DF Fans,
We had planned and prepared for a really important presentation and SAVS demo scheduled for this morning at 8:30. I had gone over and tested every part of the system a half dozen times in preparation, and as usual the DF's performance had been predicatably rock solid. The meeting was postponed until 2:00PM, and at about 3:00PM we walked outside to a hot dirt parking lot for the big climax of the presentation - a flight demo. And in perfect accordance with Murphy's Law, the SAVS refused to respond in any way. It would not arm, nor vary the red LED in response to throttle after turning on - nothing.

I had experienced similar behavior once before, on a very hot 94-degree day, and made the mistake of taking off anyway, once I managed to get it armed. In mid-flight it suddenly tipped up and dove straight for the ground. Luckily it was still responding to stick commands and I managed to get it straightened out. So I figured it had something to do with the afternoon heat and on a hunch I set it under a bush for a few minutes. It then fired up OK, and I took a chance and flew the demo:

http://www.youtube.com/watch?v=JA14lf0oPbA

SO HERE'S THE PROBLEM: The microcontroller (44-pin SMT component on the right side of the main board) will begin to run erratically above a certain temperature. With the ambient temp being above 85 or 90 degrees, the floor temperature of the IC is raised closer to its limit. If the controller is exposed to DIRECT SUNLIGHT, it will exceed this temperature and fail (temporarily). I verified this theory by turning on the DF in my garage, and verifying LED response to the throttle. I then placed it outside with the sun shining directly on the right side, and in 2 minutes it stopped responding to the throttle. Bringing it back into the shade, it took only about 15 seconds for the controller to come alive again. I reconfirmed this by spraying compressed air on the controller after it had failed in direct sunlight, and it immediately came back to life due to the cooling. This is a simple set of tests that should be conducted independently, on the off chance that the microcontroller on my board just has a bad solder connection.

Some of the behavior of the controller on the margins of heat failure sound a lot like the unexplainable glitching and Ti weirdness that some people have experienced here. The solution - paint the inside of the clear canopy with silver paint to turn it into a mirror and reflect the sun's heat. Guess it never gets that hot in Minnesota or Saskatoon!

- Bruce

Interesting report. I have been flying in 90 degree temperatures but have not let it sit idle in the sun long enough for significant sun loading. Once airborne, I think that there should be enough air flow getting thru the canopy to keep the IC cool enough. I will try to test both of my DFs to see if it can be repeated here. As I'm sure you know, a heatsink (even a small one) with a thermal conductive adhesive, could significantly reduce the chip temperature. Take a look at the bottom side of the PWB and you'll see they use a little washer as a heatsink on one of the semiconductors.

Mike

BB_DF, why is your video so crooked? I've noticed the last few being that way too.

Sky,
Actually, I had the horizon line set up perfectly in the straightforward position before I left the house. The SAVS pod is designed at a 45-off-a-45 degree angle, so that facing straight forward, the camera will be level inside the pod. In that video the camera is set downward 45 degrees just before the flight. We frequently have to change the camera angle several times during a practice mission - depending on what we want to look at. If the camera is facing 30 degrees downward, it's great when you're looking at the near side of a building, for instance, but if you want to look at the other side, you have to fly WAY over behind it, and turn around to look back. The alternative is to lower your altitude, but you can't do that because the building blocks your view. So what you have to do is aim the camera more downward, say 75 degrees.

Or let's say there's a tanker spill. On the same mission you want to fly over the scene and look down to see how much chemical has been spilled. Then you want to reset the camera level to fly down low to see if there's leakage underneath, and to read the truck's declaration plaque. You don't have the time or the setup to recalibrate the level every time you change the camera angle.

Now take that to the extreme case and say you need to aim the camera straight down, which is sometimes required. Think about this: the geometry of the mount and pod would then dictate that you'd have to rotate the camera & foam 90 degrees inside the pod. Additionally, you would not be looking forward, but at a 45 degree angle off to the right! Actually, you CANNOT aim the camera straight down! So as you can see, there are some serious shortcomings to this design. You cannot tilt the camera off the horizontal without it affecting the horizon setting. If you always have it set at 30 degrees or something, then there's no problem. Also I don't think the camera, foam and internal wires could stand up to continuous twisting around inside the pod every time we change the camera angle. We might need to re-mount the camera hardware in a simple swivel like Bergen uses.

Since we're pushing the envelope, I guess we get to lick a lot of stamps!

- Bruce

PS. Here's the ideal setup: the landing gear is raised to allow free 360 degree cameral gimbal action. We can only dream!

I understand how the ball's orientation changes when the housing rotates and how you would have to rotate the foam to counteract that to maintain a level frame. But once that is done you could almost get a straight down shot by simply pulling the foam outward near the top like I described in your [link=http://www.rcuniverse.com/forum/fb.asp?m=5939279]camera alignment thread[/link]. That goes for when the ball is in it's normal orientation too. Whatever orientation it's done in, there will be alot of foam protruding from the ball. I agree that alot of those types of adjustments would not be good for those thin wires. I guess my real question is didn't you notice the camera's angle was that far off or was that just a flight demo and not an AP demo?

I've seen that mount for those larger helos. Have you seen [link=http://helicamsolutions.com/html/rockclimbing.html]these[/link] videos of two of them at once shooting AP of a few rock crawling trucks? They shot some nice air to air of each other too. [8D] Just double click in the center of the video player and it will expand to full screen.

This demo was for SWAT. We changed the camera angle from near flat to 45 degrees downward just prior to the flights, so that we could look down on the nearby house. If nothing else it would have significantly taken away from the impact of the demo if we'd further delayed the takeoff to try and realign the horizon (with the DF already acting up!). Also it's almost worse when the horizon is just SLIGHTLY off the horizontal. It makes it look like you tried to align it, but didn't quite make it. The deputy was wearing the video goggles and was not directing me, which is why the flying is like Brownian Motion.

I've decided to set the camera at about 45 degrees downward, and try to leave it there. We'll just have to fly accordingly. I set the camera angle by aligning it with a seam on the garage floor. Here's what it looks like. Also I couldn't resist the gold color for the dome!

Cheers,
Bruce

I checked the microcontroller's factory specification and it is rated to run to +125 degrees celcius/ +257 degrees F. I would say the reason is you have a bad microcontroller or a poor solder connection on the circuit board. Heat the board up to make it stop working and use cold spray on specific parts. This will allow you to locate the bad connection. If you can't find the problem return it to Spectrosolutions for warranty.

dart340,
Thanks for that. I've done the cold air test and it's the microcontroller for sure. The solders on the SMT itself look good under the magnifier, but of course that's not the whole story. It could still possibly be a heat problem, combined with a chip that doesn't quite meet the temp spec. The DF was in the hard case in the car, so the board was probably around 100+ F. Setting the unit on the hot ground, in direct sunlight, with the canopy preventing the heat from escaping would perhaps raise the temp another 20 degrees. Now the chip itself is generating heat internally, and there is not enough ambient temperature differential to dissipate the heat, so the temperature will rise even further. The chip is flat black - pretty much the ideal black-body heat absorber in direct sunlight. Once a certain tipping point is reached, I would suspect the chip's internal temperature would climb uncontrollably. I'll also call DFI & Spectrolutions and see what they think.

Ciao,
Bruce

I'm assuming that where the foam is protruding is the top. So, if you slighlty spin the foam clockwise, you should have a level frame with that perspective. Do you think the metallic particles in that paint could affect the Rx?

It is perfectly level as shown. The camera is pointed at 45 declination from the horizon, and pointed straight forward (parallel with motor #1's mount). I'm going to shoot some test video this afternoon.
I spent quite a bit of time today on the phone with various people at DFI. Spectrolutions is in the middle of a move, so I haven't heard back from them. DFI tech support said they didn't paint the canopies because of possible "component problems", but nobody could quantify their concerns. I don't see how it could cause "interference". Flying it so far has been OK.

As far as the overheating problem, DFI believes it's not the controller, but a problem inside the layers of the PC board. That doesn't explain why it would spring back to life so quickly when cooling just the processor, but it doesn't rule out something in the immediate vicinity of the chip.

Is the configuration of the camera in your original image the way it was in the video in post #1? Not knowing the top of the frame is what's confusing me. So is the the blue dot considered the top of the frame? If not, is the yellow the top? If the blue dot is the top, then by simply rotating the foam clockwise from the blue to the red should give you a level frame. If red is the top, then never mind.

Yea, that picture is pretty confusing. Here are some pictures taken from directly in front, at a 45 degree angle looking down the camera mount arm, and directly from the right side. The last picture is how I aligned the camera. First the camera pod was set to point downward at a 45 degree angle. Then the SAVS was set on the table with its arms horizontal/perpendicular to the seam on the floor. The T-square was placed perpendicular to the seam, and directly along the forward axis line of the camera (7 1/2" right of the DF center line). The camera was adjusted so that the seam was horizontal and the camera was looking straight forward, down the ruler. This angle seems to be a very good compromise, and hopefully we won't have to change it too much. I'm editing some video bits shot with this angle and I'll post them in a little while.

Okay, so the way the camera is positioned in those images is NOT how it was positioned in post #1's video, right?

Correct. Here's some video shot with the most recent camera alignment. The 45 degree angle is very intuitive - it's pretty easy to guess what the camera is seeing, even without using the video goggles. I like the view angle, too. It's looking more downward, but has enough angle to show dimension. Also it's easier to fly over something and look back down at it from the other side.

http://www.youtube.com/watch?v=8PlmhnDTjzE

That was much better and it looked more like a news helo looking down at a sharp angle while over that house. So, the reason it was crooked in the first video is because you tried to align by sight or guess and didn't quite get it.

Thanks. The reason it was crooked in the first video is because we changed the downward tilt of the camera just before the flight. It had been carefully calibrated level/straight prior to the demo, but when we changed it to look more downward it threw everything else off. Because of the 45 degree geometry of the camera mount, the horizon and forward alignment must be reset every time the upward/downward angle is changed. In other words there are three interactive variables - change one variable and it affects the other two. I'm not sure DFI took this into account when they designed the SAVS. I would think that the simple ability to adjust the camera declination without messing up the horizon line would be a rather rudimentary design constraint. I like this 45 degree down-tilt angle, so I'm hoping we won't have to change it now that we've got it dialled in. However in a mission environment, that can't be guaranteed.

Oh yes, I spoke at length with them about these kinds of manual adjustments not being precise or easy enough. But any additional mechanics in the ball for auto leveling in any orientation adds weight and because it's suspended out there, that creates balance problems and on.. and on.. and on.

One thing to be careful of when rotating the entire camera arm is to remember that those small wires come out of that carbon fiber rod at a sharp 180 degrees at the edge of that rod inside that rubber fitting. If you twist that rod back and forth enough, the wires could eventually chafe, short out or be cut.

That's true. You can't disassemble and rebuild the assembly easily either, because they cement the pod to the CF rod when they manufacture it. I made the mistake of pulling the whole foam camera thingy out of the front of the pod once and I had a heckuva time getting it back in! I'm not sure what your options would be if you happened to break any part of that assembly.

Spectrolutions has been moving their business, but Mike Dammar called me today (Saturday) about the overheating problem. He says that this problem has not occurred in the past and they are anxious to get their hands on the board so they can identify the failure. They're mailing me a new controller in exchange for mine at no charge. It would be hard to find a company that has the level of customer service that these guys do.

Now I just need to figure how to detach the receiver after using (Old Man) Mike's super-duper adhesive on the board!