Tell us more about your plane and the mission.

http://www.rcgroups.com/forums/showthread.php?t=1035810


John,

It is the Catalina III that I've been perfecting for that last couple of years. It is intended to fly from Catalina Island to Cabrillo Beach near Long Beach. I'm begining 20 mile + testing next week. Thanks.

Jeff

Originally Posted by workshop

Bypassing your power buss shouldn’t require any skill since it just means rerouting red servo wires but I don’t feel (I say “feel” because I haven’t measured) the peak currents in a sedate 4 channel glider will present much of a load. I’ll have to stick with the digitals because the advantages are compelling and have already been installed into this plane for a year (7 flights).

PICOPILOT is perfectly capable of driving digital servos, the concern is in providing a clean power supply for the autopilot sensors. Since linear regulators can't boost voltage, once the input falls below ~5.2v, the autopilot's regulated voltage falls below 5.0v. It's the integrator's responsibility to provide a power supply that never falls below 5.2v (from voltage transients). Power supply noise is a function of transient loads (servos) and the power source impedance and transient response. The ultimate low impedance power source would be a 6v battery in this case. But, most PICOPILOTs are used with electric propulsion and that means that the integrator will probably run the autopilot from the ESC-BEC and that should work OK if the bus loads are low (small analog servos). We've received several reports of sensor noise when PICOPILOT is used with a ESC-BEC and our recomendation has been to use a seperate 3A BEC. That usually provides better transient response and a more stable power supply.

Originally Posted by workshop

http://www.rcgroups.com/forums/showthread.php?t=1035810
It is intended to fly from Catalina Island to Cabrillo Beach near Long Beach. I'm begining 20 mile + testing next week. Jeff

I checked out your Catalina thread and it looks like a challenging mission !
I couldn't tell if you intend to fly a one-way mission (20miles) or round-trip (40miles). What is the max endurance of your plane with the new battery ?

Endurance missiions are much more representative of how UAV's are used for commercial and military applications. Those operators usually expect a minimum of 1 hour endurance from their air-vehicle. "Time in the air" is what working UAV's are all about.

I've been involved in few "endurance missions" with UNAV and I've found them to be much more challenging (and exciting) than flying figure-8's around the RC field. The focus on the autopilot quickly changes from manueverability to reliability. Features like programmable altitude and speed aren't nearly as important as the navigation accuracy and reliability. These endurance missions have also shown me the value of telemetry. It's much safer and less stressful on the operator when he knows where his plane is at all times. We've done several missions over water and have found the wind is usually a major factor. We've had cases where the wind was near zero at the launch and then quickly increased to 10mph+ only 15 minutes into the flight.


Even if you succeed with a one-way flight (20 miles) that would beat UNAV's Baker Lake flight of 17miles.
Good Luck !

Originally Posted by John_Suggs

"Time in the air" is what working UAV's are all about.

I've been involved in few "endurance missions" with UNAV and I've found them to be much more challenging (and exciting) than flying figure-8's around the RC field. The focus on the autopilot quickly changes from manueverability to reliability. Features like programmable altitude and speed aren't nearly as important as the navigation accuracy and reliability.

"Time in the air" is important, but I would say payload is what working UAV's are all about. They (the boss figures) don't care how long the plane is in the air if the sensors are malfunctioning.

Accuracy and reliability are important, yes. But don't negate how important programmable altitude and speed are. Besides, it's not like you have to sacrifice one to get the other.

Several people have asked me why Picopilot doesn't use a 5Hz GPS like most the other autopilots on the forum. Picopilot uses a SiRF-III (20 chan) 1Hz GPS receiver with a RS232 interface.

It's a popular misperception that a 5Hz GPS is more accurate or that they acquire faster than a 1Hz GPS - not true. The only reason to use a 5Hz GPS is to improve groundtrack accuracy during a turn. GPS latency is caused by the delay in calculating the GPS solutions and the NMEA output. In other words, the autopilot is receiving track data that's at least one output sample old. Typically, track latency causes the autopilot to over shoot the turn-out heading. Picopilot uses a 1Hz GPS but the autopilot also limits the turn-rate to 20deg/sec so the max track error during a turn is limited to 20deg. That's not a siginificant navigation error and the autopilot can get back on the correct track in a matter of seconds. A 1Hz GPS output rate is a much more significant problem if the turn rate is faster than 20deg/sec. For example, a 90deg/sec turn-rate with a 1Hz GPS could cause a 90deg turn-out overshoot. That would result in a S turn or loss of heading control. Most of the hobby autopilots are setup for manueverability (fast turn-rate), that's why they require a 5hz update rate from the GPS.

There's a another problem with the OEM GPS receivers that Sparkfun sells - they use a TTL interface. These OEM receivers are designed to be embedded in products, not used as stand-alone parts. A TTL interface is supposed to only be a few inches of circuit board traces, it's never supposed to be used with several feet of unshielded wire. Unshielded wires are very vulnerable to RFI, especially with onboard transmitters. The interface wires are usually connected directly to the microcontroller UART pins giving any RF a direct path to the heart of the autopilot ! An RS232 interface is much more robust and is designed to protect the host microcontroller from RFI. That's why you'll find that all of the high-end autopilots either use a RS232 GPS interface or they mount the GPS receiver near the autopilot board and use a remote antenna. In simple terms, it's poor design practice to use a TTL (or SPI) interface outside the host enclosure.

Originally Posted by John_Suggs

Several people have asked me why Picopilot doesn't use a 5Hz GPS like most the other autopilots on the forum. Picopilot uses a SiRF-III (20 chan) 1Hz GPS receiver with a RS232 interface.


There's a another problem with the OEM GPS receivers that Sparkfun sells - they use a TTL interface. These OEM receivers are designed to be embedded in products, not used as stand-alone parts. A TTL interface is supposed to only be a few inches of circuit board traces, it's never supposed to be used with several feet of unshielded wire. Unshielded wires are very vulnerable to RFI, especially with onboard transmitters. The interface wires are usually connected directly to the microcontroller UART pins giving any RF a direct path to the heart of the autopilot ! An RS232 interface is much more robust and is designed to protect the host microcontroller from RFI. That's why you'll find that all of the high-end autopilots either use a RS232 GPS interface or they mount the GPS receiver near the autopilot board and use a remote antenna. In simple terms, it's poor design practice to use a TTL (or SPI) interface outside the host enclosure.

Interesting comments John, I would be more interested to hear what the experts have to say about this!

Basically John admitted that low Hz GPS is reducing agility,
I would add: even more when you have no yaw gyro (but this is not the case here).

Therefore gyro-less autopilots must avoid overly tight circles and this is why one of the first things that should be implemented for them is loiter navigation that reduces the turnrate and bank to the limits when GPS will not be completely fooled.

I would add that a cheap 5Hz GPS has less noise during maneuvers than cheap 10Hz GPS, but the latter gives better FPV experience.

I would add one clarification (based on my perspective of course.)

Hobby UAV's don't *require* 5hz gps's to operate. Hobbiests would happily tune their systems around a 1hz gps if that was all that was available. But now we have 4 and 5hz gps's that perform quite well and are inexpensive. Why not use them? The faster you receive a position update, the higher quality your navigation will be. Why not use a 5hz gps? We have embedded CPU's available that cost almost nothing with plenty of power to parse all that extra data. These CPU's have serial ports that support the higher baud rates required for the higher update rates. I can't think of any reason why any one wouldn't want to use these newer better gps's in their systems?

Regards,

Curt.

Thank you both for the education, have you any views about a remote type GPS unit connected to the AP via interface cable being more vulnerable to RFI than the system John describes?

Spitfire,

It is a noise floor issue. An unshielded cable carrying 5V data signals has a less dynamic signal that one carrying 12V signals as is the case with RS-232.

Jeff

Originally Posted by kbosak

Basically John admitted that low Hz GPS is reducing agility

I said: "Picopilot uses a 1Hz GPS but the autopilot also limits the turn-rate to 20deg/sec" ... most of the hobby autopilots are setup for manueverability (fast turn-rate).

A slow turn-rate autopilot isn't necessarily inferior to something with a faster turn-rate, it just means that they are optimized for different types of missions. Picopilot was originally designed for working UAVs where reliability during long missions is more important than the ability to fly tight manuevers. Missions like Jeff's Catalina-III and IceBear's Harbor flight are good examples of long distance flights where reliability is important and manueverability isn't. I should point out that several Picopilots performed the 200m F8 circuit with no problem.

I think you'll find that when new autopilot operators get bored with flying F8s around the RC field, they'll be looking for something more challenging like workshop and IceBear did.

Originally Posted by John_Suggs

I said: "Picopilot uses a 1Hz GPS but the autopilot also limits the turn-rate to 20deg/sec" ... most of the hobby autopilots are setup for manueverability (fast turn-rate).

A slow turn-rate autopilot isn't necessarily inferior to something with a faster turn-rate, it just means that they are optimized for different types of missions. Picopilot was originally designed for working UAVs where reliability during long missions is more important than the ability to fly tight manuevers. Missions like Jeff's Catalina-III and IceBear's Harbor flight are good examples of long distance flights where reliability is important and manueverability isn't.

Why would a 1 hz gps be more reliable than a 5 hz?

I should point out that several Picopilots performed the 200m F8 circuit with no problem.

Are the X legs of the 200m F8 circuit 200 meters long?

I think you'll find that when new autopilot operators get bored with flying F8s around the RC field, they'll be looking for something more challenging like workshop and IceBear did.

I have been playing with autopilots for many MANY years and still find that flying a really good pattern very pleasing. Maybe it's because I remember when just completing a flight was a pretty good accomplishment .

Later;

D.W.

Originally Posted by John_Suggs

GPS latency is caused by the delay in calculating the GPS solutions and the NMEA output. In other words, the autopilot is receiving track data that's at least one output sample old.

Actually the delay is less than (not at least) one output sample old. For a 1 Hz output the total delay might be ~0.3 seconds. Obviously the delay will be less for a 5 Hz solution.

Why not go straight to a 100 Hz solution? More noise. The longer you can sample the GPS signal the better the signal-to-noise ratio will be. A 1 Hz solution is more robust than a 5 Hz which is more robust than a 10 Hz. Where's the ideal performance/cost cutoff? For low cost GPS in 1995 it was 1 Hz. For the past decade it's been 5 Hz and 10 Hz is starting to become the norm.

If a 1 Hz GPS does the job then fine, but cost-wise and performance-wise there's really no reason not to use a decent low-cost 4 or 5 Hz GPS receiver like a Ublox.

Originally Posted by d_wheel

Why would a 1 hz gps be more reliable than a 5 hz? D.W.

No one claimed that a 1Hz GPS is more reliable than a 5Hz, not on this thread anyway.

Originally Posted by ehx

Actually the delay is less than (not at least) one output sample old. For a 1 Hz output the total delay might be ~0.3 seconds. Obviously the delay will be less for a 5 Hz solution.

We were talking about GPS NMEA data at the autopilot. The GPS receiver computes ground track using the previous LON/LAT and the current LON/LAT so the calculation is at least one cycle behind. Using a 1 Hz GPS means that the autopilot receives track data that's at least 1 second old and a 5Hz GPS at least 200ms old. I don't know where you got ~0.3 sec.?