Discussion in 'Instruments / Avionics / Electrical System' started by Aerowerx, Oct 6, 2014.
Why use servo-driven dials instead of LCD or OLED screens to do the same thing?
Displays just aren't the same. Lighting washes out OLED. LCD can be defeated by wearing the wrong sunglasses. One wrong thing in the cockpit and say your clipboard slams into your beautiful LCD.. and you're looking at rainbows instead of your altitude.
I don't trust anything with a backlight, or "self generated" light in high brightness situations. Most planes are flown day-vfr, so having a display that is "mostly suited for bright light" is better.
On the other side of things, many LCD and OLEDs can't be dimmed enough at night to preserve night vision. (driver issues with the CFL, or backlight LEDs, or just the non linearity of LEDs being an issue for the oled)
Because servos are somewhat less reliable than one would like, having a small LCD scrolling "are you sure" numbers would be a good idea. A little hitachi 2x16 display is $3. While a whole multi color LCD is a whole heck of a lot more money.
Some of the newer e-paper technology is offering high contrast and fast refresh rates. A white-on-black e-paper display might work for direct sunlight.
None of it is as simple or proven as servos, or syncromotors, or steppers. There's a lot to be said about displays with 3d, and shadows. Seeing the needle move (which epaper is very bad at) is something you can catch out of the corner of your eye. I'd go LCD or OLED over e-paper, from a safety perspective, as animation ~works~ on those.
Whatever you build, it should be cheap, easy to build, robust, and light weight. Trade off: single point of failure.
Military grade or precise is not necessary on something say like an ultralight. Some people fly such lawn chairs with no gauges... at all. So even something that would be sloppy inaccurate to some degree would be better than no gauge at all. Your brain will auto adjust to the error and do a mental correction.
Myself, I'd prefer gigantic 3 or 4 digit LEDs over an analog gauge any day. And used for almost everything, assuming you could recess them and make them readable in bright sunlight. One, there is no guess as to a real number compared to a needle on an unmarked dial. Two, if I were to somehow lose my glasses in flight, I'd at least have a bats chance in hell of reading them.
I could take a picture of our car dash and show you the worst gauges of all time. Needles, with a number marked at low end and high end of the gauge, with no marks between. Ditto for the digital dash... just stereo led bars that go up and down for gas, rpm, etc. Any kind of number would be superior to both. One LED bar to empty doesn't tell me a lot, but watching my fuel tick down from 010.0 to 007.0 % would tell me more.
Probably most important, your gauges should 'flash' or alert you when they are outside a low and/or high preset user configurable range of 'safety', to draw your attention to it and alert you that something isn't right. I have never in my life seen any kind of gauge do that. And then I just want to just face palm myself for humanity.
For example, a car thermostat typical restricts water flow to run the engine at around 170F. Therefore, a water temperature gauge should flash at anything below 100F (hey dummy, let your car warm up to operating temp before you drive it), or above 190F (hey dummy, your engine is heading fast towards overheating, and blowing its radiator cap past 220F).
I'm still a big proponent of hybrid gauges that use analog output with digital input and control. Enables linked systems, redundant and backup instrumentation (you could have a bluetooth link from your airplane to your phone for an emergency backup EFIS even) and yet in normal flight conditions you get a simple, clean, easy to read gauge that doesnt just look like a TV with a glut of extra crap everywhere.
For some real-world demonstration of what I'm talking about, here's an almost perfect application for bike riders:
Imagine if the primary gauges could all be lime this Omata, but with the benefit of them being linked to a central bus that makes each gauge even cheaper as they don't need their own built-in CPUs. And that central unit also might have a small screen (or bluetooth link to controller app for iOS/Android) for adjusting settings and monitoring status and a backup display.
Barring straight-up augmented reality/HUD goggles, a set of hybrid gauges and a basic HUD on the glare-shield seems like a great setup.
And IMHO they shouldn't even be noticeable unless they are notifying an out of range condition. By out of range I also include a rate of change that will put the changing parameter out of max/min range in 3 minutes. Once either condition is met an analogue gauge is the best way to present data to the human brain regarding rate of change and proximity to the limit. We vary rarely need precise numbers and when we do a simple on/off light with the preset limit is my preference.
If there was an inexpensive digital display that was really daylight visible in all conditions than it might be able to replace SBS's OMATA style gauges.
These are cheap and easy to use.
A gauge out of something other than a car will have intermediate values. And a "modern" airspeed gauge is actually pretty easy to read without calibration marks--stall is somewhere around 2:00, approach around 3:00, cruise around 8:00 or so, and Vne around 10:00.
Having a digital readout is useful sometimes, assuming that your display is reporting as accurately as it is displaying (I'm not aware of a fuel gauge on any small airplane that could anywhere close to accurate). But a pure digital readout like that is actually hard to discern trends. I've driven cars that had only a digital speedometer readout, and it's much harder to maintain a desired speed than with an analog needle. That's why even modern EFIS displays have digital readouts with a scrolling tape and sometimes a "drum" for the digits.
Car instrument clusters are often poorly designed, non-standardized, dumbed-down, and driven more by styling and aesthetic concerns than human factors engineering. But they're cars; it doesn't matter much.
Again, very common on EFIS displays. And the mechanical-analog equivalent is colored bands around the perimeter of the gauge, with (on fancier older airplanes) caution and warning lights.
If it's cheap enough, and light enough, you can carry a spare.
I'd argue with you on that. Numbers, especially LED/LCD style numbers, are easily misinterpreted. Especially with how most LED displays are run (that is mutiplexed) a quick glance can give you an extremely errornous reading.
A quick glance at an analog guage won't lie. LED bar graphs can be useful, take a look at belites offerings, most of their guages are points, so I don't think have the multiplexing issues. No guage should be trusted, on it's own, down to single percentage points.
Telling the difference between an 8 and a 0, or a 1 and a 7 in a bunch of numbers is going to be very difficult without your glasses. An analog guage (in your case, one big enough or close enough... ) would give you a high contrast mark that's going to be in the same place all the time, if the value is right. You can't mistake that.
That said, you're messing with ultralights, you shouldn't ~need~ guages.
Totally. RGB guage lights would give you a whole lot of options for that. Or even just the standard red/green leds that are ..while not quite a dime a dozen, pretty stinking cheap.
100pcs 5mm Dual Bi Color Polar Changing Red Green LED Diffused LEDs 2 Pin LED | eBay
Your car has at least one guage that does this. The check engine light, if it's flashing, you need to shut down now.
I should break out the arduino tonight shouldn't I...
This is exactly why I don't like e-paper. E-paper refreshes, even the new, non flash type, is still slow and distracting. And why I have ~some~ issues with anything that has a scanned display. If you move your head fast, scanned displays tear and take your attention. Especially in the peripheral vision, which is able to pick up much faster changes in brightness than your center vision can.
This has been discussed before, perhaps even earlier in this thread, so I repeat....
The problem with using servos is that there is no indexing. In other words you will have no idea where it is actually pointing. To get around this you attach a potentiometer to it, and then have to run the output of the potentiometer to an A/D converter to indicate the position. Then you have the problem that the potentiometer can't rotate 360 degrees---maybe 300 degrees if you are lucky. OK for a tach or airspeed, but not for an altimeter (if you want one that looks like a "real" one, that is).
With all that complexity, you might as well go ahead and use an air-core motor, which is self indexing. Just feed it with two signals that vary with the sine and cosine of the desired pointer angle. I have even come up with a scheme (on paper) of making a dual-pointer air core motor for use as a two-pointer altimeter.
Servos are "quite good" at self centering, and hobby servos contain their own potentiometer. What makes them a servo, is that you can depend on their repeatability, and position. You can also get full turn, and multi turn r/c servos, if that's what you need. A belt and pully setup could even give them a self reseting capability, if you so desired.
The only reason I don't talk about air core motors is that they're more difficult to interface with the rest of the system. Though looking at that guys website, I think I could do the servos for 1/4 or less of the weight.
Sail winch servo - 3.5 rotations: https://www.servocity.com/html/hs-785hb_3_5_rotations.html#.VyEdbuIrKCg
There seems to be a theme running through the recent discussion.
All the ideas are good, but are you designing for day VFR or IFR? If IFR, then you have to have certified instruments, even for a homebuilt IIRC.
Now, for day VFR you don't need much. Legally you don't need any instruments in a homebuilt. I recall reading about a place in Wisconsin where you can take 2 weeks and get checked out in a J-3. For your "final exam" they cover up all the instruments.
For my build all I am going to have is airspeed, altimeter, tach, oil temp/press, and compass. Probably a site gauge for fuel.
So my question is: Are you flying the plane, or is the plane flying you?
Having just bought an Excelvan android tablet, I looked on their site and saw a range of HUD gauges for autos which may or may not be of interest..
My brother just told me about some new hardware/software that is making all of this even easier for these kinds of home brew instruments. Thought I'd pass on the info. I'm having to rethink the panel for my Quickie. The box of VDO gauges I'd collected long ago become dunsel.
Almost drag and drop programming.
With the above and other Raspberry Pi modules it's getting almost as easy as assembling a Lego set.
What bothers me about this is if you watch the rabbit in the video on the linked page, it appears to be going through some uncommanded yaw. That kind of casts doubts on the reliability of the data.
Hadn't noticed that. Given the chip is from Bosch I'd suspect it's likely an external software problem? That can be fixed. Even if there is an internal software problem in the "fusion mode" (I haven't read the whole data sheet) the raw data is still available and the user can set up their own algorithms to calculate position.
It's also possible that the unit self calibrated in a less than ideal position for the demonstration and the software driving the display was making assumptions about the data actually corresponded it's these presumed axes.
But the point I was trying to make is that this is all getting pretty easy for the average HBA builder at an ever increasing rate. Makes ADS-b look pretty stone age by comparison.
Separate names with a comma.