I could swear I hear detonation around 2400 RPM mid throttle, 3rd gear, but no knock retard or knock counts show on datamaster. This sounds like a rumble/knock coming from the IP or tunnel area.

Reducing timing eliminates the noise, but also severely reduces torque, so it may simply be a driveline noise that happens at high torque/low speeds. So, what about transmission noises at high torque/low RPM?

I'm running LT4 KM and a new sensor, with 1/2 filled block. I'm not sure how the filled block affects the knock sensor function, but I am sure it changes the resonant frequencies of the block, and probably increases damping of the noise transmission to the sensor, so maybe that has something to do with it...

 

What did Vector Motosports set the max knock retard, attack and recovery to when they "tuned" it?

I'm surprised you don't have constant knock retard. The fill will transmit every possible noise directly to the knock sensor, which is now embedded in fill. Are you sure they didn't tune knock retard completely out? Are you sure the knock sensor is even working?

 

Hey, thanks for the response, Injuneer! I did not realize the fill could have such a big effect. I DID NOT request a filled block when I ordered my motor, and did not even know it was filled until I installed a brand new knock sensor and instead of getting wet, I saw the fill. At the time I thought 'nice bonus'.

I just checked vs the stock tune, and the max knock retard, attack and recovery are all still exactly stock. So is everything else, except PE threshold and A/F, Extended timing, and injector flow rate, as far as I can tell......

Yes, I'm pretty sure the sensor is working, since I see KR and get more of it when running more timing. However, the KR is inconsistant. A few days ago I had no knock on an older tune with semi-normal base timing tables, great midrange response, etc. I was pretty sure I heard detonation in the low RPM/high MAP zone, and the KR was less than 1 degree. Next day, the same tune gave me 12+ KR 2400-2600 RPM. I had to pull a TON of timing in that area to get the KR to go away. And the car got a lot less responsive down there, of course.

So, I'm starting to think my sensor sensitivity varies for some reason. Wires look OK. Perhaps expansion/contraction of the fill changing the transfer function to the sensor, or similar. I tried to dig some of the fill out with a pick near the sensor, but not easy.

Any suggestions? Less attack/decay and max retard?

 

Its a matter of how well the environment surrounding the sensor transmits sound. Solids transmit sound the best, liquids medium, and gasses poorly. And as you've already noted, who knows what affect the fill has on the resonant frequencies of the block itself. I wouldn't fill an LT1 block, unless you were looking for some outrageous HP level. I've seen unfilled blocks handle 1,000+ HP with no problems.

At this point, I would tune out the knock sensor completely, and tune it reading the plugs for the onset of detonation. Then back off on timing to allow for the variations in the quality of the fuel you are using, recognizing that race fuel is generally held to tighter octane and composition standards than common pump grade gasoline.

 

Ok, thanks a bunch for the input, I really appreciate it.

The many many sets of plug changes and 'controlled' driving cycles needed to effectively map detonation onset across RPMs would be a huge effort, I think. Plus, aside from noting shiny ***** on the plugs, I'm not sure I have the skill to read the plugs right, yet. I'll remember the advice, but save it as last resort. Sounds like getting a fill without asking for it could turn out do be a pretty significant drawback.

At any rate, I'm really disinclined to tune out the sensor until all other options are exhausted. For me, not having a knock sensor function is a drawback considering my high DCR and street driving habits, and not always having 94 octane availabe. I think I'll try to improve the repeatability of the sensor for a while, maybe play with digging out more fill and even a resistor to try less sensitivity. I'll keep trying to rule out any other, real causes of inconsistant knock.

As to the original question, then I assume it's not normal to ever have a knock sensor 'miss' hearing knock? I thought maybe some frequencies where the LT4 KM is 'detuned' could correspond to dead spots in it's sensing ability, at particular RPMs for instance.

 

My WAG on the fill is that the block is a transmission medium not a source, in which case as long as the KS is in contact with the block in the normal manner and not in contact with the fill, it should work normally, since the transmission characteristics of the iron aren't changed much if at all by the fill.


Another random thought - people use the knock sensors from our cars in retrofit to other cars (like with megasquirt), so it isn't critical that it be located in the stock position. (And I'd argue that the stock position was chosen merely for convenience.)

So it may be possible to relocate it to a part of the block that isn't filled and still have it work correctly, although this probably involves machine work, so probably not practical from where you sit now.

 

Certainly it would not be normal on a stock engine. GM isn't going to choose performance over warranty-term reliability.

On a modded car I think all bets are off.

 

Interesting thought there, Steve. Move it, hmmmm, I wonder how it would work in a coolant plug hole in the head. Probably not great due to more valvetrain noise but less from the opposite bank. Maybe in the extra oil plug location by the filter?????

Another member here recently filled an LT1 block, but was clever enough to insert a hose from above the fill line to the KS location before pouring the fill, so it will be wet. Great idea IMO.

 

The info I've read on KS retrofits says it needs to be in the block.

 

Update:

Well, a few more tunes and a day of racing, and I (think I ) have learned more.
-Car actually runs up to 10 degrees more timing than the table shows at WOT, as it is showing 32* advance but I only have 22 in the table. I have heard of 3 degrees of 'mystery advance' but 10?! I'd swear it was less a year ago, but lost those logs so can't be sure, and I don't know what would make it increase over time. Probably a topic for a dedicated post.
-I believe I have a mix of real knock at high RPM, and false knock at low RPM.
-KR at WOT was 4* on 94 octane, but 2* on 96 octane, so seems real and responds as expected.
-KR at idle and 1000 to ~2000 RPM does not seem to respond to reducing the advance. So, it is possibly false and due to the fill. I kept pulling timing every time I got KR in that zone, until I had around 7* base advance at light throttle. If anything, KR got worse with less timing. KR would start at 1-2*, and within a couple of seconds it would creep to 7* for 0 actual advance, and of course poor power.

Could this be pre-ignition due to late ignition causing overheated exhaust valves?? I checked to see if the headers were glowing, but they weren't.

I'm getting closer to taking out the KR as Injuneer suggested, but I still don't want to remove it completely. I seem to have no problematic KR in PE, so I think I'll try to cal our certain RPMs by setting max retard to zero there. Then I'll have a narrower operating band to do plug checks for. But first I want to convince myself this really is not real knock, or pre-ignition.

Moderators, this may be becoming more of a tuning post than LT1 tech, so feel free to move it.

 

I sometimes hear knock before the scanner picks it up. Pretty weird.

 

Thanks, Marshall93Z. At least now I know I'm not crazy...then again, you sig scares me a bit! Oh, wait, better than a quarter million miles, I guess blown up is ok then. Not for you, but not so unexpected, either.

With several more tunes, I think I am getting this figured out. Only around 1* knock now, once the engine is warm. figuring out the significant amount of 'mystery advance' helped a lot. And, I actually had too little advance in parts of my curve and adding advance reduced KR in those areas. I don't understand why, but it worked. I am now only taking general guidance from the data logs, and rescaling large areas of the spark map at a time, with more emphasis on the general shape as opposed to trying to exactly program in what the engine seems to 'want' and then smoothing it. I'll pull out about half of the indicated timing for several cells in each direction of a knock incident, and keep the slope of the curve smooth and always going uphill towards lower map and higher RPM, with relatively constant slope. Less bumpy map then before.

Knock sensor seems to be consistent now. For a while, anyway.

 

I have a friend with a Z06 that has this happen too!

I'm hoping I just MISTAKINGLY filled up with 87. It hasn't done this in the two motnhs I've had it running... until now.

Glad you're getting yours figured out.