NGK AFX setup ?

[orange ss]

Using the default HPT config for NGK AFX sensor, the scanner is showing 15.6 AFR in free air. I logged Raw Voltage and am showing 4.75 volts in free air. I set up the User Defined wideband config using V/.678 + 9 and now AFR in Scanner is showing 16 AFR in free air. Does anyone know what formula is being used in the default HPT config for this setup or should I not worry about it and use the default config anyway. TIA

[umrjmac]

if you arent seeing a max of 5V from the analog output on the AFX box, then you should create your own transfer function.

AFR = V*[(AFRmax-AFRmin)/(Vmax-Vmin)] + AFRmin

You can log Vmax and Vmin in HPTuners as the AFX box warms up...it should sweep low and high before it gets down to actually showing the sensor's value.

For me, that wound up being AFR = V/0.689 + 9

Now trying to get the stoich value on the AFX box to match what the narrowband O2 sensors see as stoich is another story entirely, but I am still sorting through that one...I think that I need to just adjust the offset so that it matches, but I'll mess with that a bit later this week when I have some time.

[DSteck]

Ignore the box's maximum voltage output. It'll never output true 5V, and anybody expecting it to is kind of on drugs. Also, it will never output a true 0V either... that's just the nature of 0-5V sensors.

Use this, and never look back:

Divisor: .7096
Adder: 9.102

This makes it report correctly for a stoich value of 14.68, which is what most PCMs use. Make sure you have a GOOD power and ground... (ie: straight off the battery), and just leave it be. You will beat yourself up nickel-and-diming the values to try and get them exact. Ignore the display box.

[umrjmac]

Ignore the display box.

Best advice. The box will show 14.57 for stoich....regardless of what the actual stoich AFR is for your fuel type (since its really reading lambda and translating that into AFR). My AFX box is in the glovebox and it stays there.

[orange ss]

Ignore the box's maximum voltage output. It'll never output true 5V, and anybody expecting it to is kind of on drugs. Also, it will never output a true 0V either... that's just the nature of 0-5V sensors.

Use this, and never look back:

Divisor: .7096
Adder: 9.102

This makes it report correctly for a stoich value of 14.68, which is what most PCMs use. Make sure you have a GOOD power and ground... (ie: straight off the battery), and just leave it be. You will beat yourself up nickel-and-diming the values to try and get them exact. Ignore the display box.

I have good power coming thru a dedicated circuit from the fuse box, my ground is the same ground as the ECM ground, so I believe that is all good. I'll try to log the min/max when I get home and see what happens.


Thanx for the help everyone

[S2H]

log it as it warms up...
you will see that it doesnt quite reach 0v or 5 volts...
doesnt mean that the transfer function is incorrect...just means that it cant do a dead short or 5v to the outputs..

it should center at 14.57 AFR (3.98v) or Lambda 1.0 when you set it up with the default settings..

I tune in lambda so I dont care that they use 14.57 as the stoich value...lambda 1.0 is always lambda 1.0 regardless of fuel type...

volts/ 10.40714 + 0.618
I went into the scanner config to program mine since the scanner will only allow 4 total digits with a decimal...

[DSteck]

I'll try to log the min/max when I get home and see what happens.

Don't bother.

[orange ss]

duh....I get it now, I'm not gonna need the extreme ends of the spectrum anyway....thanx fellers

[S2H]

Ignore the box's maximum voltage output. It'll never output true 5V, and anybody expecting it to is kind of on drugs. Also, it will never output a true 0V either... that's just the nature of 0-5V sensors.

Use this, and never look back:

Divisor: .7096
Adder: 9.102

This makes it report correctly for a stoich value of 14.68, which is what most PCMs use. Make sure you have a GOOD power and ground... (ie: straight off the battery), and just leave it be. You will beat yourself up nickel-and-diming the values to try and get them exact. Ignore the display box.

actually...
Volts/ .7143 + 9.11 is the proper equation for 14.68 stoich
transfer functiuon stays the same...all you do is shift the AFR over

things to remember... the manufacturer programs the multiplier into the device...it doesnt change just because an output cant reach the extremes of 0volts and 5 volts..
so if the manufacturer says the multiplier is 1.4 then its always 1.4 (or divisor of 1/1.14 = 0.7143)

[umrjmac]

Ahh great info, that changes my understanding of it...I'll go try this out on my car tonight.

I thought that a change in the min/max changed the scaling instead of just truncating the extreme values.

[S2H]

changing the minimum and max does change the scaling...which is why you dont wantto formulate your own values for the wideband...
use the manufacturers specs and shift the AFR to match

[DSteck]

actually...
Volts/ .7143 + 9.11 is the proper equation for 14.68 stoich
transfer functiuon stays the same...all you do is shift the AFR over

things to remember... the manufacturer programs the multiplier into the device...it doesnt change just because an output cant reach the extremes of 0volts and 5 volts..
so if the manufacturer says the multiplier is 1.4 then its always 1.4 (or divisor of 1/1.14 = 0.7143)

Using NGK's range of 0.62 lambda to 1.10 lambda and stoich of 14.68...

0.62 lambda = 9.102 AFR
1.1 lambda = 16.148 AFR

Based on that, I came up with that transfer function. The denominator is:

5/(16.148-9.102)

The offset is just AFR at zero volts... aka 9.102.

[S2H]

Using NGK's range of 0.62 lambda to 1.10 lambda and stoich of 14.68...

0.62 lambda = 9.102 AFR
1.1 lambda = 16.148 AFR

Based on that, I came up with that transfer function. The denominator is:

5/(16.148-9.102)

The offset is just AFR at zero volts... aka 9.102.

transfer function is hardware set and programmed by the manufacturer...
per their user manual... multiplier is 1.4
you cant come up with your own values and expect it to be accurate

AFR is 9~16
Lambda is 0.62~1.10

remember..its based on 14.57...not 14.68

technically
0.62= 9.0334
1.1= 16.027

5/6.9936=0.7149 = Multiplier of 1.3929516645


Manual says Multiplier of 1.4
and I'm sure those #'s with AFR's and Lambda have all been rounded for print in the user manual

[DSteck]

transfer function is hardware set and programmed by the manufacturer...
per their user manual... multiplier is 1.4
you cant come up with your own values and expect it to be accurate

AFR is 9~16
Lambda is 0.62~1.10

remember..its based on 14.57...not 14.68

technically
0.62= 9.0334
1.1= 16.027

5/6.9936=0.7149 = Multiplier of 1.3929516645


Manual says Multiplier of 1.4
and I'm sure those #'s with AFR's and Lambda have all been rounded for print in the user manual

You're right, I bet they have been rounded, which is why I had e-mailed NGK for more precise measurements of lambda at 0V and 5V, but never got a response. What it comes down to is whether you want to use lambda, or want to use AFR as your basis for setting up an HP Tuners function. I chose lambda. The transfer function they give you is for a stoich value of 14.57, which is not what most people use.

[S2H]

the manual gives you accurate values for AFR with a multiplier of 1.4
if you want the values for lambda you just have to work it backwards..
which results in an equation of
volts/10.4071 +0.618

[DSteck]

the manual gives you accurate values for AFR with a multiplier of 1.4
if you want the values for lambda you just have to work it backwards..
which results in an equation of
volts/10.4071 +0.618

The question is what has more rounding error... the 9.00-16.00 AFR range, or the 0.62-1.1 lambda rage.

[S2H]

The question is what has more rounding error... the 9.00-16.00 AFR range, or the 0.62-1.1 lambda rage.

the smaller values would have more error to them in general...

would you rather have a step of 0.1 in a scale of 1-10 or a scale of 1-100
the difference between 0.62 and 0.63 is quite significant compared to 9.00 and 9.01

[S2H]

and not only that... using their 1.4 multiplier in the transfer function lines right up with 9.00-16.00... henceforth accuracy based on 9.00-16.00

if they wanted you to know that the lambda values were exact and not rounded then they would have supplied a multiplier based on lambda and an AFR multiplier on top of that...which would not lend to nice math for most plug and play users

[DSteck]

and not only that... using their 1.4 multiplier in the transfer function lines right up with 9.00-16.00... henceforth accuracy based on 9.00-16.00

if they wanted you to know that the lambda values were exact and not rounded then they would have supplied a multiplier based on lambda and an AFR multiplier on top of that...which would not lend to nice math for most plug and play users

They don't figure most users care about the level of detail, that's all there is to it. 9.00-16.00 is what they show probably because it looks better than a blanket 9-16.

[carlrx7]

maybe im overthinking this, but i wana try something like this:
wb.aux=(vOut/0.7143)+9


((wb.aux)/14.57)*14.2
where 14.2 is my E10 Stiotch

would that have my AFX to display 14.2 as stioch in HPtuners? and match my commanded AFR PID?

ive been using lamda and its been working well, but id like to see some more detail, its just a big jump from .78 - .80 lamda.. it seems with the bigger numbers id beable to get a more detail