3D Robotics

AttoPilot review, Part 1: The PCB board

I just got an AttoPilot in and plan to review it in several parts. Today's part: the PCB board. As you might expect from a former Intel engineer, Dean Goedde has done a great job of PCB design. Neat trace layouts, good noise protection and robust components. The board is based on a 5Mhx Parallax Propeller chip, with 8 cores. It has a differential pressure sensor for speed and a single pressure sensor for altitude. It also has 512k of on-board memory in addition to that in the Propeller chip. Click on the photo and you'll be taken to Flickr, where you can mouse over the components to find out what they are AttoPilot board top And here's the bottom of the board. Click on this one, too, to find out what the components are: AttoPilot board bottom
E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • If I had to choose a detail to ignore, and my choices were 1. Linearity at high speeds, and 2. Ability to substitute broken tube connector for its opposite, I would have made a different choice; but no one will argue with a working device - so feel more than free to ignore my preferences. I'm just saying I think the post above which recommended the other sensor may have a valid point - you on the other hand have a valid autopilot. ;-)
  • Let me explain further my rationale here:
    1) Most sUAS that I have tested don't need to go much above 110 km/h, so 200 km/h is more than a little buffer
    2) My autopilots auto-scale the PI and D gains in the 50Hz attitude control loop based on airspeed, so I can't just stop estimating body-forward airspeed at some limit like 200 km/h. Thus, I need a method to extend the airspeed estimation above the diff P sensor
    3) Unless you are flying in a hurricane or near a tornado, the wind speed will be << 200 km/h. Therefore, if the sUAS is flying with an airspeed > 200 km/h, we can more or less ignore that there is wind; i.e. ground track speed corrected for pitch of the aircraft is ~ airspeed.

    That's my rationale. It distinctly draw a practical line in the sand and doesn't get too caught up in worrying about 2nd and 3rd order effects. Sure, armchair engineers can be technically correct and state "but there is actually wind, and it affects the airspeed even if the sUAS is moving fast, so using GPS speed is not perfect". Thinking too hard about the details isn't the problem, the problem is not being able to differentiate the dominate effects from the trivial, and getting paralyzed on details that don't matter above the noise floor.
  • The perceived problem is wind. The other problem is update speed and reliability. Finally, the exception case of straight down creates a math problem for cosine. First, the error will equal the wind speed and direction if you substitute Absolute Coordinates. I would think the Hypotenuse would be a better math choice than cosine. Finally, the GPS can "jump" and needs to be taken with a grain of salt - basing speed on GPS is just plain problematic. It seems the problem is recognized by including the extra code, but I submit that the part replacement is cleaner, provides a linear response curve through the operation envelope. Inconsistencies which are rarely tested would seem like the first place to look if a plane crashed. Why not do the looking ahead of time... didn't Airbus recently recall its speed sensor system. If it were me, I'd transition to the other part with a version change, and mark these models as sub 200km/h rated.
  • "On the subject of Airspeed and the 200km/h limit - It should be noted that GPS - aka groundspeed - and airspeed are not the same thing"

    Uhhh, yeah - that's why I described taking ground speed and converting to body-forward speed by dividing the ground speed by Cos(pitch). It works out. What is the percieved problem?

    200 km/h is about 125 mph. Pretty fast and above most sUAS unless you are talking about a funjet at full throttle.
  • Dean,
    On the subject of Airspeed and the 200km/h limit - It should be noted that GPS - aka groundspeed - and airspeed are not the same thing, and that switching over to groundspeed at 200km/h may or may not be a suitable solution - cosine or otherwise. I'm not sure how fast 200km actually Is, but I know guys that do 100mph so I'm thinking 200km is not so high as to be out of range.
    The proposal to switch components is on balance rather reasonable, low-design impact, and perhaps worth a second consideration.
    All the best to the ottopilot, and I can probably speak for most of the other armchair enthusiasts to say- we wish you success for your hard work - and hats off to a personal accomplishment - struggles and all.
  • There are two products here:
    One is a PC board stuffed with off-the-shelf components, and the other is some software which uses those inputs to guide a UAV. I'm pretty sure Dean has looked at his share of PC boards as part of his own learning process, and I'd be surprised if he hasn't learned a great deal about the software from others contributions here on DIYdrones, finally, I imagine that a large number of his buyers will have first learned of his project on DIYdrones. Taken together, I would think uav entrepreneurs would want the attention of the enthusiasts, and long-suffer their curiosities.

    Dean, You might consider opening up your board layout for people to play with - you might sell more at $300 without software than at $800 with.

    Best of luck everyone on this - Dean - here's to breaking even - Chris here's to a rockin review.
  • > I will be pretty pissy and sensitive until the day I break even.

    Another Dave Perry is born. Now some reading material 4 U entrepreneurs while U wait to finish your day jobs & get back to work:

    Tips for Startup Companies



    best way to escape 9-5 wage slavery


    Best quote: "Without using other folks' money, it is hard to do much of anything."
    Tips for Startup Companies
  • I was being immature stating it that way and trying to get a jab in. Yes, I was pretty upset at the time that Atto pics and an attempted dissection were all over DIY Drones.... The hardware is but a tiny piece of why and how my autopilot works (95% of it is hard-won code based on endless testing), but there are many many hard-won little victories even in component and layout choices and why should I just hand them over to other people? I say "let them spend $45k on components and equipment in 2008 and quit their lucrative day job in order to really focus on this.. and go into debt like I did." I will be pretty pissy and sensitive until the day I break even.

    Regarding the MPXV7002-DP sensor for airspeed... if I chose the 1-sided sensor like Jordi suggested, then how would people be able to instal BOTH the static and dynamic ports? The 1-sided sensor has a vent hole I believe on the bottom of the package... how is that usable? Besides, pressure vs Airspeed curve with my +/- 2k range gets up to at least 200 km/h... who out there with sUAS is going to fly above 200km/h? and anyway, I already thought of that and my code switches to GPS speed (with cosine(pitch) correction) above 200km/hr, so I am not limited in Atto.
  • T3
    "Jordi - I use the 2-sided MPXV7002 because in AttoPilot code, I use the absolute value of the differential pressure, so that if for some weird reason a person voided their warranty by taking the Atto out of the case and broke off one of the 2 pressure ports, they coulde just plug the Pitot tube into the other port. "
    Dean, decide yourself, are you HAPPY to see your customers doing reverse engineering or not.
  • T3
    The reveng is making possible to calculate Dean's profitability margin and to deal with him harder when going large series. This is undertandable he is angry because 80-90% of the price of every single electronics you use are software engineers. The PCB and components are nothing as they are, yet the negotiator ordering 100pcs will exploit this thread to the max, exactly going into Dean's income for the past 2 years of his work.
    All fluids and rare metals you need to make a human are under 1USD. Yet, you cannot buy it even for 1e6USD.
This reply was deleted.