evanb

I'm putting a J-spec TII engine into my '90 convertible and recycling the original N/A engine harness on the new engine.

On paper it's simple. There are only two wires that need to be "reassigned". One is pin 3R of connector EM-31. On the N/A engine this controls the 6-port injection solenoid. On the turbo engine it operates the turbo boost pressure control solenoid. I have removed the proper connector from the J-spec harness and spliced it onto the N/A harness.

The other wire is pin 2M of connector EM-32. That operates the variable dynamic intake control solenoid on the N/A engine. On the turbo harness it connects to the knock sensor. In the turbo harness, though, this wire is shielded; the wire in the N/A harness isn't.

It appears I have three choices. One, connect the knock sensor using the unshielded wire. Another is to run another shielded and grounded wire to the knock sensor. The third is to leave the sensor disconnected. The first might cause the ECU to detect phantom knocks due to electrical noise, costing me power. The second is a PITA, mostly because I have to find some shielded wire that can survive life under the hood of a TII. The third may risk engine damage.

I think #2 is what I will have to do, but I'd like to hear from anyone who's solved a similar problem. Has anyone used unshielded wire for the knock sensor and had a good outcome? How important is the knock sensor anyway? The Power FC for the FD ignores the knock sensor. Maybe it's OK not to use one on the FC.

-evan

HAILERS

Number two sounds like the RIGHT way.

That said, on my 87n/a to turbo, I used unshielded wire. I knowingly did wrong. But, I've been stepping on the pedal pretty hard for a couple of years now and it's still all together.

Blind luck? Got me. Don't think so.

Which brings up the aftermarket EMS mentioned on this forum. Do they have knock sensors? That's a good question if I do say so myself.

evanb

Well, the (potential) problem with the unshielded wire is that the ECU will see induced noice as knocks and retard the timing when it isn't necessary to do so. I'm sure it will still detect real detonation. You are possibly losing power, not risking cracking apex seals.
Yes. I'm curious to know how many aftermarket ECUs have inputs for knock sensors. I was hoping that people with knowledge about building high power engines would have something to say about the usefullness of these things in general.

-evan

evanb

Several months later....

The car is together and running, using the modified N/A harness. The modifications include a separate, shielded wire run for the knock sensor, with its shield grounded at the ECU much like the stock TII harness does.
IMG_2420.JPG
IMG_2421.JPG
IMG_2422.JPG
IMG_2422.JPG
IMG_2417.JPG
IMG_2424.JPG
IMG_2425.JPG
IMG_2426.JPG
IMG_2427.JPG

The problem now is a check engine light and error code (05) that occur most times I drive the car. Usually the light comes on when I accelerate moderately (no or very little boost although I don't have a boost gauge to tell me more exactly), but sometimes even gentle acceleration will trigger it and sometimes it doesn't happen at all. This may be heat-related, as the times the light/code do not occur are usually after the car has been warmed up then parked for 10-20 minutes. I've replaced the knock sensor with another (used) one without any change in symptoms.

I'm looking for some information on how the ECU works with the knock sensor.

Any ideas as to what I can do to solve this problem?

NZConvertible

That's not very smart...

I don't know what "test plug" you're referring to, but the FSM has all the info you need on test procedures for the self-diagnosis system. If you get a code, you follow the instructions related to that code. In this case all you can do is check is the connections and wire continuity, and replace the sensor. If you've replaced the sensor then most likely this is simply a wiring problem.

There will only be a signal from the knock sensor when there is knock occuring.

evanb

Of course. I'm obviously too much a fool to realize that the FSM contains the solution to all possible problems, and I appreciate your setting me straight.
... and if it isn't a simple wiring problem, then you have my symptoms.
Since you brought it up, the test procedure in the FSM (which I have, surprising as it may seem, read) consists of plugging an SST into the elusive three-connector test plug, then creating a false knock input by banging on the front lift hook with a mallet with the single-connector test input grounded and open.The FSM describes magic behavior by the SST that you interpret to decide what the problem is. Since I don't have the SST I'm trying to determine what the test plug's raw output is supposed to look like so I can check it with my more humble instruments.

NZConvertible

One thing I've learnt about this forum is that if someone doesn't say they've done something, 99% of the time they haven't. So unless you tell you you've read the FSM and what bits you've read, I'll assume you haven't bothered to look, because most of the time I'll be right...

I'm not sure what bit you're reading (where is it?), but the test to see if the knock sensor is working properly is described on page F2-72 and involves nothing more that a piece of wire, a timing light and a hammer. You ground the initial set coupler with a bit of wire, connect the timing light, hit the engine with a hammer and confirm that the timing retards. But I don't think that's going to help here.

The bit I was referring to is on page F2-23, which tells you what to do if you get the knock sensor error code. You didn't say that you'd checked the connections at each end of the wire, and you didn't say that you'd checked the continuiuty of the wire. Since you didn't say you'd done those checks in your second post either I'm going to assume you haven't done them.

The S5 knock sensing system is very simple. Sensor, wire, ECU. You've already elimianted the sensor, so eliminate the wire and if the problem's still there it must be the ECU.

danhaggis

iTrader: (3)

So all i have to do is change those to plug ends to make th harness opperational in my 87 n/a??? or do i have to add wires as well ???

2RotorSpeed

iTrader: (1)

Regarding your swap, see if the following link might be of any use: https://www.rx7club.com/2nd-gen-archive-72/my-na-tii-jdm-swap-635663/

evanb

I assume you're asking about using an N/A harness in a TII. I converted a series 5 N/A harness for use w/ a series 5 TII engine, and that only requires reassigning the two wires I mentioned. I have no experience with or documentation for S4 wiring, and I understand there are more differences between N/A and TII for S4 than S5.

For the series 5, though, after the modifications the US-spec harness reached everywhere it needed to on my J-spec TII engine. My engine doesn't have an accelerated warm-up system. Connecting that would require piggybacking a couple of wires (with the proper connector) onto some existing ones. The AWS and some other circuit (can't recall offhand which) are both controlled by the same pins on the ECU so extra wires all the way to the ECU aren't necessary.

NZConvertible

If you're using an S4 TII ECU on an NA harness there are only two wires to sort out.

Pin 1R (LG/B wire) is the knock control unit on Turbos and the power steering pressure switch on NA's. If you're installing the Turbo knock control unit you'll need to cut the wire ~6" from the ECU connector and extend the connector side to where you've installed the knock control unit (which also needs power and ground connections). If you're not installing the knock control unit and you have power steering you'll need to cut the wire anyway, otherwise every time you turned the wheel the ECU would get what it thinks is a knock signal and retard the ignition timing.

Pin 2K (L/B wire) is the twin-scroll solenoid valve on turbos and the split air solenoid valve on NA's. If you're keeping the twin-scroll control system operative you'll need to connect the split air solenoid valve connector to the twin-scroll solenoid valve. If you're not keeping the system operative you can leave it disconnected.

evanb

This is true. On the other hand when I include a lot of detail my posts they get so long that my actual questions get lost. Which, if you look at my original post on this topic appears to be the case here, as well.
Are you looking at an S5 manual? I don't have mine here at work, but the section on testing the knock sensor describes connecting an SST to a test plug (which is behind the battery, near the single-pin test connector) and tapping the lift bracket. With the single-pin test connector open some light on the SST should not flash, and with the connector grounded it should. I'll look up the page number tonight.

My problem is intermittent and possibly heat-related, making it more difficult to definitively diagnose. What I was really hoping for was to obtain some knowledge about when the ECU pays attention to the sensor and what it expects to see when it does. I was reluctant to start hacking away, figuring that the next step, replacing the wire, was a lot of work to undertake just to see if it fixed anything. Sitting here typing though, I realized that I can run a piggyback wire without cutting the existing one, so I will probably do that either this weekend or next.

NZConvertible

I've got the 89 FSM downloaded from the internet. Is yours a different year? Maybe they changed the procedure. The 89 test should work fine, and doesn't require an SST.

That does make it hard, particularly if a bad connection is making and breaking connection randomly. Definelty check all the connections, partucularly any you'd made yourself. Or just run a new wire.

That sort of info is really hard to get, particularly for the S5 ECU.

evanb

This weekend I ohmed out the knock sensor wire from the sensor to the ECU. It appears to be fine. While checking I discovered that the connector containing the knock sensor wire wasn't completely inserted (the retaining clip hadn't snapped into place) and I had hopes that the problem might have been related to that and, indeed, for several hours after reconnecting the ECU did not report any sensor errors. After parking for 30 minutes or so with a fully warmed-up engine the problem returned.

Owing to time constraints I did not recheck the wire at that point, but I honestly don't think the problem is a simple intermittent connection. I had the battery disconnected for a couple of hours during the check, and it is possible that this more fully discharged any lingering voltage in capacitors than my usual <1 minute reset.

I'm willing to accept that the ECU may be the culprit. Before I invest in another one, though, I'm going to try borrowing an N370 locally and see if it cures the problem.

evanb

I thought I'd wrap this up for anyone who's still interested.

The cause of the ECU code turned out to be that the shielding of my cool shielded knock sensor wire is grounded to the signal wire. I haven't figured out how or why, but until I muster some ambition to make a more elegant fix I've disconnected the shield ground at the ECU end.

So much for that bright idea. A regular, unshielded wire as used by Hailers would have been just fine.

Seeing this fix my problem makes it obvious that the ECU only pays attention to the knock sensor when detonation is possible, probably determined by some manifold pressure threshold; thus the confusingly intermittent codes.