16 posts / 0 new
Last post
srs14
Offline
Last seen: 3 years 10 months ago
Joined: 08/17/2014 - 09:11
Knock sensors with series 2, resonant and flat response?

My specific questions is for the nissan sr20det engine.  Supposedly some factory PCM's have an internal amplifier/filter used to monitor the factory knocks sensor. I was wondering if the Series 2 supports these factory sensor (flat response knock sensor) or should I source a suitable replacement from a similar engine (resonant knock sensor) to take advantage of the knock control of the EMS. I have noticed erratic reading on my particular engine and was curious if this could be related. Its been a year or so since I have messed with my car so I cant remember specific examples. 

Here is a website that im citing my info from 

http://www.enginebasics.com/EFI%20Tuning/AEM%20EMS%20Knock%20Sensor.html

AEM_NS
Offline
Last seen: 2 months 4 weeks ago
Joined: 06/06/2014 - 09:31
Yeah, there's a lot of

Yeah, there's some misleading information on that site.  First and foremost, the Series 2  has always had internal knock filtering circuitry and further, it works with both resonant and non-resonant "flat response" knock sensors.  Typically, the resonant sensors are the one wire type and are generally pretty "old school" as knock sensor technology goes.  Many times, depending on the app and the specific sensor, these happen to be excessively noisy which does not make for good knock detection as it's difficult to decipher between noise and real knock when the noise is of an amplitude that makes it look like knock.  You'll have a hard time finding a resonant knock sensor on a car built in the last 10 or so years.

The non-resonant "flat response" knock sensors are more commonly used now on newer cars.  They are typically "quite" and don't have an excessive amount of background noise while still registering knock quite apparently and clearly thus making knock control overall much better.  The old school resonant sensors were designed and limited to be used on a specific engine as it needed to be within the sensors resoant frequency thus making mixing and matching different resonant sensors on different engines problematic.  The "flat response" sensors can really be used on anything and general work very well.  

So again, the Series 2 will work with either type of sensor and the Series 2 does have a knock amp/bandpass filter.  By most standards, a non-resonant "flat response" knock sensor is usually considered to be better than a resonant one.

Kurt
Kurt's picture
Offline
Last seen: 7 months 1 week ago
Joined: 06/04/2014 - 17:17
The old forum had a thread

The old forum had a thread about setting up knock control.  It's not here anymore, but I did post it on another forum. http://www.efi101.com/forum/viewtopic.php?t=7972

spikeygg
spikeygg's picture
Offline
Last seen: 2 years 1 month ago
Joined: 08/09/2014 - 08:33
Hey Kurt! Great to see you're

Hey Kurt! Great to see you're back on the forum. I have a question for you.

This weekend I wired up a two-pin flat-response sensor in place of my original stock resonant-response sensor. While I was doing this I noticed that the stock wiring has a shield which is grounded under the dash to the rest of a ground network and appears to go all the way to the sensor. My question is should the "ground" pin from the sensor be wired to the same ground that the shield was? I ask this question because the cable that came with the sensor I bought only had one pin in it and I bought a second cable and canabalized it so I could hook up to both signals.

Here's a link to the sensor I bought: http://www.ebay.com/itm/171259604603?_trksid=p2060778.m2749.l2649&ssPage... if you look at the 4th picture in the list at the top left, you'll see that there's only one pinned connection in the plug. Anyway, I started thinking about this and I wonder if I screwed myself by making that connection touch both sides of the microphone and hooking it up to my ground network. Maybe I'm now driving noise into the ground network?

Anyway, I wanted to ask you since you have so much experience with this.

Thanks,

-Greg

P.S. thank you so much for posting that original knock sensor article. Last week I finally put it into motion and it was an excellent exercise. I have found that my optimal knock center frequency is a bit lower than the basemap's default: https://www.youtube.com/watch?v=kpUMZs2jgpg

Kurt
Kurt's picture
Offline
Last seen: 7 months 1 week ago
Joined: 06/04/2014 - 17:17
The flat knock sensor needs

The flat knock sensor needs to have one pin going to the ECU KS input and the other going to Sensor Ground (not chassis ground).  I suggest you check something first, though.  I've read that some of these flat response knock sensors will ground through the body that contacts the block.  Test this with a multimeter between the body and each pin.  If it is grounded through the body, you will have continutity between it and one of the pins.  You want a sensor that does not ground through the body and the fact that your sensor came with a harness with one wire makes me wonder if it does.  So, depending on this, you may need to get a different sensor.  If your sensor is grounded through the body and you hook that 2nd pin up to Sensor Ground then you will contaminate the signal for all your other sensors that use Sensor Ground.

Cam/Crank, O2, knock and any other low voltage sensors will usually have a shield that is open at the sensor side and all tied to a single ground point at the ECU side.  It's very important that you leave the shield open at the sensor side for shielding to be effective.  If this side is grounded in some way, the shield doesn't work.

In the case of the 2Jz engine, the stock harness has shielded single wire knock sensor cables, just like your DSM because they use resonant sensors.  It uses 2 sensors, front and back.  What I did was use the existing single wire cable to one pin of the KS and spliced two wires off of sensor ground and ran those to the other pin on each sensor.  This left me with the worry of seeing any induced noise from the sensor ground wire.  This wire is not shielded and splices to different sensors through the stock harness.  It turns out, for me, this isn't an issue.  You'll just have to test it out and see if it works for you.

From your recording, it sounds like you are picking up a lot of ignition noise.  I'd recommend you first chech out your sensor like I mentioned above.  Also, ensure that the ignition filter capacitor is in place on your harness.  Once you're certain it's wired correctly and not grounding to the block, get another recording of your engine noise.  For reference, here is a WOT pull recording of my front knock sensor.  Notice the upward sloping linear lines in your spectrogram display, that's ignition noise.  You don't see that in mine.  http://youtu.be/49LpCFK6to4

Once that is all correct, then you will want to investigate how much of an offset there is, if any, between the near cylinders and the further ones from the sensor.  It looks like Mitsubishi designed the block and braket the KS bolts to pretty well for transmitting noise to the sensor.  In that thread I linked to, pay attention to the method of inducing LIGHT knock in different cylinders.  You can do this with the cylinder ignition/fuel trims and LOW throttle.  You want to creep up on it at low load and review your logs and sound recordings each time.  From what I'm seeing with the block and braket, I think it'll work fine, but it's nice to validate.

There is a new option in the latest firmware called Knock per Cylinder.  This could be a real benefit to you, but in order for it to work correctly, each cylinder HAS to have it's own coil driver output from the AEM.  If I'm not mistaken, the 3000GT uses wasted spark, so in it's current form this won't work for you.  You could possibly change over to COP, or just put a Schottky diode on coil outputs 1-6 and pair them to the correct coils. Then set the calibration up to run like it were full sequential iginition.  This way, the AEM will know which cylinder is knocking and will only apply enrichment and retard to that one cylinder.  As it is with one knock sensor and wasted spark, you'll be retarding all cylinders when knock is detected and you won't see in the log which cylinder is acting up.  It's not necessay that you actually have COP, just that the AEM is sending ignition triggers as if it did have COP.  This new feature is REALLY nice!

spikeygg
spikeygg's picture
Offline
Last seen: 2 years 1 month ago
Joined: 08/09/2014 - 08:33
Awesome info Kurt. I will

Awesome info Kurt. I will have to do some sleuthing but I'm guessing (as you are) that my sensor is crap. Are you using a Bosch sensor? Do you by chance have the part number handy? Mine is a cheap knockoff and I guess the old adage is true, again: "you get what you pay for."

I have another question for you. I found this doc on the web that shows with just a cap or two you can wire up headphones or a laptop to record the knock signal (the same signal that the ECU is listening to). I'd like your opinion on doing this and how it might affect the sensor ground. I wired my sensor up like this with a cap on the signal side and sensor ground hard-tied to the sleeve (of the TRS/mini jack). I'm also worried that having the laptop hooked up to sensor ground might cause some issues. When you recorded your audio did you use this method with capacitors or did you unhook the signal from the ECU and plug it directly into the laptop?

Thanks for all the help. You are a wealth of information!

-Greg

Kurt
Kurt's picture
Offline
Last seen: 7 months 1 week ago
Joined: 06/04/2014 - 17:17
That recording was taken

That recording was taken using a seperate cable, so there was no connection between my laptop and chassis ground or ECU sensor ground.  After thinking about this a little, I'm going to have to go back and do a few more recordings and see what the ECU is actually getting from the knock sensor.  It will be interesting to see if there is any addition igniton noise.

Hooking into knock sensor signal and sensor ground (like the diagram shows) to your laptop won't be a problem if it's going to a high input impedance load like your laptop mic input.  I think there would be a drop in signal level the AEM is seeing if you hooked up headphones which have lower impedance and load the sensor more.  It wouldn't do any damage hooking headphones up, but I wonder if you'd actually hear anything.  You could also add a 100k resistor after the cap to reduce loading on the knock sensor and attenuate the level going into the mic input.  This may or may not be necessary, though.  I just turned my mic input level down all the way and was able to avoid any clipping.

spikeygg
spikeygg's picture
Offline
Last seen: 2 years 1 month ago
Joined: 08/09/2014 - 08:33
Get this: the recording I

Get this: the recording I uploaded to youtube was also directly to the laptop's mic input. At that time I had installed it as a second knock sensor so I was just listening to it and because it was also picking up the ignition noise that pretty much guarantees that the base was tied to the ground pin. Anyway, I pulled the engine apart tonight, pulled that bugger out, tested it and indeed one pin is shorted to the base.

So I need a new knock sensor, I just bought this one: http://www.ebay.com/itm/151311740862 it looks like a genuine bosch from a porsche so, we'll see, I guess...

Kurt
Kurt's picture
Offline
Last seen: 7 months 1 week ago
Joined: 06/04/2014 - 17:17
My knock sensors came from a

My knock sensors came from a couple of early 90's Volvos at a bone yard. lol  Cost a few dollars.  They still work fine. :)

Correct torque on the sensor will be around 15 ft/lb.

AEM_NS
Offline
Last seen: 2 months 4 weeks ago
Joined: 06/06/2014 - 09:31
The Bosch "flat response"

The Bosch "flat response" knock sensor is pn 0 261 231 006.

Kurt
Kurt's picture
Offline
Last seen: 7 months 1 week ago
Joined: 06/04/2014 - 17:17
If you ever go to a pick n

If you ever go to a pick n pull, you can easily grab them from the 1st gen Focus.  They use one sensor which is under the intake manifold and can be removed with just a 10mm wrench and wire cutters.  No need to remove anything to get to it.

spikeygg
spikeygg's picture
Offline
Last seen: 2 years 1 month ago
Joined: 08/09/2014 - 08:33
FYI: I got my Bosch knock

FYI: I got my Bosch knock sensor and it is a much cleaner signal -- no more popping from the ignition components. I created a youtube video that details my findings about the generic flat-response knock sensor just in case others want to try to do this too. Bottom line is: don't buy a generic knock sensor, they are grounded through the block.

http://youtu.be/ZSZSDDRe6Js

Kurt, now that I've done this analysis I fully understand what you were talking about when you said "1-wire vs. 2-wire knock sensors", thanks for all your help.

-Greg

Kurt
Kurt's picture
Offline
Last seen: 7 months 1 week ago
Joined: 06/04/2014 - 17:17
I just did some recordings

I just did some recordings today using the interface mentioned by SpikeyGG above.  I used a 0.01µF cap on both channels going to the mic input on my laptop and I can report that it did pull down the signal a little.  If you're not already maxed out on the gain, you can adjust for this drop while plugged in by turning it up.  Just remember to change it back when you're done recording.  

rvrkazan
Offline
Last seen: 1 year 3 months ago
Joined: 05/26/2017 - 04:34
i have sr20det black top and

i have sr20det black top and EMS series 1. what knock sensor i need for my EMS? do you now some cheap analog(bosch or other) who will work with my engine+ecu? Thank you.

 

AEM_PT
Offline
Last seen: 11 hours 58 min ago
Joined: 05/22/2014 - 08:13
The listed Bosch part number

The listed Bosch part number from earlier in the thread is about $30 on Amazon.

Solo_S14
Offline
Last seen: 1 month 2 days ago
Joined: 01/31/2015 - 09:18
Hi guys,

Hi guys,

I know that I'm raising this from the dead, but I think this is a good place to ask for clarification on a few items mentioned:

1. In Post #2, AEM_NS states "Series 2 does have a knock amp/bandpass filter"
Does the AEM EMS V1 have the same hardware?  The referenced "EngineBasics" link speaks specifically to the V1 (I run V1 so seeking clarification).

With the assumption that a resonant type sensor puts out 300-500 millivolts, I assume that both the resonant and flat-type sensors need amplification??

2. In Post 5, Kurt states "The flat knock sensor needs to have one pin going to the ECU KS input and the other going to Sensor Ground (not chassis ground).".
and " It's very important that you leave the shield open at the sensor side for shielding to be effective.".

Turns out that my SR20DET "Nissan" sensor is grounded to the block.  I want to replace the sensor as recommended, and explained in Spikeygg's youtube video (https://www.youtube.com/watch?v=ZSZSDDRe6Js).

Am I correct in stating that I would leave the Shield wires grounded at sensor ground, and Open / not connected at the knock sensor (Pics 1 & 2).  Then run a separate 2nd wire from the 2nd pin on the knock sensor to the sensor ground?  Since the Shield is connected to the sensor ground, can I just wire the 2nd pin from the knock sensor into the "grounded end of the shield" at the other end of the jumper harness? (At picture 1)?


3. Is it ok to wire in a set of head gear on the same harness as the knock sensor (ex. 3.5mm Ground and "left" (to mute left) at the sensor ground, then "right" at the KS sensor pin on the harness?)
Will wiring in a 3.5mm feed to plug into a laptop mess with the signal?  Should I instead, run a 2nd knock sensor for headgear / recording, and leave the stock location sensor for AEM EMS V1 logging?

4. Final question - if the V1 doesn't have the capability to amplify and tune knock input from a Bosch sensor, what are my options?  Are there AEM documents to reference and sort out what to do?

Thanks guys - really good info in this thread!