gitit20

  • *
  • 55
  • If it will fly I will crash it.
LibrePilot Hardware
« on: January 02, 2016, 06:51:35 am »
Are there any plans for us to do our own hardware? If so what are you guys thinking?
Do the best you can It's all you can do.

Brian

  • *
  • 119
Re: LibrePilot Hardware
« Reply #1 on: January 02, 2016, 06:12:54 pm »
I've been toying for a while with a small Revo-nano-like board, but now there are more than one Revo-nano-like boards that seem to be pretty good, so I don't really see the need there anymore.  I'm not sure where the need is for another FC board at this point.

I've suggested having a "wish" forum where we list what we wished we could have in new hardware and hope the cloners pick it up and build it for us.  :)

hwh

  • *
  • 1018
Re: LibrePilot Hardware
« Reply #2 on: January 02, 2016, 06:36:38 pm »
I don't speak for the project but there are several reasons not to.

First, as Brian touched on, there's no compelling reason to produce hardware.  There would have to be something significant and not available elsewhere to justify building hardware.  Most of the suggestions I've seen have to do with things like adding more IO ports.

Second, it's a lot of work.  Not so much designing a board, but producing it.  Several people would pretty much have to dedicate themselves to building, selling, and shipping the product.

Third, outside of China it's very expensive to build boards.  I can buy whole assembled boards from China for less than I can buy the main chip on them in the US. I looked at building a V9 gps for myself and the BOM cost went over $25 without the gps module.  All told in the US it would probably be $45-50 in parts for one.  If you're building them to sell then quantity 100 is the first reasonable price break on the parts.  That means someone has to front $4500-5000 for parts plus the labor to build and ship the boards.

In short, I don't see it happening and hope that the project doesn't try to produce hardware.  It's better served by porting the software to other people's hardware like the Sparky2 port that's being worked on.

xfce

  • **
  • 91
Re: LibrePilot Hardware
« Reply #3 on: January 03, 2016, 07:44:46 am »

I don't speak for the project but there are several reasons not to.

First, as Brian touched on, there's no compelling reason to produce hardware.  There would have to be something significant and not available elsewhere to justify building hardware.  Most of the suggestions I've seen have to do with things like adding more IO ports.

Second, it's a lot of work.  Not so much designing a board, but producing it.  Several people would pretty much have to dedicate themselves to building, selling, and shipping the product.

Third, outside of China it's very expensive to build boards.  I can buy whole assembled boards from China for less than I can buy the main chip on them in the US. I looked at building a V9 gps for myself and the BOM cost went over $25 without the gps module.  All told in the US it would probably be $45-50 in parts for one.  If you're building them to sell then quantity 100 is the first reasonable price break on the parts.  That means someone has to front $4500-5000 for parts plus the labor to build and ship the boards.

In short, I don't see it happening and hope that the project doesn't try to produce hardware.  It's better served by porting the software to other people's hardware like the Sparky2 port that's being worked on.
I come from china. Maybe for design and mass production  can do something.

hwh

  • *
  • 1018
Re: LibrePilot Hardware
« Reply #4 on: January 03, 2016, 05:22:16 pm »
I come from china. Maybe for design and mass production  can do something.

@xfce  I knew that and thought of you when I wrote about China.

But there still remains the problem of what to build.  Some feature that's so different and better than other boards on the market that makes it worth the effort to build and integrate into the system.  And although we usually don't think of it, it's a lot of work integrating a new board into the system.  A new target with bootloader and firmware for the board has to be created.  Core system modules may have to be tweaked or conditional compilation added to them to account for differences between the new board and existing boards. Go through the source using the git blame command sometime, it shows not only who changed each line last but but also when.  The amount of work and period over which it was done is amazing.  And a large percentage of it going back many years was done by people who founded LibrePilot.  Another thing that I hadn't thought about earlier is that the GCS has to be changed for each board.  Screens to configure it's ports and other options, to update it's bootloader and firmware, and probably more that doesn't come to mind right now.  Adding a board is a large undertaking that really requires some compelling reason that the board is better than anything else available.

I've been unable to think of anything I want or have heard someone suggest that I can't just go on the web and buy a board that's close enough to what I want to work.

Re: LibrePilot Hardware
« Reply #5 on: January 04, 2016, 08:58:07 pm »
If it is bad, no one buys it.  :(

If it is good, it gets cloned and sold cheaper than it is worth selling for.  :(

Just no reason for us to design / make hardware when there are already good / cheap boards available.

My opinions...  :)

gitit20

  • *
  • 55
  • If it will fly I will crash it.
Re: LibrePilot Hardware
« Reply #6 on: January 05, 2016, 07:02:52 am »
What about continue to buy op boards like the nano and revo's? cant get them anymore seems like
Do the best you can It's all you can do.

Re: LibrePilot Hardware
« Reply #7 on: January 05, 2016, 09:51:15 am »
I guess buy them used, or get mini Revo clones, or buy the TL Sparky2 board which we will hopefully soon have released support for.  I have an official Sparky2 board that I run our current test version of Sparky2 code on.  Sparky2 is Revo sized with Revo functionality (minus PWM) plus extra ports and more modern sensors (the same ones that are in Nano).

I hear that Pixhawk PX4 support might become available sometime later also.  I wouldn't buy a PX4 without a little more certainty than that though.

cato

  • *
  • 341
Re: LibrePilot Hardware
« Reply #8 on: January 05, 2016, 10:18:50 am »
Meanwhile I have two Revo Clones and one Revo original running. Actually, although the clones are different in design, I can see no difference when I fly them. So, as long as the Revo is not available, you can stick to the clones. It would be good to keep the Revo Cloen thread updated about good and bad clones.

Otherwise personally I will buy a Sparky2 soon, simply because I need the better connectivity for using a Gimbal together with a GPS-enabled controller.
Nighthawk 250, MT1806, 12A ESC OneShot125, Revo, M8n GPS, FPV
Cinetank MKII, Elite 2216, 30A Afro OneShot125, Revo Clone, M8N GPS

carbo

  • *
  • 20
Re: LibrePilot Hardware
« Reply #9 on: January 05, 2016, 10:53:45 am »
Glad to hear about the Sparky 2 since I already have one.

The one piece of Openpilot hardware that I am missing is the GPS v9. No clones of that one so far.

Re: LibrePilot Hardware
« Reply #10 on: January 05, 2016, 08:51:09 pm »
We have several possible solutions for GPS/mag, in various stages:
 - aux mag support for pixhawk gps/mag combos, coded and flying in test versions, but it needs an extra port (Flexiport or Sparky2's extra I2CPort)
 - DJI Naza gps/mag support.  LP firmware to decode the protocol without changes to GPS/mag firmware.
 - DJI Naza gps/mag support.  Changes to GPS/mag firmware to make it act like an OP GPS V9 Platinum.

xfce

  • **
  • 91
Re: LibrePilot Hardware
« Reply #11 on: January 06, 2016, 05:28:21 am »
Glad to hear about the Sparky 2 since I already have one.

The one piece of Openpilot hardware that I am missing is the GPS v9. No clones of that one so far.

Here is the clone GPSV9 under test :)

hwh

  • *
  • 1018
Re: LibrePilot Hardware
« Reply #12 on: January 06, 2016, 07:09:52 am »
It looks good.  Since you said "under test" I take it that it accepted the stock bootloader/firmware and is running. Does a revo hooked up to it tell gcs that it has a v9 and it's mag sensor?

xfce

  • **
  • 91
Re: LibrePilot Hardware
« Reply #13 on: January 06, 2016, 08:29:33 am »
function is ok now.
1) GPS now is good, can got  7 or more stars and quickly 3D fixed. AS I use NEO-M8N :)
2) mag also could give the correct data, i fix the module 5cm up from my nano, quad is 250 size. but mag health still not all green.
I suspect maybe the active antenna's big shield metal is a problem(now antenna is soldered on the PCB by the shield metal),
I will take the antenna away to have a try.
and i am considering to use LNA and SAW on board and a passive antenna revmove the shield metal.

hwh

  • *
  • 1018
Re: LibrePilot Hardware
« Reply #14 on: January 06, 2016, 07:36:01 pm »
If the active antenna has a shield on it's bottom that's made of a ferrous metal it would probably badly affect the mag sensor with only the thickness of the pcb separating them.