First Datalog

[clint78z]

Well I pulled my first datalog of the DFI on friday. It just gives you a txt file with values, but that can be imported to excel very easily. The resolution isn't on par with Tec II, and scaling the graphs is a bit of a pain. Luckily I am good with programming excel, which will make reading these datalogs much easier in furture.

 

Here's my log, as you can see it probably could stand to be running a tad richer, the dip is injector cutting of fuel when throttle is lifted.

The NPR intercooler kept temp below 50F, it was pretty chilly that day.

[Lockjaw]

All I see is a little red x.

[BLKMGK]

Try this -> http://www.geocities.com/clint78z/MISC/DATA1.bmp

 

Geocities won't allow linking like that.

 

Some of the BIGGEST beefs I had with my DFI was slow datalogging and no way to manipulate the data. Nothing like trying to graph O2 voltage that doesn't climb over 1volt against a 6KRPM engine rev! I got VERY frustrated doing this years ago It looks like you might have succeeded where I failed - terrific! Yellow is O2 voltage right? No way should system voltage go all over like that You might want to log TPS too - helps to be able to tell when you're WOT and when you're not. Battery voltage is another good one as is EGT, spark, and duty cycle on the injectors. Log as fast as it will allow, on my old DFI the data logs would show big jumps just because it logged too slowly. Grr!

 

One of the biggest improvements that's occured in the programmable EFI world has been the datalogging and data manipulation portion. The software I'm using now actually has a math library and programmable functions If I datalog temps before and after the intercooler, for instance, I can write a function to output intercooler effeciency. I intend to datalog my fuel pressure and compare it with the pressure and vac my MAP sensor sees to see if pressure is dropping where it shouldn't or find other problems. Lots of possibilities! I understand now why some racers spend so much money on data acquisition, it's worth it's weight in gold IMo.

[clint78z]

Dumb geocities I linked it before and worked, I even previewed after the post and it looked good. Yes indeed BLMGK your observations are very good. The right axis has yellow O2 voltage, and the other has RPM and MAP Pressure.

 

Yes there are some fustrating things mostly limitations that the book does not go into. This is an old system and is slower chip speed than the new ones. The main problem most ppl have is they try to datalog log too many things at a time, and the system can't handle it. Pick 3 or mabey 4 vital variables to log, that way the resolution on those variables is better (no missing gaps). The book should tell you that and it doesn't. You could even log windshield washer fluid temp BLK

 

Here's a hint for graphing quickly with Excel

1)bring in a log

2) Set up an intial graph, scale everything

3) When you do another log make it excatly same variables.

4) Paste the new log over the old one and it will update the graph for your new run.

5) If for example you did 3 or 4 chart setup initially, just paste you new logs in everything will update automatically.

 

It is a bit more farting around but will function every bit as good as the more expensive ones. I will probably make a program in Excel to give a playback in realtime, but that's down the road.

 

I am quite pleased having payed $350 for my DFI, and still be able to datalog. I like the logging feature, it is much safer than trying to drive and look at the screen.

[Tony240ZT]

I didn't realize you could put a different Y scale on the right as well, pretty cool. I figured it out and scaled my o2 meter output like you did yours with my data logging. This info was taken in from my custom data logger I built using a 25Mhz PIC Microprocessor and 3 8 bit resolution anolog to digital chips. Not sure on the exact frequency of logging, but I think it keeps up well with 9600 baud 3 digits for each.

 

I'm working on getting RPM data logged, and I could use a manifold presure sensor, just not sure where to look for one. Don't those 2nd gen rx7's use one? Maybe I could use one of them.

My air/fuel ratio is all over the place, this is due to the fact that my injectors are too large for the AFM. The car was running off of the AFM at the time the logging was done.

 

Here's a link:

http://www.zhome.com/~aktar/pics/maf-afm.gif

 

This data is only for a short period of time, from the looks of it enough to be on it twice. As more air passes the intake the MAF voltage goes up, and for the AFM the voltage drops as more passes.

I think to make the data more usefull I should put a capicitor/resitor in along with the AFM since it jumps around so much.

[clint78z]

I see datalogging has got your noodle smokin too. I find it very intresting, we can definately discuss any issues you are having and where you want to go with your project.

 

RPM is a definate must for the engine, it will help sort out fuel maps.

 

Map pressure isn't nessary if you already have MAF, however mabey you want this in the package since it is cheap and offers less restriction. It will be more calculation for you.

I suggest a GM map sensor available at any junkyard 0-5v 1bar. If you have turbo vehicle go to a jobber and order one for an 88 sunfire turbo this is a 2bar sensor.

 

I would have to say tuning on your graph looks OK, at high airflow O2 voltage hovers around .7-.8v. Then at low air flow O2 voltage drops which is good. Of course you need other info to determine what's really going on.

 

What type of I/O card are you using for logging, how is the software ??

 

Questions fire away !!

[Tony240ZT]

This logger I built is pretty custom. I'm using a PIC Micro 16 bit processor that runs at around 25Mhz, along with a couple 8 bit anolog to digital chips. And then I just have all of the data spit out at 9600 baud onto a terminal program, log it, save and then import into excel.

 

It can even drive a digital to anolog chip to generate any voltage level I'd like to trick out the stock ecu. This is the whole idea of this little piggy back computer system I'm making to convert AFM cars to run on MAF. I'm slow at getting things done, but I'll get there.

 

I'm wondering how much of the cpu power I should put to each sensor. I wonder if I should queue every sensor I have on every cycle, or maybe just some more than others.

 

I'll post more data as I get it. I think I know how to convert the nissan ecu pulses for rpm into a digital number with my chip, just wondering where to take it from. I guess right at the wire that plugs into the coil trigger.

 

If I hooked up fuel injector drivers I could probably directly control fuel. The processor doesn't have interupts, so it may be a waste of my time trying. When you don't have interupts you have to constantly check to see if an event has occured (like what degree the crank is at), with interupts the processor can be doing other things get a blip from the crank angle sensor compute that and then go back to what it was doing before.

 

We have some old Motorola HC11's here that might do the job, just not much fun dealing with 8 bit. I need a cheap processor that is easy to work with that has interupts and 16 bit computation.

 

Okay, I'm out of geek mode now. Back to motor head world. Just put a big carb on that thing, with a wild cam. Oh wait I went too far. Boost, intercooler, nos! okay, back to normal.

[clint78z]

I do find this type of stuff intresting, I snoop the DIY EFI website all the time.

 

In order to determine the chip speed you would need is going to take a little calculation before you select something as far as a chip.

 

Now looking at the new DFI software the logging is set for a default of 10samples per second and has 6 channel capability, you may want only four. That would mean the chip would have to take a reading from a sensor every .025sec minimum.

 

Just to give you a little background I have taken an instrumentation & electronics class before, we didn't go super deep into the digital stuff however I have a base knowledge of it. I have take Visual Basic, and I am about to take a C programming course.

 

I can go over some of your stuff and mabey offer suggestions. What type of program is used to take the values from the ADC through the comm port??

 

Piggybacking the ECU is probably the best way to go. Simply manipulating the AFM wires to give desired fuel curve. Take MAF wires as input voltage to your chip, then send a digital signal to an DAC to old AFM wires on the Nissan ECU.

 

I don't think you want to go into actual switching of injectors, you are opening a huge can of worms there. That is way over my head.

[Tony240ZT]

Actually I just found a new chip that I think I like a lot (just ordered one) called the BasicX at http://www.basicx.com/ it has interupts, 32bit math, built in ADC's, real time clock, multi tasking all built into it's Basic programming language.

 

The 8 bit ADC I was using is called ADC0831CCN by National electronics. I talk to it with the PIC mirco by shifting in and out bits over a 3 wire serial protocol (TTL (0v is low 5v high)). There are a couple options, but the most important bits are the 8 that give you the voltage level. The chip work from 0 - 5 volts (digital count 0-255), but you can use an op amps (like a commen anolog 324 or 740), and resitors to size up any DC voltage level to within this range.

 

I've set it up to read in the MAF info, then produce an anolog voltage level using a Texas Insterments 12bit DAC TLC5618aip, no reason for 12 bits, 8 is best I can do really. Same way of talking with it as I do the ADC, shifting bits over a serial protocal. Again, I can use an op amp to up it's output voltage above 5v if need be.

 

Currently I'm just running full balls out at computing and spitting out this data at 9600 baud. With this other chip that has multi tasking built in I'll be able to do other things while talking over the serial cable to my computer.

 

If you know how to write programs in VB to communicate with the serial port please fill me in on the details, this is something I'd like to write so that I can make a real time data logging graph. I know how to write the graphing setup (done pleanty of drawing), just haven't jumped into serial i/o yet. Also would be nice to be able to modify a visual map on the computer, then upload it to the BasicX, I should be able to do this in real time since the chip is multi tasking now.

[clint78z]

I think your skills are above mine in this area, however I think I can follow along pretty well. Most of the stuff we did is just with op amps and mostly analog. My programming is pretty decent although limited to engineering software devlopment, never done I/O from ports.

 

The new cards with Basic software are killer, and much simpler than it used to be. The new one you picked looks pretty swank.

 

So let me try and work this out in my head you, it's been a while and digital is a little rusty. take a 0-5v reading from MAF to an 8bit ADC. So this has a ribbon wire of 8 wires displaying 0 or 1 on each of these lines. The speed at which this is done is determined by the ADC.

 

Then you use a PIC microcontroller which takes the 8bit digital signal from the ADC. Can you explain to me in general about the 3 wire serial protocol, and how info gets from the microcontroller to the program variables.

 

How are you doing you graphing just a dump into a text file, then import into excel ?? I have not any serial I/O however I did save this little tidbit because I knew enough that it would be useful later on. http://www.ontrak.net/visual.htm

 

Very intresting subject, I hope to understand more of it.

[BLKMGK]

I LOVE this stuff The AEM I'm using does some incredible logging and I'm looking to add some additional sensors. Some to check out -> http://www.kavlico.com/library/pdf/p155.pdf

 

Westach sells those at a good price. Horsepowerfreaks.com does too for the AEM - tell Chris I sent you.

 

Here's another one -> http://www.tfs.net/~petek/rockets/accel/accel.html

 

Can you say linear logging of G-force in 2 axis? I was going to do that one but my ECU has fewer 0-5volt inputs than I realized. Lot's of digital I/Os though

 

Ton's of stuff you can do with this stuff, I love it!

[clint78z]

Logging the fuel pressure seems like a good idea to me, since it is critical to the engine not blowing up.

 

Logging 50g's capability, I realize you car is fast but keep dreaming BLKMGK

 

You can always take your voltage from sensor and use an analog to digital converter ADC for you AEM computer.

[Tony240ZT]

I'm thinking the digital inputs/outputs on the AEM are probably counters (like square wave inputs), or it simply senses if something is on or off. Digital inputs would be like RPM, WOT.. Digital output would probably be used to pulse a air bypass for either idle control or boost control, also it should be able to turn on NO2, or a secondary fuel pump/fans, whatever.

 

Here is the data sheet on the ADC I use:

http://www.national.com/ds/AD/ADC0831.pdf

 

It explains how it uses 3 wires to transfer the data in and out. The actual frequency that I get the ADC data is dependant on both the processor and chip. I had no problem trying to get data as fast as possible out the ADC, so my processor can't over ride the ADC's i/o speed. The ADC chip and the pic micro both talk TTL which means 0v is low (0) and 5 volt is high (1).

 

I just spit out the data in tab delimited format, with returns after each reading for the set of ADC's collected. This is easily imported as as .csv file into excel. Then writing functions in excel you can convert the numbers to voltage levels, load levels, graphs, or whatever.

 

Check out http://www.basicstamp.com/ for an easy to use pic micro setup. You can use basic to program it, and you don't need any special hardware to program the things, just a pc and a serial cable. The BasicX looks like a Basic Stamp on steriods. If you get into the stamp I can help with setting up a system similar to what I have, also I should have a BasicX today and I'll start playing.