F1 engine output approximation by sound?

[checkered]

I vaguely remember

hearing about something like this, output analysis derived by sound intensity or such, but I don't quite know where to look with this one. Is anyone doing anything like this in F1 nowadays?

[mx_tifoso]

I remember hearing something about this matter as well, I think that when I heard/read about it was the Renault F1 team talking about a Ferrari and how by analyzing the sound, I dont know in which way, it gives them an approximate estimate of the engines output. Maybe the sound intensity/level of the engine depends on the RPM's and that somehow gives an approximate output number(s). But it's definately an interesting matter to look into.

[pRo]

There was an interesting topic about this on this forum, you can probably find it, if you search for it.

[West]

Reca did an analysis last year; dunno where it is, but it was pretty close approximation.

[zac510]

Yes, Reca did some excellent analysis and summarised it in a topic on the forum here.

[Ciro Pabón]

Well, I tried after Reca showed us how it's done. In this thread he deduced RPM and speed for a 23 seconds video of Fisichella:

viewtopic.php?t=2218

What you do is to convert an arbitrary wave (the sound of the engine) into cosine and sine waves. Once you do that, you can analyze the "basic" frecuency of the engine and deduce its RPM.

It's not through intensity but through frequency. You do what is called harmonic analysis, which is the transformation into a Fourier series of a (pressumibly) periodic function: the noise of the engine. A Fourier series is an infinite sum of cosine and sine waves. You do that after filtering the sound track to elliminate noise. In essence, you use this formula:



Here is an example of how you can transform any input (in black) into sines and cosines (coloured waves):



If you can deduce RPM and you know trap speed, you can deduce gear final relations, knowing the tyre diameter. From that, you can figure out the speed from the sound. From the speed, you can deduce acceleration.

If somebody is interested, you can start by downloading a YouTube video (courtesy of vdt, answering a question from Manchild) to your hard disk using this software (you don't have to, you can use the video directly, but...) :

http://keepvid.com/

I recommend FoxSports Latin America videos: the commentator, invariably, when a lap is shown, says "Let's hear" and then he shuts up... These guys are good!

I've been trying to use the idea to help me with the tuning of karts: I filmed a lap (you have to carry the camera on board! I learned that if you don't, the Doppler effect invalidates your results!).

A simple software that gives you RPM directly can be found here:

http://www.tunelab-world.com/rpmsound.html

The output is like this:



You can see the RPM, but you cannot store them. If you wish to store the RPM vs time, to follow Reca steps, you can use PCA3 (for only 30 days before paying 24 dollars):

http://www.ymec.com/

Which gives you this output:



I've been tinkering with that program for two weeks now and I haven't been able to reproduce Reca's results. We better ask him.

I dream of having the RPM, speed and acceleration to use that figures and my battered Brembo card to get a better picture of what the kart is doing (specially the competition). You feel like Jean Todt... even if the guys at the track have started to think I'm mad when I strapped the camera to the car.

It's already been told here that you can use Bosch software to deduce ride height, tyre load and spring behaviour for a couple hundred dollars. I don't care about the jokes at my track: let's see who does a better tuning this season! At least I've learned a lot: first lesson, which I got last Sunday afternoon, is "don't tie the camera directly to the chassis: it produces its own sound"...

The other problem I had was to get the speed (I know my own gear relation, but karts have no speedometer) for calibration, so I borrowed a radar from my old buddies at the Ministry of Transportation. If you are unable to get one, you have to figure out a way to measure speed: I also found that the increase in tyre diameter and (perhaps?) some slippage of the tyre makes the direct deduction of speeds inexact: you can get 5 to 8% of error on speed, which was discouraging. I found that when I compared track times with my first approximations. Perhaps I'm making some other mistake? ANY help is appreciated, guys at SAE. How do you do it, flynfrog?

Sorry for the length of this post, I got carried away...

[checkered]

Given that

teams are now running with their rpm limited engines, it’d be interesting to get output figures to see whether there are very large differences. Frequency estimation and harmonic analysis might actually be somewhat easier nowadays since a certain pitch can be easily associated with 19.000 rpm now. I didn’t manage to get so far into the theory of it that I’d yet perceive the method completely, like choosing the appropriate sample-rate i.e. I’m nowhere close to doing this in practice and don’t know whether I’ll venture to attempt such a thing. Nevertheless, I found reca’s gear ratio analysis very interesting, too. And yes, making the distinctions necessary to weed out Doppler from a stationary microphone are most likely to prove discouragingly difficult.

Ciro, your attempts to understand the functioning of your kart are quite impressive. As I’ve understood it, kart tuning is in a way even more challenging because the chassis experiences quite a bit of below yield limit deformation under forces – intentionally. So the variables are by definition more interconnected than in a construction the properties of which are designed to be adjusted almost independently. Maybe you could make the distinction between the tyres changing from rolling to slipping resistance and back by using an accelerometer. If it’s a 3-way one it could prove useful in other aspects of tuning, too, in relation how slipping occurs over forces caused by bumps and turns. It could also indicate, with other data, under which conditions it is advantageous to slide and when to take it easy and remain within rollong resistance. I have no experience in doing that, though, so this is mere speculation.

Thank you for digging up the thread and providing the information. If anyone does happen to stumble on some current F1 engine output approximations, keep us posted.

[SystemsWiz]

Really cool thread guys. I did some work with acoustic pyrometry, which, by using acoustics, is able to derive the speed, temperature, and pressure of a gas. Fourier is a really powerful tool if you can get your head around it. It took me about a year, but it seemed to me that one day, it just made sense. Really interesting stuff.

[persovik]

The other problem I had was to get the speed (I know my own gear relation, but karts have no speedometer) for calibration, so I borrowed a radar from my old buddies at the Ministry of Transportation. If you are unable to get one, you have to figure out a way to measure speed: I also found that the increase in tyre diameter and (perhaps?) some slippage of the tyre makes the direct deduction of speeds inexact: you can get 5 to 8% of error on speed, which was discouraging. I found that when I compared track times with my first approximations. Perhaps I'm making some other mistake? ANY help is appreciated, guys at SAE. How do you do it, flynfrog?

Sorry for the length of this post, I got carried away...

Probably, the easiest and cheapest way to get accurate speed etc. is to use the Drift Box if you don't trust the accuracy you get from an AIM logger.

[ss_collins]

Ooh the archives are calling Paul Van Valkenburgh went into a lot of depth about this about ten years ago - even then it was very very detailed. You can learn a huge amount from the cars exhaust note but its highly complex