Notices

Hello everybody,

I was just wondering, if it would be possible to port Dynolicious for the N900?
Does anyone know?

Best wishes
Peter

Here is a link to the app: http://dynolicious.com/
I hope that someone besides me is also interested in the Program.

Best wishes
Peter

I think it's easier to just write one.

All it takes phone-side is a data logger with precise time stamping and decent sampling rate. There already is an app (AccDisplay) that logs and plays acc data and saves as a... xml?

A bit better recording and one can simply download the file into a PC player that replays the run.

Best part is I'm writing my own PC-side app.

I did my first trial run on my car today.

Here we go: Picture

Since quality is less than peak (picasa jpegs everything), allow me to walk you through it. Also, excuse the dirty lil' temporary excel file. I plan on doing a better job.

So. Started driving home, i was already rolling when I remembered I wanted to do the record, so it starts out driving. As you can see from the red labels, negative X is accelerating, positive is braking.

As soon as I started recording, I got a good stretch of road, but it's cold and wet here do not too much. Still, I get a peak of 0.9G with a rebound 0.85 in between because of the rebound of the spring compression. Also, the device logs at about 10 samples per second (I timed it with a watch for 10 seconds, it did 104 samples).

Also, I averaged out the first ~30 samples or so (you can see it at the top) as I got going, averaging out 0.616G.

After that, you can see it drop to zero, as I shift then a test brake the some of the second gear and the the second test brake, this one quite strong, at 0.65G.

After that, there's constant acceleration with a constant 0.5G lateral force to the right (curve).

Then there's a flat part (I waited for some old car to get in trafic, with the added bonus of self-check), then the second gear, then normal driving (for me). I'm not the hold-coffee-on-dashboard driver.

Then, at the very end, on the left-right Y axis one can see a strong left-right balance as I climbed the front wheels on the side of the road (the Z accel also shows this, cropped in the image).

I'd say those are pretty good results. The only thing that is missing (aside from a decent GUI) is that the author of AccDisplay only logs X, Y, Z and no word on current time. Precise timing would be great for playback.

I plan on doing a GUI that shows car forces, a little animation perhaps and, of course, guessing car controls black-box like (user is pulling left, pulling right, etc).

Unfortunately, I'm a Windows guru, so any GUI will be Windows only. I would be happy to share any experience with anyone willing to port/rewrite this stuff. I have Linux guy lying around, I'll ask him if he'll write for Linux, maybe even Maemo.

Hey ndi,

that looks really good! Yes you are right, it would be good to also log the time and maybe in a second step the gps coordinates.
For a start it would be nice measure 0-100 km/h (0-60 mph) times or 1/4 mile.
I'm really looking forward to the first usable program on the N900.
I would like to help you, but I don't have enough knowledge about programming to help you, sorry.

Thanks Peter

Nice app. Interesting to be ported on maemo 5

Didn't carman from the Canola guys do something similar ?

I would love to get similar program on my n900, Carman would be nice. Hopefully someone makes it happen soon...

If you have fine enough GPS coords you don't need the accel data, since acceleration is variation of speed. However, GPS is not accurate enough to give anything else than a somewhat average speed, since errors prevent it from detecting sudden changes, like peak acceleration and stuff. A compound data log would be nice, but apps need to be rewritten (or running at the same time). Precise time stamps would be much more useful.

0-100 and 1/4 mile is a lot easier with a watch, since the sampling of the program is imprecise and to get speed you need to add acceleration every second, a process that compounds errors. I'm pretty sure you get a lot better with a watch.

What you CAN so is start recording, floor it to 100 and then suddenly take the foot off at 100, and then see timing, so you can clock yourself. However, I usually do this via camera (video) because frames are a lot more precise. used my N80 that did 15 frames per second and counted frames from the speedometer first nudge to 100 Kph (60 mph). However, you can time yourself to A speed (not necessarily 100) by adding acceleration and delta timestamp. It will be AROUND 100, and probably fairly constant through several runs, but unlikely to be exactly 100 (kph).

What the accel log will show your is a power curve as they like to call it (actually it's the torque), and thus profile engine and gearbox (and driver), see where it's less than efficient and either tune it or learn to redline race. The dyno is better because the measure "at the wheel" which is good but has nothing to do with what's in the owner's manual.

What's even more useful for is profile the route for a race, log attack angles and lateral acceleration limits, profile tires, stuff like that (just like pit crews, except they log speed from speedometer not GPS and have may other sensors (you don't need) like suspension excursion). The GPS is useful at high speeds and averages only at current sampling rate.

As for HP and stuff, that's marketing alone. Even the original app says "guess" your horsepower. Because of how one's HP is calculated, it has nothing to do with acceleration (hint: locomotive). Dynos make use of an adjustable hydraulic or electric brake to variable-load your engine - something you can't do alone.

Also, since it's passed through the transmission to wheels, torque is also hidden unless you know the gear ratio, final drive ratio and exact tire specs (including pressure).

What it's really good to have an accelerometer for is profile tires (which give me better traction), limits (skid, lateral and forward), and tuning. That's where the "dyno" thing comes in - if you drive really well you can see if a new sports filter gives you a torque increase at what RPM - something that would be lost with other measurements.

You can also compare runs of the same track and see what late or early braking brings, for example.

For me, it's going to be a race for peak G at the start line - I'm a bit of a forward G junkie. At 0.9 G stuff that isn't bolted practically fall to the back of the car. That kind of stuff you can't (and shouldn't need to) explain.

Oh, and, the car those figures came from is a 3-liter V6 and RWD on wet, cold asphalt. I can do better.

I'm back.

Good news is, I have a working draft.

Bad news is, it's not working.

The problem lies in both the imprecision of the accelerometer and the sampling frequency/technique, IMO. The problem being that if you start recording, then you slide the phone forward, then backwards, the numbers don't add up.

In theory, the math is simple. 1G is 9.8 M/s/s, so a 0.1G acceleration yields a 3.6*9.8 M/s if sustained FOR ONE SECOND. Problem being, I have no idea what the mean acceleration was for that second because I have 10 samples that vary wildly.

In theory, going forward one second at 1G gives you 10 samples with +1000, and going backwards gives you 10 samples with -1000 in the XML. In practice, the don't add up.

In the screenshot I posted, I drive first and second gear, up to some 70-80 KPH, then brake to a standstill, then a long stoplight, seen as the uneventful line in the middle. Problem is, the sum of all acceleration, even corrected by integration, shows that I'm still doing 30, because the braking is not as violent and low acceleration isn't as fine-grained or as important to the equation as high acceleration.

One could also see what happens when standing still. Namely, that there's quite some noise from the accelerometers, and even that noise doesn't add up. That means that over the 30 seconds I stood still the error adds up to AT LEAST 0.5 M/s. This means that the compound error is never going to add up or make sense over a long period of time.

Further compounding the problem is the fact that the car actively leans backwards when accelerating hard, some of the acceleration being spread out between axes, with integration necessary.

I'm still working on it.

I think I need a trial run to calibrate errors, the one I have is a standard drive around, with curves and stuff. And I'm half snowed in.

Volunteers?