Electronic level/horizon indicator

[Fabrizio Sciarra SOC ACO]

Steadicam newbie/non-newbie roboticist here...

 

You can get very small 9 axis IMUs that give you 3 axis accelerometer, 3 axis gyro and 3 axis magnetometer readings in real time. These are used in UAVs (flying robots basically) as a matter of course these days, where the need to tell the difference between what direction really *is* down and the apparent direction due to local acceleration is pretty much an unavoidable requirement, or you end up with the UAV going into an uncontrolled spiral and a crash. It would be pretty straightforward to build something based on one of those. I might try it myself, actually.

 

Another buyer here

[Sarah Thompson]

Hmmm... looks like I should probably be knocking up some kind of prototype.

 

One question, for the experienced operators -- how would you like to receive the information?

 

Heads-up display overlayed on the video? If so, composite, HDMI, HDSDI, SDSDI, something else?

Separate display?

Something else?

 

Oddly enough it's the display part of this that would be the trickiest rather than the IMU and the filtering of the telemetry it generates.

[JobScholtze]

The transvideo level works just fine. ( there is a upgrade now that helps/prevents the issue we had when fast moves or pans push'd the level into the corner, thats all gone now )

Betz made a very nice extra for that level. With his tool, all electronic's and button's AND sensor is now in the middle / under my post. Very helpfull. As the betz topstage has a small level, its so easy to calibrate the electronic one. Now becose the sensor is under your post and not inside the monitor i never have to calibrate it again. It stay's calibrated forever.

 

 

My 2 cents in this unbalanced matters ;-)

[William Demeritt]

An overlay to the HD-SDI would be fantastic, especially if it had negligible frame delay (or none whatsoever). An HDMI version might benefit a few people, but I think you'd find them less interested in spending the money. An HD-SDI version, however, would sell like hotcakes!

 

A separate display option would be nice as well, something similar to a bubble level that you can physically place on the monitor, matte box, wherever...

[Thomas English]

To be honest when your first starting to build this thing make an LED readout with a green LED in the middle and orange and red ones either side. This can be velcroed to the top of any monitor and the sensor to the bottom of the sled.

[Sarah Thompson]

To be honest when your first starting to build this thing make an LED readout with a green LED in the middle and orange and red ones either side. This can be velcroed to the top of any monitor and the sensor to the bottom of the sled.

 

That was what I was thinking. I could do HDSDI overlay, but it would need an FPGA and is not something I could make from off-the-shelf parts. I'd need to design a PCB, have it fabricated, etc. There are reference designs for HDSDI input and output, so it isn't crazily hard starting from that point, but it's still quite a bit of work. A line of LEDs is drastically simpler.

[Fabrizio Sciarra SOC ACO]

To be honest when your first starting to build this thing make an LED readout with a green LED in the middle and orange and red ones either side. This can be velcroed to the top of any monitor and the sensor to the bottom of the sled.

 

Like it!

[James Davis]

Sounds great, please put me on the list for one if/when you make it.

 

Also a couple of points:

 

Housing should be machined metal for durability

 

Power cable/connection should be external not built in so we can carry spares if a cable gets trashed.

 

Should be easy to calibrate

 

My two pence

[Sarah Thompson]

Sounds great, please put me on the list for one if/when you make it.

 

Also a couple of points:

 

Housing should be machined metal for durability

 

Power cable/connection should be external not built in so we can carry spares if a cable gets trashed.

 

Should be easy to calibrate

 

My two pence

 

I could machine the housing from a solid block of T6 aluminum, which should be about as strong as anyone would ever want.

 

As for power, I could use military-style connectors with a locking ring. Hard to break, very reliable.

 

As for controls, what I was thinking was just having a single button and a knob. The button would be push on/off, with holding the button down for 4 seconds (say) putting the box into calibrate mode, where it would then wait until everything settles down, then take an average reading over a few seconds and then go back into normal run mode. The knob would set sensitivity.

 

So basically, in use you'd just tweak the knob to set the sensitivity you prefer. To calibrate, trim out the rig using conventional means (spirit levels, visual reference, digital angle gauges, whatever you prefer), then hit the button for 4 seconds, back off and let it do its thing. I'd have it write the settings to EEPROM, so next time you power up it'll all be there like you left it.

 

Next question is: put the sensor in the box, or have a separate sensor module. The best place to mount it would be as close as possible to the gimbal from a dynamics point of view, but it could potentially go anywhere.

 

Last question: how many axis? Roll only, or roll & tilt? Would you want separate controls for each axis?

 

One idea: I could make it figure out if it's in low mode automatically so it doesn't need recalibrating.

[Jess Haas SOC]

You can get very small 9 axis IMUs that give you 3 axis accelerometer, 3 axis gyro and 3 axis magnetometer readings in real time. These are used in UAVs (flying robots basically) as a matter of course these days, where the need to tell the difference between what direction really *is* down and the apparent direction due to local acceleration is pretty much an unavoidable requirement, or you end up with the UAV going into an uncontrolled spiral and a crash. It would be pretty straightforward to build something based on one of those. I might try it myself, actually.

This has been on my to-do list for a while now. Even have some of the parts in my shop.

[Peter Sheppard]

Put me on the sales list.

I'd say roll only, I usually tilt the sled for framing, headroom etc and wouldn't really want another LED moving about distracting me.

For the housing question, I'd like the box with the sensor to also have the controls on it which I could mount onto the top-stage so close to the camera. Then a cable to a small LED strip I could mount anywhere above or below my monitor. It would be great for the cable to be a BNC so it could be run down one of the extra lines down the post, and the power to be a lemo so pretty standard to everything else.

I'd reckon if you made and priced these well you'd be able to retire to the Bahamas on this invention.

Shep.

[Nikolay Kerezov]

Call me crazy , but I would like the sensor at the top of my sled where the camera sensor/film plane is (as in the Christian Betz's sled). Is in it the camera that captures the wrong horizon?

If you are using bottom of the sled method ,it should be as close to the post (the best would be right under or inside)

[Peter Sheppard]

Yeah that's what I said. On the Top-Stage, close to the camera.

[Sarah Thompson]

Hmm, doable. So sensor/main CPU/control box at the top, machined T6 case with a few 1/4-20 holes and maybe some unthreaded 1/4" holes for bolting through to give you some mounting options, display connector via 75 ohm BNC. I'd probably recess the controls into one edge of the box slightly so they are somewhat protected (and would look cool too, I suppose). I could do a phantom power-like trick to send power and data for the display down there so you only need to hook up power to the box at the top of the sled. Tricky, but doable. What are people's preferences for power connectivity? Would you want a loop-through?

[Eric Fletcher S.O.C.]

Something to consider, if you are using a HD-SDI signal and you pass it thru your box you are going to introduce considerable delay. To prevent that you have to have your level interface part of the LCD driver. Otherwise you are going to introduce at least 3 frames and up to 6 frames of delay.