Another one from Sparkfun, which is quickly becoming a serious UAV parts supplier: the power monitoring sensor board that Dean Goedde uses for AttoPilot. If you want to add motor current and battery voltage measurement to your autopilot, this is a good way to do it. Analog output, so you'd want to add it to a port with an ADC (ArduPilot has six free ones so it would work well with that). Oddly expensive at $26, given what's on the board, but perhaps that reflects the low production run. Woohoo! Now $12.95! (and half sold out immediately). Go Dean!
There are lots and lots of neat scenarios like that that you can come up with and try when developing an autopilot at the code level. For me that is the fun of all this!
The kind of thing you describe is along the lines of how I started with Atto in 2007. In the end, GPS and power sensing are the types of measurements that would only work with a very self-stable plane like a high-wing polyhedral motor glider... you might think a telemaster or LT-40 trainer would be OK with this... and they might fly fine for a while, but your margin against instability and spiral dive are lower than you might think. I speak from experience. I have seen even the Miss2 flown with GPS-only autopilot in lite winds have a problem when autopilot was applying rudder and a puff of wind hit just from the right angle and at wrong time... the two together pushed roll angle too far and plane went into a spiral dive.... I realize that you are talking more about pitch and climb rate control and my example is centered around roll and heading control... but my point is that unless your system has some idea of attitude with low time lag, the risk is high.
You would need to do experiements and ather real-world data from flight to see if the power draw is significantly different between the two scenarios you describe.... plane speed matched to propeller versus more of a stalled flight. The potential problem is the two conditions might not be well seperated... the observability could be low.
My fatal IR experience had nothing to do with power or airspeed. Also, that plane had very little dynamic stability with no dihedral and very small size... it also could easily cruise at 60 mph (100km/h)... a little rocket. The power scenarious you propose I REALLY doubt would work in this plane.
Gary - that yellow/blue/white airframe is long dead! It is what I called the "Python V-tail on RCGroups". It died when I decided to fly thermopile autopilot when heavy raindrops were already coming down... LOG file showed during flight why plane suddenly nosed up +90 pitch... pitch axis thermopile data max'd out instantly. I have never seen this on any other flight, and conclude that front sensor got a raindrop. Plane was 400 meters away, so RC control was impossible... plane freaked and spiralled to its death. That was in June 2008. I have the CAD .dwg files, so might make another copy of this plane. It flew great at 12 ounces and a little 30 watt brushless, and awesome on a 200 Watt brushless at 16 ounces (110 km/hr!)
bGatti wrote: "Assuming you get very little voltage diff across the shunt - what do you use to bring the voltage into the range of a 5v ADC?"
OK, I see what you are asking now... the current sensing is done by Texas Instrument INA-169 which measures voltage drop across the shunt, does it's thing then give an analog output. The voltage determination is done seperately by precision divider which measures from high side all of the way down to GND... not merely across the shunt.
bGatti - Nathan at Sparkfun has agreed to lower the Sparkfun re-sell price to just $12.95. I am now selling them to him at cost or a slight loss, to use this as an advertisement for AttoPilot in hopes that maybe 1:100 people might buy an AttoPilot system through this advertising.
The scaling is actually setup for 3.3V ADC, meaining at max stated volts and amps measured, the analog output is 3.3V on each of the 2 channels. Of course you can also measure it with a 5V ADC.
The voltage is done by a precision divider. I used a 10k/1k 0.1% tolerance divider in SOT-23 package and to raise the range a bit I added a precision 4.7k Ohm resistor on top of that. I made sure that the temperautre resistivity coefficient of the two resisto packages is the same, so that across the temperatures the output is a constant 1:15.7 ratio. So of the three resistors from Vdd to GND it goes 4.7k 10k 1k with the output tap between the 10k and 1k in the SOT-23 packaged part.
bGatti - Current is through a solder-flooded path, not a thin resistive board. This is not a proposed design, but rather fully field proven and suitable for use in very small models. See spec sheet posted on Sparkfun for "installation" details: The 12 gauge lead is fed through hole, and area up to shunt is solder flooded top and bottom of PCB. The GND-to-GND is also a low resistance solder flooded path through the large via hole in center pad.
While 130µΩ of a Hall sensor is higher than 500µΩ of my design, it's important to look at what this penalty is in the total cost model of size, weight, performance. At 90 amps, my very thin and lite sensor would have 4 Watts heat generated in the shunts... but at 90 Amps the typical RC plane would likely be running at least 6s LiPo which is over 25V, for total power of 2250 Watts. That's only 0.17% loss and 99.82% transmission. The 130µΩ Hall sensor in same scenario loss is 1 Watt, for 0.04% loss and 99.96% transmission. The loss difference will not be noticeable, as it equates in a 30 minute flight to 75mAh versus 18mAh... planes of this size are using at least 5000 mAh packs. I argue that even on high power systems, power loss is in the noise. Small models the difference could easily be in the 10mAh amount for a 30 minute flight. Because I like to fly 10 ounce MAVs with little fuselage space, this super thin sensor works great.
Seeing that none have sold in almost 24 hours (though I think at $26 it is a good value based on tested performance, size, and competing products) I have e-mailed Nathan at Sparkfun... this board is also an advertisement for my main product, so I proposed to Nathan that if he drops the Sparkfun price to $15 or less, I will sell it to him at my cost. After all, if just 1 person out of 100 ends up buying an AttoPilot system, then this was a good advertisement, and more people are likey to buy it as an impulse to try it out. It will work great in the low-cost ArduPilot on a spare ADC channel. To boot, making it low cost will enable more people to gain value from it in the ArduPilot.
uh huh, uh huh. $26 seems perfectly reasonable for anything tied to the insanely popular Attopilot. Sparky can still charge $69 for obsolete gyros which only cost $22 in bulk.
Comments
The kind of thing you describe is along the lines of how I started with Atto in 2007. In the end, GPS and power sensing are the types of measurements that would only work with a very self-stable plane like a high-wing polyhedral motor glider... you might think a telemaster or LT-40 trainer would be OK with this... and they might fly fine for a while, but your margin against instability and spiral dive are lower than you might think. I speak from experience. I have seen even the Miss2 flown with GPS-only autopilot in lite winds have a problem when autopilot was applying rudder and a puff of wind hit just from the right angle and at wrong time... the two together pushed roll angle too far and plane went into a spiral dive.... I realize that you are talking more about pitch and climb rate control and my example is centered around roll and heading control... but my point is that unless your system has some idea of attitude with low time lag, the risk is high.
You would need to do experiements and ather real-world data from flight to see if the power draw is significantly different between the two scenarios you describe.... plane speed matched to propeller versus more of a stalled flight. The potential problem is the two conditions might not be well seperated... the observability could be low.
My fatal IR experience had nothing to do with power or airspeed. Also, that plane had very little dynamic stability with no dihedral and very small size... it also could easily cruise at 60 mph (100km/h)... a little rocket. The power scenarious you propose I REALLY doubt would work in this plane.
You have to make a profit Dean.
Even if its a tiny one.
What you make will not be everybodys cup of tea, let them make their own. This is DIYdrones after all.
Cash flow is vanity, profit is sanity.
OK, I see what you are asking now... the current sensing is done by Texas Instrument INA-169 which measures voltage drop across the shunt, does it's thing then give an analog output. The voltage determination is done seperately by precision divider which measures from high side all of the way down to GND... not merely across the shunt.
The scaling is actually setup for 3.3V ADC, meaining at max stated volts and amps measured, the analog output is 3.3V on each of the 2 channels. Of course you can also measure it with a 5V ADC.
The voltage is done by a precision divider. I used a 10k/1k 0.1% tolerance divider in SOT-23 package and to raise the range a bit I added a precision 4.7k Ohm resistor on top of that. I made sure that the temperautre resistivity coefficient of the two resisto packages is the same, so that across the temperatures the output is a constant 1:15.7 ratio. So of the three resistors from Vdd to GND it goes 4.7k 10k 1k with the output tap between the 10k and 1k in the SOT-23 packaged part.
While 130µΩ of a Hall sensor is higher than 500µΩ of my design, it's important to look at what this penalty is in the total cost model of size, weight, performance. At 90 amps, my very thin and lite sensor would have 4 Watts heat generated in the shunts... but at 90 Amps the typical RC plane would likely be running at least 6s LiPo which is over 25V, for total power of 2250 Watts. That's only 0.17% loss and 99.82% transmission. The 130µΩ Hall sensor in same scenario loss is 1 Watt, for 0.04% loss and 99.96% transmission. The loss difference will not be noticeable, as it equates in a 30 minute flight to 75mAh versus 18mAh... planes of this size are using at least 5000 mAh packs. I argue that even on high power systems, power loss is in the noise. Small models the difference could easily be in the 10mAh amount for a 30 minute flight. Because I like to fly 10 ounce MAVs with little fuselage space, this super thin sensor works great.
Seeing that none have sold in almost 24 hours (though I think at $26 it is a good value based on tested performance, size, and competing products) I have e-mailed Nathan at Sparkfun... this board is also an advertisement for my main product, so I proposed to Nathan that if he drops the Sparkfun price to $15 or less, I will sell it to him at my cost. After all, if just 1 person out of 100 ends up buying an AttoPilot system, then this was a good advertisement, and more people are likey to buy it as an impulse to try it out. It will work great in the low-cost ArduPilot on a spare ADC channel. To boot, making it low cost will enable more people to gain value from it in the ArduPilot.
-
1
-
2
-
3
of 3 Next