Goblin

Wow. This thread got REAL interesting real fast.

I've just got my car running again. I'll try duplicating Rusty's test tomorrow if I can manage to get a couple hours. I wonder if we're all 0.2 volts off from the supplied graph.

13brv3

Any progress?

I should have my new oil pressure sender today, and as long as that magically gives me oil pressure, I'll be testing my new polynomial this weekend. Of course if I still don't have oil pressure, I'll spend the weekend drinking heavily to drown my sorrows

Cheers,

DavidDeco

I'm without my FD until late this afternoon.

I wasn't able to do it Christmas Eve or Christmas. I was tuning back from my office tue night but didn't want to change my polynomial because I was comparing runs.

May get a chance to check it tonight or when I get back from my weekend trip. I was hoping Fred from FJO would have replied to my email or this thread which I sent him.

13brv3

Sorry, but my first polynomial is waaaaaay wrong. I was getting .75 when the FJO was reading 18. It didn't take too long to realize that I had the axis reversed. The new file will be attached, and I have not tried it yet.

What I did try, was putting in two points from the data I collected the other day. I put in 1.16 V at 11.4 AFR, and 1.41 V at 12.0 AFR. This isn't much of a range, but the curve is really very flat from 10 to 14.5 anyway.

I'm happy to report that putting in these two numbers gave me an almost perfect result for what I could test. With the camera rolling, I was able to get the FJO to stabilize at 11.5 and 12.0. At those points, the Datalogit read 11.49, and 11.96. I guess I could probably live with this but I would like to do some more testing to see if it's still good from 10 to 14. I suspect it will be.

This just may be a winner. Try it and let me know if it works for you too.

PS- the oil pressure sender looked very good inside (I cut it open), but thankfully it wasn't good. The new sender gives me 60 psi at 3000 rpm, and 30 psi at idle. I feel better now

Cheers,

13brv3

Well, the new poly didn't work either, so don't waste your time. I just don't know what's wrong with it. If someone could look at the Excel data, and figure out what I'm doing wrong, I'd appreciate it.

In the mean time, I went back to my two points, and fought the "interface not connected" battle for a while. Once I got it working enough to actually drive around, I was able to compare the FJO to the datalogit again. What I noticed is that sometimes it's perfect around 12 AFR, and sometimes it's a couple tenths off. I also noticed that it's a good bit off until the car gets up to normal temp.

I just don't know what to think about this variability. If it stays within a few tenths of correct, that's probably good enough for my purposes, but I'm starting to wonder what correct really is. I'm afraid I'm going to have to complicate matters some more, and install a tie-breaker. I have a new Tech Edge that I can install in the stock bung, to get another opinion. I'm not sure when I'll get to try that. I shredded a PS/AC belt today during one of my logged runs, and I will hopefully have the new belts tomorrow. It's pretty sad when you can't even buy a damn belt at the local auto parts store

Cheers,

Goblin

Blarg... I've been unable to get time to check things out on my setup. Hopefully tomorrow (if it doesn't storm as bad as the weather forecast is making out), or the day after (supposed to be bright, sunny, 60's... <g>).

However, I went looking around for more information. Seems we're not the only ones having problems:

http://www.aempower.com/bbs/viewtopic.php?t=28 <-- Anybody know what the "gain" on the AEM unit does? My limited understanding of dsp filters doesn't seem to jibe with the way they're using the term.

Also:
http://www.aempower.com/bbs/viewtopi...&highlight=fjo
http://www.aempower.com/bbs/viewtopi...&highlight=fjo (<-- Install for Supra with AEM... Might be useful, might not. Haven't read it yet.)

13brv3

Greetings,

The following is a quote from the log that I keep for the car. I'm also attaching a spreadsheet with all the FJO AFR vs Datalogit AN1 voltages. I never got a polynomial to work, but as you can see, I'm happy with what I have now.

Cheers,
Rusty



I tested the FJO and Datalogit AFR reading again, this time by using the 1000 rpm injector control to vary the fuel. With the air pump off, you can get from about 10.5 to 12.6 AFR, and with the air pump on, you can get 13 to 15.

The following numbers were recorded with the AFR using delta between AN1 and AN2, and the following two points in the calibration boxes: 11.4 AFR = 1.16 V and 12.0 AFR = 1.41 V.

FJO Datalogit AFR Datalogit AN1 (AN2 is always 0)
10.5 10.73 .88
10.7 10.92 .96
12.3 12.19 1.49
12.5 12.33 1.55
13.7 12.85 1.76
14.0 13.42 2.00

From this, I created yet another dysfunctional polynomial, and immediately pitched it. I then put in the following two numbers for the calibration: 10.7 AFR = .96 V and 12.5 AFR = 1.55 V. The following numbers are the result.

FJO Datalogit AFR Datalogit AN1 (AN2 is always 0)
10.8 10.70 .96
11.1 11.00 1.06
12.3 12.32 1.49
12.6 12.56 1.57
14.2 13.99 2.04
15.9 14.70 2.29

This is pretty usable, and seems to be within .1 AFR from at least 10.8 to 12.6, so this is where I plan to leave mine.

For the sake of keeping records, the following numbers are valid, actual data points that have been collected along the way.

FJO Datalogit AN1
10.5 0.88
10.7 0.96
10.8 0.96
11.1 1.06
11.4 1.16
11.5 1.22
11.8 1.31
11.9 1.39
12.0 1.41
12.3 1.49
12.5 1.55
12.6 1.57
13.7 1.76
14.0 2.00
14.2 2.04
15.9 2.29

13brv3

The formatting got a bit screwed up on the previous post, but hopefully, you can figure it out.

Cheers,

Goblin

Alright, well, I got tired of messing with the thing.

The different values we're all getting between the FJO and the PFC are disconcerting to me. Even 0.1 afr difference is too much inaccuracy for my taste. I mean, yeah, I'm only tuning to within -0.2 to 0.2 of my targets in general, but it's the principal that's bothering me. Assuming the FJO itself is solid and reliable, I want that solidness in the numbers I pull from it for the DataLogIt. If I hadn't been so concerned with that accuracy, I would have kept playing with my old OzDIY unit (More a learning experience in building electronics than anything else, for me... I messed it up pretty badly, I think... <g>).

So, with all the fiddling it occurred to me, again, that the FJO has serial output. There's no equations or anything that raw voltages need to go through, it's just pure, this is it, this is your number, textual output.

With that thought festering away in my mind, and a lot of other ideas about things I'd like to be able to do with the DataLogIt, I started a new project. If it moves forward well enough, I'll post details and share results (until then it's all just vapor, like a lot of things). I'm hopeful, though. It's progressing quite well.

Now if I can just ignore the insistent urge to get my car tuned properly long enough to get this done.

13brv3

So instead of having 2 readings that don't match, you'll have 3

Just kidding

I'm still deciding if I really want to know what the Tech Edge would say... probably not. As much as it irritates me to know this isn't exactly right, you have to draw the line somewhere. Just think of all the folks that are trying to tune without a WB, and here we are quibbling over .1 AFR

Cheers,

Goblin

Heh heh... Well, I'll have three readings that won't match, but one of them is the direct output of the FJO controller, and it won't have been passed through any potential weird grounding or impedance issues and then run through a formula. It's just a text representation straight from the horse's mouth, as it were. <g>

I'm thinking a couple other things could be inline with the project, too. Specifically auto tuning and dyno graph style output of afr numbers and estimated power.

If I hadn't had to get a copy of WindowsXP for the laptop in order to run the port monitor, I'd have had more information on the feasibility of this by now...

twokrx7

iTrader: (1)

Guys,

Keep up the good work. One question, how do you know which of the delta features to use if the logged voltages of AN2, AN3, and AN4 are different? I looked at a few logs of mine and find AN3 and AN4 to be almost identical but AN2 is different.

Kyle

13brv3

I don't quite understand what you're trying to do Kyle. What do you actually have hooked up?

The delta feature requires two inputs for one signal. In other words, a wideband "analog out signal" would connect to AN1, and the wideband "analog out ground" would connect to AN2 (not to ground). You would then check the box at the bottom that's marked Delta AN1-AN2, which is literally AN1 minus AN2. This should be a more accurate way to record the analog output, since tenths of volts can make a difference in the AFR reading. BTW- you can of course substitute AN3 and AN4 for AN1 and AN2 respectively in all of the above.

Did that help at all? I'm not sure I actually answered your question.

Cheers,

twokrx7

iTrader: (1)

I think that cleared it up a bit. I have been using the UEGO wideband which does not have a ground wire per se, so I only attach the 0-5 volt output of the UEGO to the Datalogit. I guess the system is grounding through the cig lighter? Not sure how I would use the Delta feature with the UEGO.

I have measured the voltage output of the UEGO and compared it to the Datalogit/FCEdit raw voltage reading and get about 0.05 or 0.06 of a volt difference. This results in the Datalogit/FCEdit reporting about 0.1 AFR leaner than the Wideband. I did not consiser this an issue because it's in the conservative direction, there is usually a greater spread than 0.1 AFR in each datalogged P/N cell anyway, and who knows if the WB is spot on to start with. Been tuning with it like this for awhile now.

Kyle

es

From what I remember on the DIYWB list the Datalogit box has an input impedance that is low enough to actually drag the sensor voltage down when used with the display.

Since the sensor is designed to produce a voltage to one device@ a fixed impedance vs: 2 in parrallel the two input impedances would create too much of a load...

The DIY fix was to remove a resistor (sorry that's all I remember).

I guess what I'm getting at is check the FJO display without the Datalogit connected. Then connect the Datalogit and see if the FJO reading drops. If it does then you know the problem is with over-loading the sensors output signal.

After all, the first step to correcting a problem is to deturmine where the cause is.
Eric.

13brv3

I do intend to test this when I get the chance, but I don't expect the display to change on the FJO. I do suspect that the Datalogit is pulling down the analog out voltage of the FJO, but if the FJO is as well designed as I think it is, this won't have an effect on the it's display.

The best test I can think of is to let the car idle at a fixed AFR. Note the FJO display, the datalogit delta V, and the actual delta V using a meter. Next, I'll disconnect the FJO from the datalogit, and see if the FJO display changes. I'll also measure the voltage on the analog output of the FJO. I'm guessing these will now be .2V higher than when they're connected to the datalogit.

In addition to the above test, I also have a call into Fred at FJO, who will likely be able to tell me exactly what's going on. I'm sure they must know if the datalogit box causes them a problem.

BTW, I may not get a chance to report on any of this until late next week.

Cheers,

13brv3

Great news, I got a call back from Fred at FJO, and things started getting interesting in a hurry.

First, their measurement circuit for the display is isolated from the analog voltage output, so even shorting the output will not change the display accuracy. That's one question down.

I was asking about the possibility that the datalogit was pulling down the analog voltage output, and he asked me to do a test. He wanted me to power up the FJO and datalogit, but not crank the car. The goal was to have a stable voltage on the analog output of the FJO. Next, I measured the analog output from the FJO while connected to the Datalogit. This was 4.19 V. Then I removed the wires from the Datalogit, and measured the analog output voltage again. This time it was 4.91 V. So, the Datalogit is pulling down the FJO, which is where my .2V went before.

Now here's the really good news. Fred took this reading, and produced a new AFR to V curve to compensate for the impedance of my Datalogit. I would assume they should all be the same, but if the correction doesn't seem to work for your unit, you should probably run the above test yourself. Between the two of us, we even managed to produce a working polynomial between 10 and 14.5 AFR. It is:

A= .4011
B= 2.0519
C= 8.4241

I tested it some this evening, and it seems to be spot on

Finally, I'd like mention how great it is to be able to call someone like Fred at FJO to work out problems like this. Though it wasn't any problem with their unit, he spent a good bit of his time, and their phone bill to help me get this working. Was the FJO worth the money? Yes indeed

Now, to try to figure out why I keep getting this "interface not responding" message...

Cheers,

Goblin

Cool. Nice action. Glad to see a resolution to the issue, if it proves to work for others. And very nice of FJO to spend time helping.

And thanks for sharing the data so quickly. Looks like I'll be able to do some tuning this weekend after all. <g>

As for my thing, I think I'm gonna keep moving forward with the protocol hacking. I've got some things I want to do that aren't doable with FCEdit, and new versions of the software aren't really coming fast and furious with new features. I kind of like the idea that if I want to do something (like, say, see all the values the PFC is using as standard conventional units, rather than weird Apex numbers) in the software, I can just do it, rather than havingto ask for the feature to be added, and then waiting for a consensus, and then waiting for a revision, etc, etc. And I'm sure those guys are busy enough building the boxes and ramping up for the PFC's on other cars that they don't need me whining to them about new features anyway.

13brv3

I'll be real interested to hear how it works out for you.

I'll be playing with mine some more today, mostly to try to figure out why I keep getting these interface not responding message intermittently.

Good luck. Hope it works out.
Rusty

DavidDeco

Fred is the man. He's the one who provided the calibration data for me with like 1 day turn around when I first bought the unit. It's nice to know this wasn't an isolated incidence. Great product support from FJO.

As far as that interface not responding....I'm getting it more and more. Especially now that it's cooler out.

Boostn7

WOW !!! thanks guys

Gonna plug in the new numbers and may go for a ride tonight

Thanks again to those involved.

Goblin

Curiously, that poly didn't work for me... Can you describe the process you guys went through to generate it? I'm guessing you just built a new voltage curve, but did you just drive around and sample the voltage while watching the FJO display to do it?


This is slightly off topic now, but, if anyone else is interested, I've cracked the PowerFC/DataLogIt packet format.

(WARNING: Programmer tech speak follows...)

I've got the command/response formats figured out. Most of the number formats figured out. And a lot of the commands figured out and mapped to data. There's maybe a few dozen request for information commands. The big 20x20 maps were easy to spot and decode (that's where I started). The rest, it's a matter of converting the data to readable numbers and then tracking down those numbers in FCEdit to see what settings that match up with. In addition to the maps, I've got about %20 of the settings exposed by FCEdit so far (I'm not sure if there's stuff that FCEdit hasn't exposed. For instance, the "Normal", "Option 1", "Option 2", etc, strings in the PIM data look like they're probably changeable (Maybe the boost control unit changes "Option 2" to say something different?), but those aren't exposed as being changeable by FCEdit.).

The rest should come pretty quickly. I think there's another number format I've yet to figure (probably a short word percent value...)

I've got a working document describing everything pretty well.

And I've got a gui app worked up to start playing with. Hopefully cross platform, but I've only got PC's to compile on (unless I get it in my head to do a port to XBox for some reason... <g>).

Maybe this belongs in its own thread now. I'm not sure. It may be too much a niche for anyone but me to be interested. <g>

DavidDeco

The poly seems to be working for me. I've spot checked it and at least at lower rpm & boost it's on pretty tight.

I'm going to get a co pilot sometime this week and try it out.

David

es

Goblin, please do start another thread about this so I can at least check for your updates, screen captures...

DavidDeco

On closer inspection, mine seems a bit off but slight. It seems on in the lower ranges but sometime in the 12's and higher (leaner) my datalog goes 0.2 or so HIGHER now than the FJO display. That's not too bad and probably safer for me anyways.