reyk

I'm looking for an aftermarket knock sensor that produces a digital output (knocking, not knocking), or a way to turn the 951 KLR knock sensor output into a digital on/off type output.

Found stuff this site and elsewhere that shows the 951 sensor KLR output produces lots of fast short pulses on knocking, not logic level.

Only thing I found was something that lights an LED, might be alright, at
http://www.prostreetonline.com/buy/turboxs_knock_lite/. Anyone use one of these?

I'm running a Motec M4 (really nice) with all analog inputs used, but can use digital input to retard timing, increase fuel if I could find way to get knock signal.

Or, I suppose I could get off my butt and build cheapo circuit one shot that stretched out the edges. Anyone else want one?

Thanks
Reyk

m42racer

I have one that outputs a 0 - 5 V signal. PM me and I will tell who sells them. really good units.

jgporsche

clear your pms, so i can pm you, or pm me please. thanks.

reyk

Hi Guys,

I'm not so hot on the PM stuff. I tried some replies but maybe it didn't work.

My email is jreyk@comcast.net - Thanks

cpt_koolbeenz

How do you know those sensors are "tuned" for our engines? Knock sounds different for different engines (different frequency range)...

Zero10

Also worth noting, is that in the traces shown in past threads, it was discussed that perhaps the sampling rate is too low to show the actual output, and it could be a 5V digital output that is not being properly read due to it's short duration.
I plan to build my own 'knock sensor' setup. It will consist of 2 seperare circuits, and might answer whether this is a digital signal or not. One will just be a LED tied to the output, the other will be a LED positioned after an OR gate. If they behave exactly the same, then it's digital.
I just haven't gotten around to wiring it up since I'm a little bit afraid to cut my engine harness right now, and I injured one of my fingers badly at work, so wiring is out of the question for a little while.

reyk

I'm using the stock sensor with the 951, whose raw signal goes into the KLR box. So, it's the sensor that Porsche installed. The KLR box has a knock signal output, that I think only goes to some test harness, not the DME.

On the sampling rate issue, what I'm looking for is a 1/2 sec to 1 sec digital signal that I'll use to retard timing for ~ 1 sec - pretty drastic - so what would work would be something that took those many short logic signal edges and stretched them out to 1/2 sec - 1 sec. A 555 timer circuit can do just that - called a retriggerable one shot. So, what's needed is a small voltage regulator to make 5 volts from the 12v battery, 78L05 would work, and the 555 timer triggered by the KLR knock output. The 555 output is then the logic level knock signal.

But I'm hoping to get some info from the above posts on something already made

Thanks

mtnman82

reyk - IMHO you're on the right track with a one-shot. I think a scenario could arise where the knock pulses are too short duration and too infrequent (i.e. at the onset of knock) to light up an LED long enough to be seen. I also think in this case triggering a one-shot (or some other digital circuit/logic) is the proper approach vs. a 'sampling' approach.

The knock sensor is more than likely an analog sensor. Using it directly will mean that the digital part of the circuit will not be triggered until the logic threshold is reached (~1.4v?), which means knocking under this level will not be detected. One solution to this would be to use a comparator circuit, set to trigger at less than the logic threshold level (1.4v). Here you would want to set the comparator threshold to just above whatever noise you see from the car running normally under various conditions. Anything above normal engine noise would be considered knock, triggereing the digital circuit.

cpt_koolbeenz

The stock sensor is analog. The KLR listens for the proper signature of knock then retards timing. You WILL get a lot of noise out of the knock sensor... some of which may be above whatever threshold you set with your comparator... Not only do you have to look amplitude of the waveform out of the knock sensor, but you also need to look at the frequency.

Hook up a scope to the sensor output and watch it when you are driving... Then park it and hit parts of your block with something metal and see the different signatures. You would have to get the engine to knock reliably to find the correct signature. After which you wouldn't want to run that engine ever again!

I was talking with a fellow lister about doing some experimenting just like you guys are doing. He did scope it out and sent me the wave forms. I don't know if I still have them, but I will look.


Maybe a more reliable way would be to tap into the KLR and get the knock yes/no line to do what it sounds like it should do: output high or low when it detects knock.

badass951

If you need a wideband O2 as well, but the PLX R-500. You can connect the knock sensor to it and it does really give you knock output voltage, as well as a warning LED if you choose to set it as so.

Zero10

I think we are slightly unclear here.
The KLR knock sensor output is a digital output.
The knock sensor itself is analog, rather similar to a microphone.
No idea what the KLR looks for in the signal, perhaps a sudden spike in noise?...
Anyways, if you want to tie a retriggerable timer to the knock output, it shouldn't be hard. There are lots of 5V regulators that take from 7-18V input, pick one that can handle ~200mA, since you never want to stress your voltage regulator.

I'm worried you're going to find that the KLR emits a knock signal somewhat randomly, and that you will get a lot of false alarms, but that is just a guess.

reyk

Thanks for the input, all.

Looking at the plots of knock (Klopfen), throttle position (DK-Winkel), etc previously posted I have a guess as to what's going on. I agree with Zero10 on the knock sensor inputs and outputs from the KLR. The sensor input is low level analog, like a microphone. The KLR knock output is digital, in that it's either at 5v or 0 v.

Looks to me like the KLR amplifies the sensor signal, then filters it to exclude all but the frequency range where knock occurs. Then it runs the filtered signal to a comparator which compares the filtered sinewave to a fixed threshold. When the threshold is exceeded, the comparator switches.

So, when knocking occurs, the filtered signal increases dramatically in amplitude, and everytime it swings above the comparator threshold the comparator switches. If this is true, then we should see comparator switches between 5v and 0v at the knock filter frequency. This is kind of like what the plots look like. The second plot, with no knocking but still wide open throttle and max airflow, doesn't have any comparator switches, so hopefully the false alarming isn't too bad. Depends on how good the filter is on the raw signal and how much engine noise there is in that frequency band. Knocking is going to be a short impulse in the time domain, so it will have lots of high frequency components in the frequency domain, so maybe they're doing a high pass or high frequency bandpass filter on the knock sensor

CPT_koolbeenz' suggestion to do it in the KLR also sounds like a great idea - more robust than doing it externally. Think it only takes 1 chip, which can be deadbugged onto the printed circuit board.

I'll give this a try and post results. Think we can test it by banging on the block near the sensor. Also, maybe I can dig up the KLR schematics which will show us exactly what they are doing.

Thanks - Reyk

Zero10

If you bang on the engine using a metal rod and a hammer, that will usually trigger the knock sensor. If you do it at idle, you can hear the engine stumble for a second as timing is retarded and whatever else it does is done.

The KLR is a well designed piece of equipment, despite it's age, and it is pretty darn good at detecting knock. I'd trust it. Let us know what you build, and how it works. I plan to build a small circuit of my own to test it out as well. I would like some kind of knock warning light.

reyk

Thanks - I've ordered the parts and will test it soon.

The 555 timer will source > 100ma, so I'm running its output plus ground wire to high brightness LED array for warning light also.

As a side note, I'm controlling both fuel and ignition with my Motec, not DME or KLR, but banging on the block at idle still causes stumble. So, this might be due to injectors or something else not liking being banged, rather than KLR?

Zero10

Is the KLR still plugged in?.... I do not know exactly how your setup is, so I can't help too much.

555 timer drawing over 100mA?.... I sure hope not. I always thought they were in the 10-15mA range. Perhaps the capacitors and such used to alter the timing increase the draw?