a better CM4 + FMUv6U board. The current board isn’t meant to suit every need, I put as much stuff on it that I could and made sure it met the needs of my projects. Now is the time to make a custom board for yourself. The PX4 and ArduPilot software support is for the FMUv6U module so it should cover all custom designs.

We have stock of the sensors and components necessary to build the FMUv6U. I encourage everyone to give Upverter a try and then feel the pride and accomplishment of using your own custom flight control board. You can check out my talk giving some tips on how to quickly throw together a design.

Thanks again for your support.

A recent post on this blog discussed the release of a new Pixhawk, but there was no indication when that might release. Moreover, I would like to know more about the FMUv6 version 6s. Do you know when it will be released?

This is off topic. If you have follow on questions, open a new thread.
“fmu v6” is basically just a number denoting a flight controller with an STM32H7 microprocessor. There are 4 or 5 variants from different manufacturers, some are already released - fmu v6 isn’t really a standard, more just a tool for marketing agreements. Cube Orange, mRo ControlZero-H7, CUAV X7, Holybro Durandal, Qiotek Zealot H743 (all of which have been released to the market) are equivalent.

1 Like

Much needed comment. So many are getting confused recently!

Hi @smithandrewc , this is Varun from Bangalore India. hope you are doing well, I wanted to understand the first time flashing of the board [The Pixhawk Side for FMUv6U, cuz I heard there are some issues with it] if I manage to custom design a board based on yours, could you please help me with that information so that I can make a decision ? Thanks in advance!