Shop OBEX P1 Docs P2 Docs Learn Events
Relay for switching small signals (serial and I2C)? — Parallax Forums

Relay for switching small signals (serial and I2C)?

max72max72 Posts: 1,155
edited 2015-01-03 11:30 in General Discussion
I have a friend with a drone and he's doing the necessary certifications.
Italian norm requires the possibility to switch off the drone, and to do that independently (separated power supply, switch transmission...).
His drone main board sends the signal control to the ESC board using I2C and serial 5V signals.
I would like to be able to cut the I2C and serial signals. I started considering an I2C repeater or a similar solution for the serial (I asked about that in the forum).
Due to the separated power supply requirements the solution is not optimal. I'm wondering if relays could work in this case and allow a correct signal trasmission when closed.
Obviously your insight (including other possible solutions) would be greatly appreciated.

Thanks in advance,
Massimo

Comments

  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2015-01-01 08:35
    I am a bit wary of the idea that you must use relays to solve this, but if that is really what is required -- you might consider latching relays.

    A latching relay accepts a pulse to change state, and doesn't require constant power to the coil to hold position. You may find that a conventional relay would create excessive drain of power on the battery. After all, even a small 5VDC relay coil might demand 30ma or so.

    Jameco has some small ones that will give you an idea of what is available. http://www.jameco.com/1/3/5v-latching-relay

    I wonder why one would bother cutting the I2C signals when cutting power to microcontroller or the I2C chip would do the same thing. You might consider just selectively cutting power.

    There are chips that could put the I2C bus into a high-impedance mode without bothering with relays or cutting power. And a flip-flop could toggle that state on and off.

    So there are alternatives that you may have overlooked.
  • max72max72 Posts: 1,155
    edited 2015-01-01 09:43
    Thanks,
    I was thinking of using latching relays, to increase safety.
    Accessing directly the board is very difficult, while cutting trasmission would require a board to be placed in the middle of an ongoing trasmission between two separate boards.
    The I2C tranceivers are very nice chip, but I have to power it, either with the main power or with my separated battery.
    In the first case there could be problems with the certifications (they consider a copter and a plane the same, forgetting that with no power a multi copter flies like a stone...).
    In the second case a failure of the secondary power would shut down the drone, and I want to reduce the danger to have the safety system causing mayhem: in case of a failure of the secondary system the thing must fly.

    Another alternative is cutting the main power (tested and worked fine), but the currents involved are very high, so I was looking for alternatives operating on other systems.

    There also is the possibility to overload the switching regulator of the main board, but it could damage the microcontroller, and while the system is expected to be used in extreme cases, I cannot exlude that it could be tested, with no damage.

    Massimo
  • ercoerco Posts: 20,256
    edited 2015-01-01 10:20
    I really like latching relays for their efficiency in most applications. Obviously there is an internal latch/release mechanism which is cycled electromechanically, intended to stay in its last configuration. In this mission-critical scenario, I would be concerned about mechanical vibration combined with G-forces accidentally releasing the latch. That could certainly be a show stopper. Flying things need to be bulletproof.
  • tonyp12tonyp12 Posts: 1,951
    edited 2015-01-01 10:24
    https://www.google.com/#safe=off&q=kill+switch+Rc+toys

    What are the "legal" requirements?, does drone need to drop out of the sky if radio signal is lost?
    Here is the difference with drone vs chopper/plane, with drones it's better that it stays airborne and try to return close to home base.

    Do you need to have a kill-switch radio signal button? signal going in to the main board LDO-Enable and it proves that the motors shuts down if main mcu board is not on.
  • PublisonPublison Posts: 12,366
    edited 2015-01-01 11:22
    I would think a "Return To Home" routine would be better than "Drop out of the sky".

    Of course that would require a GPS unit.
  • max72max72 Posts: 1,155
    edited 2015-01-01 13:23
    Go home is not compliant (and already present).

    The requirement is an independent system. Consider it a last resort, in case everything goes crazy, and possibly never used.
    It must use a different radio and receiver, and a separate battery. It must kill the system and let the drone fall down. The requirements are the same both for a plane and a multirotor, so something looks silly because of that.
    For instance the separated battery makes sense in case of a plane that can plane and go a long distance even with no main battery. In a multi rotor a problem in the main battery means falling immediately.

    So it must be triggered from land. The flight zone is less than 300 m, so little issues with loss of signal, but in case of signal loss it must not automatically activate.
    In case of a failure of the emergency system it should not be triggered.

    Massimo
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2015-01-01 14:10
    For gating small signals from a separate power supply, you might consider a PhotoFET. Here's an example:

    -Phil
  • ercoerco Posts: 20,256
    edited 2015-01-01 14:43
    Can you deploy a parachute when the "fall from the sky" command is received? That's a win win for everyone concerned.
  • frank freedmanfrank freedman Posts: 1,983
    edited 2015-01-01 16:25
    erco wrote: »
    Can you deploy a parachute when the "fall from the sky" command is received? That's a win win for everyone concerned.

    Like the guy on the ground perhaps?
  • max72max72 Posts: 1,155
    edited 2015-01-01 23:21
    Yes, a servo can deploy a parachute when there is the system is triggered. I'm planning to use an RC receiver, so the servo action comes "for free".
    I found there are latching relays suitable for high vibration environments, so it there is not adverse effect on the signal quality that could be worth a test.

    A possible solution using the I2C tranceiver and being still compliant, would be to have a backup battery operated from the main power supply, but it would add complexity to the system.

    Thanks Phil for the PhotoFET link! I searched for suitable solutions but found none... this one looks very nice. I have to provide a normally closed solution, but this wouldn't be too difficult. I'll prepare a test circuit with this one.

    Massimo
  • max72max72 Posts: 1,155
    edited 2015-01-02 04:47
    Frank,
    in Italy we had a free for all approach until recently. Now things are changing, and they are discriminating between drones to be used in non critical areas and critical areas.
    One the discriminats is the presence of people below. In this case the drone must comply to other requirements, including maximum mass and impact force, propeller protection, and so on.

    Massimo
  • jmgjmg Posts: 15,173
    edited 2015-01-02 18:55
    max72 wrote: »
    ..
    The requirement is an independent system. Consider it a last resort, in case everything goes crazy, and possibly never used.
    It must use a different radio and receiver, and a separate battery. It must kill the system and let the drone fall down. ..

    So it must be triggered from land. The flight zone is less than 300 m, so little issues with loss of signal, but in case of signal loss it must not automatically activate.
    In case of a failure of the emergency system it should not be triggered.
    That's a little self contradicting, having a dead-man type design, that is not allowed to activate if the link goes down ( thereby disabling the crowbar)

    Still, if that is what they really want / spec, then you do not need to switch serial and i2c, you need to remove the total system power (A failure could be a power-device failure, or a zapped controller, so link switch is not enough)
    - for that you could try a * Photovoltaic Output Optocoupler and a suitably rated Power MOSFET, wired to power ON when there is no current from the optoCoupler & Optocoupler over-rules the gate resistor when activated.. That gives total power removal, so covers all failure modes, and also gives you a brief re-boot option as well.
    It also ensures no driven props on falling, which should be safest impact wise.

    * Photovoltaic Output Optocoupler - typically source 5-15uA when driven, and behave like a chain of small solar-cells.
    eg TLP190B specs 8V clamping (Io = 0uA) and 12uA Iout, (shorted)
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2015-01-03 04:01
    erco wrote: »
    Can you deploy a parachute when the "fall from the sky" command is received? That's a win win for everyone concerned.

    In theory parachutes are a great idea, but in practice you may need to have a small explosive charge deploy an emergency parachute. That's a bugaboo. What happens in the parachute deploys accidentally while someone is handling the aircraft during pre-flight?
  • max72max72 Posts: 1,155
    edited 2015-01-03 05:34
    Thanks for the insight.
    I guess the idea is to have a backup safety system.
    So you should reasonably avoid single point failures (in this case my safety system). The basic system has a lot of redundancy, and the idea is not of break the redunded system by adding a "safety" feature. For what I understand the system has double flight control, extra propeller, multiple power supply, go home and various safety functions.
    The extra system that we must add is to be used if something extremely unexpected happens. And in this case you simply drop the extremely expensive drone. So in principle the failure of the extra safety system alone should not drop the drone, because we have many other options.
    While I like the idea of explosive bolts,I think the only viable solution for a parachute is a servo driven lid. Let's say if we are lucky we'll reduce the damage.

    I have another more general option that cuts all the power, but currents involved are quite big, so I was investigating a small signal/voltage alternative...

    Massimo
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2015-01-03 06:24
    No explosive bolts, just a blank cartridge of an appropriate size -- a big parachute might use a shot gun shell, a drone might use something like a .32 cal handgun blank.

    The problem is that you are deploying an emergency parachute close to the ground. Just waiting for it to slide out of a tube and catch air might be too late.

    Frankly, I think the best safety mode is to attempt to go into a hover and a slow decend before you run out of power. I can't grasp the concept of you need to comply with. I do understand that it is Italian, and that may just be a legal requirement that isn't well concieved.

    I would first comply minimally to get certification and then see if the local authorities change their minds. The world has a lot of rule makers that really don't understand what they are doing or asking.
  • Mark_TMark_T Posts: 1,981
    edited 2015-01-03 06:35
    I would avoid a latching relay, that is clearly not fail-safe! You want something that needs power
    constantly to enable the flight system.

    The power required to operate a relay during flight is tiny compared to the power to the motors (100W is typical
    of a mid-sized drone). I would suggest a dual reed relay if you can find as the lightest option. is there any
    reason against opto-isolation? An opto-FET might be possible.

    Can't see a parachute being reliable from a tumbling craft BTW, too likely to wrap round things, though
    it could be deployed from a vertical mast to increase its distance from CoM.

    [ I am presuming the reason for the cut-out is when control signal is lost or a separate manual override is
    selected - you want to prevent it moving outside the operating zone should this happen - the cut-out itself
    has to be high reliability/certified, then the rest of the craft can be experimental ]
  • max72max72 Posts: 1,155
    edited 2015-01-03 11:30
    Thanks,
    I have opto FETs on order. That would be the best solution. I only have to invert the logic.
    The requirements are rather strange, but they grouped rc airplanes and multicopters, so things are messed up and sometimes illogical.
    Anyway this is what me must comply with....
    Massimo
Sign In or Register to comment.