otlski's blog View Details
Posted by otlski | Mar 22, 2013 @ 09:00 PM | 7,742 Views
To start, I see a lot of misinformation regarding night vision, infrared cameras, and various CCD / CMOS technologies. I have made several posts throughout several forums. The following is a compilation of some of them.

A few months ago, i was approached by an Open Pilot member inquiring about IR camera choices. Below was my reply

I've looked at FLIR cameras for my own use as well. Having one would keep me familar with the problems our customers face. The model I keep coming back to is the FLIR PathfindIR. It is available in 30 fps with a NTSC output (if you are a US citizen). The output is a monochrome image (grayscale). They are 320 horizontal by 240 vertical resolution; for low-end FLIR, that is decent resolution. The FOV might be an issue at 36 X 27-deg. Because of the FOV, you would have to deal with the "tunnel vision" as well as optically amplified affects of airframe vibrations whether with or without a gimbal. These cameras were used by Cadillac and BMW in their night-driving option package.
you'll likely need their proprietary cable connectors to make hook-up easy.

I also seem to recall seeing that the FLIR Firstmate series has a NTSC / PAL output. They offer differing resolutions and FOVs. All outputs are grayscale.

A few weeks ago I picked up a FLIR I7 hand-held for various purposes: home cold air infiltration, attic critter detection, wood stove troubleshooting, general mechanical
...Continue Reading
Posted by otlski | Mar 02, 2013 @ 10:19 PM | 7,150 Views
A reoccurring theme on the hobby forums is when a post is made regarding how much the government is paying for UAV flight hardware. Invariably, it is always the same. Someone says "the idiots overpaid; I can do it for half the price and still be rich".

I've thought about responding but never did. A month or so ago I saw this topic on DIY Drones and was happy to see a response by Toby Mills. Below is his response followed by my thoughts on the subject.

Comment by Toby Mills
Just to re-iterate what Bart says.
It really annoys me when we as a group look at the cost and go, I could build one cheaper than that.
The reality is... no you couldn't.
First you have to gather the requirements.
Then you have to design it.
Then you have to production-ize it. Keep in mind this includes all the tooling costs which can run into hundreds and hundreds of thousands of dollars.
Then you have to build a ground station, by the look of the ground station in the picture, its proper milspec gear built to last and endure. Hundreds of thousands of hours go into software development and testing of the software in that type of gear.
Then you have to write a user manual.
Milspec user manuals are not the one page "made in china" type stuff we normally get in our boxes. These are multi-volume manuals that take a week to read let alone write. The cost of employing a team of technical writers to do this is insane.
Then you have to get it tested and signed off by the customer. This
...Continue Reading
Posted by otlski | Feb 06, 2013 @ 08:46 PM | 6,285 Views
Some links to a few technical papers on gimbals at the Space Electronics website.

The first is written by my colleagues as an overview of gimbal balancing.

The second is by myself as the primary author and was written for optical guys as a way to introduce the subject of mass properties.

The third paper was written for mass properties guys and introduces imaging and ancillary concepts.

Please note that there is a typo in the first paragraph of page 11 where is says...
"A reasonable approximation says that when jitter causes an image to shift less than 20%, the image will only be marginally impacted. Above 20% will degrade MTF".

what it should say is this...
"A reasonable approximation says that when jitter causes an image to shift less than 20% of a pixel, the image will only be marginally impacted. Whereas a shift of above 20% pixel will degrade MTF".
Posted by otlski | Feb 02, 2013 @ 10:56 PM | 6,559 Views
So far our text, drawings, and videos have shown the yaw axis reaction to disturbances. In fact, everything that applies to yaw also applies to pitch. For this next example we switch to the pitch axis because it presents an easier to imagine example. But again, this newest example applies equally well to the yaw axis or any other named gimbal axis.

When we suggest that a gimbal axis is immune to the affects of the 3-DOF of angular acceleration, we actually mean to state a very specific case where the rotational center of the angular disturbance is coincident with the axis pivot under analysis. However, this is not the case for most disturbances. Consider a gust of wind impinging on an airframe (shown in brown in the video). The airframe will pivot in reaction to the gust around its own CG (larger CG symbol). When this happens the entire gimbal will move through an arcing motion that approximates a linear translation for our intents and purposes. This approximation acts upon the gimbal axis shown because the axis CG (the mostly hidden smaller symbol) is not coincident with the axis pivot.

offset angular (0 min 10 sec)

You might observe the video and think that by purposely setting the CG low on the example pitch axis will help provide a natural stability against pitching disturbances. In reality this will not work. Although it appears to provide the needed correction, in practice it will be impossible to set it up for all rates of accelerations. Furthermore, the practice of purposely setting the CG low will absolutely harm the stability for relevant translational accelerations. Again, the single best thing that you can do is to correct the axis CG so that it is coincident with its respective pivot.
Posted by otlski | Feb 02, 2013 @ 10:47 PM | 6,504 Views
What happens when you model real bearings that have friction? Unlike the previous video where we used frictionless bearing pivots, this video will use commodity bearings which indeed do have friction. The result of which is when an outer frame or axis is rotationally disturbed, friction in the torque domain will transfer some of that rotation undesirably to the successive inner axes.

Angular with bearing friction (0 min 11 sec)

Thought exercises:
-more bearing friction means more undesired coupling
-more coupling means more unintended motion
-lower axis MOI means more unintended motion
-higher axis MOI means it is harder to command a purposeful axis slew

Bearing friction is a virtual given for the vast majority of gimbal configurations but it can be reasonably managed with proper selection and by design. Proper selection of cabling and the routing of that cabling is equally important. Cabling likely will increase the unwanted coupling between axes in a similar fashion that bearing friction does.

Lastly, attached is a zip folder with an Excel Spreadsheet that can be used to explore the relationships desribed in this post.
Posted by otlski | Feb 01, 2013 @ 09:49 PM | 6,844 Views
So far out of the 6-DOF, we've discussed the 3-DOF of translational acceleration disturbances. Let's turn for a minute to the 3 -DOF of angular acceleration disturbances. Keep in mind that our gimbal so far has no closed-loop torquer motors to restore position, and let me also say that the example gimbal has frictionless pivot bearings and no cabling between aircraft frame and the outer or inner gimbal axis.

Observe the video here;

Angular Accel Only (0 min 13 sec)

What happens? Under angular acceleration where the center of the acceleration is coincident with the axis pivot, the gimbal axis will remain stationary, fixed staring at its intended target. This is because the frictionless bearings transfer no torque from the moving frame to the camera axis.

T=Ia. Or torque equals inertia times acceleration. Where torque is in Newton meters (N-m), MOI is in Kilogram meters (Kg-m), and angular acceleration is in radians per second per second (rad/sec-sq). Simply rearranging we find that a=T/I. Thus for angular acceleration disturbances as described above, in the presence of any MOI, and the absence of torque transfer; there will be no angular acceleration imparted to the inner camera axis.
Posted by otlski | Feb 01, 2013 @ 09:15 PM | 6,560 Views
Four drawings clarifying the initial math described in the previous post. In the first three drawings, the gimbal axis is shown complete with its undesirable CG imbalance. The 2D CG is in the same location in these three drawings, only the vector of acceleration is changed.

The fourth drawing shows the corrected CG is now coincident with the axis pivot. The axis becomes immune to translational accelerations of any sort. The attached video demonstrates the forth condition

Balanced Translation (0 min 3 sec)

Posted by otlski | Jan 27, 2013 @ 09:01 PM | 6,353 Views
In our previous example, we demonstrated how an unbalanced gimbal wants to respond to translational accelerations. In this example we will explore how that same gimbal responds to vibrations or oscillatory motions. Vibrations are accelerations, they happen to have acceleration vectors that change sign as the gimbal is shaken back and forth.

07 Oscillation with Imbalance (0 min 16 sec)

In this case the gimbal still rotates clockwise from a forward acceleration toward the target, but then rotates counter clockwise as the gimbal experiences a backward acceleration. The torque force are calculated the same as in the straight line example.
Posted by otlski | Jan 27, 2013 @ 08:51 PM | 6,426 Views
Continuing with the 2D view, we will now show what happens to the imbalanced innermost axis (the camera) when a forward acceleration is applied. The applied acceleration in this case is a straight-line translational acceleration. For our example, we have no restoring force normally provided by closed loop torque motors. We are simple applying acceleration disturbance and observing the reaction motion.

06 Acceleration with Imbalance (0 min 4 sec)

The 2D CG can be expressed in moment units (torque units), say Kg-mm for example. Let's say our camera weighs 1 Kg and the CG is 10 mm offset from the pivot relative to the direction of motion. This creates a normal moment of 10 kg-mm. Now, let's say the acceleration experienced is 3 G (gravity), the torque initially created would thus be 30 Kg-mm.

Let's take note of a few important things here. First, in the absence of translational acceleration, there is nothing making the gimbal want to rotate off target. Second, any torque from acceleration disappears once a steady state velocity is obtained. Third, the torque via acceleration becomes less and less as the gimbal rotates in response. This continues until the CG is located "behind" the pivot in the acceleration shadow. Forth, if the CG was coincident with the pivot, there would be no moment arm for the acceleration to act against, thus the gimbal does not respond to translational accelerations.
Posted by otlski | Jan 27, 2013 @ 08:21 PM | 6,323 Views
Concluding the basic features of the example gimbal, we reorient the image to provide a 2D view looking down on the gimbal. Notice the Center of Gravity (CG) symbol as the video concludes. The CG symbol is a circular "checkered" symbol. In this case is represents the 2D location of of the CG which is behind the axis pivot. The 3rd axis of CG is not important for the upcoming demonstrations.

05 Transition to Top View (0 min 13 sec)

Posted by otlski | Jan 27, 2013 @ 08:15 PM | 6,260 Views
The pitch motion is exercised by articulating both the innermost axis (camera) and the outer axis. These axes are pivoted about the axle shafts attached to the frame.

04 Gimbal Pitch Demo (0 min 13 sec)

Posted by otlski | Jan 27, 2013 @ 08:09 PM | 6,200 Views
A quick demonstration of the Yaw axis articulation. The innermost axis (camera) is suspended by the outer axis via axle shafts pivoting on low friction bearings.

03 Gimbal Yaw Demo (0 min 12 sec)

Posted by otlski | Jan 27, 2013 @ 08:04 PM | 6,302 Views
A quick visual overview of a two axis gimbal. This configuration features a rectangular camera body and cylindrical lens as the innermost axis. The outer axis is the next rectangular frame out. And finally, the rectangular frame that might mount the gimbal assembly. Simplified "dowels" designate the pivot axis.

02 Gimbal overview short (0 min 10 sec)

Posted by otlski | Jan 26, 2013 @ 05:54 PM | 6,366 Views
When conversing about motion mechanics we frequently refer to the six degrees of motion, or 6-DOF. 6-DOF includes three axes of translational motion and three axes of angular (rotational) motion.

In the attached video a cube-like object is pictured. You can see three straight-line arrows depicting the cube's three DOF translational motion options, and three arced-line arrows depicting the cube's three DOF angular motion options. The translational motions occur "along" the depicted axes, the angular motions are said to occur "about" the depicted axes.

6 DOF (0 min 16 sec)

Along with defining an object's type of motion, 6-DOF also can refer to the application of multi-DOF disturbances forces imparted to an object.
Posted by otlski | Jan 26, 2013 @ 04:48 PM | 6,408 Views
A common occurrence on RCG and other forums is to to refer to rotational propeller physics as "inertia" whereas the poster really means to refer to moment of inertia. I myself am guilty of casually interchanging the terms on a regular basis but when I'm feeling like I'm in an exacting mood, I will use the correct terminology.

Moment of inertia has a few aliases/abbreviations, among them are:
  • rotational inertia
  • MOI
  • mass moment of inertia

The same misnomer is often applied to the discussion of gimbal axes where the posters are describing a gimbal axis resistance to rotation as inertia, where they actually mean to say MOI.
Posted by otlski | Jan 26, 2013 @ 04:20 PM | 6,479 Views
I am Daniel Otlowski. I have been in the RC multicopter world for five years. What attracted me to this hobby was the diverse skill set required to become proficient. The skills naturally dovetailed with my line of work at Space Electronics where I've worked since 1979.


At Space Electronics we design and build instrument that measure CG, MOI, Gimbal Balance, POI (dynamic balance), Blade balance, etc. In many respects my work experience puts me in a position to advance the ball, spreading the knowledge to beginners and advanced hobbyists. I find that a lot of my posts are related to teaching how physics applies to multirotors and to convey the subtleties of terminology. My infrequent posts are scattered about as I usually try to gently push information out there. I thought I would start this blog to gather these snippets in one place.

The blog is meant to be read in order

Mass Properties are fundamental physical properties of matter. All manner of objects from the atomic level to large man-made assemblies and naturally occurring objects can be characterized by their mass properties.

They include:
  • Mass and Inertia (provided as two separate thoughts although essentially the same)
  • Mass – the amount of matter – often thought of as ‘weight’ when in the presence of a gravity field
  • Inertia – resistance to a change in motion – Inertia is that quantity which depends solely upon mass and vice-versa
  • Moment of Inertia – resistance to
...Continue Reading