EMF / EMI causing MCU to hang? Help Needed

[donperry]

Good day guys,

I'm building an arduino based water injection controller for motor vehicle. It Uses the Airflow sensor and manifold pressure sensors as input. On my test bench the device works fine, doesn't hang and could go on for hours.

In the car, on ignition only (car not running) the device works fine. However, once the car is running it freezes after a few seconds/ minutes, just randomly whenever it feels like.

I opened up the pressure trace (to disconnect from circuit) and the device still hanged.
I opened up the Air flow trace and then the problem goes away.

The pink line is the MAF line, the other is pressure

So now I'm thinking the MAF (air flow ) sensor input is causing this. I'm suspecting some noise or spikes or something is causing this. What approach would be best in keeping things stable here?
I have attached complete and partial schematic, PCB print, and actual device

 

[MikeMl]

I dont understand What FET2 is doing.

The "isolated" power supply does nothing. Why is it even there?

I dont understand your method of snubbing the coils?

Your problem is related to improper snubbing of the inductive loads, and the common ground between the mcu and everything else...

 

[donperry]

The coils and everything work well on the bench, in the car the coils isn't even activated as yet before errors, so it's not the coil. Its 100% in the input trace as when that trace is removed i have no problem.

Please note that the ZENER symbol is not in the diagram as i couldn't find it in the program i'm using. Just know that a zener is there in the snubber.


the snubber design is this one:

 

[large_ghostman]

I might have got this totally wrong..........Is there any chance water is coming into contact with the sensor and shorting it to ground or whatever?

You dont have much in the way of caps and I cant see a ceramic 0.1uf decoupling cap anywhere, I could have missed this though.

 

[donperry]

NO chance of water. That very sensor drives the car.

I only have one decoupling cap, top right of screen near the power source, but no caps in the signal line. What do you have in mind for me?

 

[large_ghostman]

NO chance of water. That very sensor drives the car.

I only have one decoupling cap, top right of screen near the power source, but no caps in the signal line. What do you have in mind for me?

I dont know anything about the sensors so i am reluctant to suggest anything on the signal line, I would be tempted to beef up the caps a bit on the supply though and add a 100nf ceramic into the mix, I doubt it will cure it but it helps cut down further issues cropping up.

Its possiable it could be something simple like board layout and the ground plane, I think for cars etc its better to have large ground planes and plenty of copper fill.

I recently redid a board for servo control that kept resetting and the only way I could sort it out was a board redesign, it can be a bit of a black art. Sorry its not more helpful!

 

[large_ghostman]

I cant tell on my screen but is this a short

Please dont think i am criticizing but some of those joints dont look 100%, may or may not be a problem now but I would sort them out so they dont cause a problem in the long run.
I hate commenting on peoples hard work doing boards!! But sometimes its as simple as layout or board design

 

[kubeek]

Could you try redrawing the schematic? I can´t really tell what goes where, and the colours don´t seem to match anything and change arbitrarily over the course.
Use lables such as VCC and Ground point wherever possible and minimize crossing wires, all I can see now is for example a box of 8x8 wires all crossing about near the second lowest sensor.
Some places there are the junction blobs, some places there is nothing, some places the junction point is just hanging in the middle of a wire, how is anyone supposed to make heads or tails of it? And then there´s small junctions and larger juctions, what is that about?

 

[large_ghostman]

Could you try redrawing the schematic? I can´t really tell what goes where, and the colours don´t seem to match anything and change arbitrarily over the course.
Use lables such as VCC and Ground point wherever possible and minimize crossing wires, all I can see now is for example a box of 8x8 wires all crossing about near the second lowest sensor.
Some places there are the junction blobs, some places there is nothing, some places the junction point is just hanging in the middle of a wire, how is anyone supposed to make heads or tails of it? And then there´s small junctions and larger juctions, what is that about?

To be fair to the OP we get alot worse! He has at least put a fair amount of effort and work in IMHO.


Out of interest is that lead free solder? I would be tempted to redo alot of those joints with real solder

 

[kubeek]

LG and donperry, sorry if that came out too harsh I didn´t mean that. But really I cannot make heads or tails of what the schematic is supposed to od or how the sensors and fets are wired into that.

 

[large_ghostman]

Thats ok its easy to sound harsher than intended online. I think ultimately its hard to spot the problem with so many 'little' things that need correcting.

 

[ronv]

Good day guys,

I'm building an arduino based water injection controller for motor vehicle. It Uses the Airflow sensor and manifold pressure sensors as input. On my test bench the device works fine, doesn't hang and could go on for hours.

In the car, on ignition only (car not running) the device works fine. However, once the car is running it freezes after a few seconds/ minutes, just randomly whenever it feels like.

I opened up the pressure trace (to disconnect from circuit) and the device still hanged.
I opened up the Air flow trace and then the problem goes away.

The pink line is the MAF line, the other is pressure

So now I'm thinking the MAF (air flow ) sensor input is causing this. I'm suspecting some noise or spikes or something is causing this. What approach would be best in keeping things stable here?
I have attached complete and partial schematic, PCB print, and actual device

Both the ECU and micro run off 5 volts?
You could try a .1 ufd cap across the 10k resistor. That zener is very high voltage did you mean for it to be that high?

 

[Mikebits]

Is this one of those HHO gizmo's? As for your circuit malfunction, the following article from EETimes describes the signal integrity in automotive issues and some of the methods to prevent them. **broken link removed**
Does your Arduino have configurable pullups? If so you might want to try that for starters. Along with suggestions by other members here.

**broken link removed**

Give it a read and good luck with your endeavor.

 

[large_ghostman]

One other small thing, I know I said copper planes were good, but little islands of copper all over the place and not connected to ground via a cap are not good. I am not having a go but this is why its better to take some care and a bit of time making a board, it sure saves agro later.

 

[donperry]

THanks but that via actually connects right on to that pin of the jack. And the diode pad i agree is a bit uncomfortable there, but it was fine too.

I cant tell on my screen but is this a short View attachment 97631

Please dont think i am criticizing but some of those joints dont look 100%, may or may not be a problem now but I would sort them out so they dont cause a problem in the long run.
I hate commenting on peoples hard work doing boards!! But sometimes its as simple as layout or board design

 

[donperry]

First of all, thanks for the replies. I didn't get any email notification, i didn't know you guys were replying.
Secondly, I'm just learning but I am an enthusiast so i'm able to do all this in a few months of know how, but still a lot left to learn
finally, let me anwer a few question.

Can the schematic be drawn over?
I'd love to. I know there has to be a better layout. But still learning.

Is this lead free solder?
Yes. lead free,

That zener is very high voltage did you mean for it to be that high?
Yes, the power supply on the board can manage up to 30v, a car's alternator puts out around 14, an 18V zener i guessed should suffice there.

Is this one of those HHO gizmo's?
This is a water-methanol injection controller.

 

[large_ghostman]

First of all, thanks for the replies. I didn't get any email notification, i didn't know you guys were replying.
Secondly, I'm just learning but I am an enthusiast so i'm able to do all this in a few months of know how, but still a lot left to learn
finally, let me anwer a few question.

Can the schematic be drawn over?
I'd love to. I know there has to be a better layout. But still learning.

Is this lead free solder?
Yes. lead free,

That zener is very high voltage did you mean for it to be that high?
Yes, the power supply on the board can manage up to 30v, a car's alternator puts out around 14, an 18V zener i guessed should suffice there.

Is this one of those HHO gizmo's?
This is a water-methanol injection controller.

Biggest tip I can give you for now, especially when first learning only use lead free solder on boards you want to eat! Its horrible stuff to use and joints can be hard to make with it, I am pretty certain you will have no end of problems with that board using the lead free stuff, real solder is MUCH much easier to use and get a decent joint. Dont feel bad I do alot of soldering and struggle with the lead free ****.

 

[large_ghostman]

If I have all the symbols I will try and redraw the schematic later for you, I am not great with them as i have my own way but if I have the symbols I might try and give it a go later.

 

[donperry]

Alright buddy

 

[large_ghostman]

The packages i normally use dont have all the parts, if I get a chance I will give it a go at the weekend but will need to make some symbols for the software.
If I do manage to get it done I might as well do a board layout for you as well, what size PCD you got in mind?