90 745 Regina code 3-2-1, REX code 1-4-3, low power & shaking at idle/during acceleration.

[DET17]

So you refuse to check the signal?

It would make sense to me to confirm if it is putting out a high signal, low signal, or no signal. The code is very vague.

I found a troubleshooting blog for a 2.3L engine from an 850. Someone provided a complete test sequence for the gent..... wiring and connector testing essentially identical to my REX green book data. However this test sequence also included actual measurements of the signal. I will check if the 850 uses a Bosch sensor like these tractor engines. I know most 850s are Mo_Tronic ECU so likely the monitoring is different that what VOLVO Engineers spec'd for these 8V iron engines. Nice article, thanks for adding this to my thread!

IDK, there's not much info available about Rex knock issues

Bob, if I used one of the cheap O-scopes, would I connect across the pair of pins at the REX ECU to obtain a waveform? How about a data log trial, highway run? Do I need special hook connectors on the scope so not to damage the connectors? Wait a minute, there is also an inline connection under the hood at LH strut brace, perhaps I could reach these connections for a driving trial? I have not used a scope in, oh, about 45 years (Physics lab) so I'm going to need some assistance on this one.

Several articles claim the spark knock is symptomatic of LEAN conditions in the combustion chamber...... and my vacuum signal is still nearly 14 inches HG. It is plausible that my undetermined vacuum leak (perhaps leaking intake gasket which I reused) is sneaking fugitive air into one of the runners and these phantom knock signals are being caused by that?

I'll post a separate thread question about used intake gaskets & vacuum leaks thereby.

 

[ZVOLV]

Would I connect across the pair of pins at the REX ECU to obtain a waveform?

Knock sensor | Automotive oscilloscope test

The purpose of this test is to evaluate the operation of a piezoelectric knock sensor when subjected to a simulated engine knock.

www.picoauto.com

 

[bobxyz]

bobxyz said:
How long was it working OK before the CEL&code?

DET17 replied:
About 225 miles on the current tank of fuel

Oh, that's a lot longer that I thought. Let me look over the greenbook info and my notes and I'll get back to you. An off-the-cuff guess is that your sensor is working OK but the engine isn't noisy enough at higher RPMs, and the Rex box thinks there's a problem.

If you disengage the overdrive and run some higher RPMs around town, does the CEL go away on its own? I forget what the greenbook says for detection - is it something over 2500 or 3000rpm?

 

[DET17]

If you disengage the overdrive and run some higher RPMs around town, does the CEL go away on its own? I forget what the greenbook says for detection - is it something over 2500 or 3000rpm?

The ONLY thing that clears the 1-4-3 CEL is a battery cycle. The diagnostic port black momentary button clears everything else (test pin #2 or #6, hold button 5 seconds with key in position II, and codes are gone) but does not reset only this exact fault code.

The GB manual states the knock sensor is polled by the ECU when the RPM is >2000 RPM and engine hot.

I quickly scanned the REX fault codes expecting to see one for KNOCK DETECTED.... but it doesn't exist. Seems odd to me, but the 1-4-3 CEL code is defined as "knock sensor input missing or faulty". Per GOOG AI, the fault is generated when the speed & temp conditions meet the spec and the signal generated from the knock sensor "are missing, weak, or determined implausible by the ECU". It is worth noting, both my original REX and the newer revision REX both are generating this 1-4-3 CEL.

Another fun fact with REX/Regina.... all other CEL codes are resettable. That is, once codes are set (save 1-4-3), if the next time the key is placed in position II and the correct input value is received, the ECU resets operations to the standard (even tho the CEL code remains).

I drove the wagon this morning a nice jaunt at 70mph, then 55mph, no faults. Arrived at my destination, then idling into the parking area and the CEL was again generated for 1-4-3. I have seen this twice now..... while decelerating to a traffic light I have seen the CEL illuminate. In both instances, the fault registered while the RPM was below 2K. Yet the GB states the engine RPM must exceed 2K to poll the knock sensor for a signal. Perhaps the CEL doesn't register until the engine moves back to idle? I can't recall ever seeing one generate at speed and with RPM in excess of the 2K threshold.

 

[bobxyz]

I looked, and there are lots of details on Rex/EZK knock handling in TP31397 (pg 28+), but not the exact details that trigger the 1-4-3 diag code.

Knock detection is like listening to the engine with a microphone, but detecting vibration with a piezo sensor instead of sound waves. As the engine load and RPM increase, the vibration increases too. Knock has a distinctive vibration pattern that is different enough from the normal vibrations that it can be detected. Knock usually occurs over a fairly narrow range of crank rotation, say 15deg ATDC to 40deg ATDC. I'm guessing that the Rex knock detection measures the average vibration over the ~15-to-45 degree range and compares it to the average vibration over the non-knock range of ~50-to-100 degree range. By comparing the 2 ranges, instead of using a simple threshold, it's easier to detect knock as load&rpm change.



For the 1-4-3 code, I'm guessing that it's triggered when the detected average vibration is too small for the current load/rpm, i.e. the engine is too quiet for the conditions based on a factory set noise level. A broken, or unconnected sensor, would easily trigger this. For your failure, I'd try clearing the code, warm up the engine, and either unplug the sensor or swap it to a loose spare sensor. Then rev the engine to over 2000 rpm to see if the code triggers immediately. You might need to gently drive around a bit to trigger the code. What you're trying to figure out is how sensitive it is when there's no sensor, or a loose sensor. I'd be surprised if the code doesn't occur almost immediately.



If you can figure out better what leads up to the code with the sensor normally connected, that could help. I'm guessing that the sensor, wiring, and Rex circuits are mostly OK, but there's some small operating region where the average detected vibration level is a bit too small. You could get a couple old salvage yard sensors and try swapping them in to see if there's a sensor-to-sensor variation, or if the new sensors aren't quite the same as the old ones.

On other stuff, detecting knock and briefly retarding timing is normal, so there's no special diag code for it.

The knock sensor cable is a shielded cable -- under the green insulation, there's an outer woven wire sheath that surrounds an inner white(?) wire. The outer woven sheath is used as a ground wire, and the inner wire carries the signal.

One other thought - when did you check the valve clearances last? Overly tight clearances might be too quiet.

 

[DET17]

Thanks Bob, that is some excellent information you have shared.

I've been thinking about knock sensing, and the relationship to regular combustion events. For our beloved 4 bangers (2 up, 2 down), there is a combustion event every 180* of crank rotation (and a matching "exhaust event"). All of us know who have torn these engines apart, 1 and 4 pistons/rods move together, as do 2 and 3. However the Volvo engineers chose to measure knock, they know the normal combustion/exhaust events have a unique signature of sound/vibration, and the frequency of same follow the RPM exactly. That way, the REX ECU must be monitoring those waveforms and frequencies and listening for "exceptions".

Regarding my repeating 1-4-3 faults, I'm leaning toward these faults being actual detections and not ghost faults or a system failure.

I have a couple spare BOSCH knock sensors, but I did throw away my original. The spares I have are actually Porsche spec parts (they may have come off a SAAB 900T engine, can't be sure the source). I did run one of those sensors, and it also faulted quickly.

At some point in the discovery, I started the engine sans knock sensor. From memory, it faulted before I drove the wagon 1 mile.

To the continued prodding of ZVolv, if one were to data log the knock sensor.... without matching the recorded data to exact RPM data, how would any data analysis be able to identify a specific knock event? In my old engineers brain, without the matching RPM data, measuring knock data would be next to worthless. Said another way, if one could data log the knock sensor output AND synch those with the actual crankshaft RPM, it might be possible to pour over the logs and identify an actual knock event.

I'm confident that Volvo red block engineers provided the same acceptable/NO GO specs to both BOSCH and BENDIX. It was their engine that both were controlling..... an alarm generation must have required the same anomaly to generate the CEL.

Yesterday I poured over a plethora of the last 20 years of 1-4-3 knock fault threads on TB. Many folks mentioned fuel and specifically "summer recipes" causing their misery..... some noted that when the "winter recipe" hit the local pumps... .their CEL faults stopped. So we have that data.

I am overdue to run a hot engine compression test, in order to ID if I have any mechanical deficiencies with my recent overhaul.

One more thing to consider; several of those ancient TB threads were concerned about failing injectors as the possible root cause of the random knock faults. In my case, I am running one rebuilt injector and 3 "cleaned 94 injectors", which are of unknown provenance. All I did with the 3 injectors was run them thru the cheap Amazon injector tester/cleaner. To eliminate this possibility, Ive ordered 3 more rebuilt injectors from the same vendor as the first.

What are your thoughts about data logging the knock sensor / engine RPM signals, in an attempt to identify the smoking gun causing these repeated faults?

Finally regarding valve adjustments: I targeted setting mine to the "Tom Fritz aka StealthFTi" settings. I was however limited by shim selection at my disposal..... I know I had some Intakes as tight as .012", and at least one exhaust was as loose as .018". Once all were set, I did install hush-er-s so likely the valves are as quiet as possible with the red block SOHC design.

 

[ZVOLV]

I don't know what strategy the computer uses to set that code, but that is why I said you should compare the signal to a like-vehicle. Do you have any other bricks in the barn?

 

[DET17]

I don't know what strategy the computer uses to set that code, but that is why I said you should compare the signal to a like-vehicle. Do you have any other bricks in the barn?

Just my 84 242 that is gutted, will be WB powered and likely aftermarket ECU or otherwise Mo_Tronic 4.4
Nothing to compare it to.....

 

[bobxyz]

Per this old post, the knock sensor signal is pretty small and gets distorted when you measure it. cleanflametrap monitored the internal knock circuitry (the integrator that sums the filtered signal over most of an engine stroke) and managed to catch a trace with hammer-induced-knock detection in the middle of the trace.

Post in thread 'Lh2.4 knock sensor issues'

Apr 2, 2020

Even the 10 megohm input impedance on my scope threatens the accuracy of the knock sensor's output. I looked for the signal downstream in the EZK, and found where it is blanked during the expected detonation. Just some fun. Same end result, though, one sensor performed well, while the old original set the code on #6. These traces show the knock signal after it is integrated.

• cleanflametrap

[4cylinder 4 stroke engine has each cyl firing once over 2 engine revs, or 180degrees between sparks.]

 

[DET17]

Per this old post, the knock sensor signal is pretty small and gets distorted when you measure it. cleanflametrap monitored the internal knock circuitry (the integrator that sums the filtered signal over most of an engine stroke) and managed to catch a trace with hammer-induced-knock detection in the middle of the trace.

So the electronics in the ECU filter, integrate, then ignore (at a specific portion of the rotation of each cylinder piston, when not in the combustion power stroke).

Once my next batch of parts gets here, I'm going to retest NO SENSOR (which should fail as soon as the engine hits temp) and then torque a bolt/nut/fat washer on the back side of the sensor (to simulate block surface contact on the back). I'm betting the engineers were smart enough to code the sensor circuitry that if it doesn't see the "expected normal" data, it will fault again.

 

[ZVOLV]

Did you ever load test the two wires on the circuit?

Like I always rant about, you need to use a test lamp to confirm a circuit is good. You cannot just find close to zero ohms on a wire and assume it is good enough to carry a load. One strand of wire could show zero ohms, but it won't carry a load.

 

[DET17]

Did you ever load test the two wires on the circuit?

No, I did not. We discussed in another thread..... sensor teeny tiny signals don't need current..... Bob can state what the voltage is on the signal from the KS.

Sure, devices that "do work" need to be load tested. Continuity is only one test.... but inputs from sensor are not carrying current larger than milliamps IMHO.

 

[bobxyz]

I haven't probed the knock sensor on a running engine, but I'd guess that the signal is maybe 100mV peak-to-peak at low rpm&low load, going up to a few hundred mV at higher rpm&load.

The Bosch datasheet says "self-impedance > 1Mohm" which means that you'll need a standard 10x 10Mohm scope probe to look at the signal without hugely distorting it.

 

[DET17]

A surprise discovery this morning.... buried inside my Green Book for Rex/Regina, the knock sensor is spec'd to be torqued at 8 lb. ft. ......

This smaller torque is half of what most online sources list, as well as possibly the BOSCH specs in those flavor of Green Books which state the knock sensor torque is 15 lb. ft.

The question.... could clamping the new knock sensor at twice the REX spec be causing my CEL faults?

 

[bobxyz]

Doubtful, but give it a try. Some of the early knock sensor designs did need very specific torques, but I think they almost immediately went to a better design where the hold down torque didn't effect the knock signal.

 

[DET17]

The knock sensor PN listed in the same green book for REX/Regina is exactly the PN I'm using. That printing says 88- on the cover. It does state on the cover page that all revisions thru May '91 are included, but nothing beyond.

When I pull the intake to install an NOS PENTA intake gasket (to ensure 100% leak free intake runners) I'll take the KS loose and retorque at the lower value.

 

[DET17]

Some of the early knock sensor designs did need very specific torques, but I think they almost immediately went to a better design where the hold down torque didn't effect the knock signal.

Your spec sheet states 20Nm torque.... which converts to 14.75 lb. ft. Surely the specs match, but our RB sensors of the latest BOSCH version I believe end in 046. I did test a couple of the 006 style which for some reason were in my 745 parts tote.

Likely the half torque spec I found was in reference to the earlier PN as you surmised.

 

[bobxyz]

From TP31397:

 

[DET17]

From TP31397:

Thanks Bob, always a joy to keep FACTS in these discussions.

Just back from 2 weeks in Italia; the search resumes for solution(s).

NEXT Actions:

Install the newish & cleaned/flowed late 960 type injectors, along with an NOS PENTA intake gasket, so to remove any chance of a vacuum leak (every other stone has been turned). Will investigate use of sealant around the 4 intake runner ports to be 110% sure we are leak free.

Will perform a hot cranking WOT compression test to determine if any mechanical deficiencies exist in valves/pistons.

WANTED listing posted for a Dale's old school 6* advance round tooth timing belt sheave.

More to come......

 

[DET17]

Status update:

Engine HOT, performed WOT cranking compression test on all cylinders.

Results - 175, 175, 172, 175 psi ( 1 thru 4 in order)

THEREFORE, engine mechanical elements (pistons/rings/valves) are AOK, excellent condition..... what I expected from a 77K 94 engine starting point.

NEXT (week) Actions:

New intake gasket (with sealant), rebuilt 960/850 NA injectors, charcoal canister & purge valve.

The wagon is being prepped for its longest mission to date... and 8 hour drive to Ohio next Friday.