Random USB link loss on engine cranking despite isolation. [FIXED]

[ITman496]

Hello all!  I'm making an airplane!

FIXED EDIT:  I grounded the datalogger to the engine block and now all is well. 

I am also having a problem.  Normally I make nice all put together oshpark circuit boards but in the interest of doing a quick test of the engine to make sure the cooling system is up to snuff, I've put together a very rough data capture system with an arduino mega.  It connects to four thermocouples, two thermistors, a manifold pressure sensor, a fuel pressure sensor, a throttle position sensor, and 3 hall effect rpm sensors. I am, however, having an issue.

Despite the system running fine for hours on its own, and being completely electrically isolated, grounds and positives, from the starter motor on the engine, when I crank over the engine for a few seconds, the serial usb connection to the computer appears to hang up and despite the TX light still blinking, the computer will not receive data until I unplug/replug.  This behavior was verified on a raspberry pi as well, so its not just my computer.  And again, all the sensors are electrically isolated from the starter motor and engine block, not a single ground shared, even the thermocouples are isolated.  And the starter motor and associated wiring are on the other side of the engine block!

My question is.... what gives!  And how can I deal with it?  I also want to say, again, please ignore the mess, it's meant to be a temporary system.

A further note is that it does not lose serial connection every time, and it appears to lose it upon releasing the starter button.  Arcing inside the starter solenoid maybe?  What can be done to help the system out?  Another quirk is that when I connected the engine block to my lab supply, which is also an isolated power supply, I just checked with my meter to wall ground, no direct connection... the thermocouple amps start freaking out and intermittently not being able to read data.

What can I do?  Nothing makes any sense.  It's all isolated and yet it affects everything dramatically anyway.  Why would connecting an isolated power supply (megohms measured to ground) cause it to freak out?  The system itself never appears to crash, but the usb connection to the computer.  I have yet to directly tie to serial and see if it actually reboots or not.

Would grounding the datalogger to the engine block help in any way?  Also there is currently no ignition system attached.  So it can't be interference from the spark plugs firing..  The CDI and coils and such are all laying on the table in the background. 

I just really hope its obvious because I'm at my wits end troubleshooting these intermittent issues and am mostly just hoping that someone goes 'ha idiot you just have to do X' and it will all be better.  This is why I should have gone to college..







And have some pictures of nicer things..

[langwadt]

USB can be iffy with noise, do you have something like an FTDI cable so that the wires are serial and the USB part is in the plug right at the computer?

[wraper]

Is it original Arduino Mega? Asking because it is crap with broken USB connection. Designed by some idiot who has no clue how to properly make GND connection and ESD protection.

[amyk]

Another quirk is that when I connected the engine block to my lab supply, which is also an isolated power supply, I just checked with my meter to wall ground, no direct connection... the thermocouple amps start freaking out and intermittently not being able to read data.

Something is picking up noise. Any long enough cable can act as an antenna.

[coromonadalix]

my 2 cents

Could a supercapacitor hold the supply while it crank     use it as a rechargeable battery with an anti reverse diode ...

Done that for an obd2 scanner with an 2 farad / 5 volt cap, dont recall the resistance in seriaL to avoid a good surge while charging  ....

Or some usb isolators  wiith included dc-dc

[ITman496]

Is it original Arduino Mega? Asking because it is crap with broken USB connection. Designed by some idiot who has no clue how to properly make GND connection and ESD protection.

It's a knockoff of the original mega, I think..

USB can be iffy with noise, do you have something like an FTDI cable so that the wires are serial and the USB part is in the plug right at the computer?

I will try that!

Another quirk is that when I connected the engine block to my lab supply, which is also an isolated power supply, I just checked with my meter to wall ground, no direct connection... the thermocouple amps start freaking out and intermittently not being able to read data.

Something is picking up noise. Any long enough cable can act as an antenna.

The weird thing with this is that the engine is not cranking when this issue happens.. it just freaks out while sitting there.

my 2 cents

Could a supercapacitor hold the supply while it crank     use it as a rechargeable battery with an anti reverse diode ...

Done that for an obd2 scanner with an 2 farad / 5 volt cap, dont recall the resistance in seriaL to avoid a good surge while charging  ....

Or some usb isolators  wiith included dc-dc

Sorry, I should have been more clear, they are not powered by the same source, when I say isolated I mean isolated.  The data capture is powered by usb from the computer, the starter is powered by a battery.  No grounds shared (though I'm going to try sharing them and see what happens) and no power shared either.  I can plug and unplug the battery and the data capture system stays running.

[ITman496]

Update!

It's fixed now.

All I did was, at my electronics teacher's advice, ground the datalogger to the engine.  Now no matter what I do, what power supply I clip on or don't, crank or not crank... no errors, buttery smooth.

So despite none of the sensors having any electrical connection to the engine and no meter measurement being possible from engine block or +12v on the engine to the data logger (infinite resistance) it certainly does matter.  A bit counter intuitive but...  hey whatever it works now.

[bson]

Arduino? Cut the trace connecting the USB shield to signal ground...

[ITman496]

Would that help?

[m3vuv]

just out of interest is that a rotax microlite motor?

[wraper]

Arduino? Cut the trace connecting the USB shield to signal ground...

And how disconnecting USB shield is supposed to help to its operation in presence of transients? If it was original arduino, I would say connect USB shield to GND directly, since they made something ridiculous with it. But as it is a supposedly clone, it depends on how it was made.

[ITman496]

just out of interest is that a rotax microlite motor?

close!  It's a kawasaki 440A with a conversion kit to make it into an aircraft engine.  Similar vein to the rotax engines.