ngk afx not matching HP tuner read out

[MNR-0]

To scale mine, I warm up the car and turn it off, then read the AFR (it will be steady) and modify the voltage divisor to match. Only needed to do it once - then it was all good from there for every other car I tuned

[S2H]

you dont scale it....
you do the math and do it correctly

again...read my post on the math and follow..
you want 14.7 as stoich?

Volts / .7086 + 9.114

[04blackgmc]

make a custom EIO(AUX) PID
Volts / .7495 + 9.102

then your HPT will Show Lambda 1.0 as exactly 14.681...exactly what is specified in your tune.

ok so u suggest trying this? to get the car to show correctly on the scanner?

[DSteck]

I know damn well how a wideband works, and know that it is based on giving a lambda value in terms of a voltage output, which people can convert to an AFR based on whatever they want to define as stoich. You are missing the point.

9.00 (with 14.57 as stoich) = 0.6177 lambda
16.00 (with 14.57 as stoich) = 1.0981 lambda

The NGK AFX is defined as having a range of 9.00 AFR to 16.00 AFR, but is also listed as 0.62 lambda to 1.10 lambda. Those two don't completely match each other since 14.57 is defined as stoich in the controller, and the question is what are the actual limits that the controller operates between. Obviously the (Vout * 1.4 + 9) isn't completely valid, and I think people deserve to have the right information so they can have the right transfer function. There are two possibilities that I see... either it operates between 0.62/1.10 or it operates between 0.6177/1.0981, and people can set it up to make stoich whatever they want (although lambda is more important). Ultimately, I just want to know exactly where 0.85 lambda falls on the voltage output. I don't give a damn about stoich or the gauge.

Please don't talk to me like I'm stupid. I'm a mechanical engineer that has actually done research studies in the automotive field pertaining to engine operation through Ford and U of M - Dearborn, and worked as a powertrain engineer for Chrysler on top of all of this as a personal hobby. We're all here for the same goal and to help each other, so I think you need to adjust your attitude and lose the superiority complex.

[MNR-0]

you dont scale it....
you do the math and do it correctly

again...read my post on the math and follow..
you want 14.7 as stoich?

Volts / .7086 + 9.114

Hey Soundengineer,

I completely agree with you - mathematically it is sound. That is, of course, presuming the printed material performs exactly as stated

Now in the real world I observed that the max. voltage my AFX will produce when connected to the EIO is 4.77v. Therefore it aint a true 0-5v range we are dealing with here. Yet if I use your math for that value the scanned reading is even more way out.

Now it does produce a true 0.01v (metered at the EIO and scanned) so a voltage offset IMO is not needed. It's the slew rate of the voltage divider that needs work

There is also a slight skew only at higher AFRs (no problem for me as its only lean cruise) but since I want stoich to meter correctly (as my dial-in point) I just dialled in my voltage divider to match the gauge at stoich (engine warm, not running). Remember that each car has its own characteristic earth sinks

Now the AFX is accruate to within what? +/-1%. So if we get it within +/-0.15 we are within tolerance

My reasoning here is that the voltage will be linear between stoich and richest AFR, and that's the range I am most concerned with for accuracy

I would trust a steady state reading from the gauge over and above my EIO inputs as I have more faith in the ECM guys to get it right than us. Metering at a steady state negates the fact that it is a reasonably slowly responding controller

Now my AFM1000B model follows your math EXACTLY as you state. But that's an order of magnitude costlier and has much higher quality electrical engineering inside than the AFX

The AFX is what it is - a cheap entry level unit - don't expect precision out of any unit at this price point

I also feel your frsutrations about people making a song and dance about "it must be balls-on accurate or there is somethign wrong". You just won't get that at this price point.

ALso, when you turn your AFX on, it undergoes a warmup procedure and it outputs various voltages allowing you to test exactly this. It outputs 0v, lambda and max. volts. Meter it during warm-up and you will see what I mean. Lambda is not 4.09v as your (quite correct) math suggests

FWIW...

My voltage divider came out to be 0.701 on my AFX. My AFM1000B came out to manufacturers specs using the same method for diallign in I used for the AFX

[superwagon]

ok so u suggest trying this? to get the car to show correctly on the scanner?

Soundengineer had a typo in that formula. For 14.681 afr it should be (volts / .7095)+9.102 This is what I will be trying!

[04blackgmc]

Soundengineer had a typo in that formula. For 14.681 afr it should be (volts / .7095)+9.102 This is what I will be trying!

yea i tried his and was 14.3..

[JimMueller]

Not knowing much about engineering... if the widebands were able to output lambda to HPT, would that get around the voltage offset/slope issues? Would that make tuning more difficult?

[DSteck]

Not knowing much about engineering... if the widebands were able to output lambda to HPT, would that get around the voltage offset/slope issues? Would that make tuning more difficult?

They do put out lambda... in the form of a voltage signal. That's why we use a transfer function to read the voltage and make sense of it.

[superwagon]

I used sound engineers formula and tried it last night. I found that my actual afr reading on hpt was very close to the same that the ngk controller showed. I went and logged for about a half hour to check my ve afr err, when I went to view afr err I found that histogramm said not supported? When I logged with the default ngk wideband it registered ve afr err but not with custom set up wideband setting. I can view actual afr from wideband and delivered but not the error %.

[DSteck]

I used sound engineers formula and tried it last night. I found that my actual afr reading on hpt was very close to the same that the ngk controller showed. I went and logged for about a half hour to check my ve afr err, when I went to view afr err I found that histogramm said not supported? When I logged with the default ngk wideband it registered ve afr err but not with custom set up wideband setting. I can view actual afr from wideband and delivered but not the error %.

Your AFR % Error PID probably doesn't reference your new NGK AFX PID.

[superwagon]

So does that mean I need to add the afr err pid to the table when logging or is there a setting I need to change in my ngk afx pid?

[DSteck]

If you aren't logging the error PID, then the histogram won't work. I don't know how you have your scanner set up, so I can't really comment on what exactly you need to do.

[JimMueller]

They do put out lambda... in the form of a voltage signal. That's why we use a transfer function to read the voltage and make sense of it.

What I meant was a method to output the lambda reading without having to use a transfer function.

[Frost]

What I meant was a method to output the lambda reading without having to use a transfer function.

log raw voltage I guess!

[DSteck]

What I meant was a method to output the lambda reading without having to use a transfer function.

As mentioned by Frost, you can look at raw voltage... but you'd have to know that 3.9V was stoich. The only other way would be to program the wideband to output a lambda value equal to a voltage value... ie: 1 lambda = 1V out. A 5V out signal gives more resolution, which is why it's done the way it is done.

[schnellTTZ06]

I had a similar problem with my AEM, but I found a simple way to make it pretty much dead on from what I can tell. On the Aem, I dont know if the NGK does something similar but if you disconnect the O2 sensor it will read 14.8 on the guage. Then I just set the scanner to read the raw voltage, then enter that voltage in the list section. As long as you know the lowest reading your wideband puts out you can just enter that in and the same for the upper limit. The scanner will then automatically populate the correct afr correlation.

[muayyadf1]

Should we use 14.57 as commanded stoich AFR?

[S2H]

heres a spreadsheet I made to help make it a little easier to do the math

plug in your info into the yellow boxes...it spits out results
Math based...
put the wideband's stoich value in the "stoich" box and it spits out what the voltage should be based on manufacturers specs

this also assumes that their math is correct... just because a device doesnt output a full 5 volts doesnt mean that their transfer function math is wrong...just means that it will never spit out 5 volts..disadvantages of budget parts

*File updated with a second page that converts from known Lambda range to an AFR equation for HPT*

[mr.prick]

Should we use 14.57 as commanded stoich AFR?

This would be the easiest way, if the AFR expression of (V*1.4)+9 was right.
With the Lambda expression of (V*0.096)+0.62 & AFR expression of (V*1.4)+9:
(3.958*0.096)+0.62=0.999968 Lambda
(3.958*1.4)+9=14.5412 AFR
(3.958*1.3987)+9.033=14.5690546 AFR

(V*1.3987)+9.033 is closer to Lambda 1.00=14.57 AFR.
The AFR expression of (V*0.096)+0.62 is no good unless your Stoich AFR is 14.5412
Ignore the AFR expression the WBO2 manufacturer gives because it is rounded and
doesn't match to the Lambda expression.
Use the Lambda range with your Stoich AFR to make a new AFR range and
create a custom PID from that if you want AFR to match the Commanded Stoich AFR set in the the PCM.