I believe that reliably detecting knock is the hard part - noticing when it's happening, and not triggering on false events. It's all down to tuning the filters to match the engine. Might be beyond us mortals!
Actually doing something with the knock signal is the easier part.
knock sens
Moderators: JeffC, rdoherty, stieg, brentp
-
Oliver Sedlacek
- Posts: 68
- Joined: Mon Sep 13, 2004 8:45 am
- Location: The Chalfonts
- Contact:
hi lads,
no doubt there are issues with exactly capturing the threshold between no/acceptable knock, and undesirable knock, and that is both due to diffentiating signals, and sensor placement. even major manufacturers, however, face that problem and i wouldnt assume they have it quite so clinically solved as one might imagine. seems to me that even mortals can capture useful information, even if only at the extreme ends of the scale. after all, if knock is really bad, you can hear it with your own ears, so at that level, it you could adjust your detection device by ear. having a feature to retard ignition in the face of only extreme knock would be useful. i am, however, sure we can do better than that.
there are already devices you can buy and which are being used, which indicate engine knock. how close they get to indicating THE desired threshold is unknown, but they are out there, and if mjlj can take their output, as an input, with only a software change, that has to be a good and useful thing.
re the frequency issue: this is perhaps not as complicated as one might think. intuitively, the frequency of the knock is going to be related to the dimension of the bore, and that is exactly what i was asked when i purchased the item i am waiting for, with the express purpose of having the filtering preset. someone on another thread quoted an engineer mate as saying that the rule of thumb is that 520000/bore in mm, yield knock frequency in Hz. eg 80mm bore would give 6500Hz. 110mm gives 4700Hz. i note that the Delphi website (GMs supplier) shows 4 sensors with centre frequencies from 5200Hz up to 8000Hz, so that rule of thumb seems quite credible. bosch seem to be the world's main supplier, and i have read nothing about their sensors having different frequencies. i have, however, read that the trend is moving from narrow to wide band sensors, with all the filtering done at the processor level.
obviously brent has a never ending list of enhancements and amendments to work on, but i really think this would provide a lot of value for the effort involved.
(brave statement of course, not actually being able to quantify that effort! so, who knows were to start on the task of understanding how to amend the code?)
regards
alexander.
no doubt there are issues with exactly capturing the threshold between no/acceptable knock, and undesirable knock, and that is both due to diffentiating signals, and sensor placement. even major manufacturers, however, face that problem and i wouldnt assume they have it quite so clinically solved as one might imagine. seems to me that even mortals can capture useful information, even if only at the extreme ends of the scale. after all, if knock is really bad, you can hear it with your own ears, so at that level, it you could adjust your detection device by ear. having a feature to retard ignition in the face of only extreme knock would be useful. i am, however, sure we can do better than that.
there are already devices you can buy and which are being used, which indicate engine knock. how close they get to indicating THE desired threshold is unknown, but they are out there, and if mjlj can take their output, as an input, with only a software change, that has to be a good and useful thing.
re the frequency issue: this is perhaps not as complicated as one might think. intuitively, the frequency of the knock is going to be related to the dimension of the bore, and that is exactly what i was asked when i purchased the item i am waiting for, with the express purpose of having the filtering preset. someone on another thread quoted an engineer mate as saying that the rule of thumb is that 520000/bore in mm, yield knock frequency in Hz. eg 80mm bore would give 6500Hz. 110mm gives 4700Hz. i note that the Delphi website (GMs supplier) shows 4 sensors with centre frequencies from 5200Hz up to 8000Hz, so that rule of thumb seems quite credible. bosch seem to be the world's main supplier, and i have read nothing about their sensors having different frequencies. i have, however, read that the trend is moving from narrow to wide band sensors, with all the filtering done at the processor level.
obviously brent has a never ending list of enhancements and amendments to work on, but i really think this would provide a lot of value for the effort involved.
(brave statement of course, not actually being able to quantify that effort! so, who knows were to start on the task of understanding how to amend the code?)
regards
alexander.
-
Oliver Sedlacek
- Posts: 68
- Joined: Mon Sep 13, 2004 8:45 am
- Location: The Chalfonts
- Contact:
I guess it's also a matter of managing user expectation. Retarding the ignition during the occasional times that the engine starts knocking significantly can probably be achieved with a crude, empirical setup. Keeping an engine safe during extended running close to the detonation limit is a different kettle of fish.
well said, oliver.
support for you statement is found in this interesting article, which i have found in the meantime.
http://autospeed.com/cms/article.html?&A=0348
different formula, i note, but which yields a similar answer for knock frequency.
support for you statement is found in this interesting article, which i have found in the meantime.
http://autospeed.com/cms/article.html?&A=0348
different formula, i note, but which yields a similar answer for knock frequency.