Turbobricks Forums

Turbobricks Forums (http://forums.turbobricks.com/index.php)
-   performance & suspension (http://forums.turbobricks.com/forumdisplay.php?f=13)
-   -   High speed data logging fun (http://forums.turbobricks.com/showthread.php?t=255382)

monkecmonkedo 03-22-2012 11:41 PM

High speed data logging fun
 
For kicks and grins I used a Labjack U6, which is capable of 50kHz data streaming, to see what's going on with my car during WOT runs. I figured I'd start with the AMM and got the following results for my 016. For the record, the car is a '92 245+T AW70 running pretty much stock except a 15g @15psi and a 954 ECU with Mike's chip (EZK chip too).

This is a 0 to ~80mph run collecting data at 2kHz (20 data points per crank rotation at 6000 rpm). At first glance, I thought all the "noise" was due to using unshielded wires, but no... you can actually see the air moving into each cylinder (pics 2 [idle] and 3 [WOT])! While I didn't measure RPM directly, I used the peak and valley of each intake pulse to calculate engine RPMs in pic 1. The response time of the hot wire in the AMM is pretty amazing. I assume that the ECU sees only the maximum of this signal, since I seem to be maxing out my AMM at ~3500 rpm (my AFRs peg at 10 and I've got no boost leaks).

Once I get my 012 AMM installed I should be able to calculate VE using the intake volume for each pulse (the area under the curve) plus the AMM transfer function.

I think there's lots of interesting things to note about this high speed data and I'm curious to hear people's thoughts.

http://i301.photobucket.com/albums/n...T/Picture1.png

http://i301.photobucket.com/albums/n...T/Picture2.png

http://i301.photobucket.com/albums/n...T/Picture3.png

Gralin 03-23-2012 01:55 AM

COOL:nod:

CAPT_BLOTTO 03-23-2012 02:47 AM

I would think the ecu takes some sort of rms/averaging function before interpreting a value.

AJS 03-23-2012 03:29 AM

I like this. Be good to log some different Volvo setup's

740volvo 03-23-2012 06:08 AM

A/F ratio of 10!! Seems like you can get a little more out of her :) once you get the 012 in.

crogthomas 03-23-2012 06:43 AM

Great stuff. I was about to try doing something similar myself to see how close to the limit of the AMM I am. My logger only reads at 10Hz though, so I won't get such detailed information. I may try finding a better logger. What would you say would be the minimum sample rate I could get away with?

It would be very interesting for you to remove the AMM screens and then try the same test again to see if there is a difference in the shape of the curve (once the ECU has learnt the new setup).

Karl Buchka 03-23-2012 06:48 AM

I'm surprised the AMM is that fast. You should log the VR sensor output. Would be interesting to see the crankshaft acceleration during a power stroke.

crogthomas 03-23-2012 07:30 AM

Quote:

Originally Posted by Karl Buchka (Post 4107160)
I'm surprised the AMM is that fast.

Surprised that it is possible, or surprised that Volvo bother?

Hot wire anemometry is commonly used to measure turbulance in airflow. The sensors are usually a lot smaller (i.e. lower thermal mass) than those used in our Volvo AMM, but I see no reason why they would not be able to measure these very slight variations in airflow. What surprises me is that the measurement need be that precise, especially as it is then probably averaged out in the ECU.

monkecmonkedo 03-23-2012 11:18 AM

I think the ecu must only look at the peak. After all, that is the max airflow (Valleys are when the valves are closed). Furthermore, the ecu obviously thinks the airflow is off scale, since it's going so rich. Note also that all the peaks are around 5.2V, but the Valleys change. Has anyone identified this circuit in the ecu? modifying the circuit to look at the average or valley might be one way to trick the amm into being more useful for higher air flows. You can certainly flow allot more air through a 2 inch tube without being choked (super sonic) than an amm can measure. Even at idle, airflow through an amm it's turbulent according to Reynolds number calculations. I think intake velocities up to 36 psi of boost are still well below choked flow. Note that due to the hot wires small size, 70 um, flow around it is laminar even at really high velocities. Note: I amm not an expert it electronics our fluid dynamics, so my assumptions could be wrong.

monkecmonkedo 03-23-2012 11:21 AM

Karl, what do you mean by VR sensor?

Homer 03-23-2012 11:26 AM

Crank position sensor.

Karl Buchka 03-23-2012 11:32 AM

You can either read the sinusoid directly from the sensor or you can tap pin 7 of the LM2903 in the EZK for a square wave output.

monkecmonkedo 03-23-2012 09:49 PM

More high speed data logging fun
 
So I modified a NA EZK box and put my chipped Turbo EZK guts inside:

http://i301.photobucket.com/albums/n...-47-52_707.jpg

That gives me access to clip onto each terminal. In addition to the AMM signal (direct wire), I hooked on to pin 13, knock sensor signal, pin 16, ignition pulses (had to use a 2x voltage divider as it is a 10V signal), and pin 17, conditioned VR signal to ECU. I upped the collection frequency to 9kHz, which should give me 4 degree resolution on the crank at 6000 rpm which means 2 degree resolution for the cam (again at 6000 rpm). Resolution will be better at lower rpms. I can go up to 50kHz, which would give me sub 1 degree crank resolution at 6krpm, but only for a single channel, maybe two. I'll learn more as I play around. Again, I idled the car for a few seconds, then floored it until I hit 80ish. This was not a launch... that will come after the local track opens. Here's the results:

http://i301.photobucket.com/albums/n...atalog1png.jpg

Idling:

http://i301.photobucket.com/albums/n...12Datalog2.jpg

And WOT:

http://i301.photobucket.com/albums/n...12Datalog3.jpg

I generated over 14MB of data in just 16 seconds. Now I just have to analyze it all. I should be able to get ignition advance vs rpm, see if I'm having any detonation, and plot air intake vs. rpm, plus??? This is just getting started, there will be more (intake pressure, intake temperature, EGT, various AMMs, ???)

First though, does anyone know where the missing tooth (teeth?) is for the VR sensor. I assume it is a hall effect sensor that measures a gear in the bell housing? Is the square wave leading edge at TDC for cylinder 1? It looks like there must be a second missing tooth, maybe at BDC for cylinder 1... I need to know since it is the reference for where the crank is. Thanks!

Karl Buchka 03-23-2012 09:57 PM

The crankshaft position sensor is a variable reluctance (VR) sensor that reads a 60-2 trigger pattern drilled in to the flywheel.

Don't measure pin 17, that's useless. Tap the LM2903 like I suggested and it'll tell you everything you need to know.

monkecmonkedo 03-23-2012 10:03 PM

I should have listened... tomorrow I'll clip on pin 7 of the LM2903 like you said. I'll see if I can log at 50kHz to get good resolution. At the very least, 36kHz.

So I'm guessing I was measuring the speed sensor (diff?) that feeds the speedometer? The signal is definitely speed dependent.

Karl Buchka 03-23-2012 10:07 PM

The signal on pin 17 is sent to the fuel computer so it knows the engine's RPM.

monkecmonkedo 03-23-2012 10:50 PM

Quote:

Originally Posted by crogthomas (Post 4107158)
Great stuff. I was about to try doing something similar myself to see how close to the limit of the AMM I am. My logger only reads at 10Hz though, so I won't get such detailed information. I may try finding a better logger. What would you say would be the minimum sample rate I could get away with?

Depends on what you want to do. If you just want to know if your AMM is maxed, the info on TB seems accurate (At 15 psi with a 15g, the stock 016 AMM is maxed by 3500-4000 rpm). Honestly, what I'm playing around with is interesting, but is probably overkill for a 160k tractor motor running 15 lbs of boost. For checking sensor function, etc, a low data collection rate is probably fine. If you want to start looking at ignition mapping, etc, you need to have a data rate above what you're trying to measure. For example, 6000 rpm is 100Hz. If you are measuring crank rotation, you'll want a minimum of 360 points per rotation or 36kHz for 1 degree resolution.

That said, check out Labjack.com They aren't plug and play data loggers and they're not exactly cheap, but I'm pretty impressed with what they can do.

monkecmonkedo 03-24-2012 10:41 AM

Quote:

Originally Posted by Karl Buchka (Post 4108013)
Don't measure pin 17, that's useless.

Karl - Actually it is useful... but as you said, not for crank acceleration. It looks like pin 17 outputs a square wave for TDC and BDC. See my post about ignition timing. Thanks for the help! Dave

944Turbo 03-24-2012 10:03 PM

Oh man, that is seriously effin cool!

monkecmonkedo 03-26-2012 09:25 PM

So here's the VR sensor output (pin 23 in the EZK), the engine speed output to the ECU (EZK 17), and the Ignition pulses (EZK 16) at idle and acceleration. Data acquisition rate 13.5kHz. Count em' if u can: 58 teeff an two missin' uns. So it looks like TDC may be the trailing edge of the square wave (when it goes low)? That makes me wonder if the spark is triggered high or low. Anyone know for sure? This means my ignition timing calculations in my other post were off. Still, the engine speed pulses occur at the same spot regardless of engine speed so relatively speaking, the results are the same, my car seems to be pulling timing.

I'll have to data log at a much higher acquisition rate (50kHz, which means only 1 channel, the VR) to be able to show the crank accelerating through a power stroke. I need to be able to better resolve the time difference between the peaks created by the teeth. I'll get to that in a few days.

http://i301.photobucket.com/albums/n...T/VRSensor.jpg

http://i301.photobucket.com/albums/n.../VRSensor2.jpg

DNAsEqUeNcE 03-26-2012 09:37 PM

I think it goes low, to fire the bosch 124 module.

Karl Buchka 03-26-2012 09:42 PM

Brett is correct. Spark advance and dwell clearly increase with RPM. TDC is the rising edge of your "speed" trace.

EDIT: Typo.

gross polluter 03-26-2012 09:43 PM

Spark fires on the falling edge of of the coil output, the time that wave is high is the dwell time. Also, the missing tooth is offset by about 60 degrees from TDC.

monkecmonkedo 03-26-2012 10:05 PM

Thanks guys. BTW - the scale is different in the two graphs. Top one is 160 ms, bottom is 35 ms. Dwell does increase from 4ms (top) to 6ms (bottom). I'll have to go back and recalculate my ignition timing.

Hen 03-27-2012 08:57 AM

Referring to the start of the thread and the varying AMM signal, is there any way to check that it really is intake pulses you are seeing? And not interference from ignition leads, etc?

It seems impressive that the intake pulse can still be measured after passing out past the throttle, through the intercooler and most importantly out through your turbo (especially on the WOT run where the turbo would be spinning at full speed).


All times are GMT -4. The time now is 11:45 PM.

Powered by vBulletin® Version 3.8.8
Copyright ©2000 - 2014, vBulletin Solutions, Inc.