Ground loop problem?

[dnb]

I have an annoying little problem with laptop serial comms.  I'm sure it was never a problem before, but it is now.  (I have a slightly different, but related problem on standard ECU'd Imprezas too, so it's not all the fault of VEMS)

When I connect the laptop to the ECU all is well when the laptop is running on internal batteries.  (When the laptop is not connected at all, all is well too, but that's what anyone would expect...)

However, when I use my 12v to 20v boost converter (bought from Maplin) to power the laptop, the CLT jumps straight to 101C (but sometimes drops back a few degrees briefly) and the IAT jumps up by 10 degrees or so.  Other sensors may be affected, but I haven't looked that hard.

I'm sure this didn't happen before the car doors (where the fag lighter socket is) were retrimmed & the wiring in them repaired.  But then there's no accounting for TVR wiring!!

Any bright ideas as to how to stop this?    Short of optoisolating or making the serial port into a bluetooth port, I am out of ideas - A perfect ground plane for the whole car is just not going to happen...

Sounds like a good time to look at these:  http://www.brainboxes.com/product/bluetooth/BL-730.aspx

[hilly]

It sounds like the connection in your VEMS loom between GND and GND5 is higher impeadance than optimal.

The RS232 port uses GND (as do the ADC references) and when you hook it up to the laptop it links it to the PC ground.
When you are running on batteriues it is not a problem as the laptop is islolated.
As soon as you plug in the power cord the RS232 ground is linked to the power cord ground.
As the laptop is pulling a fair bit of current to charge the batteries there will be voltage drop in the power cord and associated TVR earth wiring (being a fibre glass car).
This voltage drop added to the RS232 ground (and ADC ground) will effectively add and offset to the input signal.

If you hook the laptop power cord to the same earth point as the VEMS do the offsets lessen ?

Try measuring the resistance between GND and GND5, is should ideally be less than 0.5 ohm.

Hilly

[dnb]

If I ground the laptop from the block I'm sure the problem would at least reduce if not go away.  However this isn't the most practical of solutions...

Oh the joys of a fibreglass car!

[rob@vems.co.uk]

Are all your grounds joined 10 to 15cm from the EC36?
I ground the RS232 shield to the join point.

[dnb]

They're joined 20cm from the EC36.  Best I can do in the space... 

[hilly]

Having the GND and GND5 joint at 20cm shouldn,t really cause a problem if you used a decent gauge of wire.

My own personal rule of thumb is to use all wire for grounding at least twice the gauge (more if possible) than anything else, have lots of them and make sure it is a very good connection to the car as poor ground impeadance causes so many problems.
Remember all the currents from the different supplies (injectors, coils, relays, idle valves, lambda heaters etc) ALL return through the grounds of the ECU.

Where is the ground for the RS232 lead connected to ?

Next thing to try is check the GND and GND5 just in case the current ones have failed somehow.

Hilly

[dnb]

I've had VEMS in the Griff for nearly 2 years now, without this happening.  So something has changed that I haven't noticed yet.  Maybe a joint has got caught up in something and broken.

Believe me - the engine loom I made for the Griff is over engineered    

It's a shame the rest of it (the stuff I didn't do) is so poor.  So unless I rewire the whole car I am unlikely to suceed at getting a good ground plane that has equal potential to GND5 everywhere in the car.

This adds another reason for wanting to have wireless comms with the ECU - it removes a long wire and gives total immunity from this ground problem.

[hilly]

If there is a problem with your grounding (which it appears there may be) then it is at least worth having a quick peek at the loom as it may have suffered a bit of damage which could eventually take out something else................

Hilly

[dnb]

Exactly... 

I've just looked at the engine loom & there is no problem.  Also, the datalog from this morning shows minimal noise.  (Laptop on batteries again)  Whatever is causing the problem must be in the body loom - which is virtually inaccessible without taking most of the interior of the car out.

I'll probably find the cigarette lighter wiring that you can't see is the automotive equivilant of bell wire!

[cliffb75]

If you expect to use the laptop power supply in the car often I'd suggest installing a power jack with a better connector. We tend to use banana plugs, wired straight from the battery (fused of course ).

For continuous use, we find the cigarette lighter plugs tend to overheat and melt the plastic. Plus they have an annoying tendency to work loose through vibration, and you suddenly find your laptop has gone flat when you thought it was charging.......