Servers by jDrones

Funding for UAVCAN DIY UC4H ESC nodes [Closed]


(mike kelly) #1

Topic: Buy prototype UAVCAN UC4H ESC nodes for Tom Pittinger Mark Whitehorn, Francisco Ferreira and Leonard Hall.

Proposal type: Hardware [X ] , Software [ ] , Other [ ] : _________________

Description: UAVCAN ESC nodes for Kiss 32 ESC’s to explore integration of low cost DIY UAVCAN devices into the Ardupilot environment.
UAVCAN brings simpler wiring, easy monitoring of hardware status, single cable type with daisy chained devices and eliminates ringing out cables to get RX/TX etc correct. A full UAVCAN aircraft can have only three connections to the flight controller - CANbus, Power and RCIN for a completely functional aircraft. A simpler system to construct for beginners and a more reliable and easier to maintain system for professionals. A set of 4 ESC’s and an SLCAN “network sniffer” will give a developer a significant hands on introduction to UAVCAN and the benefits to the Ardupilot ecosystem.

A group buy has been organized to purchase a minimum order to pay for solder masks etc with ShenZhen2U a circuit board and assembly manufacturer used previously by the BBBmini project.

A functioning system for developers would consist of 4 each UC4H ESC carrier/nodes in which 4 Kiss 32 ESC’s are attached.
https://www.rcgroups.com/forums/showpost.php?p=39225194&postcount=526

The ESC node/carrier boards will be $52+shipping per set of 4 and the Kiss 32 ESC’s run around $29 each for about $120 + shipping per quad aircraft. Hence say $175+$35 shipping depending on the destination(s).

Also an SLCAN adapter is needed to troubleshoot and inspect the CANbus.
It can be made DIY for $15 in parts
https://www.rcgroups.com/forums/showpost.php?p=38060727&postcount=12
or purchased from Zubax, the Babel for $59
https://shop.titaneliteinc.com/index.php?route=product/product&product_id=1004.

4 Canbus JST-GH 10cm cables $20
https://store.mrobotics.io/product-p/mrc0220.htm

Planned amount $$ (USD): $250-$320 Per set

Estimated time for completion:1 month for build and delivery from Asia and three weeks to
test the boards before reshipment to developers.


Dev Call Feb 26, 2018 2300 UTC
(proficnc) #2

I think this is great!!!


(Olivier Brousse) #3

Thanks for the proposal Mike.
Currently discussing, we’ll update soon.


(Fnoop) #4

It’s a shame this isn’t open source, but it’s still a brilliant community effort to bring on uavcan, hope this gets funding :slight_smile:
The proficnc CAN stuff in the unconference hardware talk is super interesting, good times ahead!


(Francisco Ferreira) #5

So is this funding for some UAVCAN hardware to some devs to do what? Just getting familiar with UAVCAN? If it is just that, I’m not against it, but I’m sure there are more devs that would like to have it, I certainly would.


(James Pattison) #6

Let’s add Francisco!


(Khancyr) #7

About the proposal, it is really insteresting, but I think it need a little more details !
How much kit will be funded (the funding commitee can only vote for 2000$ max ) ?
And as Francisco underline : what the goal ? just get some hardware ? or get some hardware to make demo, uavcan wiki page (that would get my vote)


(mike kelly) #8

The total cost depends on the shipping destinations but the total for four developers would be $1000 to $1280 depending on making or buying a finished SLCAN adapter.

I see CANbus as a very significant improvement for the Ardupilot ecosystem but there are still decisions to be made on the philosophy of parts of the implementation. With any of the projects within Ardupilot it takes an introduction, interest and a certain familiarity to get support and then make progress.

The purpose of this proposal is to allow developers to gain first hand experience with UAVCAN to allow them to understand the issues and see the benefits.

Obviously gaining a wiki etc are important outcomes but I think we need to get over some hurtles first. At that point I would be happy to write/edit a wiki on CANbus for the Ardupilot ecosystem, with some review by subject matter experts.


(ACPUK) #9

Is this to develop on Ardupilots UAVCan support, or develop the ESC side?


(James Pattison) #10

There are many benefits, but specifically for this I think ArduPilot’s support for and use of esc feedback should be a key goal.
Olli has already written some code to consume the esc.status message (https://github.com/ArduPilot/ardupilot/pull/7790), next step is to use it rather than just log it.


(mike kelly) #11

Everyone may have a different idea but my reason for the proposal was to develop/refine Ardupilot UAVCAN support.

But, of course, having initial hardware offerings are important to gaining broad industry support and yielding low cost, mass-produced peripherals that enhance the ecosystem by having something for people to start with/learn about new technologies.

OlliW does not make commerical products for profit but his designs are available for commercial hardware vendors to use in their products. This does help the ecosystem. So in some regards the answer is yes to both your options. The proposal is to help enhance Ardupilot and get CANbus features codified and the hardware in this proposal can aid that effort but it can also provide a launch platform and sample implementation for commercial hardware vendors.


(ACPUK) #12

Ok I would also like involved. I can dedicate some considerable developer’s time to this as I bereave this will be beneficial and improve Ardupilot support. I would also like a hardware ‘kit’ to work with and feed back development. I will pay for my portion of the costs.

Can we get a complete list of all developers that would benefit from hands on to this hardware. And which of them are willing to contribute. Finalize a total quantity and let me know what to total cost will be and what funding has been raised.

Once we have an outstanding balance, I will see what I can do about it financially.


(mike kelly) #13

https://www.rcgroups.com/forums/showpost.php?p=39251655&postcount=542

I think it is germane to copy a note from OlliW on RCgroups regarding his latest work on the UC4H ESC node:

"hey folks,

I’d like to share with you my today’s evening entertainment .

Exploring the new possibilities offered by the UC4H ESC KISS node, i.e., the advantage of allowing us to measure the rpm and current (and some other quantities), I thought that this should allow us to also estimate the thrust characteristics of the ESC-motor-prop combination. This should be highly useful, since the thrust characteristics is among the most important properties determining the behavior of a copter, but is not trivially measured. Usually relatively sophisticated thrust measurement stands are needed.

In fact, since that characteristic can be so important, ArduPilot has some parameters specifically for adjusting the motor output to the specific thrust characteristic. It’s called MOT_THST_EXPO. You can find some very insightful explanations and comments by Leonard here: https://discuss.ardupilot.org/t/usin...e-expect/26172.

Problem is, as mentioned: Normal mortals like us have little chances to measure it. So we stick with the default value. This may change fundamentally, thanks to UAVCAN, and the substantially lowered entry price.

So, with a crucial hint by Pavel (thx, sir!!), I wrote a python script to ramp up and down the motor+prop, measure the rpm and current, estimate the thrust using some basic blade momentum and motor theory results, and fit the result to a curve specified by ArduPilot, to yield an estimate for MOT_THST_EXPO. I have to admit that the numerical value doesn’t yet make sense to me, so, this needs to be further investigated. BUT: That’s the key message, it’s now all doable, and easily doable, for anyone of us "


(ACPUK) #14

Superb, we are also just starting to build a ESC, motor, thrust test system. Thrust, Watts, vibration with full logging. We can compare results from the estimates with real measurements.


(Coby Leuschke) #15

It seems exposing as many core devs to UAVCAN is a good step to wider UAVCAN support. And perhaps not just for ESC’s in the future…

Hacker has released the Ditex line of servos with telemetry. Sadly, closed to Jeti so far. Hitec also has a new line of digital servos with telemetry but they also appear to be sticking to a closed protocol. It would be awesome to see an open alternative using UAVCAN for servos. Closed loop for simple gimbals and tilt rotor servos comes to mind. Assuming continuous rotation could also be handy for rovers.


(mike kelly) #16

I believe OlliW is interested in adapting UAVCAN via UC4H nodes for servos. Check his UAVCAN thread in RCGroups.

https://www.rcgroups.com/forums/showthread.php?2869828-Uavcan-for-Hobbyists&highlight=UAVCAN


(ACPUK) #17

Thought this may be a reasonable place to ask as this is for ESC development. For our test rig I need a reliable PSU. has anybody had any experience with EA Elektro-Automatik 6230263 Digital Bench Power Supply:?

https://uk.rs-online.com/web/product/products/8069717/

Thanks


(Khancyr) #18

I am not against this funding nor UAVCAN experiment.
But ArduPilot isn’t rich unfortunatly, so I don’t want to spend much money on project without direct counter part for everybody. The project money isn’t dev money, but communauty money. So I cannot spend money on something without direct counter part for the communauty, other we can be accused to only use the money for ourself, called devs.

So I would expect, some wiki doc, and some blog post on the funded hardware in 1 to 6 month ! That is my condition, that is not big …
That is my personal opinion on this proposal, it engage only myself. But I prefere to be clear from start.


(James Pattison) #19

@khancyr FYI http://www.olliw.eu/2017/uavcan-for-hobbyists/
There’s already a rather comprehensive blog :wink:
But yes, if we pull support in upstream, the wiki needs attention. I think that’s a fair stipulation.


(Coby Leuschke) #20

Thought I remembered something similar.

https://discuss.ardupilot.org/t/next-gen-esc-validation-and-integration-vesc-declined/12534