We have a new TPI installation, speed density with a 7427 ecm on a mild 350. The problem is the knock counts continually count up to 255 then start over as long as the motor is running. We have tried everything imaginable to cure this but no change. Thee is clearly no actual knocking. Here's whats been done:
1. Wiring to the knock sensor has been rung out and is correct.
2. Lifting the wire from the knock sensor has no effect.
3. Grounding the knock sensor wire has no effect.
4. Grounding the knock sensor input at the ecm has no effect.
5. Changed ecm and memcal from a running car and still the same problem. Put them back in the other car and it ran fine. (No excessive knock counts)
6. Checked knock sensor ground and ecm grounds. All good.
7. Pulled the alternator output wire to eliminate alternator noise confusing things. No help.
7. Checked ecm power and appears to have no excessive noise.

The ecm does pull timing in response to these knock counts so it runs crappy under load.

My conclusion (so far): These counts have to be noise (electrical noise that is) related. With the engine not running but the ecm on, there are no knock counts. But the noise must be coming in elsewhere than the knock sensor input, the ecm power or ground. Could there be another input with noise on it which would couple over to the knock sensor input?? Any other possibilities? A beer to the guy who figures this one out!

Steve

 

Not quiet sure but If ya have Serpintine Belts take them off and start the car and see if ya still have knock counts.

I have heard people say that the serpintine Belts have caused knock counts. Also bearings in altanator or water pump. Taking the belt off will See if any of these is the problem.

Just My .02

 

An ignition problem. An open wire or large enough air gap to generate RFI. You running a MSD?. if so try bypassing it.

 

To MTPFI-MAF: It cant be actual mechanical noise (from serpentine belts, bearings, etc) cause we unhooked the knock sensor and it kept reading counts.

Grumpy: Yep, that's what we are thinking also, ignition. In fact I cant see it as anything else. No MSD, just a stock gm HEI dist. Tomorrow we are going to change out the dist innards. The owner thinks they are of questionable quality. But it does fire right, and the timing seems to behave normally except for the knock retard. I would think that anything in the dist that would generate enough rf to tear up part of the ecm would also cause some erratic operation in the firing, either misfire or screwy timing.

Also, remember that rf cant get into the ecm thru the knock sensor input wire cause we grounded it (at the ecm) so it would have to be coupling into that circuit inside the ecm. Ya spose that's possible??

Steve

 

Last guy that had this problem was using a '165 MEMCAL in a '730 ECM, so no knock filter was present.

RBob.

 

Yep, I saw that thread also. But in this case we swapped memcal and ecm (separately and together to eliminate a weird combo) from my own car which runs fine with the identical setup. Still gets the knock counts in his vehicle.

I'm upping the ante to a 6-pack now guys!

Thanks,

Steve

 

Just wanted to double check

In your first post the ECM is a '7427 and it is a conversion. Is the ECM really a '7727 (just a typo)?

Going on this being a SD system ('727) there is only the sensor wiring going to the ECM. By grounding that wire the knock counts would have to stop, even if caused by mechanical or electrical noise.

If they don't stop then the wire to the ECM is bad/incorrect. The knocks sensor wire should go to pin A11 on the '727.

RBob.

 

RBob....
Yep a typo on my part. It is the 7727, the weather pak ecm. The knock sensor does connect up to A11. Checked it a bunch. Even when that pin is grounded AT the ecm, it still counts up knocks.

Another test we did was to pull a vacuum hose while watching the knock counts on the scanner. This way the engine would die without changing any of the power to the ecm. And sure enough, as soon as it died, the counts stopped.

It is a tough problem

Thanks,

Steve

 

Tough one is right. Check pin A17 and make sure that it is open.

Other then that I'll bet it comes down to a wiring problem. Most likely in a ground someplace. All the ECM ground wires need to go to the engine block.

Then the usual heavy ground from battery minus to the block. From the battery minus to the fender, then ground strap from engine to firewall.

RBob.

 

A17?? Will check that one tomorrow. But what is it? I dont show that pin on my diagram. You must have a complete pin out list with functionality? Where do I get one?

In my experience weird problems are more often than not ground related so that was the first place we went. Frame, block, and body are all tied together with ground strap and connections are good.

I did come up on one more clue while sifting thru the scan data. The vehicle speed as shown in the data is about double the real speed and is very jumpy, moving as much as 10mph between succesive scans. We are scanning every half second. No speed is shown while at a standstill. The vss we are using is one I used to run in my vehicle and there it worked fine and showed the exact correct speed. Now I'm sure his gearing and tire size is different than mine but not enough to double the speed. Noise on the vss input?? Could that couple into the knock circuitry? Will also check that one tomorrow.

Steve

 

The knock and VSS signals go to the same chip within the ECM, and the chip uses counters on both. I'd say that there is an electrical issue. Maybe a power or ground lead missing from the harness to the ECM.

Or, something in the vehicle is creating a lot of electrical noise. May want to check the cap, rotor and verify the plug wires are of a supression type and are not arcing anywhere.

RBob.

 

Ignition wise, I was talking about a secondary side problem, generating enough RFI to fubar the ecm.

The MSDs can get into a false triggering mode, where the timing is random, and cause knock becuase the timing is so far advanced.

 

Ok, after a couple of days to collect thoughts here's where we are:

Checked all power and ground wiring, again. Connections are good, single point ground is good, checked for noise on the power inputs with a scope and also for noise appearing on the ground leads, and nothing out of line.

Took distributor apart, replaced modules inside, ohmed all grounds, checked rotor and cap, all good. All plug wires good, no funny firing, missing, etc.

Thinking maybe the injector firing was somehow causing the problem, we pulled each injector power fuse separately (ran surprisingly well) to see if that stopped the knock counts. Still kept on counting.

Pulled the speed sensor and ran it with a drill motor with the engine not running to see if the speed on the scanner was steady. It was. Then ran the engine while spinning the speed sensor with the drill and the indicated speed was still steady. Problem was the knock counts kept on going.

Conclusion: Although the problem intuitively seems to be electrical noise, I am slowly becoming convinced it is not. We have checked wiring continuity and have looked for noise with a scope and there is very little there. My gut feeling is that GM designs ecm's to be pretty tolerant of noise. And we have run the vehicle with the ecm hanging underneath, with test leads clipped all over, etc. and the knock counts have never varied once, as long as the engine is running they count up like a clock. It is too consistent for noise.

So what can cause that knock counter inside the ecm free run?? I wish I knew.

In the meantime I burned a chip with knock sense timing corrections disabled so we can run and do some tuning and maybe have an inspiration on the knock stuff.

Ideas??

Steve

 

Oh, and RBob pin A17 had no connection.

Steve