drive it

iTrader: (1)

I'm looking for ideas about this one-My new engine has the block filled past the knock sensor hole and I can't seem to "desensitize" the knock sensor enough.
Yes it's false knock; already ruled out real knock with all the usual, and made sure nothings rattling/vibrating.
I wrapped the threads with teflon tape, tried an 86 and then 91 sensor. Tried real light tq to factory spec. tq.
I know I can ramp the spark back in a lot more rapidly to help deal with it, but if anyone else has run into this and found a different knock sensor that helped (LT1???) or another solution I'd really appreciate any advice! Oh Yea-8D with aujp. Thanks!

ULTM8Z

That's wierd. My ZZ4 block has the the same type of filled hole, but I don't have an oversensitive sensor. I'm also running 8D, but with ANHT. I'm also using a 90-92 350ci type sensor.

I'm not sure about a solution, I simply offered this as a data point in your investigation.

I suppose just make sure that your sensor corresponds to your displacement, either a 350 or a 305. From what I understand, they're different.

87 Vette Owner

I'm not sure if your set up is similar, but on my '87 Corvette, I was able to install a much less sensitive ESC Module (GM part # 16038331) and that solved my false knock problems. Here is the tech tip link from the Corvette Forum:

http://www.corvetteforum.com/techtip...D=99&TopicID=2

884+3

Another possiblity I think Grumpy has suggested in the past is to use a 1\4" service 45 to get the sensor farther away from the block. I never had occasion to try it myself but if Grumpy suggests it im not going to argue the point.

Grim Reaper

Just limit the maximum amount of Knock Retard within the eprom. If it is only in WOT where this is occuring, you can just use the Max KR vs RPM (PE) table and leave the other (Max KR vs MAP) for regular driving.

I know a number of people who limit the Max KR vs RPM (PE) to just 1* and then control the spark curve themselves. In fact, by limiting the KR, they can control the spark curve better and actually run a little more spark advance (if it makes more power).

JP86SS

iTrader: (1)

Wrapping the threads with teflon tape will not allow the sensor to ground to the block. could make matters worse.
Limiting the amount of reaction to the signal is probably the best thing until you find something better.
Could try adding a resistor in parrallel with the sensor to reduce the sensitivity. The 91 sensor should be 3.9K ohms, try reducing that by 10-15% and see how it fares.
Jp

midniteplowboyy

iTrader: (1)

Where did the 91 knock sensor come from or off of?

Fast355

iTrader: (2)

I am running a Corvette LT4 knock sensor with the addition of a parallel resistor on my TBI 350.

dimented24x7

iTrader: (2)

If the block is filled, then Id probably go ahead and just disable the knock retard, and just use it as a reference. The acoustics have totally changed now that its filled and theres no telling if it would actually properly detect knock.

HaulnA$$

Check out this thread. https://www.thirdgen.org/forums/diy-...k-sensors.html . I posted a file that shows how to properly de-sensitize the knock sensor with resistors. It serves to de-sensitize the knock sensor by 25% as shown but can be reconfigured for even less sensitivity if you recalculate the resistor values. The circuit retains the original resistance that the ECM sees so it wont throw a code. I found it on the diy-efi site. HTH

Grim Reaper

Again, why all this work, cost and effort to do something that is not as quick and effective as controlling the Knock Sensor within the eprom?

Besides, I thought this is DIY Prom, not the Electronics Board.

Fast355

iTrader: (2)

The LT4 knock sensor has to have its resistance values changed in order for the TBI ECM to properly pick up the signal and not set a code.

dimented24x7

iTrader: (2)

The object, at least to me anyway, is to get the knock sensor to pick up only what its supposed to, and not every little engine vibration. If the KS picks up everything, then you might as well disable it and have no retard at all as its of little use if its constantly sending false signals to the ECM.

Grim Reaper

D, I've always attacked the KS within the eprom. Sometimes, disabling it is the most effective method. I prefer to disable it only where it causing me a problem and retain as much as possible. I figure why spend time, money & energy when a 5 second change in the eprom fixes the whole thing? Especially when "filtering" is not really fixing the "harmonics" issue anyway but "deafening" the KS.

I also assume the poster has the eprom burning equipment as he has posted this on the DIY Prom Board not the Electronics Board. However, if he doesn't have the eprom burning equipment, then electronics is a definite way to go.

HaulnA$$

By de-sensitizing the knock sensor externally to the ECM and then by adjusting the knock parameters in the chip you can experience the best of both worlds. Think of it this way. Most everyone runs their TV audio through their stereo in some way. If the TV volume is turned up too high and the the stereo amplifier is turned down too far, the audio quality suffers. The inverse is also true. If the TV volume is too low and the amplifier is up too high, the audio suffers as well. If the proper levels are set on both, you get the best sound. The same applies to a knock sensor with the sensor playing the part of the TV and the ECM the amplifier. After all, the knock sensor is an acoustic device, a hydrophone (underwater microphone) to be specific. I believe this applies to the DIY PROM theme as it deals with handling an input to the ECM in the best way necessary to optimize the PROM calibration. JMHO

drive it

iTrader: (1)

Replies to all in no particular order-this is a 730 speed density, so no esc module to update, yes sensor is for a 350, I really would like to keep the knock sense function so dont want to just disable it, yes I have all software and hardware for tuning and tune myself; thanks for all the replies! At this point I'm going to try the 45deg adapter and then the resistor. Unless anyone has found a table that addresses the knock sensors sensitivity I don't see any other alternatives at this point. Thanks!

Grumpy

Dinkin with these probably will get you what you want.
I say probably cause they can be a real exercise in patience to work with.
If you come up with a quick answer, 50/50 chance you went too far, or did something wrong.

BE EXTREMELY CAREFUL USING THIS INFO., detonation can at times instantly transition to preignition, and wipe out a piston dome, in msecs.. Yes, just a few combustion events can totally wipe out a piston dome.

I did some playing around with the old 60 project, and dinked with these to be able to use various knock sensors. It seemed to work, but gads, it was time consuming.

AGAIN, BE FRICKIN CAREFUL WITH THIS INFO.!.........

;--------------------------------
; ERROR 43, KNOCK
;--------------------------------
L8298: FCB 100 ; 10 SEC TIME REQUIRED
L8299: FCB 181 ; 3.62V UPPER VOTLAGE THRESH
L829A: FCB 72 ; 1.44V LOWER VOTLAGE THRESH
L829B: FCB 1 ; Add 0.3 deg SA for test
L829C: FCB 180 ; 95c, Min cool for for ERR 43B
L829D: FCB 75 ; If MAP < 75 THEN DISABLE ERR 43B
L829E: FCB 71 ; 24.9 Deg MAX S.A. for ERR 43B TST

Grumpy

Being an acoustic devise, it's going to lick up every noise, and if it's the right frequency, then the ecm is going to use that input as being knock.

You can fne tune the K/S by using different memcals, with different filtering on them. But, again, it's a PITB to do, and can get pricey for min results.

The frequency is all about the flame front's speed through the chamber, as it goes Mach. So different filtering is based on cylinder bore, and the *ring factor* (noise from the supersonic wave hitting the piston, and making it rattle within the bore). Obert talks about it in detail in his book.

rgarcia63

A couple of observations

1.) If the ECM needs to see e.g. sense the total original resistance then the circuit shown will not work because it changes the total ressitance.
A code 43 is not thrown because the original resistor was retained (can't remove it - it's internal to the knock sensor,) but because changing the total circuit resistance changes the signal amplitude to an acceptable level.

2.)
http://www.corvetteforum.com/techtip...D=99&TopicID=2
In this post James Kovitch states;
Actually;
The system performs a functional check once per start up to check the ESC system. To perform this test the ECM will advance the spark when coolant is above 95 degrees C and at a high load condition (near WOT). The ECM then checks the signal at "B7" to see if a knock is detected, The functional check is performed once per start up and if knock is detected when coolant is below 95 degrees C (194 degrees F) the test has passed and the functional check will not be run, If the functional check fails, the "Service Engine Soon" light will remain "ON" until ignition is turned "OFF" or until a knock signal is detected.

dimented24x7

iTrader: (2)

As shown above, the ECM will be looking for the correct range of voltage across the KS. These bounds can be changed within the prom if need be to prevent false code 43's. I dont know how it is with the TPI ECMs, but in the later PCMs, it looks as though they eliminated the forced knock test. I never liked that test, anyway.

dimented24x7

iTrader: (2)

It doesnt seem as though anyone really truely knows what causes engine knock. To me, detonation is the point where the mixture, or remaining mixture, no longer needs an ignition source to ignite. It can then, theoretically, ignite all at once, which would cause the cylinder pressure to immediatly shoot up, which would shock the surrounding cylinder walls, piston, head, etc. This also fits with the observations of engine tuners when an engine goes into detonation. Its basically the same thing as an explosive going off. It always seemed to be a better theory then supersonic flame fronts, colliding flame fronts, etc.

dimented24x7

iTrader: (2)

I took that same approach, just disabled it and kept it as a reference. If the original poster wants to retain the functionality of the KS, then it might be worth a shot to try and adjust the resistance, which would in essance be adjusting the gain of the sensor.

Grumpy

Only because so many *experts* have posted *opinions* as opposed to Obert, and Heywood, who did the research, designed tests to prove their ideas, and documented what they did in a scientific manner.

Read Obert's book the Internal Combustion Engine, and Air Pollution, and you'll read what the truth is.
Another source of accurate data is the NACA works.

Since the days of the original research a great many authors in order to write new text books ignored true data, and replaced it with their opinions.

And if you want to read up on the actual combustion reaction, Glass is the author to read.

HaulnA$$

The circuit posted does not change the total resistance seen by the ECM. The formula for total resistance (Rt) in a parallel circuit is Rt=1/[(R1+R2)/(R1*R2)]. Using the 3.9k circuit as an example where R1=the 3.9k of the sensor plus the 975 ohms from the added resistor then you get 1/[4875+19500)/(4875*19500)] which equals 3900 which is the exact resistance of the original sensor. I have used this circuit several times to tame down the gain of the sensor in various levels and it works flawlessly without a code 43.

JP86SS

iTrader: (1)

L829C: FCB 180 ; 95c, Min cool for for ERR 43B "TEST"
Setting this to maximum will not allow the forced knock test to run.
I do that just because I hate it.
Experimenting with resistors in parallel to reduce the sensitivity has been known to work. I was informed by a tuner that frequents a local dyno shop that the newer PCMs have settings to allow this adjustment. The $8D doesn't have anything except the error thresholds to adjust.
Trial and error (slowly incrementing) should show results either worse or better and allow you to get the sensitivity in the range where knock is actually occurring. Changing the error thresholds to determine if the sensor has gone out may need to be tweaked once the final value is obtained.
Might not take all that much.
The guy told me that when installing headers on cars he has done, he always has to desisitize around 10-15% to be effective.
He was certified and extremely sharp on details so he didn't hit me as a run of the mill chip burner at all.
Just my 0.02

vernw

iTrader: (2)

Pls excuse my ignorance on this, but which way would desensitize the knock sensor say, 10% ? Increasing the resistance to lower the voltage seen by the ECM?

And why would it have to take place at the ECM? Seems like being nearer to the KS would be better (plus you don't have the extra wire run to have to make and protect)

What am I missing? There has to be something....

HaulnA$$

If you look at the circuit, the de-sensitizing is determined by R1. In the 3.9k example, R1 is 975 Ohms or 25% of 3.9k hence the 25% de-sensitization. R2 is only in the circuit to change the resistance "seen" by the ECM. If you only wanted 10% de-sensitization, R1 would be 390 Ohms and R2 would be 42.9k Ohms. Determining any level of de-sensitization then solving for R2 in the formula posted will get you any level you want.

The circuit does not have to be added at the ECM. The resistors can be placed anywhere in the circuit as long as the proper connections are made. Doing it at the ECM just makes it easier to break the circuit than crawling under the car or doing it under the hood. HTH

drive it

iTrader: (1)

I just went looking for resistors at radio shack today and saw from 1/4 to 10 watts. What did you use that worked? Does it matter? Thanks!

HaulnA$$

1/4 Watt is plenty. No need to go with a higher wattage than that. HTH

rgarcia63

The circuit and formula HTH posted are correct as well as the correction for amount of sensitivity.

"Lower the voltage"
Care must be taken not use this type of circuit for changing voltage as parallel resistance circuits are current dividers, not voltage dividers, to change voltage a series resistance circuit is needed. Although the KS is in series with R1 they are seen as one resistor in parallel with R2 which decreases the current seen at the ECM.

If by "run extra wire" you mean from KS to ECM there's no need for that, as HTH said, "break the circuit" (cut KS wire) at the ECM. if amount of densitizing is not known then this would make it easy to use resistance pots say R1=3k, and R2=45K for a range of 10-75% and then replace with resistors of the desired values.

HaulnA$$

First of all, HTH is an acronym for "Hope this helps". My screen name is HaulnA$$.

Second. You're missing the most important point. The total current draw from the ECM never changes as long as the original resistance is maintained by the circuit, 3.9k in this case, and it is. According to Ohms law, it cannot. The current through the sensor actually goes down as the sensitivity goes down and the value of series resistor R1 goes up. The loss of sensor current is taken up by the parallel R2 to maintain the original resistance. It would seem that "vernw" is confused enough without having to deal with incorrect information.

vernw

iTrader: (2)

LOL

Yep, that's me!!

Seriously though, I do understand what you're saying and the logic and mathematics, after all, I'm a civil engineer with a math minor by degree. I just don't do that kind of work any more.

I appreciate the info. Once I get the new heads and MR and cam ugrade finished in the next few weeks I may be trying to desensitize mine a little bit (10% to start testing).

Thanks, Guys!!!!

HaulnA$$

No flame intended by any means. That series parallel resistor stuff can be confusing just like some of the other DIY PROM stuff. Now I'm jealous of your new MiniRam.

rgarcia63

First, I apologies for assuming HTH were you're initials.
Second, I welcome criticism, and corrections, this reply is to help me find my mistakes by showing how I arrived at my conclusions, unfortunately I seem unable to see the forest for the trees.
If the resistance (R) doesn't change then, for any given voltage (E) produced by the KS, the voltage will be same across any parallel resistance in the circuit, and will produce a corresponding proportional change in the current (I) draw at each parallel node.
For calculating purposes in the attached table I used 12 vdc.

The node between (KS+R1) and R2 is where the ECM is picking up the knock signal.
The current drop across R2 is not seen by the ECM.
- R.J.G.

JP86SS

iTrader: (1)

Since we know the Supply voltage 5V (yes?), We know resistance of the KS (3.9K in this instance), and we know the resultant voltage drop range (error ranges above), that gives us the current draw.
It would make sense to use that info in sizing the resistors for the current value (or range) that the circuit used as a basline for changes.

I never put a meter on the KS connector to see what the supply actually is. I'm assuming it to be a 5V but could possibly be anything up to 12.
I have a difficult time beliving the KS is passing 3Amps. If so then I could see there being a shunt in the Memcal for the voltage drop measurement.
Dunno, but interested to find out.

vernw

iTrader: (2)

Hey, no flame was received or perceived either, so don't worry about it. Appreciate the concern, though!

Yeah, I'm starting to get anxious about it and the heads. I'm solid on the 195 Eliminators and MiniRam choices. Wish I could see as clearly on the cam bit (230/237 HR or ???/??? SR....)

vernw

iTrader: (2)

Uh, not to be an @sswipe, John, My Friend, but look at the units in the attachment again. I think it says mA, so the current would be .003 amps unless this Old Phat Phart is missing something here.... (which is very possible )

Another question - how did you not lose your EGR with the HSR? Not have one on your 86 to start with, or some creative plumbing?

JP86SS

iTrader: (1)

needed that one!
That's why I specialize in hydraulics.
Just piddle with the electrons when needed (with some help obviously)
Dumped the EGR with HSR even though I drilled the heads.

vernw

iTrader: (2)

Yeah, I'm finally going to have emissions legal drilled heads and I'm putting them on with an intake that doesn't use EGR.

rgarcia63

A Knock Sensor is a self-generating piezoelectric device requiring no power to the sensor. Some late model Knock Sensor designs do use a supply voltage of 5V, but I don't think those will work in 3rdgens.
I had not done any research on Knock Sensors for the calculations, I just used 12v for no particular reason.

I did find a Pico PC O-scope Knock Sensor waveform.
Excerpt:
The frequency of knocking (pinking) is approximately 15 KHz. As the response of the sensor is very fast an appropriate time scale must be set, in the case of the example waveform 0 - 500 ms and a 0 - 5 volt scale. The best way to test a knock sensor is to remove the knock sensor from the engine and to tap it with a small spanner, the resultant waveform should be similar to the example shown.
http://www.picotech.com/auto/wavefor...ck_sensor.html