What is the I2C Octo Isolator
- Thread starter JLO
- Start date
Bartman
Welcome to MultiRotorForums.com!!
JLO
The Mikrokopter flight controller sends all motor commands to the motor controllers via two small wires. The motors know which commands are for them by the address included with the command. Commands addressed to other controllers are ignored. This single stream of information works great but when one controller or the stream itself gets corrupted, all of the controllers can become corrupted causing everything to shut down. To be fair, I've never heard of this actually happening but it's possible.
The I2c isolator receives the I2c stream and breaks it up into individual command streams for each motor and send them directly to each motor controller (BL in MK speak). If an ESC goes bad, the isolator continues to send good information to the remaining ESC's and also prevents the bad ESC from corrupting the I2c stream.
It's a worthwhile addition to an MK stack, I've got one on my bench that will go into my next build.
Hope that helps,
Bart
The Mikrokopter flight controller sends all motor commands to the motor controllers via two small wires. The motors know which commands are for them by the address included with the command. Commands addressed to other controllers are ignored. This single stream of information works great but when one controller or the stream itself gets corrupted, all of the controllers can become corrupted causing everything to shut down. To be fair, I've never heard of this actually happening but it's possible.
The I2c isolator receives the I2c stream and breaks it up into individual command streams for each motor and send them directly to each motor controller (BL in MK speak). If an ESC goes bad, the isolator continues to send good information to the remaining ESC's and also prevents the bad ESC from corrupting the I2c stream.
It's a worthwhile addition to an MK stack, I've got one on my bench that will go into my next build.
Hope that helps,
Bart
MK uses an addressable bus where all the BL controllers are listening to the same data stream for commands from the flight controller specifically for themselves. If something happens that shorts one of the lines carrying the data then all BLs lose communication with the flight controller resulting in all motors doing the same thing which is generally shutting off or something equally catastrophic.
Standard ESCs use PWM (Pulse Width Modulation) on their own data lines from the flight controller and a problem on one data line does not cause problems with the remaining data lines so they do not need to be isolated, they already are.
The idea behind the octo isolator board is to take the single I2C data bus from the flight controller and split it into 8 buffered data lines so that a problem on one of the BL controllers I2C bus will not affect the communication with any of the other BL controllers. Should one of the BLs fail downstream of the isolator the remaining ones can still work and theoretically the multi will still fly rather than crash. The reality is while an octo is usually capable of continuing to fly when a motor fails, it is the only non-coaxial configuration that can do so reliably.
That said if you feel you need an I2C isolator board I have one I'm not using and no plans to use again that I will sell for a reasonable price.
Ken
Standard ESCs use PWM (Pulse Width Modulation) on their own data lines from the flight controller and a problem on one data line does not cause problems with the remaining data lines so they do not need to be isolated, they already are.
The idea behind the octo isolator board is to take the single I2C data bus from the flight controller and split it into 8 buffered data lines so that a problem on one of the BL controllers I2C bus will not affect the communication with any of the other BL controllers. Should one of the BLs fail downstream of the isolator the remaining ones can still work and theoretically the multi will still fly rather than crash. The reality is while an octo is usually capable of continuing to fly when a motor fails, it is the only non-coaxial configuration that can do so reliably.
That said if you feel you need an I2C isolator board I have one I'm not using and no plans to use again that I will sell for a reasonable price.
Ken
Thanks for the explanation.
So what you are saying is this is only an issue with MK controllers. I have standard 70 Amp speed controllers for a large octocopter platform and will more than likely use Arduino. So am I correct in saying with my setup the I2C isolator will serve no purpose as each speed controller has it’s own data line?
One other question. My speed controllers will be powered from a ring main supplying all eight speed controllers. In the unlikely event one of the speed controllers fails with a dead short on the line side surely it will bring down all the other controllers. Is this an unlikely scenario or should one consider fitting a fuse on each line tap supplying each speed controller?
Thanks.
So what you are saying is this is only an issue with MK controllers. I have standard 70 Amp speed controllers for a large octocopter platform and will more than likely use Arduino. So am I correct in saying with my setup the I2C isolator will serve no purpose as each speed controller has it’s own data line?
One other question. My speed controllers will be powered from a ring main supplying all eight speed controllers. In the unlikely event one of the speed controllers fails with a dead short on the line side surely it will bring down all the other controllers. Is this an unlikely scenario or should one consider fitting a fuse on each line tap supplying each speed controller?
Thanks.
Bartman
Welcome to MultiRotorForums.com!!
whirlwind,
if your arduino board has individual ESC outputs (which I assume it does) then you don't need an I2c isolator. I believe the isolator is unique to MK.
as for a short in an ESC, maybe a fuse would work? I haven't seen that suggested before but it doesn't mean it isn't worth a try. I don't have enough experience with electronics to know but maybe someone else can comment? if you had a dead short, the motor would fail and if it's a quad you'd crash immediately. a hexa would probably give a good wobble and you'd be in a rush to land probably. an okto would give a less noticeable wobble but how long the batteries would survive is anyone's guess. Wouldn't the short burn through given the relatively thin wires and high amps thus allowing the batteries to recover?
if your arduino board has individual ESC outputs (which I assume it does) then you don't need an I2c isolator. I believe the isolator is unique to MK.
as for a short in an ESC, maybe a fuse would work? I haven't seen that suggested before but it doesn't mean it isn't worth a try. I don't have enough experience with electronics to know but maybe someone else can comment? if you had a dead short, the motor would fail and if it's a quad you'd crash immediately. a hexa would probably give a good wobble and you'd be in a rush to land probably. an okto would give a less noticeable wobble but how long the batteries would survive is anyone's guess. Wouldn't the short burn through given the relatively thin wires and high amps thus allowing the batteries to recover?
With MK it can actually be far worse. If a BL fails and starts to burn a FET the result usually takes out at least three of your motors and probably 4 Bl's in about 5 seconds. Also the PDB. Seen it happen a few times now. The biggest problem with MK is that everything is open plan and gathers dust and debris every time you fly. Anything that is conductive has a million places to lodge and start a chain reaction. Just my luck to have seen it a few times I suppose. They don't call it the fireworks kit for nothing. I have seen it happen when the copter was sitting on the ground acquiring sats.
Last edited by a moderator:
Bartman
Welcome to MultiRotorForums.com!!
what i don't understand is that I've heard a lot of times about BL's and cascading failures of the I2c but Shawn's Oktokopter that survived a motor failure was actually a BL failure that didn't cause any further failures. was it just luck or is the I2c vulnerability overstated?
Thanks for all the replies, great info which all helps!
I am in the process of putting together a Skyjib 8 and I want to build in as much safety / redundancy as possible hence my questions. What you are saying regarding the I2C Isolator and MK ESC combination makes total sense as I did not realize MK ESC share a common signal wire so for my application the I2C isolator is not necessary.
Regarding my second question concerning fusing each speed controller. I believe your comment about the wires burning through within the failed ESC is a likely scenario which will allow time for a safe landing. Providing the current draw is not so high that the remaining ESC’s shut down.
As a matter of interest I was at the local flying field this last weekend. A guy comes to the field with a brand new fixed wing foamy electric park flyer. He plugs in the battery and leaves the plane lying on the ground while he does an aileron / flap mix on his transmitter. The next moment the plane launches forward destroying the prop and the motor is left spinning at high RPM. I immediately picked up the plane and unplugged the battery. The ESC was smoking and melting its way through the foam. This all happened within seconds, I do not know if he had calibrated the ESC. He did not increase the throttle but maybe he did do something else wrong, I should have asked him!!!! The point is ESC’s do fail from time to time and I think it is important to have some kind of redundancy. I am guessing but in a little park flyer like that the current draw during the failure could be in the region of 30 amps or more. On a big octo with the same type of failure the current draw could be in the region of 200-300 amps. That is a lot of power which could easily destroy a very expensive piece of kit. Maybe fusing each ESC with lets say an 80 Amp fuse is not such a bad idea. I have found some very small quick burn fuses plus holder that would do the job.
I am in the process of putting together a Skyjib 8 and I want to build in as much safety / redundancy as possible hence my questions. What you are saying regarding the I2C Isolator and MK ESC combination makes total sense as I did not realize MK ESC share a common signal wire so for my application the I2C isolator is not necessary.
Regarding my second question concerning fusing each speed controller. I believe your comment about the wires burning through within the failed ESC is a likely scenario which will allow time for a safe landing. Providing the current draw is not so high that the remaining ESC’s shut down.
As a matter of interest I was at the local flying field this last weekend. A guy comes to the field with a brand new fixed wing foamy electric park flyer. He plugs in the battery and leaves the plane lying on the ground while he does an aileron / flap mix on his transmitter. The next moment the plane launches forward destroying the prop and the motor is left spinning at high RPM. I immediately picked up the plane and unplugged the battery. The ESC was smoking and melting its way through the foam. This all happened within seconds, I do not know if he had calibrated the ESC. He did not increase the throttle but maybe he did do something else wrong, I should have asked him!!!! The point is ESC’s do fail from time to time and I think it is important to have some kind of redundancy. I am guessing but in a little park flyer like that the current draw during the failure could be in the region of 30 amps or more. On a big octo with the same type of failure the current draw could be in the region of 200-300 amps. That is a lot of power which could easily destroy a very expensive piece of kit. Maybe fusing each ESC with lets say an 80 Amp fuse is not such a bad idea. I have found some very small quick burn fuses plus holder that would do the job.
Bartman
Welcome to MultiRotorForums.com!!
it's a tough call to make based on a foamie doing the puff and burn routine. which foamie was it? i've seen lots of comments lately regarding hobbyking RTF stuff where the ESC's and motors are just junk. their RTF motor gliders are cheap cheap cheap but the motors and controllers are practically unuseable. the chinese ESC's (Hobbywing, Turnigy, Maybach, ExceedRC, etc.) aren't expensive but i don't recall ever hearing about them being unreliable, surprising as that may be. In fact they've been surprisingly reliable for myself and others that use them.
all of this stuff works based on speed of control signals and soundness of connections. you've got to really ask yourself if something is necessary before introducing another layer of connections into the system.
regards,
bart
all of this stuff works based on speed of control signals and soundness of connections. you've got to really ask yourself if something is necessary before introducing another layer of connections into the system.
regards,
bart
Mactadpole
Member
I have to agree with Bart and your response, more connections definitely equals more points of failure. This is one of the reasons I got rid of my Futaba TX/RX and went to the MX-20. With Futaba I had to use a pwm-ppm converter between rx and FC. Between signal translation and many many connections I considered it the weak point of my setup. I too considered the I2C isolator in the beginning but then went with the "keep it simple" theory as well. My in-flight BL failure resulting in the motor stopping was scary but all was fine. Was I lucky? YES! Did it convince me to change my setup? NO. I think we all have to realize that these are expensive flying machines and we have to be content with the fact that they could fall out of the sky at any moment due to user or hardware/software error. All we can do is be faithful at doing pre- and post-flight checks!
My 2 cents,
Shawn
My 2 cents,
Shawn