Subaru Impreza 98 --> 2002-ish

[kb1gtt]

Some progress on the suby, ambient temperature finally got up to 10F (-12C) and I got through my snow removal duties. After getting to play with the red suby, it looks like my wiring is shot and can't be used. I was hoping part of this harness was going to be use-able, but it appears the connectors are just to far gone. Blah, I'll have to re-wire pretty much the entire car = blah for labor efforts.

Here's how I connected my CAM, it goes through the E1/B20 engine connector, then the E15 connector on the CAM sensor, it has been 100% disconnected from up stream components, and was originally tested when connected to the Frankenso board, with the same results as noted below. The DMM claims 1.742k ohms between red and black wires, not connected to the frankenso, just red and black.

CAM.jpg

suby_cam_noisy.png

Here's how the crank looks, it's better but still very noisy. The DMM claims 1.622k ohms when connected to the green / white as well it was connected to the Frankenso.

Crank.jpg

suby_crank_noisy.png

Both VR sensors are measuring the same ohms, so perhaps the sensor is some how broken. I really shouldn't have this much noise on the line. The ohms is also not un-reasonable, but is a bit high for most VR sensors, which implies a bad connection. However I don't know the expected ohms, so I don't know if I'm really measuring high or if it's with in specification.

[kb1gtt]

I see no other way to fix the subys wiring, other than to gut it entirely and start from scratch. Which is kind of frustrating.

I took the CAM sensor out and clipped onto it with the scope. I got a very nice and clean signal when I waved a wrench in front of the sensor. I then re-installed it into it's position sensing the CAM, then I disconnected both engine harnesses (which disables spark, fuel, solenoids, ect) I then cranked it over with the scope connected. Ch1 was one lead, CH2 was the other lead, both scope probe GND straps were connected together and connected to the frame near the sensor. However I still get a noisy as hell signal. Best guess is that starters brushes must be making some noise. I really need to find a new battery for this DSOQuad. Mine blew up like a balloon, and I stopped using it before it popped. There is a chance this noise is coupling in through the USB and laptop GND's. The laptop was floating relative to the car.

CAM_no-engine-harness_direct-clipped.png

I guess I could have tried capturing it to see what rusEFI would consider a noise pulse vs what I see on the scope. However I really want to make it better on the scope first.

Any how, still fighting with noise issues caused by a really sketchy wiring done by a prior installer.

[puff]

is the cable from vr shielded? what if you connect the scope directly to vr cable (GND & CH0)

[kb1gtt]

Yes cable is OEM shielded twisted pair. I used a alligator an clip to GND it as the GND was broke when I cut the cable. I originally cut the wire and connected Frankenso R0.2 to it. I got many noisy pulses in the digital sniffing. I then connected the scope and saw a huge amount of noise. I've been trying to isolate that noise and remove it from the equation. To do this I've been working my way closer to the sensor and sniffing in a variety of ways, as well I've been disconnecting harnesses to remove potential noise sources.

This engine is an engine swap, which uses 2 different harnesses. The harnesses were spiced together with these scotchlok connectors.

41WttNxCnmL._SL500_SS100_.jpg

That is a blue one, and I used to joke about how the fellow used brown connectors (blue and red that got hot) I just learned that they actually make these connectors in the brown color. Point being it's some really dodgy wiring.

I have sniffed in several different ways, which includes direct connecting the scope to the sensor. I still get the noise. My plan is to remove the existing electronics even more. I plan to remove the starter battery lead, then connect my battery power direct to that and use a relay to engage the starter. This will prevent powering the existing ECU and existing radio, ect.

[kb1gtt]

This red beast is still an electrical nightmare. But I got out there and did some testing today. The short version is that if you have a suby, with after market ECU I would suggest a 1.6kohm to 2.2kohm 1 watt resistor be installed at the ECU or just inside the ECU. This loading resistor will help get a better electrical signal to the ECU. See below for a slightly more detailed analysis below.

In this case, the slow cranking pulses are above 85mS, so the MAX9926 zero-crossing threshold will self adapt to about 4mV to 30mV, typically 15mV. The chip typically self adapts to 33%, so when in these low cranking RPM situations, you want your VR signal to be at least 30mV * 3 = 90mV, which is 180mV peak to peak. By adding a resistor to the circuit at the ECU, you can decrease the effects of noise. However if you put on a loading resistor with small ohms, you can also prevent slow RPM tooth detection. In my case, I have a very noisy signal, and I'm looking to decrease the noise by adding a loading resistor.

I was hoping that I could minimize the noise and get valid readings from my suby with questionable wiring. I set out to see what ohms of impedance I could put on the VR to minimize the noise, and decrease false tooth detection.

Here is what my suby signal commonly looks like while cranking. Please note, this is probably not the typical. the prior owner did some really messed up wiring, it's a 98 with a 2002 engine harness. The battery has been relocated to the trunk, and I have many brown scotch connectors. This wiring is very sketchy and bad. I don't drive this thing, as it's just waiting to catch fire. I need to gut it completely and wire it properly. However I have not gotten around to doing that yet.

620ohm_5ms_0.3Vpp_noisy.png

The above is with a 620ohm resistor across the VR.

Here is what my noise looks like with the key on and not cranking.

620ohm_no-crank_1ms.png

You can see that as soon as I start cranking, I get massive noise issues. As well you can see that even with a 620 ohm resistor, I'm still getting about 480mVp-p which is more than the required 180Vp-p. You can see the noise floor is very low, so this noise is not from ambient stuff, it's from the car. Also the hood was open, and the scope was placed outside the car near the front passenger fender. Such that the wires were as far away from potential noise generating wire as possible. The shield was connected to the frame ground, so the shield should be mostly functional. Also I have disconnected any devices I can get my hands on. However there are many things buried in the engine.

Here's a picture of where I captured the signals. I'm showing the scope probes on 10X and that I'm on that side of the wiring connector. When I did the testing, I ran the wires 180 degrees from how shown, such that they went over the fender and left the engine bay right there.

Setup.jpg

Any how, despite my attempts to keep the noise down, it's still there for me. I don't see enough of a change from 10k to 2.2k to prevent false tooth detection's. To me it does look like the 10k is more noisy than the 2.2k. As well it looks like 620ohms is less noisy, and I expect that lower ohms will decrease the effects of the noise.

So I wonder, how low in ohms can I go? I really don't know. It depends on a bunch of details which I do not have. I don't know the gaps, or inductance of the VR, etc. I don't know how many mJ of energy the VR will produce at full RPM. I know a 1/8 watt 620 ohm resistor, could handle about 9Vrms, so about 12Vp-p. I expect VR's to typically get up to 50vp-p ish. So I would think that 40Vrms ^2 / 0.125W = 13k ohms is all you can get out of an 1/8 watt resistor. I would expect a 1W could handle 1.6k ohm. However I don't know the voltage that would be produced by this sensor at full RPM, so that's just guess work. It would be handy if I could get this sensor spinning at 6kRPM, then I could measure it's voltage and make good suggestion for what size resistor it could handle.

Any how, the conclusion I would make from this is that if you use a 1.6kohm or a 2.2kohm resistor, you want to install it across the leads at the ECU or if you can inside the ECU. This will help decrease the electrical noise and will help prevent noise from generating false triggers. I also know you can electrically go down to 620ohms and you would still have sufficient signal for the MAX9926 to detect, however you are likely hitting thermal limits when you go below 1.6kohms. If do install a loading resistor, I suggest using a 1 Watt resistor, and that you provide it with thermal cooling. Granted you aren't going to be full RPM all the time, so you would probably be OK with a smaller resistor. However keep in mind you may end up doing engine braking as you are heading down a mountain or long hill, and it would suck if it overheated. Also there really isn't a difference between 1/2 watt and 1/8 watt, so you might a well spend the extra $0.50 and get the 1W resistor.

Also for archival purposes, and to leave a bread crumb trail for kb1gtt here's my notes of this testing.

Subaru_Crank_Sniffing.7z

[sleepingAwake]

Posting to subscribe, offer my help if I can be of any at some point and to say thanks for the detailed information.

--- I'm kind of a Subaru nut.

[Rhinoman]

That's the worse VR signal I've ever seen, where is the noise coming from? surely that's not straight out of the VR sensor? Can you remove the sensor and spin it with a drill or something? that way you could measure the output voltage and determine what current you need to lose and what resistance you need.

[kb1gtt]

Wires are bundled in the engine, I'm to lazy to tear the engine apart. I'm sure this is coupling from near by wires buried in the engine. Probably that flapper valve thing, or with the dodgy wiring, perhaps a pinched wire or rouge scotch connector. That emissions valve thing has not been deleted yet. I have no idea why the guy I bought this from put on a STX turbo but did not remove the air restriction is beyond me. I would guess he didn't have the $ for the delete plates. There are many oddities with the security system, and lots of things blink and do odd things. I understand the UTEC ECU modification device was programmed with a gameboy. It's really a dodgy system.

I could buy a sensor and spin some metal near it, My Millwakee 5378-20 drill claims to do 2500 RPM, it's a hammer drill and does 40,000 BPM on the hammer pulses. Any suggestions on were to get a cheap crank wheel? Perhaps one on loan? I could mount it in a drill and do a mock up system which would allow me to take measurements and figure out which resistor and what ohms could be used.

[sleepingAwake]

Which one would you like or both?
Pm an address and I'll send them your way