switch between manual/autopilot?

[danstrider]

I'm trying to build a switching mechanism for manual vs autopilot flight so I can have manual control for takeoffs/landings and give control to the autopilot for piloting where the ground isn't so close :-)

Does anyone have a circuit schematic for such a safety switch or a link to commercial boards? I can give more details of my personal requirements, but I don't want to limit responses!

Thanks for the help, and I figure others here will also benefit from the circuit too!

Thanks!
Dan

[Rhathid]

Cam Man's project should be able to turn something on and off if it's a soft button.

http://www.rc-cam.com/project1.htm

If not, you can use a silicon diaode with it to open and close a cicuit. You'd need two of them I believe, one to open, and one to close.

[danstrider]

I'm not sure if the mechanical switch will work for pwm (it might, I have concerns in the switching time the servo would spaz or something). I found a prototype circuit, but it wasn't confirmed yet to work.... there has to be something out there that can do this.

I have toyed around with AP before, triggering cameras etc. The best solution came down to programming a basic stamp or buying a prism switch. Gotta be a commercial board........

Dan

[LukeZ]

What exactly are you trying to switch? The signal from the receiver to your controller?

[danstrider]

Yeah, exactly that. Switch input from a receiver (i.e. manual mode) to a computer (i.e. autopilot mode).

Dan

[LukeZ]

Well one way that it's often done, as you know, is to have the output of the receiver be an input to the computer. In manual mode you have the computer simply pass the signal on to the servos, in autopilot mode, you don't have to turn the signal off, you just have the computer ignore it and issue its own signals to the servos. Of course in that case manual mode is still in some ways dependent on the proper operation of your computer.

As for switching the PWM signal itself, like you I'd wonder if a mechanical switch might not work very well, given the noise it would introduce. I suppose you could try it and see what happens. I don't know of any commercial switches that would do it, but your circuit looks pretty good.

[Gary Warner]

I can't put my hands on it right now, but there is a two op amp circuit that switched over at the 1.5ms range. It was typical to use these circuits to start and stop motors before ESC's became popular (now I'm showing my age :eek: ). Use the output to drive a micro 5 volt relay. I did this to dissable my AOA controller on the first version, as I didn't have the code 100% stable.

Today, I'd just code it into the PIC I was already using or use a standalone PIC to switch the control. At a buck or less each, the PIC solition is cheaper than a relay.

In PBP, it's dirt easy.

test control port for high, loop (re-test) if high.

PULSIN port,value

test value for over/under 1.5ms

branch as needed from test.

Jump/include correction code lines or drive relay, FET's, photo couplers or what ever as desired.

This process works well on my slip/skid project where I must shut off the sensor input for launching a sailplane.

Another 'hammer' for the thumb tack is to use a servo to move a mechanical switch. I've done that too on the AOA project.

Gary
--

[danstrider]

In this case, I can't pass the signal through the autopilot (for reasons beyond the scope of the question), though I did think of that too.

I have a sinking feeling about the mechanical relay, so I'm going to seek different solutions before I try the relay....

Programming a pic seems robust enough. I'll look into that for a bit to see if I can figure out the specifics.

If you can remember that two op-amp circuit, I'd really love to see the schematic!

Thanks for the input guys.

Would anyone potentially be interestered in a commercial switch for 8 channels? My contact that sent me the circuit above is looking to eventually make a small pcb for the purpose.

Dan

[Gary Warner]

In this case, I can't pass the signal through the autopilot (for reasons beyond the scope of the question), though I did think of that too.

I have a sinking feeling about the mechanical relay, so I'm going to seek different solutions before I try the relay....

Programming a pic seems robust enough. I'll look into that for a bit to see if I can figure out the specifics.

If you can remember that two op-amp circuit, I'd really love to see the schematic!

Thanks for the input guys.

Would anyone potentially be interestered in a commercial switch for 8 channels? My contact that sent me the circuit above is looking to eventually make a small pcb for the purpose.

Dan

I wouldn't worry about the relay. The most you'll get is a one frame 'glitch' in the transition. Hardly a reason to crash.

I don’t have a schematic available from here. Try to fallow this:

First op amp: one input is 1/2VDD (reference), the other input is from the receiver. Output to low-pass filter (LPF).

Second op amp: one input is threshold adjustable reference (fixed once the value is known). The other input is the LPF output from the first amp.

As time 'on' increases from the receiver, output on amp one climbs. Once it's over the threshold reference voltage of the second op amp, the second op amp toggles its output. Reverse logic for decreased 'on' time from receiver.

One big problem can be dropped frames, from weak signals or RFI. They can screw with the RC time constant of the LPF. Using a low enough LPF can reduce the chances of there being a problem, while slowing down the response time of the command to change state.

Gary
--

[mattweisz]

Here is a link to a commercially available one

http://www.microboticsinc.com/servctrl.html#TOP

I think it is about $450...which is about $400 more than what is would cost to make a similar one yourself. Anyways, that link should give you some ideas about how to build one. Buying all the servo connectors is the expensive part.

Matt

[danstrider]

Ouch! $450! Well, I took a look and it's beast robust for what I'm looking to do! But it is indeed good reference at this point.

I took a good hard look at the circuit I posted above and I think there was a mistake. I've contacted my source and told him, but it'll be tomorrow before he gets back to me about it. I'll post what he thinks.

Still chewing on your opamp idea Gary. At first glance I see how it works; will need more thought to grasp it fully. Thanks for the post though!

Dan

[Gary Warner]

Here's the basic circuit. Don't ask for the values of the components - I'm not an engineer ;) . I can't find a copy of the original schematic on the web.

Gary
--

[Gary Warner]

Oops, forgot the file.. :rolleyes: - senior moment.

[Tom in Cincy]

What AP board are you using that does not have on/off control channel?
http://www2.towerhobbies.com/cgi-bin/wti0001p?&I=LXCJD6&P=ML
Tom

[JayFrancis]

These two diagrams illustrate what I think are possible solutions for what you want to do. If there's enough interest, I can design either or both.

The first diagram shows a simple setup. You would choose a receive channel to control the selection of R/C Receiver or UAV Controller to the servos. There are easy ways to implement the control logic, but they may prove troublesome in the presence of glitches or temporary loss of signal.

The second configuration would use a Microcontroller to decode the R/C Receiver outputs (deglitching, loss of signal detect, etc.). Servo outputs are generated by the Microcontroller as well. The UAV controller would be able to read the R/C Receiver's outputs, and provide its own servo settings. The Microcontroller would still monitor one of the R/C Receiver channels to determine which device has final control the servos.

Interfacing the UAV Controller to the Microcontroller would be through a serial, I2C, or other connection. This eliminates the need for the UAV Controller to generate servo pulse timings.

Let me know what you think...

--Jay

[Mr.RC-CAM]

From what I understand, you wish to switch a servo(s) between two signal sources. A mux switch is the typical solution for this.

There are a number of ways to create the SPDT mux switch that you need. You could use purpose specific IC's like a MAX4053, or create something from generic CMOS gates like a CD4011. For example, the attached schematic shows how to create a 4-CH Autopilot/Rx switch using the CD4011 solution. It can be expanded for more channels just by adding one 4011 for each new channel.

For the common Remote On/Off control signal I suggest you use the PanCam PIC. It has an Aux control output (pin 3) that is just fine for this. For this application just ignore the PIC's other features. http://www.rc-cam.com/pancam.htm

RC-CAM

[kd7ost]

I don't have a way to post schematics. Sorry. Old school. But I use the 74HC367 to switch between two devices to a single servo. You can download the data sheet from here.

http://www.semiconductors.philips.com/pip/74HCT367.html

Dan

[danstrider]

Just to let everyone know, I haven't disappeared. I'm talking over all these ideas with a friend who is my local electronics guru.

Thanks for the overwhelming response! Perhaps we can eventually work something out to release to the public....

A better response to come tonight when I have a moment away from work.

Dan E. (to differentiate me from kd7ost)

[sesat]

Not a real switch in that sense, but the Paparazzi code for the FBW MCU essentially switches between driving servos using instructions from the AP MCU and instructions from PPM_IN from an rc rx.


Ram.

[chrisgood]

Take a look at this circuit:

http://www.chrisgood.com/uav/web_components/avigator/mux.htm

This was designed to do exactly what you are looking for. It uses an RC (resistor/capacitor) filter to convert the switch PWm signal to an analog value, and then uses a comparator to compare that resulting analog value to a known analog value created with a voltage divider. The discrete on/off output of the comparator goes into the selector input on the multiplexer chip. Unfortunately, after building this, I found that the analog swing created by this circuit's RC section is not always enough to cause the comparator to switch from A to B, and vice-versa. I combined the the above circuit with the front end of this circuit:

http://www.capable.ca/stefanv/rcstuff/onoff.htm

from Stefan Vorkoetter's website. The front end of that circuit is very good at differentiating the analog signal created by the RC filter against a known analog value. We have been using this combined circuit for several months to control the switching between a commercial autopilot and remote control for a project at work. We wanted a way to take control completely away from the autopilot. The pass-through method (RC signals into autopilot, autopilot decides which signals get to servos) is great until the autopilot locks up and nothing gets through.

I will draw up the total circuit tonight and post it later.

Chris Good

[danstrider]

Thanks again everyone for your response! I will try to address everyone's comments in a giant post....


Matt Weisz:
I think the $450 is definitely overly expensive for my use, though I'm sure it has a place and a time. Made for a good read flipping through the manual.


Gary Warner:
I see the logic behind your circuit, and it is definitely a nice, compact, and simple circuit. Exploring this circuit is on the back burner b/c of the issue you brought up about dropped frames. Possibly a similar approach with faster components will do the trick (and I don't know low pass filters well enough to design up the rest of the circuit...). Thanks for making up the schematic though, I may get back to it.


Tom in Cincy:
I am using a PRISM switch for a non-related project. It does have on/off switching, but it's not a y-harness for muxing two signals into a single output. It would be possible to use a standard AP switch to select the input of another circuit (the "control logic" in Jay Francis' first drawing), but if I'm building a seperate circuit anyway, I'll see if I can't incorporate a switching circuit without buying and hacking a commercial AP switch.


Jay Francis:
Perhaps some more clarification is needed here of my particular use of the switch. My autopilot (a Micropilot 2028g in one case, a BX-24 with a MINI SSC II pwm board in the other case) already produces pwm signals. My switch will be used to hand control between two PWM sources. Also, I have a computer controlled camera mount that I would like manual override capability. My switch will also handle this task.

The second drawing you have is effectively a better version of the MINI SSC II and could definitely be marketed in its own right. Similar serial servo controllers sell for ~$40, but this additional functionallity I have not found on any of the other controllers.

The first drawing is my current need, where two seperate PWM sources are inputs, a selector somehow pics which signal is to be used, and a single output goes to the eventual recepient of the PWM signal. A possible solution to the "Selector" circuit of this flow chart is represented in Mr RC Cam's schematic.

I am not sure exactly how much interest there is in a commercial switching board, but the responses here are likely representative of the potential customer base, beside the diy'ers :-)


Mr RC Cam:
Exactly! That is such an elegant and easy solution, I am flabbergasted. The combination of four NAND gates makes perfect sense, and with a response time measured in nano-seconds, it should be plenty fast for PWM. Using your PanCam PIC is a good idea also. I like how the 4011's can be on common power to the servos and thus function even with loss of a (seperate) autopilot battery (that also would power the PIC switch). I am moving forward with the 4011 circuit and finding a friend with a PIC programmer.


Kd7ost, Dan:
I think a bunch of the schematics posted were done in MS paint :-)
The 74HC37 looks like it could do the job. I think I have to go with the simpler hardware of Mr RC Cam's design in this case though. Six or one-half dozen, accomplishes the same goal.


Chris Good:
I laugh in horror at the following quote from Steffan's site:
"two people have reported a problem where the controller would refuse to shut off during a flight"
I likewise think a non-pass-through circuit in an order of magnitude better, especially if the circuit is relatively fault tolerant. UAV's need to have safer reputations to be accepted by the public.

After a good look-through of your website, you and I are going on such similar routes, it's nuts. I also have started with the BX-24 and a u-blox receiver and have access to aerial video equipment and a bob II osd... The only difference is I am aero engineering, soon to be a masters student, and that I'm not going to design my own aircraft for my project. Are you doing your masters through your company? I think I'll send you a PM later with some more specific questions :-)




So that wraps up this horrendously long post. The project is ongoing, so feel free to suggest more ideas or schematics or whatever! Hopefully pictures will soon follow!

Thanks all,
Dan

[JayFrancis]

Hi Dan,

Here's a first pass at a simple switch. Using a PIC as the selection control logic allows programmability of switch setpoints and some intelligence for glitch/loss of signal detection. Control of the switch would be hardwired to whatever is connected on port 8. Ports and channels are not required to be the same, wire them up however you want, just make sure they pass through to together (i.e. ailerons could be channel 1, you hook them to RC4, and make sure the aileron servo connects to SERVO4).

I would estimate the board size to be around 2" x 2" or smaller. The connectors (or holes for soldering in servo extenders) take up the most room.

--Jay

[note] RC8 will need to be clamped before directly connecting to the PIC. There will probably be more items to fix... that's why it's a first pass :o

[chrisgood]

Here is a quick drawing of the mux we have been using. This has been the autopilot disconnect on a $10k electric helicopter UAV for sever months. I have this drawn up nicely somewhere and will try to locate it this week.

R1 is a current limiting resistor for the RC signal input.

Z1A is a buffer for the signal.

R2 and R3 are a voltage divider to give a reference voltage.

R4 is a pull-up resistor.

R5 and C1 act as an RC filter. It takes the PWM signal and converts it to an analog value. You need a scope to see how the RC filter charges and then discharges, but the idea is that when the PWM output is 2 ms, the voltage is higher (by about 200 mv) than when the PWM output is 1 ms.

The resulting analog voltage from the RC filter goes into Z1B and Z1C that compare it to the reference voltage created by the divider. Those two comparators output 0 or 5 volts, based on the two inputs.

The resulting 0 or 5 volts feeds into Z1D. When tied to Vcc through R6 and allowing a discharge/charge though C2, it acts as a missing pulse detector.

That logic signal goes to the select input on the MUX chip. The MUX chip sends the A or B signal to the Y outputs, depending on if the select input is high or low.

Most of this circuit was take from Stefan Vorkoetter's website: It Works!

Chris

[kd7ost]

Sweet Chris. That is both as simple, and as elaborate as it gets isn't it? 4 Channels switched in a single chip too. I thought my 74HC367 was cool but the best I could get was 6 channels in two chips. Thanks for the tip on the 157 mux. I'm going to get some of those. :D

Dan

[danstrider]

Dang! And I just ordered parts for RC Cam Man's circuit.... guess UPS gets more of my money (shipping companies gotta love internet ordering).

That is a pretty cool chip, the 157 mux. I also like the alternate hardware for the 0/5v switching, instead of programming a pic or something. Though, a pic would be less components... at any rate, I don't have a pic programmer, so I'll be trying your circuit first.

Thanks a bunch for sharing that circuit; thanks everybody for sharing circuits. This has been a fun learning process for me :-)

Dan E.

[chrisgood]

You are welcome. Mr RC-Cam's CamMan is a great little device. I have made many of those for friends and installed in several cameras. If you have the PIC on order (already programmed) use that as the switch select device and feed the output into the 157 MUX. This link shows all the cameras I have installed the CamMan into.

http://www.chrisgood.com/rcplanes/aerial_photos/index.html

Chris

[Mr.RC-CAM]

The HC157 mux with a PanCam (not the CamMan) PIC would be about as low parts count as it gets for a 4-channel solution. Just three parts (PIC, mux IC, and decoupling cap).

If you want to try this combo and need a programmed PanCam PIC, then let me know. I think I have all that you need for one proto; I should be able supply all the parts for about what you probably paid to ship that other order. :)

RC-CAM

[LukeZ]

This link shows all the cameras I have installed the CamMan into.
Chris
Wow Chris! Mr. RC-CAM should put you on his frequent-flyer program or something! ;)

[JayFrancis]

Some nice solutions folks!

I'm still working on one (for myself at least). Here's the latest schematic and a picture of the PCB. The layout is complete and I'll be ordering boards and parts this week. I'm trying to decide whether I want pretty (more expensive) or ugly boards. :rolleyes:

This design supports up to 7V battery input (5 cells). The servo/radio/auto headers will probably be left unpopulated and the appropriate types of male/female pigtail wires soldered into their holes.

Initial software will be a hardcoded switch point. Future software may include a configurable setpoint and some form of handshaking/messaging with the autopilot.

The other application I had for this board is on a non-aerial vehicle (UGV), which is why I wanted mounting holes.

--Jay

[chrisgood]

Sorry about that MR RC-Cam. I was confused as to which PIC version I was identifying. I actually had a slightly different circuit in mind; the CamMan would toggle off like it does normally to close a shutter, but would feed into a flip-flop which would latch and keep the MUX select low or high. The PanCam AUX OUT is a much easier way to do this. Thank you.

Chris

[danstrider]

Got the parts from Mr RC Cam (many many thanks!) and it works :-D

I'm trying to get a whole bunch of servo pigtails now so I can hook this baby up for real. It'll be a little bit of time, but I'll have a pictures of it then. I definitely will also post some instructions somewhere about how to build this thing. It's really easy. It's so easy, even I can do it (and I'm aero engineering by background).

Thanks everybody!
Dan

[JayFrancis]

Dan,

Let us know if you find a good source for pigtails. The best place I've seen so far is Dymond Modelsports (www.rc-dymond.com). Another option is to buy connectors and build them up. When you add up the time, it might make sense just to buy them prebuilt.

Thanks,
Jay

[danstrider]

Hey Jay,

I'm not sure of a cheap source either.... my UAV club at school has been switching to locking molex connectors:
http://www.newproduct.molex.com/datasheet.aspx?ProductID=93966&BV_SessionID=@@@@1013732675.1114952667@@@@&BV_EngineID=cccfaddehghhgflcflgcehedffgdfmk.0
http://www.newproduct.molex.com/datasheet.aspx?ProductID=91404&BV_SessionID=@@@@1013732675.1114952667@@@@&BV_EngineID=cccfaddehghhgflcflgcehedffgdfmk.0
(product # 70107-0002 and 51030-0330 if the links don't work)
These things don't vibrate loose and survive most crashes even. I don't know why RC components are using those non-locking ones now :-)

We also order the towerhobbies 50 foot roll of servo wire for $15 and usually go through that in a semester (including 200 some odd crimps), so I'm not exactly sure where single-item pigtails can be found cheap. For the switch-to-receiver connections, I'll need 7 standard JR male plugs (sockets, not pins), but we have a ton of old snipped JR connectors from the change to locking molex.

For the pure reference of it, I'd like to know where to get cheap pigtails too. I'm running out of my personal busted servos to steal test plugs from!

Dan

[mattweisz]

If you modify your receiver to output the PPM signal on one of the channels....you could eliminate all but one of the connections between the receiver and manual/autopilot switch. Ofcourse, you would then need to decode the signal with something like a PIC...

Matt

[danstrider]

True, but that defeats the commercial-off-the-shelf nature of the switch. Have to have a modified rx, modified ap, extra parts .... though I imagine there are projects where this would be useful!

(to note, I think it's cool that the UAV forum has turned into a collection of people interested in meshing airplanes and computers. Definitely a good outreach to the electronically savvy people among us!)

Dan

[danstrider]

Anyone know Eagle to layout a pcb? I have the layout drawn in mspaint, but I'm struggling to climb the Eagle learning curve... I was hoping to use www.sparkfun.com for making pcb's, but pcb123 or similar would be okay.

I made a perfboard version of this and it looks okay (dimensions are smaller than a receiver: 1.25" x 1.75" x .35"). Was hoping to get a shiny green pcb :-)

(edit: to note, I had planned to just solder the cap onto the voltage reg, but adding it into the board somewhere would be awesome too! It goes between v_in and gnd)

(another edit: I don't have power shown for the 8th switched channel ... I personally didn't need it, but it could be added in there somewhere)


Dan

[JayFrancis]

Hi Dan,

I've used locking wire-to-wire connectors similar to the Molex product in robots. I agree with you, non-locking always seemed a bit risky!

On the mux I'm building, I used Samtec TSW-108-08-G-T-RA headers to hook up to the servos (3 pin by 8 right angle header).

I did pick up some Molex 50-57-9003 housings and pins from Digikey (WM2801-ND). They're not much cheaper than picking up connector kits at hobby stores until you get into higher volume.

Matt,

Yes, that would be the most efficient method!

--Jay

[JayFrancis]

Hi Dan,

I've used many different packages for schematic and PCB design. Unfortunately I only did a simple design in Eagle to evaluate it a few years back. Maybe I can still help if you have general or process related questions.

--Jay

[danstrider]

I'm just totally lost :-)

I've never done any electronics this low-level before. I tried to follow some Eagle tutorials yesterday but couldn't get through them after a few hours, so I had to put it down. Sparkfun has a little intro about how to configure the files they need for making boards (http://www.sparkfun.com/tutorial/PCB/eagle-output-tutorial.htm .... wow, they went up to $5 per square inch from $2.50 per square inch just last week!).

This circuit is pretty simple ... one layer will even suffice. I just don't even know where to start. Thanks for any help you can give Jay.

Dan

[JayFrancis]

If you've never done this before, starting with PCB123 might be a good idea. Their tutorials look reasonable. If you're lucky, all the parts in your design will exist in their libraries.

On second thought, I just looked at their board pricing... pretty steep...

This is a pretty good general tutorial: http://alternatezone.com/electronics/files/PCBDesignTutorialRevA.pdf

Did you see this Eagle tutorial?
http://www.ee.ualberta.ca/~vjsieben/pdf/eagle.pdf

It looks like an intro designed for a student lab.

I hope this helps a little,
Jay

[mattweisz]

Sparkfun still has the $2.50 / sq in PCB Deal.....but, you must have previously successfully submitted a PCB without problems to receive that price. They have a pretty good support forum on their website( which also explains why they raised the price).

http://www.sparkfun.com/cgi-bin/phpbb/viewforum.php?f=16

I ordered some boards from them...and they turned out really nice.

Matt

[danstrider]

Thanks for the encouragement and links guys. After looking at Eagle again today with fresh eyes and a bunch more tutorial reading floating in my brain, I got much much closer to a workable pcb design... It's still being revised, but here's what she looks like now...

Thanks again guys :-)
Dan

[Mr.RC-CAM]

Got the parts from Mr RC Cam (many many thanks!) and it works That is great news. It was my pleasure to help out.

For best operation, a .1uF decoupling cap must be directly across the power pins of the PIC. The traces going to C1 are too long and too skinny to satisfy that requirement. It would probably be best to place C1 at the top of IC3 and use at least 25mil power and gnd traces to all IC's. Your Vreg will need bypass caps too (info will be on data sheet).

RC-CAM

[JayFrancis]

Dan,

Glad it's getting somewhere.

Make the power and grounds between the receiver and servos as large as possible. They could see an amp or more surge depending on the servos and loading.

I would treat Mr. RC-CAMs suggestions more like requirements. One thing about hardware design is that there are all these seemingly picky little details. Unfortunately, if you skip one (or more), Murphy's Law will come into play.

Add bypass caps to the 74157s as well (place them close to the VCC pin, and use beefy power traces).

I don't know how well Eagle does floods... On a two layer board, I like to route as much as possible on the top layer and flood the bottom layer with ground. Depending on the board, I will then flood the remaining top space with VCC.

--Jay

[danstrider]

Super suggestions! I actually was also thinking the cap shown was buried too far away from the reg, so I must be learning something :-) I'll go read up on the data sheets and start a new design tomorrow after work. Thanks again guys, you really are an asset to this hobby!

Dan

[JayFrancis]

Hi Folks,

I've been writing the code for my control PIC, and came up with a philosophical design question.

First, a couple of assumptions (correct them or feel free to comment...)

The switches we've been discussing are more like "remote controlled buddy boxes". The "instructor" is on the ground, and the "student" is in the plane. That being the case, we would assume that the plane is always in contact with the instructor (never flown beyond view).

Now for the question: In the case of glitches or loss of signal on the control channel ("instructor" channel), should the last valid command rule (hold "instructor" or "student"), or should the switch default back to the "instructor".

I'm leaning towards the default to "instructor" position. I would hope that some erratic behavior would signal the "instructor" that there is a problem, and have them take control before it becomes a fly away situation. Ideally, a downlink from the plane back to the "instructor" would warn of a problem (maybe a downlink should be a requirement?).

What do you think?

--Jay

[danstrider]

Jay and other readers,

We have discussed this in my club and at competitions at length as a very big safety concern.

For the competition safety requirement, ANY loss of manual signal (the ability to take full manual control) must initiate a "hard over" manuver to stall/spin the airplane in a more or less controlled crash. To operate under this requirement, my rx failsafe mode is hard-over and revert to manual control.

The default to manual means I see the plane start to deviate and click my switch to manual control to take over (pilot's first assumption is the autopilot goofed, must take over manual control) and then correct the sticks as needed. If the problem was indeed RFI related, my failsafe will hold the hard-over until the signal regains link and then the manual transmitter as the primary signal source (pilot just holds sticks in a correcting position for *hopefully* eventual return to manual control).

Hopefully (and this has happened in competition) the RFI is short lived and the plane once again allows the manual TX signal to pull out of the kill manouver.

(I digress: in competition, we saw the airplane at over 1/4 mile away point straight down and start rolling. The RFI was short-lived and the pilot pulled safely out of the manouver. Three other teams had similar occurances, but one team wasn't so lucky. The judge's base-station radio frequency analyzer never picked up the RFI....)


BUT, this is not my ideal situation. The autopilot I use can detect loss of manual signal link and trigger a pre-programmed error path such as "return to home." This trusts the autopilot is the best source for valid pwm signals and assumes the autopilot can make it back safely to the home waypoint. This also gives the best possible chance to regain manual signal link.

However, if my manual link is gone, I can't command a controlled crash. For the competition safety folks, they'd rather have immediate and unwavering ability to take over manual control at any point in time. Thus the reason for the "hard over."

I'll bounce back to group discussion, how does this philosophy sound?

(Jay, I think you should start a new post :-) I'm sure this discussion could go on for ages!!??...)

Dan

[JayFrancis]

Thanks Dan -

See, we help you with electronics, you help us with UAV info!

For initial experimentation, I think the "hard over" philosophy is the safest. I hadn't thought all the way through, but I guess I need a receiver that either has a failsafe mode, or doesn't have "last good frame hold". It would be a shame to have the receiver lose signal and then hold the autopilot on...

Were you guys using PCM or PPM radios? What club/school are you involved with?

If the control philosophy discussion gets too lengthy, I'll start a new thread.

--Jay

[danstrider]

I'm president of the NC State Aerial Robotics club in Raleigh, North Carolina. You can read a bit about us at: http://www.ncsu.edu/stud_orgs/ar
We are heading to competition in late June to compete in the AUVSI Student UAV competition: http://www.bowheadsupport.com/paxweb/seafarers/default.htm
We have done well in the past (2nd and 5th overall) and hope to regain our lost Safety Award this year, too :-) That's where the safety switch comes in...

We're using PCM with a programmable failsafe. Our receiver also supports an "all hold" failsafe mode. Not sure of the rx number, but whatever came with the JR 10x.

For non-competition testing, we use the "all hold" mode because we're likely to regain RC control after passing through an RFI area or the RFI stops. We've never had a flyaway, though potentially this could be a concern. Our test field is away from population, so the aircraft will most likely crash itself before it could fly far enough to hit someone. I should state this is not necessarily my stance, it's field rules.

Dan

[william541]

Interesting topic guys, I had never really thought about the safety aspect of a UAV out of control.

One thought that popped in my head would be to have a timer associated with loss of control. If control was not re-established during the time interval, then power would be cut and a parachute deployed and the locator beacon activated.

I know, extra weight and hardware, but it may save the craft and minimize property damage at the end of its fall.

I like the idea of return to base, but if it got back to base still without control, then what? Deploy the chute!

[JayFrancis]

Very impressive Dan!

I've started a thread on Hobby UAV safety:
http://www.rcgroups.com/forums/showthread.php?t=367031

I hope we can exchange specific techniques (i.e. Dan's details) as opposed to rambling theoretical discussions... :p

--Jay

[JettPilot]

DanStriedr,

That philosophy is flawed. In the event if a loss of signal, you should have the plane level out and at a low power setting in a controlled glide. If you do not have an autopilot, you should have slow glide trimmed with idle throttle. Most signal interference is temporary in nature and you want to buy as much time as you can to get control of the model to PREVENT a crash instead of your hard over that causes one in short order. If it does crash, less damage will be sustained in a controlled glide than a stall spin. Not hitting as hard is always better, and from a safety standpoint its better.

Many people get so narrow minded and paranoid thinking about "safety" that they ending up doing the WRONG thing and making BAD choices. The hard over option is a clear example of overthinking "safety" to the point that you have lost common sense, it actually creates more of a hazard than it solves.

Give your self every oportunity and time to prevent a crash. Minimize damage to your plane and whatever it may hit. Its simple....

[JayFrancis]

Here's my switch so far. It works. I'll be heading to the local hobby store tomorrow to buy the servo pigtails.

--Jay

[Mr.RC-CAM]

Here's my switch so far. It works.Beautiful work!

I'll be heading to the local hobby store tomorrow to buy the servo pigtails.In case it helps, I'm selling off a bunch that I don't need. http://www.rcgroups.com/forums/showthread.php?t=367315#post3705737

RC-CAM

[JayFrancis]

Thanks Mr. RC-CAM,

I'll keep you in mind if I need to buy a bunch more pigtails.

--Jay

[danstrider]

oooooh pretty :-)

I'm hoping to get my pcb order in tonight. I'll be sure to post a pic when it comes in... it won't be as pretty as that one though!

Dan

[kd7ost]

Here's my switch so far. It works. I'll be heading to the local hobby store tomorrow to buy the servo pigtails.

--Jay


Very, Very Nice Jay,

I'm trying to make out the www.??????.com on your picture of the board but can't make it out. This is a little off topic but what board house do you use and do you just e-mail them the Gerber file? Do you mind if I ask the cost? I Have a small amount (~ 20) of little boards I need to get run and am shopping for the best price. I have the software and info for PCBExpress. I know there are other board houses and just don't know where or what price. I keep etching them by hand at home and it's starting to lack decent productivity even though I only use them at a slow pace.

Once again, Very professional looking work sir.

Thanks Jay

Dan

[danstrider]

www.reactivetechnologies.com it looks like....

[JayFrancis]

Hi Guys,

www.ReactiveTechnologies.com
I'm a freelance hardware engineer, and that's the name of my business. I don't know if this will end up a business related project, but it's easier to get all the info into the design files now.

These boards were done at Advanced Circuits in Colorado (www.4pcb.com) using their 3 for $33 each promotion. I had them shipped 2-day UPS, with the final total around $120. I've also used the "bare bones" service (www.barebonespcb.com), as well as their regular production service. Their "free DFM" checking service (www.freedfm.com) is a good gerber sanity check.

All the design files are uploaded through Advanced Circuit's website when placing the order. You will need gerbers, nc-drill, and (depending on your tools) aperature files. You need to make sure to read the capabilities of each of their services. Line/spacings are different, as well as soldermask/silk/plated holes/etc. They batch the lower cost PCB runs, so everyone has to follow the same specifications.

AP Circuits (www.apcircuits.com) is another place that does small quantities, but they get pickier about how many drills you use (and what sizes).

If your boards don't need silkscreen and you can live without a soldermask, the barebonespcb service is a pretty good deal.

Hope this helps,
Jay

[kd7ost]

If your boards don't need silkscreen and you can live without a soldermask, the barebonespcb service is a pretty good deal.
Hope this helps,
Jay


Perfect Jay.

Many thanks.

Dan

[JayFrancis]

And where are my manners?!?!?

Thanks for the compliments guys,
Jay

[danstrider]

Yeah, my switch didn't work.... The pancam chip worked fine, but the 157mux didn't like me. I can't get the mux chip to pass through servo signals. *sigh*

Anyone else have luck? Anyone have a switch I can buy?

Dan E.

[Mr.RC-CAM]

I can't get the mux chip to pass through servo signals. We can get the HC157 to work. Show the details of what you have done and a solution will probably pop up.

RC-CAM

[chrisgood]

There are a few pins on the 157 that are not used, but they must be grounded. Check to see that every pin is connected to something.

Chris

[pjsmith]

Hi,

I'm planing on developing somthing for exactly this in the next couple weeks. My idea was to use the processor itself and a servo controller. The processor (bx24) would sample the incoming channels (rudder, elevator, motor). In passthrough mode it would simply relay these to the servo controller. In autopilot mode it would ignore them and do it's own thing.

The advantage of this method, that I saw, was that I needed no failsafe compatible controller. Sanity checks on the sampled data will filter out/record glitches and could turn on the autopilot in the event of loss of signal. So I get all the benifits of an expensive receiver without the cost, and I don't need to add any more components other than what I was already using.

I realise that this method is not suitable for the original starter of this thread, but I'd be interested to know if anyone else has used this method, and of any problems they encountered.

[kd7ost]

I haven't tried that but I think it's a good idea. I play around with Parallax Basic Stamps, but the radio's I fly with are all PCM. When I do fly a BS2 it's to do pretty rudimentary functions. Bumping the Camera servo to prevent sleep mode etc. (I do have other plans but haven't employed them yet). It's a little over kill to have it on board but I have the room. It would be cool to see someone do what you propose.

Dan

[danstrider]

So I follow up on this thread (it's been a while, sorry),

I'm working with Jay on his switch board. The prototype brought out a bunch of neat quirks with my initial assumptions. Both of us have learned a lot from the experience. I don't know exactly what Jay's plans are for the future, but I can say that the prototype he pictured above worked wonderfully in my application.

So you all know, this type of switch is an excellent safety device and is definitely worth the investment. Take the autopilot out of the critical path of the servo signals from the receiver to the servos!

Dan


I do think JettPilot's application ideas will differ; if you want to rely on an autopilot as a reliable system, things will be a bit different. This safety switch is geared more toward experimental autopilots or at least experimental applications.

[akcarl]

How about running the auto pilot on the dsc port of the receiver and just switching that input on and off?

[srik901]

I just came upon this thread and I was wondering if it is possible for me to buy one of these auto/manual swsitches. My email is

srikanth.saripalli@gmail.com

[JayFrancis]

The switches aren't for sale right now.

Dan has made a few feature requests that I'm considering. If anyone else has specific features to add, please let me know. The simplicity of the current switch has its advantages.

As for future availability, I guess it depends on how many people are interested.

--Jay

[JayFrancis]

I'm going to build some more switches. How many I build will depend on if anyone else is interested in them. Price will also depend on the quantity built.

Send me a PM with your e-mail if you're interested. I'll tally up the requests and let people know the status in about a week.

--Jay

[sesat]

Carl, that sounds like what I've wanted all this while...

[LukeZ]

Jay, earlier you had mentioned the possible addition of some other features. Is the mux you're going to make basically still like the one pictured earlier, or have there been some changes?

Luke

[JayFrancis]

Hi Luke,

It will be very close to the first one. I will be adding:

1) Weak pullup (or down, depending on which works out best on the PCB) resistors on the inputs. With the first switch, I needed to jumper unused inputs to power or ground.

2) A power connector. Power can still come in through any of the servo connections as well.

3) Repin the microcontroller I/O so that I can program it in circuit.

4) If I can pull it off, I may add an extra row of ground pins to the input and output connectors. This would allow you to use a ribbon cable header with just servo signals and grounds.

5) I'd like to add a servo output pulse that is a "reverse" of the input selection signal. Dan requested this to interface to the Micropilot.

My goal is to keep it simple.

--Jay

[danstrider]

Howdy gentlemen,

I was the prototype tester for Jay's original board. I figured I'd post some installation photos that were representative of what can be done with his product. My original installation had probably $15,000 riding on it, so I wouldn't have installed it unless I trusted the circuit :-)

Feel free to ask questions of me about the functions and I'll take the load off of Jay so he can work on the new revision.

Enjoy the pics!
Dan

[LukeZ]

I think you guys sold me on this one... :)

[voidpointer]

Hi,

I seem to be a bit late for this discussion but maybe you guys have a look at this idea:

Most RC receivers deliver the complete PPM frame at the batery connector. This signal contains the pulse witdth of every servo channel. With the use of a microcontroller one can decompose the PPM and control the servos. A dedicated channel (e.g. channel 7) is used to switch between auto and manual mode. In auto mode the switch can be controlled via I2C or serial connection. In manual mode it could deliver current servo positions to the main controller.

The benefit is small size and little weight. A prototype looks like this:
http://voidpointer.de/servoswitch/mounted_front.jpg
I documented the thingy at http://voidpointer.de/servoswitch/index_en.html. It is already working but the code needs improvement. I'm still looking for someone who will have a look at the ASM code and tell me what to improve. Maybe someone is interessted.

Achim.

[JayFrancis]

Hi Achim,

Nicely documented project!

I don't believe all radios provide easy access to the PPM stream. It is certainly a cleaner connection if they do. Weren't some of you guys looking at the using the DSC port on the Futaba radios? Any luck there?

--Jay

[voidpointer]

I don't know about the DSC port except from what Google told me. But this concept seems to work at the transmitter side, not at the receiver :eek:

Of course, there are receivers without easy access to the PPM signal. Which leads to the most efficient way of implementing a servo switch - hacking the receiver. If you cut the input and reset line of the 4017 and put a microcontroller in between you will save most space and don't need to provide additional servo connectors, because the receiver already has them. The disadvantage is that you gear into the RC receiver...

Achim.

[Mr.RC-CAM]

DSC is a hardwired Rx connection from the Tx's buddy jack. It is used to control the servos without the need for a RF signal. The more expensive Rx's have the DSC signal on the unused pin of battery input jack.

Using the Rx's DSC feature, and a spare I/O pin on the microcontroller, a manual/auto "switch" could be created that avoids a mux switch IC altogether. It would just be a bit of software trickery with really no other parts.

[voidpointer]

Yes, this way noone would need to open the RX and do some odd soldering. I wonder how the RX then distinguishes between auto and manual mode. There might be a kind of electronic detection that decides which signal (wireless or cable) to choose. One would have to deal with that.

[Mr.RC-CAM]

I wonder how the RX then distinguishes between auto and manual mode.Periodically during the autonomous mux state, the DSC injected PPM signal would need to be disabled for a moment so that the native PPM could be measured. This solution would need to be done with tight cooperation of the host controller. In other words, operation is shifted to mostly software rather than hardware.

[JayFrancis]

Hi All,

I have a batch of RxMux switches available for sale. Please see the attached document for technical details.

They are available at $99 (US) each plus shipping. PM me if you're interested.

Thanks,
Jay

[rclinks2002]

subscribe

[danstrider]

I was eating lunch and thought I would update everyone on the status of this project.

It is alive and well! As far as I know, there are a good number (5-10) of independent users of Jay's hardware.

This level of manual override has become nearly standard at the AUVSI Student UAV competition. The control requirement levied is to have manual control override outside of an autopilot pass-through. I know specifically of two teams that used Jay's switch, but I suspect there were many more.

I spoke with one gentleman from University of Texas at Austin who said the hardware saved their UAV more than once when their autopilot turned off.

From personal experience, having the RC system separate from the autopilot power supply saved NC State's aircraft at least twice when a power blip reset the entire payload, including the autopilot. Had the signals passed through the autopilot, the aircraft surely would have hit the ground before the system rebooted.

The most recent changes have been mostly implimentation updates. There is now provision for a redundant RF system including an autopilot. The hardware has always been top-notch as delivered from Jay, so no changes noted there.

Feel free to ask questions! There is a lot of info in this thread, but I don't mind answering repeat questions about the final system (if nothing else, simply to have a more compact q/a section).

Dan

[phatmonkey]

I am looking for something quite simple just to switch between autopilot/manual for two channels, switched by another channel on the receiver. What would be the best solution? JayFrancis's board seems to be a bit of an overkill for what I want to do, especially as it'll probably only be temporary for development.

[Unterhausen]

When I was looking for recievers that would work with the xbow micronav, I discovered that the Futaba receiver demux their ppm signal with a shift register. Seems like this might simplify a manual/autopilot multiplexing circuit, particularly on the wiring side.

[jbeanelliott]

A few months ago I saw a small advertisement in Servo magazine for a $19.95 board by Pololu which would select between 1 of 2 banks of 4 servos. The intended application was for robots to switch between RC control and onboard computer control.

Unfortunately, I seem to have misplaced that issue. Also, I cannot find this multiplexer on the pololu.com website so I am not sure if it is really available or not. As I recall from the magazine advertisement the board used all surface mount components and the main IC was a single 74AC157. I believe the dimensions were about 1.0 inch by .75 inches. I have emailed pololu requesting availability of this product but have not heard back yet.

Is anyone familiar with this servo multiplexer by Pololu?

I am looking for something quite simple just to switch between autopilot/manual for two channels, switched by another channel on the receiver. What would be the best solution? JayFrancis's board seems to be a bit of an overkill for what I want to do, especially as it'll probably only be temporary for development.

[dalbert02]

Hi All,

I have a batch of RxMux switches available for sale. Please see the attached document for technical details.

They are available at $99 (US) each plus shipping. PM me if you're interested.

Thanks,
Jay
Are these still available for sale?
-dave

[jbeanelliott]

I just received an email from Pololu saying that their new servo multiplexer might be available later this month (July 2006). Below is their email:

Hello John,
Thank you for your interest in our new multiplexer. We hope to have them up for sale later this month; the price will be $19.95.

Candice Kamachi
1-877-7-POLOLU
(702) 262-6648
www.pololu.com

At 12:49 PM 7/3/2006, you wrote:
I recently saw your advertisement for a servo multiplexer board in a recent
issue of Servo Magazine (I seem to have lost that issue). This product
allowed you to switch between 2 groups of 4 servos and cost $19.95.
However, I cannot find this servo multiplexer product on your website.

Are you presently selling this servo multiplexer board?
John Elliott

[JayFrancis]

As of January 2007, I'm putting together RxMux boards on a build to order basis. Lead time is about three weeks and the price is still $99 plus shipping.

PM me for more details. I've attached the latest documentation.

Thanks,
Jay

[jbeanelliott]

I recently purchased one of Jay's RxMux boards and was very impressed by
the quality of the construction. Excellent documentation also.

I had been waiting for a 4 channel servo multiplexor from Pololu for 6 months but finally gave up. I think Jay's RxMux will better suit my needs anyway.

-John Elliott (Seattle)

[danstrider]

Update: Jay is now selling the RxMux through another vendor. He posted the information here: http://www.rcgroups.com/forums/showthread.php?t=829241

It's fun to look back and see where this all came from! I sort of laugh at my attempts at circuits in the beginning. Thanks all you guys for humoring and helping me climb a learning curve. I'm happy now though moving on to doing things rather than getting stuck in the details of making this board.

Anyhow, trying to keep all the info in one place.
Dan