I've been running this thing for a little while and I am getting pretty consistent results as far as noise levels and response from Sensors 9-12 in the logs.
My issue is that I am not sure how to interpret the noise levels under some conditions.
For example, I'll have a region in cruise where RPM is about 3500 and MAP is transitioning from 40 to 60 or so kPa and the noise level increases dramatically around 50-60kPa. I don't feel my fuel and spark are aggressive (14.7 AFR, 34-38 degrees advance). It could just be an engine resonance or noise. I did try to trim the noisier cylinders (there were some pretty sizeable differences from cyl to cyl in some cases) and that helped some.
the good news is that under boost (I'm only running ~9 psi on the gauge/150 kPa on MS) there is more noise but no large spikes. Actually that may not be true 100%. I need to get some SD logs and verify none of the spikes are clipped. See note below...
Some places are easy to interpret. I turned off (raised tpsdot limit) enrichments to work on tuning EAE and I get occasional tip-in spikes when the mixture goes lean briefly. And seeing individual per cylinder knock is useful as I mentioned above for tweaking per cylinder trims.
I have NOT seein the issue where knock level "sticks" to a certain value when cold starting. Mine always hovers around 10-20 at idle, depending on the cylinder.
Note to James: Since knock can be very short duration in a datalog, can the logs show highest knock over the last "t" time period? Perhaps like a third (300ms) or half-second (500ms) rolling peak indicator like your stereo equalizer that shows you a max level for any given previous third or half second or something. NOT an average... but a peak. So if a spike falls between log values, it still captures the maximum knock level and displays it. Even having "knockin" done this way would work since generally I want to be overly cautious when reacting to knock and don't mind an extra half second of retard.
I've made some of these additions ready for the next release.
I added new gauges/variables for knock per cylinder (instead of hijacking sensor09+) and made them peak/decay.
The SPI knock initialisation code also validates the replies it receives, however this isn't working correctly with the pre-prototype board I have installed on my car, so more investigation required. (Works fine on the bench with a board the same as the ones I sold.)
Does the per cylinder knock work only with this module or with other modules? I have the Saab unit my car came with and get a pull to ground when knock is sensed. Will this also give per cylinder information?
Keithg wrote:Does the per cylinder knock work only with this module or with other modules? I have the Saab unit my car came with and get a pull to ground when knock is sensed. Will this also give per cylinder information?
Keith
I'll have to look into that, I'm pretty sure it won't at the moment.
I have to ask because there is no activity in this thread anymore - are the proto-boards not working? or are they working so flawless that there is nothing to write about?
I think a lot of people would like to know how it goes with this and how the further development will be...
Thanks!
--------------------------------
fun is not a straight line
Zaphod wrote: are they working so flawless that there is nothing to write about?
!
This is where I'm at. I was about to make a post last night saying "put 5k on this board so far and it's working flawlessly." But felt that it didn't really contain enough content to post. I think right now this is one of those "no news is good news" type situations. In my case, I had to make some modifications by adding resistors inline with my knock sensors to get the board to work properly. Now it's working absolutely flawlessly. Really nothing else to report on my end.
Any updates on the "intermittent incorrect values" until powercycle? I made a change for beta19 firmware that may (or may not) have addressed it. I've been unable to replicate this problem on the bench myself.
James i have done some testing as i am still having HELL with the strange reading i was getting.......
One of the problems seem to be gone with your code additions. This was the problem knockin not always being 0 with the engine off and ignition on.
However i still get HORRIBLE readings when cold, it is ether the sensor it self, or the wiring i am starting to think. I had it running for a long time, very long and the readings very very VERY high. So i put on some heat gloves and went to work. One of the tests i did was to disconnect the the knocksensor when running, and with engine rpm there was a moving knockin reading, so somehow noise is getting into the system.
First thing i am changing is the knock sensor as it is a single wire, it grounds through the sensor body, will change it to a bosch sensor, and see what happens, if that stops i will change the shielded wire..
Still cant figure out when turning off and turning on will fix it.....has to be a dam ground...