Chassis earth vs sensor earth

[Gareth]

I have wondered about this for a while but as it is fresh in my head from wiring up some LS coils (and I need to have a break from staring at bunches of wires ) I thought I'd post about it and get others opinions on this.

So over the years I have built hundreds of custom engine harnesses and have always followed manufacturers diagrams for earthing of various engine sensors, coils, etc... and have always been of the belief that certain sensors create electrical noise that could effect the signals of other sensors sharing the same earths.

So recently I built a custom harness for a bike running a M130 motec, the harness I was replacing was a 'milspecwiring.com' custom harness (that cost thousands) and I noticed that the ignition coils (inbuilt igniters so sensor earth and chassis earth) where earthed together and terminated at the engine block.... So I went against what I know and copied that, we found no issues with the bike or ignition system or other sensors.

So I suppose my question is, when sensors are earthed at the ecu, is there some circuitry that prevents noise from effecting other circuits? Why would a well known harness maker not use them?

Thoughts?

[Dylan]

I've made a fair few coil looms for LS coils and yes they earth with the ECU for power supply through the ECU for trigger. I've also done a few for the Holley smart coils and every other brand of that same coil and they have an extra earth wire that is to be seperate and they state that it's to be under 300mm and must be fixed to the same cylinder head as that coil is firing on.

No direct answer but just some observations.

[vlad01]

Yes there is decoupling caps and some filtering on the input/output on the ECU. Look at the 808 and all the rows of caps right on the connection pin pads.

Basically when I do my grounds I runs more noisy devices on a separate ground to the more sensitive devices but I still couple the grounds close to the 2 grounding points in most cases so they have a balanced potential and also for redundancy if one ground point on the block develops resistance or open circuit . Also just below the 2 grounds coupling joins I run heavier gauge wire to the block points so there drop is less and one ground is more than capable of providing next to 0 drop for both earth lines in case of failure. I often run the heavy gauge up to the common sensitive ground junction point for lest drop. See diagram

grounds.PNG (7.17 KiB) Viewed 5613 times

[Gareth]

Basically when I do my grounds I runs more noisy devices on a separate ground to the more sensitive devices but I still couple the grounds close to the 2 grounding points in most cases so they have a balanced potential

I don't follow what you mean by 'balanced potential'? If the earth at the same point then surely noise will transfer between circuits.

Sorry my electronic engineering knowledge is limited.

[vlad01]

There is 2 separate grounds for load and potential balancing as well as redundancy. As long as the wire is thick enough there will be little drop and therefore little interference, 2 connection points also doubles the current carrying potential and balances any discrepancies between the 2 points. The fatter wires below the junctions ensures one ground can't lift the voltage on another, similar the coupled grounds offer half the drop to any load at any time independent of which main ground line it comes down without lifting the voltage of any of the others.

For ground to be interference free it needs to at 0 resistance and 0 voltage drop, in real world due to physics this is not possible so you can only improve it to close as 0 as possible. Its really just simple ohms law when it comes to grounds. Just treat the ground wiring as a resistor voltage divider circuit and imagine a signal generator placed in parallel across the loads and ground circuit. You can then infer the peak to peak signal across the ground and how it adds or detracts from the desired voltage dividers that are the sensor circuits. What you want is all the signal to appear on the sensor side and none of it to appears on the ground. If some does it offsets other sensor circuits on the ground by the amount of voltage drop.

1. Good wire gauge lowers the drop

2. Adding 2 or more grounds multiplies the current carrying potential, further lowers drop. Iron has significantly more resistance than copper, more points means more contact surface area which compensates for this increased resistance.

3. Tying all grounds together 0s out the ground potential for all grounds (makes all of them as one and the same big ground so there is no voltage differences between them)

4. Tying them at the thick gauge part of the ground closer to the ground point further improves on the above point, like a bus bar basically.

5. having sensitive grounds and noisy grounds on separate groups ensures voltage drop of such typically long run section of the ground wires in the loom don't lift/drop voltages on the sensitive group. Once they joint the main "bus bar" section the drop therefore on towards the grounds are negligible and has next to no effect.

Hope that makes sense.

[VL400]

No thats not really correct, and its absolutely not for redundancy in the case of the two earth connections in the factory ECU configuration (noisy/high current and the quiet/signal earth) either. Its called star grounding and is the most common way to ensure that there are no earth potential differences induced by high current switching of injector and ignition drivers, relay switching, solenoids etc and low current high sensitivity signals for things like MAP and temperature sensors. GM didnt run multiple wires (of different gauges) for no reason to the same or similar points on the block, they'd skimp on anything if they could so you know it was for a reason. One carries high current and the other low current, its nothing to do with balancing anything. The injector driver IC ground is not the same ground inside the ECU as the signals ground. The common point, or star ground point, is on the block (there is a crap load on star grounding online so not going in to details here unless people want it). There would be absolutely no point running two wires of different gauges to the block if they just joined back up in the ECU either.


In the case of LSx type coils that bigvl was asking about, there is a signal earth and switching/driver earth as noted. They should be both earthed independently using two wires of different gauges, the switching earth to the block and the signal earth to the ECU signal ground. The switching earth carries all the current from the driver IC to ground during dwell so needs to be heavy gauge wire, the signal earth only a very tiny current for the logic level trigger. You can get away with and not notice anything at all (within reason) if the high current ground wire drops 1 or even more volts, but if the signal wire drops the same 1 or more volts due to being tied to the same ground at the coil while the coil is dwelling (ie a single wrongly sized ground wire used) it can cause massive issues. If you multimeter the two earth wires on a LS coil they are not joined together inside the coil, this is all part of being able to star ground the coil and not induce noise (from a raised ground potential) in to the device switching the coil or have the coil false trigger itself from a bad grounding

[Dylan]

Out of interest on an LS1 loom the 2 earth point's are joined together inside the loom from memory

[immortality]

Out of interest on an LS1 loom the 2 earth point's are joined together inside the loom from memory

Yes that would seem correct. See

02a.pdf

(486.55 KiB) Downloaded 311 times

[VL400]

No they are still split earths. The coil harness on each bank combines the signal and driver earths to one pin each, the signal earth goes back to the PCM and the driver earth goes direct to the block.

LS1 Coils.png (68.45 KiB) Viewed 5531 times

[Dylan]

I may of explained it wrong. The 2 earth point's on the harness are connected inside the loom with a wire, I think.

I have one pulled apart I'll double check