high octane fuel for establishing knock threshold

[Yves]

I was wondering if it would be possible to use a high octane race fuel (non leaded - 109 MON) to see what 'normal' levels of knock are which can then be attributed to engine 'normal' noise.

Anyone ever try this ?

[Matt Cramer]

I haven't myself, but it should be workable.

[elutionsdesign]

Isn't it easier just to back off your timing with your current fuel?

[Yves]

Isn't it easier just to back off your timing with your current fuel?

Well, how do you know what's knock and what isn't ? I mean is 50% knock or 20 % ?

[nathaninwa]

I look for the spike in knock percent.

[Yves]

I look for the spike in knock percent.

hmm, lots of spikes.

[nathaninwa]

Could be lots of things. Timing chain, piston slop, solid bucket cam followers, exhaust tinging against block

Maybe try pulling some timing globally or adding some octane and see

Are the settings correct for your sensor?

Dyno in the area? The tq curve will show kiock

[Yves]

Well I think it's mechanical noise since the overal knock% is increasing/decreasing with rpm, but that remains to be seen after some further testing. However i'm seeing 100% so if it's mechanically induced, then it's useless I think.

I think the settings should be ok, at least the frequency, unless this link was providing incorrect info. http://www.phormula.com/KnockCalculator.aspx.

Yes, a dyno would be great, but I do not want to spend countless hours on it because of the cost involved so at least the tune should be close before that becomes a possibility.

[grom_e30]

can you post logs? also have you adjusted the gain at all?

[Yves]

can you post logs? also have you adjusted the gain at all?

I'll go through my logs this evening and will try to post them.

No I have not adjusted the gain.

[grom_e30]

i had adjusted the gain as originally i was reading a fairly high number just at idle say 30-40% or so lowered the gain to bring the % down at idle to a reasonable number say 10%. then i backed the timing down so the engine was not going to knock like -10deg from the whole table and did a few runs so i could determine the background noise from the engine and used the max knock% i logged from those runs with a small margin added on top as the % that is real knock.

on a log a real knock will look like a really high sharp peak over a small time.

[Yves]

Here you go

[grom_e30]

just a quick thought as you are running 2 sensors and linking certain cylinders to specific sensors one on each bank id guess, you will need to adjust the gain as cylinders further from the sensor will read differently to the ones closest. this is something i have not experimented with yet.

[Yves]

yes, but what amount of gain ? I have no idea how to set it.

[grom_e30]

my guess would be to either set come gauges for knock% to all the cylinders on one bank then use the gain up or down as if it was a volume control until they were all pretty uniformed in there reading or take some logs and look at them in mlv.

another thing for you to look at if you can easily remove a sensor do a data log with the engine running and then tap the removed sensor with something metal and look at the log you should see a very big spike and will give you an idea of what to look out for don't rely on the knock % you read from this but more as a visual representation of what you are looking for.