pgmfi.org

Here are some details on how i connected up my Innovate LM-1 to a USDM P28 and read the full 0-5v range using the O2 input.

I initally had some problems with the LM-1 output spiking, along with wanting to be able to read up to 5v so i combined it all into one. The reason i modified the board in the way i did was because i had a flying plug in my OBD0-1 conversion harness that i plug my wideband unit into.

http://www.marklamond.co.uk/tech-honda/pgm-fi/o2-input/o2-input.htm

I've no doubt that this can be applied to the rest of the OBD-1 series of ecu's (and probably OBD-0 also) - its technically not the best way to process a signal but it does work well in reality. You must include the switch as if you leave the modified input in circuit with a stock O2 sensor the car will still run closed loop, but the ECU would appear to track the cycling of the O2 sensor voltage in closed loop very closely (kind of expected that) and occasionally during certain periods of cruising you can get an O2 CEL. You could drop the value of the cap down to 6uf or thereabouts but this allows some mild fluctuation to be seen. But as far as im concerned you may as well fit a switch to the ecu and switch back to the original input when you have finished your wideband session as you know you're not going to get any problems.

This is a good route to use until spare ADC channels are made available, of which i have got a lot drawn out - however i can't really do anything when it comes to the code side of things.

Edit - URL updated.

Interesting write-up. But in the end, I don't think you actually gain anything from all of that work.

Why? Because the stock circuitry actually amplifies the stock O2 sensor voltage. That's why you max out the A/D reading when you put more than 3.8V on the input. So what? Well, that means if you set up your meter to output 0-3.8V for 10.0 to 20.0 AFR, and used an unmodified ECU, your actual resolution (after the A/D conversion) is going to be the same as if you had gone to all the trouble of bypassing the the op amp and inputing 0-5V for 10.0 to 20.0 AFR.

Even if you did gain resolution, how much did you gain? Let's say the Innovate uses a 10-bit DAC to produce a 0-5V output. If 0-5V corresponds to 10.0 to 20.0 AFR, that's a resolution of better than .01 AFR per count (assuming no noise, which is not true). If we instead use 0-3.8V for 10.0 to 20.0 AFR, that's a resolution of about 0.013 AFR per count. That's 30% worse, but at that level of resolution, what's the difference? In any case, your experimental data seems to indicate that any real signal at that level that might be coming from the Innovate is totally swamped by noise that is one to two orders of magnitude greater than this resolution limit.

Finally, and you didn't mention this, there's the issue of having to modify the ECU program to understand the WBO2 sensor signal, as it is totally different from a stock sensor and has a completely different interpretation than the signal the stock program expects. I'm not saying it can't be done - I've done it myself - just pointing out that the necessary accompanying firmware mods need to be made.

Thanks,
91crxsi

Okay, so its actually fairly pointless on paper when you look at the accuracy you require in terms of AFR for tuning purposes (being accurate to .1 in terms of AFR would probably be enough, and how accurate is the WB itself anyway etc) and yes i take your point about the noise as well, essentially defeats the purpose of modifying anything in the first place - im going to have a look again at where that was coming from as it doesent add up. I just thought it would be nice to figure out a way round this "limitation" for not just the purposes of logging AFR, and also the people who have a fixed 0-5v output range on their WB.

Following what you say about the amp effectivley rescaling the O2 input so 3.8v in = 5v to the ADC, if i max out the O2 input i see 3.8v datalogging, but if i disconnect the input to the mux and pull it to 5v, my datalogged value reads 5v which would prove the ADC isnt maxed out. If the logic is true it should read 3.8v due to what the code is doing when the actual ADC input is 5v.

I'm sure when i pulled the O2 input to 5v i was only getting 3.8v at the mux, and the datalogged value was also 3.8v, which is why i made the change originally. It was almost a year ago i did this so im going to set it all up again and check tonight.

Regarding the code though all i was trying to say was the ecu isnt very happy if you connect up the stock O2 sensor even with a little smoothing, and try to run closed loop, im not suggesting you try to run closed loop with a 0-5v WB input. With the wideband set up to 0-5v and closed loop disabled i didnt have a problem in logging AFR's.

Hi,

It's been a long time for me, too. I thought it was amplifying everything so that 3.8V at the ECU pin would result in 5V at the A/D input. Either that or they're doing a rescaling in the firmware. Maybe I'll have to check it out, too. If it's the latter, then all you would really need to do is change the scaling coefficient.

As for using the O2 sensor input with a WBO2 with closed loop disabled, then I get why you did what you did. That makes perfect sense to me now.

Finally, do you think that the noise in the Innovate could be due to real fluctuations in AFR or a defective meter / bad design? Anyway, good luck with figuring it out.

91crxsi

I think almost everyone sees those fluctuations... I even see fluctuations in my PLX WB and the AEM unit I had previously.

It will be interesting for me to get the chance to try this and see how it works.

i did this modification to see how well it would work. i do not believe it is the firmware that limits it to 3.8 as you can use that circuit and it will just max out at 3.8v. once you apply the 5 volts around the circuit it will read 5 volts. all the software is doing is reading the voltage. what i did was used methods black box and the output of that to the input for the O2. i get a max voltage of 4.98, even though my aem will go up to around 7 volts when in cut-off. instead of the resistors stated by methods to use, i used the 2.2k and a 100k. i have a difference of -.2 volts between the input and output of this circuit. i have not tried a larger resistor yet, but i do plan to when i have time. also, i noticed that using the filter cap helped stabalize the signal, compared to what i used to have.

http://www.hondata.net/forum/viewtopic. ... highlight=
http://www.angelfire.com/electronic2/methods/

The only limiting factor is the input stage to the ECU's A/D, which is what i am bypassing with this mod - 5v applied directly to the mux that feeds the A/D will read 5v / 255 via your datalogging software (or very close to it).

I've also looked at this setup on a JDM P30 and its exactly the same, just a bit more fliddly to change because its all surface mount lol.

Simple pictorial of how to do this mod (non switched)

First, remove R5 resistor

Below C70 are 3 thru-holes on the ground plane. Use a soldering iron to heat the middle one, so you can suck the right hole clean. This will be your grounding point for the 10uf cap.

Insert the negative lead thru the grounding point, and the positive lead through the right hole of R5. Underneath, bend the lead of the cap in the R5 thruhole up and to the right, and tack it to the C3 pad. Solder both thru-holes for the Cap. cut excess leads off.

Edit: Tested and confirmed. Voltage read at WB output matches the value I read in FreeLog (With O2 input set to Volts). I can now read from 10 AFR to 20 AFR with the 5v input!!

You should post this in WikiMe. Which ecus does this mod apply to?

I looked at a P06, P28 and P61 (the P61 I had being a 93 GSR California version). They all have the same configuration. I believe it applies to most OBD1 versions. (except JDM small-box style or some OBD1-II oddballs)

wiki'd

Nice!!! This is a great mod...

This is a awsome mod, this gives me a lot more resolution for datalogging my wideband I have done the mod and now have 5v of input signal my signal in Crome is more stable too I am using the LM-1. Thanks for the contribution.

Can someone add this to the wiki?

http://www.pgmfi.org/twiki/bin/view/Lib ... rdwareMods

Finally got off my ass and Wiki'd it!