Jamie all too well knows my frustration with the documentation. Very understandable. Sounds like you are getting to the finish line!

Too tired to play outside tonight so I edited the wiring diagram for sensors so I have something to work with tomorrow night.
I have hidden all the wiring except crank, cam and knock sensors to make it easier to follow.

Based on Jamie's comments and the later MS3 wiring diagram, I think I have the crank sensor wiring sorted now.
A question was raised about the knock sensor wiring and its multiple ground paths. I have knock sensing disabled at the moment, but before I start using it I may need to revise that wiring. Suggestions?

http://forums.pelicanparts.com/uploa...1402324705.jpg

Sensor grounds vs all other grounds
 

Somehow I remember that all sensors need to go to the same ground at the MS3. Your Flywheel sensor ground goes to pin 1 and its shield goes to pin 2, cam sensor ground goes to pin 8, and the knock sensor grounds go to pin 14. 15-19 go to engine ground, and 7 and 9-13 are not used. The diagram does not show the rest of the sensor grounds, such as CLT, MAT, etc.

The documentation says that all grounds on the card are common, but that all sensors, and ONLY sensors, should go to the same ground pin to insure common ground plane for all sensors.

So I used pin 19 for sensor grounds. All other pins went to a single lug on the engine, to which all other actual grounds were connected. From the body lug (to which the transmission is connected), a single large ground went to the ground terminal at the battery using the same type of mesh wire as the ground between the transmission and the body.

Somewhere I read that you were chasing ground problems. Maybe this is part of it?

Brian

Look at the diagram here I first posted:

Bill: well, you are correct, and my post was not consistent with MS-Extra information. Turns out my knowledge was based on planning for a Sequencer with non-Extra information, then switching when the MS3 came out. And since the planning was already ingrained, I did not pay attention to the Extra guidance you posted. Sorry for posting the wrong path (pun intended) to grounding.

Brian

Brian,
Not a problem, I also didn't pay close enough attention to the later version of the diagram that Lapkritis posted as far as the grouping of sensor grounds is concerned. I am reviewing that arrangement right now, but it will be a real PITA to implement without undoing some serious harness work. I will post something on this later.

I can't tell you how many times I've hacked my harness and rearranged things. Looking at the schematics for the V3.57 board, there do not appear to be any differences in any of the ground pins. I doubt there is any difference in which pins you use. The most important point is to ground the shield/s only at one end (at the MS3box). I have one common ground on the engine for the MS3/sensors and a separate ground lug for the WB, coils, IAC, boost valve, etc..

Before I start cutting and soldering wires, this is my understanding of how the grounds should be hooked up.
The only wires I am uncertain about are the termination of the two knock sensor shields. I guess the only options are either at G1 or G2. November, got any hints?

Collecting all the sensor grounds at the engine and running one wire to pin 7 sounds simple but requires a lot of harness rework that I would like to avoid. What I have shown below is a grouping of the sensor grounds at the ECU end of the harness and then connecting to pin 7. This is about 30 minutes work at most, whereas revising these grounds at the engine end is a lot more work and will result in a cut and taped harness shroud. I'm sure you can understand my reluctance to head in that direction unless I really need to. Any comments?

http://forums.pelicanparts.com/uploa...1402406840.jpg

Bill Generally I like to ground all shields to the same spot. In this case, you can ground them all to Pin2. The purpose is that all the noise picked up by the shield is sent to ground at the same point and more importantly all have equal electronic potential values at ground.

These sensors (crank, knock, cam) all have both a positive and negative component to the signal. This is why you have a + and - side to the signal. The ECU is looking for a transition from high to low through a zero point (or vice versa). These signals are often very low (mV or lower) and any bit of noise picked up before the amplification will cause a lot of mis-readings. This is why they must all use shielded wires. The shield will collect all the noise like an antenna and then send it to ground.

In most cases you can send all sensor and ECU grounds to the same point. However, the low voltage signals require special consideration. Its just something that we have to deal with using modern digital electronics that are reading very small signal levels. Cars are electronically noisy.

Another thing to pay attention to is the position of the wires. Make sure that the engine harness, ignition wires and starter wires do not lay parallel to each other. If they have to cross or touch make them perpendicular. This will lessen the antenna affect and help reduce noise.

Let's just hope you don't also have MAP sensor noise issues with the ITBs. :) That is another area of frustration when tuning.

Bill, have you read the ground presentation from MegaSquirt? It's a simple PowerPoint I drilled into my head.

I may differ over the other guys, but I grounded all sensors to an intake lug. I vaguely remember MegaManual or the presentation stating to do so (stating not to use the chassis). I then took one of those fat grounds from the engine lid and attached at the same lug.

My sensors generate very little noise compared to some of the MS'ers on the Extra forum.

I'm so glad I drilled the ground requirements into my head as it paid off.

You can further reduce noise when building the harness by rolling the bundle of wires. Then proximity isn't such a big deal to other items. You'll see this technique more in high end motorsports harnesses as an advanced method for reducing interference.

Tippy- I remember the same ground all to the engine drill. I built my first msextra car in 2006-2007 following those instructions. About two years later they had changed the manual to rearrange the grounds. I've updated two older build harnesses since then which I fortunately had the foresight to include expansion capacity wires within.

Finally got some spare time today to revise the wiring harness and run a few tooth logs to check the result. Result is basically no change to what I had before on previous logs. I'll post the new logs below.

Wiring of sensors now looks like this. The only point I'm not sure about is the wiring of grouped sensor ground wires at G2. I have them connected to pin 7 as well as engine ground via a heavy spare wire I had in the harness ... not sure if engine ground is required.
It was a lot less work to join together all the sensor grounds at the ECU end of the harness rather than at the engine, however the changes to wiring haven't produced any positive results, so can moving this junction (G2) six feet towards the engine really have an effect? I just can't imagine how it can.

EDIT: The red wire from G2 to engine ground should be deleted.
http://forums.pelicanparts.com/uploa...1402827597.jpg

To summarise changes, I have ...

• grouped all sensor grounds together (at G2) and connected them to pin 7 AND engine ground
• grouped crank sensor and knock sensor shields (at G3) and connected to pin 2
• pins 14 to 19 connected together (at G1) and wired to engine ground
• for testing, I have wired up the starter and solenoid to a separate battery

What I haven't tried yet ...

• original Bosch starter (although the current HiTorque starter is electrically isolated)
• turning up R56 and R52 past four turns and one turn respectively
• switching the ignition input capture to rising edge (I will try this when I remember, but I don't see how this would have an effect on the tooth log)
• reducing crank sensor gap
• hooking up to oscilloscope (had a lead on one for sale last week but seller not responding now :()
• testing crank sensor (however, it must be working or there wouldn't be any signals!)
• changes to the noise filtering software (don't know how, yet)

What I have tried/checked ...

• connecting the starter and solenoid to a separate second battery
• adjusting R56 and R52 from fully CCW to four and one turns respectively (I guess I could crank them further and see what happens)
• 12K series resistor in crank sensor input signal and 2.2K resistor in parallel to ground (no significant difference with or without)
• cranking with and without compression = big difference in logs below
• crank sensor gap = 1.0mm

http://forums.pelicanparts.com/uploa...1402830953.jpg

http://forums.pelicanparts.com/uploa...1402830977.jpg

Have I missed anything important or got something wrong?

Operating table today ... seems like some of you might be familiar with this scene as well!
http://forums.pelicanparts.com/uploa...1402833155.jpg

Good list, keep cracking an you'll get it. The sensors ground to engine should be disconnected. Otherwise the benefit of isolation is negated.

Another option is a resistor on the power supply to all coils across the engine ground. OEM manufacturers ran these for a number of years. The manual has a chapter on this topic and sources from the bone yard. My suspicion is the Highlander coils. You could disconnect the coils with plugs in place so there is compression and disable injectors so you don't wash the cylinders with fuel while you confirm. If the noise is not present with all coils disconnected then we have isolated the source.

Yes, I thought that sensors ground to engine didn't look right after I had done it.

As far as coils are concerned, I have had them disconnected for all tooth logs except maybe the first few. I don't believe the coils can be causing the apparent noise problem in the flywheel sensor wiring as the coil circuits are totally dead (no fuses).

Do you mean a capacitor?

Eureka! It was the goddam flux capacitor all along! :rolleyes:

http://forums.pelicanparts.com/uploa...1402839969.jpg

I had to go look...



Be sure you can pass this test:

if you unplug MS box and measure resistance from any sensor's (MAT, CLT, TPS) ground to car/engine ground , you should not have continuity.

Interesting read.
I have pins 14 - 19 grounded directly to the engine but with a 2.8mm2 wire which I think is about 14 or 12 gauge. Not as heavy as rickb suggests, but certainly not light.

I don't have a relay for power supply to the ECU which could be a potential problem, but then my issue seems to be noise in the crank sensor circuit with starter, injectors and coils either on a second battery circuit or disabled. So relay or not, I don't see any noise source coming through the ignition switch that could cause a problem.

Not sure what he means by "And you do not want the input signal returns to be connected to anything but a sensor or be grounded to the chassis or provide a ground for anything other than a signal return."

"if you unplug MS box and measure resistance from any sensor's (MAT, CLT, TPS) ground to car/engine ground , you should not have continuity."
Yes, I can see how that would be the case with my wiring once I remove the ground wire from G2 to engine. I'll check that tomorrow.

If you have inadequate grounding (sounds like the case) for pins 14-19 then the path of least resistance is across the board and down the sensor grounds to the block.

I just spent an hour playing around with black wires again.
First of all, I replaced the 2.8mm2 wire from pins 14-19 (G2) to engine ground with a 6mm2 wire (a bit larger than 10g). This had no effect on the tooth log, it still looks the same as log 12 above.

Then I did a bit of continuity testing with the following results.
With ECU unplugged and all sensors connected, the collected sensor grounds at pin 7 in the harness are not connected to ground. :)
When the ECU harness is connected and ground continuity is checked at the sensor connectors in the engine bay, ALL are grounded. The ground path is into ECU at pin 7 and out to engine ground via pins 14-19.

This prompted me to check continuity between pins at the ECU MS3 connector and I found that all pins 1,2, 7 to 19 are connected internally. Hence when the harness is connected, the ground path for sensors is completed.

I am confident that all the sensor grounds are connected correctly so I am now out of ideas on that front.

One thought I had was to run a completely new shielded twin wire from the crank sensor to the ECU outside of the existing harness. But really what will that achieve? All the recent logs have been run with unpowered ignition and injection circuits and the starter and solenoid are using a completely separate battery ... so where is noise coming from? :confused: