Kestrel autopilot problem

[kostixx007]

Hi everyone!
I am participating in a project with the main objective to build a high/performance and low cost UAV airframe (comparable with the available solutions on the market). The size of UAV is very close to an Aerosonde, calculated weight is around 12 Kg.
All right, now about the problem. Kestrel autopilot was chosen for its good price, mainly. It took a lot of months to do all the paperwork and the communication with Procerus was extremely good and fast.
We have received autopilot around 3 months ago. I can say that I was quite shocked by the quality of some components. My opinion is that the components of the system that were outsourced are of the good quality and all the small bits, that Procerus decided to do in-house (simple soldering and assembly mostly ) - were of the low-DIY quality.
Okay, cosmetic issues are all right with me if all the system does work. It seems that everything is working, but we have made a shocking discovery.
In the mode when the RC transmitter is used for guiding an airplane (RC over comm. Mode), there is a delay in servo reaction to the stick movement . Delay could be of the order of 1/4 second. That means , for example, that if we have super fast digital servos with the speed of 0.11 s/60deg, then the delay is almost twice the time of servo movement. Okay, we have installed autopilot in ZAGI flying wing and tried to launch the model in the 'RC over comm.' mode. 3 times (out of 3) we have crashed during the first 3-5 seconds - because of the delay in servo movement. I need to say that the pilot is a professional RC pilot and ZAGI wing flies beautiful with the ordinary RC system.
Now, the interesting part. I am still trying to get support from Procerus. I can say that for the last 30 days I have sent them around 7 emails and I have received only 3 emails from them. They are just ignoring me - that is the only thing I can say . What they did said, however is that there is a delay of approximately 1/4 of the second and that everybody is happy with it.
The bad bit is that it will be dangerous to attempt a landing/launch with a delay in servo reaction of an UAV that weights 12 kg, going at the speed of at least 60 km/h and costs tens of thousands of dollars.

Some other notes and impressions:
a) Manual is not the best. A lot of uncertainties.
b) Impression is that system is thought for 'recreational use' in small park flyer/zagi wing airframes.
c) Customer support is useless a soon as you receive the product.


The only solution I see at the moment is using something like RxMux from http://reactivetechnologies.com/RxMux_files/RxMux%20V1.3.pdf
To switch between the 35 MHz receiver and autopilot.
But, unfortunately it is at backorder at the moment.

Does anyone expirience the same problems ? What are the solutions ?

P.S. pictures of Kestrel autopilot and zagi wing for autopilot testing .

[airmcn_3]

OH crap! I'm about to spend $24,000.00 with them.........


Anybody got a better suggestion on autopilots? Cloudcap is very nice just waiting on a return E-mail on cost and functionability.

Thank you so much for the heads up.

Chris

[reedchristiansen]

Kostix,

Sorry to hear about your bad experience. This is extremely uncommon.

I think the root of your problems are the European ac4868 modem. They have an extremely large latency which makes it almost impossible to fly in RC mode. It can be done, but only if the airplane is in trim and rate damped. Those modems are meant to be used in a vehicle where the gains are tuned and the RC controller is no longer needed.

For gain tuning (where RC over ride will be required) I recommend switching to the aerocomm ac4424 or even better to the maxstream 9xtend. This will more the latency in half. Still noticable, but very useable - even on smaller aircraft. I am sure we can get you a set of use able modems for your troubles.


As far as the "hobby like" workman ship - almost all of the items are hand made to save as much weight as possible:

1. GPS : uncased with a very thin ground plane. We could go with a housed ant with coax, but you are talking about doubling the weight. Customers with bigger planes sometimes use the garmin GPS 18 - not a bad solution if you can carry the weight.

2. Dipole ant: Again hand made with nitinol wire. Very light. Again, we sell a bigger/nicer antenna (maxstreem.net - standard 900 mhz with plastic housing). It is much heavier, but not a bad way to go on bigger airplanes. BTW, even though that dipole is not much to look at, it performs very well and was hand tuned using a network analyzer.

3. heat shrink covering on autopilot: - saves a ton of weight. We are building a nice carbon box with a milspec connector on it - similar to cloudcap and micropilot - great if you can carry another few ounces.

Are nitch really has been small and light airplanes (does not mean cheap necessarly). We have catered to this market by handmaking basically all the accesories (gps board/ant/cabling/enclosure) to keep the weight down.

However, as we get more and more customer flying bigger airplanes we are starting to provide more traditional antennas/enclosure/wiring harnesses. The all up weight with these items is about tripple the weight of the stuff you have, but it definatly looks nicer.

Sorry about the bad experience. I think once we get you a different set of modems with the lower latency you will start having a better experience.

Please PM me and I will send you my personal email so we can get this worked out.

Reed


Hi everyone!
I am participating in a project with the main objective to build a high/performance and low cost UAV airframe (comparable with the available solutions on the market). The size of UAV is very close to an Aerosonde, calculated weight is around 12 Kg.
All right, now about the problem. Kestrel autopilot was chosen for its good price, mainly. It took a lot of months to do all the paperwork and the communication with Procerus was extremely good and fast.
We have received autopilot around 3 months ago. I can say that I was quite shocked by the quality of some components. My opinion is that the components of the system that were outsourced are of the good quality and all the small bits, that Procerus decided to do in-house (simple soldering and assembly mostly ) - were of the low-DIY quality.
Okay, cosmetic issues are all right with me if all the system does work. It seems that everything is working, but we have made a shocking discovery.
In the mode when the RC transmitter is used for guiding an airplane (RC over comm. Mode), there is a delay in servo reaction to the stick movement . Delay could be of the order of 1/4 second. That means , for example, that if we have super fast digital servos with the speed of 0.11 s/60deg, then the delay is almost twice the time of servo movement. Okay, we have installed autopilot in ZAGI flying wing and tried to launch the model in the 'RC over comm.' mode. 3 times (out of 3) we have crashed during the first 3-5 seconds - because of the delay in servo movement. I need to say that the pilot is a professional RC pilot and ZAGI wing flies beautiful with the ordinary RC system.
Now, the interesting part. I am still trying to get support from Procerus. I can say that for the last 30 days I have sent them around 7 emails and I have received only 3 emails from them. They are just ignoring me - that is the only thing I can say . What they did said, however is that there is a delay of approximately 1/4 of the second and that everybody is happy with it.
The bad bit is that it will be dangerous to attempt a landing/launch with a delay in servo reaction of an UAV that weights 12 kg, going at the speed of at least 60 km/h and costs tens of thousands of dollars.

Some other notes and impressions:
a) Manual is not the best. A lot of uncertainties.
b) Impression is that system is thought for 'recreational use' in small park flyer/zagi wing airframes.
c) Customer support is useless a soon as you receive the product.


The only solution I see at the moment is using something like RxMux from http://reactivetechnologies.com/RxMux_files/RxMux%20V1.3.pdf
To switch between the 35 MHz receiver and autopilot.
But, unfortunately it is at backorder at the moment.

Does anyone expirience the same problems ? What are the solutions ?

P.S. pictures of Kestrel autopilot and zagi wing for autopilot testing .

[airmcn_3]

Kostix,

Sorry to hear about your bad experience. This is extremely uncommon.

I think the root of your problems are the European ac4868 modem. They have an extremely large latency which makes it almost impossible to fly in RC mode. It can be done, but only if the airplane is in trim and rate damped. Those modems are meant to be used in a vehicle where the gains are tuned and the RC controller is no longer needed.

For gain tuning (where RC over ride will be required) I recommend switching to the aerocomm ac4424 or even better to the maxstream 9xtend. This will more the latency in half. Still noticable, but very useable - even on smaller aircraft. I am sure we can get you a set of use able modems for your troubles.


As far as the "hobby like" workman ship - almost all of the items are hand made to save as much weight as possible:

1. GPS : uncased with a very thin ground plane. We could go with a housed ant with coax, but you are talking about doubling the weight. Customers with bigger planes sometimes use the garmin GPS 18 - not a bad solution if you can carry the weight.

2. Dipole ant: Again hand made with nitinol wire. Very light. Again, we sell a bigger/nicer antenna (maxstreem.net - standard 900 mhz with plastic housing). It is much heavier, but not a bad way to go on bigger airplanes. BTW, even though that dipole is not much to look at, it performs very well and was hand tuned using a network analyzer.

3. heat shrink covering on autopilot: - saves a ton of weight. We are building a nice carbon box with a milspec connector on it - similar to cloudcap and micropilot - great if you can carry another few ounces.

Are nitch really has been small and light airplanes (does not mean cheap necessarly). We have catered to this market by handmaking basically all the accesories (gps board/ant/cabling/enclosure) to keep the weight down.

However, as we get more and more customer flying bigger airplanes we are starting to provide more traditional antennas/enclosure/wiring harnesses. The all up weight with these items is about tripple the weight of the stuff you have, but it definatly looks nicer.

Sorry about the bad experience. I think once we get you a different set of modems with the lower latency you will start having a better experience.

Please PM me and I will send you my personal email so we can get this worked out.

Reed

Reed,

Thank you for coming in and explaining the issue.

Can you tell us what the latency time is with the changed Modem?

Thanks,
Chris

[reedchristiansen]

Chris,

With the Maxstream 9xtend the latency averages around 80 ms. This is due to the hop sequence of the radio. The Maxstreams transmit on odd bins/receive on even, so you need to wait till the next hop sequence to send data. We send the RC stick commands at about 15 hz, which is up to 66 ms delay - then add the radios natural latency and which pushes to between 60 and 100 ms. Very usesable but you will notice the delay especially on really small/squirely airplanes. Adding rate damping helps a ton.

If you want better response you can use a higher end modem like a freewave. Slightly better latency especially if you use priority packets in the TDMA mode. I think going to freewaves for range and military frequencies makes sense, but probably not needed for the latency issue. The maxstreams are sufficient. They don't provide the crisp control for precision aerobatics but certainly enough to take control of an unstable vehicle and prevent a crash.

Most pilots get used to it withen seconds. We had no reports of it being a show stopper or even a major problem.

However, the latency issues mentioned above with the european aerocomms are for real. 250 ms is pretty much unusable for "stick-to-surface" flight on an airplane without significant rate damping. I have done it before but it was pretty harry.

Reed





Reed,

Thank you for coming in and explaining the issue.

Can you tell us what the latency time is with the changed Modem?

Thanks,
Chris

[airmcn_3]

Chris,

With the Maxstream 9xtend the latency averages around 80 ms. This is due to the hop sequence of the radio. The Maxstreams transmit on odd bins/receive on even, so you need to wait till the next hop sequence to send data. We send the RC stick commands at about 15 hz, which is up to 66 ms delay - then add the radios natural latency and which pushes to between 60 and 100 ms. Very usesable but you will notice the delay especially on really small/squirely airplanes. Adding rate damping helps a ton.

If you want better response you can use a higher end modem like a freewave. Slightly better latency especially if you use priority packets in the TDMA mode. I think going to freewaves for range and military frequencies makes sense, but probably not needed for the latency issue. The maxstreams are sufficient. They don't provide the crisp control for precision aerobatics but certainly enough to take control of an unstable vehicle and prevent a crash.

Most pilots get used to it withen seconds. We had no reports of it being a show stopper or even a major problem.

However, the latency issues mentioned above with the european aerocomms are for real. 250 ms is pretty much unusable for "stick-to-surface" flight on an airplane without significant rate damping. I have done it before but it was pretty harry.

Reed

Reed,

Thank you again for the time.

It sounds manageable, my concern is the testing of the system. I do not want to have any issues with latency if I need to recover a squirmy Autopilot.

Sounds like some talk with you may be necessary to make sure we do not run into this issue.

I am very pleased with the size and function ability of the Kestrel system, it would be great to continue our plans without having to reconfigure for a different Autopilot.

Thanks again,
Chris

[reedchristiansen]

Chris,

Can you come out to utah for a day - we can put you on the controls and let you feel the latency for yourself. We can also "train" you on the system so you get a feel for everything it can do and get confidence that this is something you can do yourself. We do that for lots of customers and would be happy to do it for you.

Reed

Reed,

Thank you again for the time.

It sounds manageable, my concern is the testing of the system. I do not want to have any issues with latency if I need to recover a squirmy Autopilot.

Sounds like some talk with you may be necessary to make sure we do not run into this issue.

I am very pleased with the size and function ability of the Kestrel system, it would be great to continue our plans without having to reconfigure for a different Autopilot.

Thanks again,
Chris

[airmcn_3]

Chris,

Can you come out to utah for a day - we can put you on the controls and let you feel the latency for yourself. We can also "train" you on the system so you get a feel for everything it can do and get confidence that this is something you can do yourself. We do that for lots of customers and would be happy to do it for you.

Reed


Reed,

Great! I will contact you on this.

Thank you,
Chris

[kostixx007]

Reed,
Thank you for providing a detailed explanation of the problem.
Reed has emailed me and suggested to personally fly to Latvia (northern Europe) in order to hand deliver the right modems and provide help with autopilot integration- the best customer support that is possible. It seems that the solution of the problem will be found.

Konstantins.

[avandelay]

kostix, I am glad you got it worked out. The whole integration of an autopilot is a very, Very time consuming, expensive and a potentially frustrating process. It really pays off when the plane starts flying on its own!
I have seen autopilots from a many companies fly on many ocassions and can say that they have all had their share of crashing. I have only noticed a couple of concerns with the kestrel. The reason I would choose the kestrel is that the size and functions are great for smaller uavs therefore I would choose the kestrel for a uav that small.
As far as the quality goes, I agree with Reed in that the materials used have to be decided wether to go "all out" or to go "slim and light".
The workmanship is something that can be argued because if another company delivered a comparable autopilot with robotic manufacturing quality that company would have the more reliable product. Reliability is a must!
The customer service is a hot issue as well. You have to balance out wether you are giving out free customer service or should it be paid for. I am not sure how Procerus does it but the way you described dosen't sound like a very pleasent experience until you voiced your opinions on here.

[jetblackaircra]

You guys should stay tuned for the autopilot being developed by me (Jet Black Aircraft), and Synappsys Digital Services. Just a run down of the features:

5Hz GPS (lat, lon, alt, GS, ground track)
Pitot airspeed from 0 to 200 mph with 0.5 mph resolution
6 DOF IMU using 300 deg/s rate gyros and linear accelerometers
Sonic Altimeter from 0 to 20 feet in 1 inch incriments
3 axis magnetometer (tilt compensated)

Option to control directly with 72mhz/2.4ghz R/C radio. (ZERO LATENCY, direct control via solid state switch/multiplexer)

Failsafe to R/C upon (unlikely) autopilot failure

Other failsafe modes, return to home upon R/C Tx signal loss, circle over waypoint upon RF link failure, return to home upon RF link failure, return to home upon GPS failure, etc.

Augmented control mode (automatic airframe stabilization with control inputs, like F-22, B-2, F117A)

Fully autonomous mode including auto takeoff, fly to waypoint, circle over waypoint, auto landing, programmable course/flight profile etc.

RF link for digital video (up to 1080/24p Hi def)

PC based ground station including low latency camera feed, heads up display of primary flight instruments, debugging/flight test mode with raw sensor data feed

Configurable for conventional, tail forward, flying wing/tailless, tandem wing, V-tail, rotorcraft, tilt-rotor, etc.

Just wait, it'll be here soon....

[brojo]

ref: autopilot
PLEASE LET ME KNOW WHEN YOU'VE GOT IT!
broderick2000@yahoo.com

[brojo]

ref: autopilot
let me know if you come out with it asap . thanks.
broderick2000@yahoo.com

[Pants]

I'm glad you got a response from one of the Procerus guys, cause I personally have nothing but praise for their after market service (both personal and professional), depth of knowledge and quickness in sorting out problems.

I have also found that 90% of problems are due to the user, not the equipment, so if you haven't done the procerus course with them I recommend that you do, it certainly taught me many aspects and functions of their system, uses, potential issues, things done well, limitations etc, that are not so obvious when not previously having had the proper training.

cheers

[airmcn_3]

I'm glad you got a response from one of the Procerus guys, cause I personally have nothing but praise for their after market service (both personal and professional), depth of knowledge and quickness in sorting out problems.

I have also found that 90% of problems are due to the user, not the equipment, so if you haven't done the procerus course with them I recommend that you do, it certainly taught me many aspects and functions of their system, uses, potential issues, things done well, limitations etc, that are not so obvious when not previously having had the proper training.

cheers


There is a lot of truth to that.

[CenTexFlyer]

Kosti,

Lose the doral fin. You (and your autopilot) will be much happier with the flight performance.

Gene