mike_kelly's blog View Details
Posted by mike_kelly | Jan 30, 2017 @ 12:10 PM | 9,351 Views
I am constantly surprised at how hard it is to interface things together in multirotors. An example is putting together a Gimbal where the manufacturer seems to have no interest or plan for attaching the gimbal to a multirotor. Many have no method to attach other than bolting it in place to the bottom of the multirotor. A gimbal often needs to be able to be moved back and forth to balance the camera/gimbal on the frame. Tarot has their two carbon tube rails system but most gimbals have no way to attach to it.

No holes to mount the Tarot hangers:

One of the international aspects of the hobby is the creative ways people find to solve such problems and their willingness to share what they have done. One such person is Alan from the United Kingdom who designed an adapter to adapt the DYS NEX gimbal to a Tarot two tube rail system.

...Continue Reading
Posted by mike_kelly | Jan 20, 2017 @ 06:02 PM | 17,672 Views
Gimbals 101

A gimbal is a device used to stabilize a camera in one or more axis. When we fly an aircraft, it is subject to a lot of movement. Winds, thermals, turbulence and the vibration of props and motors can cause video and still images to blur. A gimbal senses the movement of the aircraft and moves the camera, via a brushless motor or servo, to keep it level even though the aircraft has moved. As the aircraft moves, the gimbal keeps the camera level in one or more axis. When the aircraft moves back to level, the gimbal returns the camera so it remains level in the new position of the aircraft. It does this so fast that the camera appears to sit motionless. Brushless gimbals are much more popular and costly than servo gimbals. The brushless motor can move much faster to keep the camera level than can a servo. But a servo gimbal might be just the ticket for still photography because it is lighter.

For the DIY gimbal builder the first requirement, not unlike a flight controller, is to get the camera perfectly balanced on the gimbal. If you roll, tilt or pan the camera it will stay where you leave it if it is balanced. If the gimbal motors have to assist in keeping the camera balanced they may run out of power to keep the camera level when the aircraft is bouncing around. That is why you need to choose a gimbal that has balance adjustments in all axes. You will need to slide the gimbal arms in and out to balance the weight of the camera. Also, gimbal designs...Continue Reading
Posted by mike_kelly | Sep 11, 2016 @ 10:47 AM | 13,570 Views

In the search for the lightest-highest quality camera for aerial video and stills I have searched high and low. One of the more interesting candidates was the Sony QX10. It is under 100grams and has a Sony "G" series lens. (This is compared to a Sony Nex5N which is 400grams with a fixed 16mm lens. ) It is a 18 megapixel camera that is essentially a Sony Nex without the extra weight and with a fixed zoom lens. So that makes it pretty attractive for AP. It is high quality and light weight. It has a nice remote smartphone app designed to be an add-on for a smartphone to improve it's camera capabilities. So it connects to the smart phone via wifi.

This is fine if it is attached to the smartphone but not so good if you are 1000ft away. This problem pretty much made the QX10 unusable.

But I have found a couple of solutions. The first is a product from VP systems of Sweden. It is a wifi interface for the Sony QX10, and other cameras, that connects via wifi and then translates that into signals you can control via your RC transmitter switches via PWM. It is called the Camremote V3 and it includes their Camremote trigger product and the wifi interface in the same small module. Here finally is a way to remotely control the Sony QX10. Unfortunately it does not get you live video out. It just controls the camera but you would have to use a tiny fpv camera to "see" what the QX10 sees for framing.

I just discovered another solution. The Xiaomi Mi wifi range...Continue Reading
Posted by mike_kelly | Aug 31, 2016 @ 03:39 AM | 20,155 Views

My last project was a H4 680 folding quad. I really like this frame, available in a variety of slightly different versions, because it is light weight and folds into a small package. It commonly yields builds where users are getting 30 minutes of flight time. It has a more logical main box that is rectangular and seems easier to mount typical electronics than the popular frames out there with circular main plates. I built the "two clamp" version and I was very satisfied with the build.

But Daya has just released a new version of this popular frame. In my opinion they have improved many of the features and fixed weak spots in the previous versions.

Even though it is a 680mm frame it folds smaller than the previous versions. The frame is shorter but the center box is wider making it even easier to get the electronics needed inside.
...Continue Reading
Posted by mike_kelly | Apr 07, 2016 @ 04:28 PM | 22,493 Views

Comparison attached as PDF at end, Compare_minis_3.pdf

02/17/17 added zip file with original spreadsheet and corrections for MicroPix.

It is very exciting to be at the beginning of a new technology. It is a time of great growth and new ideas but also a time of chaos.

I found myself needing to substitute one mini Pixhawk variation for another, in a finished build and started to check the pinouts of the similar connectors. I was surprised to find how different and impossible it was to substitute one for the other. I decided to survey the mini Pixhawk field and see just how different the cabling is between them, compared to the Gold standard of the original Pixhawk designed by Lorenz Meier (Pixhawk.org).

I found out why it is so hard to simply get a GPS connected to a mini Pixhawk.

The industry desperately needs a 500lb Gorilla to come into the field and force everyone to conform to a standard so we can get rid of the ridiculous chaos.

I should make a comment about connectors here. The original Pixhawk used DF13 connectors and they have what I describe as a square hole as opposed to the similar Molex Picoblade which has slot like holes. The two can be interchanged but it is not a perfect fit. The last and most commonly used in the mini Pixhawk field is the JST-SH 1.0. It is noticeably smaller than the other two.

From left to right, the JST-SH 1.0, the Molex Picoblade, then the DF13.

...Continue Reading
Posted by mike_kelly | Feb 05, 2016 @ 11:01 AM | 17,897 Views
* The Beaglebone organization has just released (Spring 2017) a new Beaglebone with sensors built-in for robotics https://beagleboard.org/blue
* Mirko Denecke once again ported Ardupilot to work with the Blue so we have a single board computer running linux that does not require an
* additional cape to provide sensors and sells for less than $100 US.
* Here is Mirko's documentation on using the Beaglebone blue with Ardupilot: https://github.com/mirkix/ardupilotblue

The RC world has come a long way in the last few years due to the advent of flight control computers. The abilities and features of these boards have been improving and changing at a rapid rate. So much so that some flight controllers have become quickly obsolete because the boards ran out of room to store the programming on board or ran out of computing power to run the more sophisticated mathematics the allows smoother more reliable flight.

Case in point is the APM flight controller from the open source Ardupilot project. This 8 bit computer based flight computer was a great upgrade from it's arduino beginnings but as time passed and more features and improvements were added it began to be hard for the developers to fit their code on the board, constantly running out of room. The cpu just was not fast enough for the new mathematical routines that were needed.

...Continue Reading
Posted by mike_kelly | Jan 29, 2016 @ 11:41 AM | 18,391 Views

One of the more advanced aspects of RC flight is FPV. FPV stands for First Person Video and it means you are looking at a video display from a camera on the aircraft giving you the view that a pilot would have in the cockpit. This experience is as close to actually flying a plane as you can get without leaving the ground. It is also a way to always fly with the orientation of the aircraft constant. That means that if the aircraft is flying away from you or towards you the aircraft right is always your right and aircraft left is always your left. Because it is like you are in the aircraft. One of the more difficulty aspects of normal RC line-of-sight (LOS) flying is that your orientation, relative to the aircraft, is always changing as you steer the aircraft. When you fly away from your self left is left and right is right but when you turn the aircraft around and fly back to yourself everything is reversed on the aircraft. In order to bank the aircraft right you need to use the left roll on your rc transmitter. It can be very confusing. But with FPV you can always maintain the same orientation with the aircraft.

There are challenges though. First you are looking at some kind of display. That might be a video screen or a pair of goggles. Either way it is a two dimensional image so you have no depth perception. It can be hard to tell how high you are or how close you are to objects.

When you fly LOS (line-of-sight) your frame of reference is the aircraft itself....Continue Reading
Posted by mike_kelly | Dec 05, 2015 @ 09:40 AM | 14,272 Views
A reference for those looking for details. My scale may not be perfect but is seems to compare to claims well.

TL65B01 (650i ironman) claimed frame weight on the box=450g, my weight for the whole kit including plastic bags but not the spares =540g
Land Gear alone=100g

TL65S01 (650s sport)claimed frame weight on the box=750g, my weight for the whole kit including plastic bags but not the spares =755g
Retractable Land Gear and controller module alone=165g
Main PDB plate = 80g.
Arm Length 231mm

TL68C01(690)=claimed frame weight on box=600g, my weight for the whole kit including plastic bags but not the spares =765g
LG alone=115g
Main non-pdb plate = 45g
Arms fixed=323mm Folding=333mm

TL68P00(680Pro)=claimed frame weight on box=780g, my weight for the whole kit including plastic bags but not the spares =945g
Landing Gear alone=160g
Main PDB plate =85g
Arm Length 264mm

Tarot T960 TL960A no landing gear 1045g, unconfirmed

Tarot XS690 quad TL69A01 675g claimed, with or without landing gear?
Posted by mike_kelly | Dec 05, 2015 @ 09:39 AM | 13,347 Views
**** Ardupilot has ended it's affiliation with Dronecode.org and 3DR so it's funding comes completely from Donations from Partners and people like you!! There is a donate page on the Ardupilot.com website now. DO IT NOW, go and support Ardupilot so we all beneft from new features and bug fixes. Small donations from lots of users can make a difference.
http://ardupilot.org/copter/docs/common-donation.html ****

I am going to reserve this page to introduce the people we have depended upon in the Ardupilot project. I am not an insider but I think it is important to start to learn about the project that we use and the people that make it happen. These are really smart really valuable contributors. Please post additions. These are only the people I have figured out are important assets in the project. In no particular order and forgive me if I am not aware of your contributions, information is very hard to find.

Lorenz Meier:

Swiss Federal Technical Institute of Technology, Department of Computer Science Computer Vision and Geometry Group
I'm working on mobile localization and 3D reconstruction on smartphones and micro air vehicles. I started my aerial robotics project, PIXHAWK, in 2009 as a master student. It is by now an on hardware autopilot anyone can use (open hardware website, research website). I participated in the sFly EU research project 2011-2012 and I'm currently mostly working on obstacle avoidance on drones, funded by a gift from Amazon Prime Air. I maintain...Continue Reading
Posted by mike_kelly | Aug 07, 2015 @ 12:35 PM | 14,294 Views
ZMR250 Coaxial Octocopter

Now that I have a few builds under my belt, I did a review of what worked and did not work for me. When I first came to RC, with no experience whatsoever, I was surprised to see things used like the horrible Dupont servo connectors. They may have been simple and cheap in the beginnings of RC electric planes but they are really bad connectors for multirotors that vibrate and dive down through prop wash.

When I first started, I decided to do away with all those lousy connectors and such by soldering everything. Using connectors is notoriously unreliable. I produced a nice tidy build with every wire the right length, neat and tight by soldering. No cables 12" inches too long all folded up to stop them from tangling in the props.

Trouble is, the first time you crash and something fails inside the frame it can be a major major problem trying to repair it. Sometimes having to completely tear the build down to original parts to get at something.

I have been disappointed in the quality of the parts commonly used in RC but I should not really be surprised. The technology we use is absolutely amazing for the low price. A video transmitter for $30 or an autonomous flight controller for $100 is incredible. But the assembly and soldering are often poor, and the parts are probably "seconds" much of the time. I just have to live with it because I am clearly not going to pay the high cost of milspec parts which might cost 100 times more.

This...Continue Reading
Posted by mike_kelly | Jul 04, 2015 @ 12:14 PM | 13,169 Views
*********** This article is now a year old. As with any technology this item is now obsolete and too slow for the latest versions of Tower etc. There are other low priced tablets on Ebay that are faster and a better choice now. ***********
***** Oct 2017, I feel compelled to add a note that the last two tablets I have puchased off Ebay have come "rooted". What this means is that they were installed with a permanent back door that would allow someone to remotely control and remotely steal accounts and passwords from the tablet. So I will not be buying anymore smartphones or tablets from Ebay.
************************************************** ************************************************** ************************************************** **********

I'll say it right up front. I hate tablets. I don't like gestures and I miss my good old keyboard. The first tablet I bought had a resistive touch screen which was horrible. I found the early versions of Android to be buggy and unreliable.

So why did I get another tablet? I must have been tired late at night.

Or I heard the siren song of the foolish. $29 for a 7" Android tablet.

I bought a Maylong M-295 from Ebay:

I just knew I was going to be sorry.

But to my surprise this little tablet actually works. It is not the fastest thing in the electronics world but it is satisfactory. I needed to pause just a moment rather than zooming around. I...Continue Reading
Posted by mike_kelly | Jun 09, 2015 @ 08:36 PM | 60,911 Views
I just reviewed the Pixhawk LITE a low cost small form factor Pixhawk clone offering from GoodLuckBuy. It seems like it will serve well as a replacement for APM.

But for my big multicopters I am more and more looking for reliability. I have learned a few things from my first builds and the top three are:
1. The quality control on the common parts we buy is poor.
2. You are going to crash and break things.
3. Because of #1 & 2 you better be able to fix your aircraft without tearing it down to bare bones because you will be doing it often.

If you don't want to be frustrated think reliability from the start of your build.

That is where the Pixraptor comes in for me. You might think why do we need another Pixhawk clone, especially one that is more expensive than other clones? Well this new Pixraptor is not your take-a-picture-and- copy-it-exactly type board. Somebody really took a look at what could be improved and tried to make a better product. I won't know if they met those lofty goals until the build is finished but the attempt was definitely made.

First thing I notice is the connectors all come out the sides. I like this. It is hard to make a build with clean wiring if all the connectors are pointing straight up. Next they used molex connectors instead of Df13. Some people hate DF13 connectors. The deal is the DF13 connectors require you to tilt the connect on edge and kind of rotate into the socket. If you don't do that installing and removing you risk...Continue Reading
Posted by mike_kelly | May 20, 2015 @ 05:49 PM | 297,656 Views
************************ Notes or errors and omissions **************************
I started this thread a year ago. So be advised that although there is invaluable information in this thread the manufacturers can and have changed the product over time. Hence although I am not aware of it as of Jan 2016 they could someday fix the problems making the information here on page one obsolete. If you receive a board that does not seem to fit the data send me a PM and I'll update page one.

NOTE There was a version one of this board that had the power module pins backwards. You can not use a "normal"power module cable with that early version. The current verson GLB is calling V2. They fixed the power module connector. It now conforms to the Pixhawk standard. But all the other problems talked about below still exist.

Whomever created the pinout pictures on the GLB website got a number of things wrong, at least as of this date. Hopefully they will correct it.
The order of many of the pins are backwards from the same connectors on the 3DR Pixhawk. See the drawing labelled "corrected" for the actual pin-outs.

Since the I2C connector is backwards, but there is an expansion board for more than one I2C device, (I2C devices are compasses, LED external lights etc) jesmail suggests using a reverse cable from the Pixhawk LITE I2C port to the I2C expansion board. This has the effect of fixing the problem so all the devices you plug
...Continue Reading
Posted by mike_kelly | May 19, 2015 @ 08:38 PM | 13,085 Views
Tarot T shaped landing legs are pretty common throughout their line, as well as many other brands. Tarot uses hollow carbon tube skids. If you land at an angle and touchdown on only one of those legs first, they tend to break off right at the T joint. It seems like a pretty common problem. Some people put wooden dowels inside the carbon tubes but I did not want anything rattling around in my skids.

I first tried some fiberglass skids but I was concerned that with something much stronger it would just transfer the implact further up the landing gear. Sure enough the next flight with the fiberglass skids I did not break the skid but instead broke the T Joint and the landing leg and the attachment at the top for the landing leg.

Not a good trade-off.

Then I found some nylon 10mm rod on Ebay. It comes in just about the right length. It bends, unlike the fiber glass and will absorb some shock. I dont know if it will be enough but time will tell. It flexes under stress but is not too soft.

If you have a better idea please post a comment, thanks.
Posted by mike_kelly | Apr 13, 2015 @ 11:22 AM | 16,621 Views
*** a note ***
i think that the following rant is a bit harsh and part of the problem is how I look at things. I see the rotation of the motors as beng the reference point but if you are using the threads on the quick detach as the reference then the labels are correct. The threads on the quick detach are the opposite of the rotation of the motors. Even when you finally get them on correctly there is a possibility that if the props are large, 15" or larger, and the motor stops suddenly in flight the props can have enough momentum to unscrew.

************************** Mis-labelling *******************************
I think I have figured out what is wrong. THESE ARE LABELLED WRONG!! I am so tired of products with sloppy incorrect documentation. On each quick detach is an arrow, which you might reasonably assume indicates the direction of the motor they are designed to go onto. Nope the labels are backwards and the description is backwards.

In most ads the TL68B36 is described as a ccw and the photos have an arrow in the ccw direction. My Tarot prop adapters that are labelled as such go on the CLOCKWISE motors NOT on the ccw, in order to work correctly i.e. not spin off during flight.

Please see Post#59 at https://www.rcgroups.com/forums/show...foxtech&page=4

Originally Posted by strips View Post
Unless you have a genuine foxtech prop mounts I would not trust the colour to be
...Continue Reading
Posted by mike_kelly | Mar 19, 2015 @ 05:15 PM | 19,081 Views
One of the most important build issues on a multirotor is reducing vibrations. The motors and props make a lot of vibrations and the flight controller does not like this. The flight controller has a number of sensors that are vibration sensitive. So if you want your multirotor to fly smoothly and loiter accurately you need to get those vibrations down. The problem is that the flight controller sensors need to be firmly attached to the multirotor to sense the movements. So you have a problem, you need to dampen the vibrations from the motors yet you don't want the sensors to get too isolated, using too soft a spongy foam/gel, from what the frame is doing.

That is why a lot of people spend so much time balancing motors and props to reduce vibration at the source.

Another way is to use anti-vibration motor mounts. I bought a set of Tarot TL68B33 (black) and TL68B34(red) to test on my Tarot 650 Sport quad and see if they would really be any help. They isolate the motor vibrations from the arm by adding a little rubber bumper between a motor plate and the motor mount. It did not look like it would do much.

But much to my surprise the reduction in vibrations is quite significant. Vibrations on an APM flight controller are considered acceptable if you can reduce the x-axis and y-axis vibrations to below +-3 units and keep the z-axs to -5 to -15 units. It is not easy to get your vibrations to this low level. I ended up using the RCTimer anti-vibration flight controller mount to...Continue Reading
Posted by mike_kelly | Feb 14, 2015 @ 04:49 PM | 20,704 Views
I just got through with my Tarot 650 Sport build. I was very careful to check that all the rotations are correct.

1. Motors are rotatiing in the right direction on each postion on the multirotor according to the 3DR wiki. Note that the motor numbers are not simply clockwise around the quad, but the motor test in Mission Planner does not follow the motor numbers and simply starts one motor at a time in a clockwise direction.

2. The Clockwise (CW) and Counter Clockwise(CCW) props are on the right motors A clockwise motor needs a clockwise prop. The reverse rotation props are the clockwise, sometimes known as "pusher" and sometimes marked with an "R" like 9045R. The so called normal props are the counter-clockwise.

3. The Radio Calibration is correct
in Mission Planner and the sticks move in the correct directions. Note that the Pitch must move the slider bar DOWN in Mission Planner when you move your pitch stick "forward". All the other sliders move in the same direction as the stick moves.
...Continue Reading
Posted by mike_kelly | Jan 27, 2015 @ 05:52 PM | 19,461 Views
>>> Feel free to post your Tarot travel case solution as a comment at the end of this blog post <<<

Now that I have my Tarot built I need a travel case. The aluminum ones from Asia are way too expensive for me and the Pelican cases are worse. After looking and following other posts I finally came upon the Plano 114400 Bow Case.

This case is 41"x17" x7" which is enough room for the Tarot 650i 650s 680 680 Pro and 690 even with extended arms. It is lockable and has room for my gimbal too.

It comes with a thin layer of pluck foam so I plan to put a layer of solid foam underneath the pluck foam.

It is not a perfectly rigid case like the Pelcan cases. It flexes a bit and it is definitely a cheaper build quality but it is lighter and a lot less expensive.

I got mine off Amazon for $75 shipped.
Posted by mike_kelly | Jan 26, 2015 @ 10:07 PM | 68,393 Views
************************************************** ************************************************** *******************************
* Be advised that this article is now obsolete for Pixhawk. I have written a new article for Pixhawk.

************************************************** ************************************************* *******************************

I have been wrestling with this issue for weeks and I thought I should document what I discovered. Many people these days need to use more channels to control external devices like Gimbals, Lights, Retracts, Camera Shutter and Video switches. For more traditional buillds it is not hard to take the pwm outputs of your reciever and directly connect them to your devices, by-passing the APM flight controller...until you run out of channels. But for newer builds with PPM or S.bus receivers you may not have any pwm outputs to use directly from your receiver. Yes with many Frsky receivers you can do both s.bus and pwm but that is a different story.

So if you are using ppm input to your APM how do you control extra devices?
Mission Planner provides a optional hardware page in setup for APM, it is labeled "Camera Gimbal". It allows you to map input channels from your transmitter to an output pin on the APM board. This is where the documentation gets a little tricky. First APM does not support more than 8 channels. Pixhawk does but not APM. Second the documentation is poorly written and uses...Continue Reading
Posted by mike_kelly | Jan 22, 2015 @ 11:32 AM | 23,789 Views
One of the fun things about multirotors right now is that we are at the very beginning. Someday soon there will be mostly RTF (ready-to-fly) multirotors out there with all the problems solved. But right now we get to figure it out ourselves and have the fun of designing and solving problems.

One of those problems for me was how to get a video and shutter control cable connected to my Canon S100. I went to Gentles in the UK, who makes lots of different shutter video solutions for RC and he said it was not possible with the Canon S100. What?

So I did some digging and this is what I found. The Canon Powershot series uses a mini USB cable for connection to a desktop computer and for video out. They make a cable for video out, the Canon AVC-DC400 and the Canon AVC-DC400ST. The only difference is the 400 is single-channel audio and the 400ST is two-channel stereo. But why do they also make the USB Cable IFC-200PCU? Why two cables? It turns out that our favorite connector, the mini USB, has many versions. The typical cable that we use for connecting our flight controllers only needs to use a few pins. It needs power and ground and transmit and receive. But the mini USB can have up to 11 pins. It has five on the top side and six on the bottom.

11pin mini USB connector

This is what Canon does; they use the top five pins like any normal USB for communications with the PC and they use the bottom six for video out....Continue Reading