Dallas J

iTrader: (1)

Sampling at 10x input frequency is more than plenty to build a frequency plot. You're measuring position to calculate rate. 10x may miss the peaks of position at high bump rates but that value doesn't really matter anyways. 1000x is overkill.

Robevo RS

iTrader: (16)

My opinion is based on my track - hill climb and rally experience, which is now over 25 years as we speak. So here it is.
I am 100% sure data logging is some what essential to getting a base set up for your suspension, and for the car you have . (since all high end coil overs and cars in racing are different so there is not really a universal set up for everybody neither two similar cars in my experience) Specially when you start playing different springs etc. You def. need a little note book in your car as i know now. Where you can write down your settings etc, depending on surface - - weather - tires etc.
But i dont think you need to go too far with science here after you nailed down your base set up for the spec you have on the car each time.
SO 500mhz or 1000 mhz or even only 100 mhz , is fine. I am sure about that.

The reason is , because what ever your builder, and suspension Guru set up for you to START with, the driver will be the ultimate decider of the shock adjustments. In the real world the best set up is where the driver feels comfortable in the car. That is depends only the driver habit - style and attitude in racing. So that is why every driver has a different set up.
I dont see too much need for higher and higher details, because you can lost when you focusing too much on them, and you can loosing sight of the goal..
The fact of the matter is, get a good base line to start, for a driver. Then you basically follow him, you might find things weird sometimes as an engineer, but makes sense for a drivers.
Example :I am always wondering why everybody like hard set ups, like 12K 14K 16 K springs etc. When i do the opposite, try to be soft as possible but still stable, let the suspension work too, not just the tires .... In fact i got an EU suspension race tuner set up my suspension. Did a great job, but after he did his home work get a great base set up for the coils with a springs i have and tires, at the testing it was all about how the driver feel, and set it up for that. No more science, it is all about feel.
I hope you guys get my point about HZ , and i explain myself clearly ... lol

KevinD

iTrader: (56)

Rob, your philosophy is shared by many. Ultimately it's the driver who decides on setup.

My opinion is this: people at amatuer level racing aren't good enough to know the difference, most aren't very good at giving real feedback, and science is always right

So if your driving an f1 car and can run ever lap within a 10th of a second, if you ask for a suspension change I would believe it would make you faster. But they are backed up by the data I am collect along with a dozen other parameters I don't have the means to collect.

A good analogy is this: would you pay a guy to tune your engine that doesn't record any data? He just goes by "feel", and driver feedback?

My objective is to show the science behind spring and dampers and will be able to give people insight into what they need to look for and how to set up the cars based off science and not just "feel". I know our cars can get more
Mechanical grip with correct data then going purely off feel because some people "think" having enormous spring rates, or way overdamped shocks are faster when it may actually be their driving style isn't conducive to going fast.

The second thing I want to shine a light on is the term "custom valved" shocks. Since almost no one has any idea at all what a damper even does, I don't think they can properly inform the shock builders what they are looking for. (And we already know less then1% of cars on the race track actually measure suspension parameters) My hope is to educate people how different damper settings effect the cars response, so that people who do care, can give the needed feedback to ensure their shocks act in a proper mannor.

Iowa999

iTrader: (1)

This looks very interesting, even if I have nothing to add. But I suggest correcting the labels on the plots of the spring-rate tests (you have front vs rear backwards), so no-one gets too confused. As to recording rate, the Nyquist rule only requires twice the input frequency to avoid aliasing; 10x is overkill and 100x is silly.

Robevo RS

iTrader: (16)

agreed. I just try to explain, why i dont think is really a big deal if your record something with 100hz vs 1000hz.
Because end of the day what ever you come up with your data logging, the deciding factor will be the driver feel, even in amateur level.

Now you mentioned F1. That is a different discussion entirely. There you need a 1000 hz and there is a totally different environment - cars - suspension - tires - drivers - team - funding etc etc etc. Has nothing to do what i am saying here in this car forum.

I am talking about Mitsubishi Evolution, and the suspension which comes with it. So as a driver.
To measure the driver i think the throttle logging is more efficient then a suspension

I could have share stories about that next to a beer.. LOL
Like when a guy complains the car is not going , not fast enough. Open a logger and you find he did 72% of his run off throttle basically.. lol Explain that

griceiv

iTrader: (2)

I think you guys missing the point of the high sampling rates. For position measurement 10x+ sampling is fine. When you take the derivative of that positional data to get shock velocity the lower sampling rates adds noise, and that noise can have a substaintial effect on the fidelity of your measurement. Taking the derivative of data sampled anywhere near nyquist is going to give you squat.

Iowa999

iTrader: (1)

We are getting way off into the weeds, but I completely disagree. Most of all, recording at too low a frequency does not add noise (in the technical sense of "noise"). However, when you sample at 10x the actual input frequency all you're getting in the additional data is noise. Finally, if you plan on taking the derivative, you only need to double the recording frequency; 10x is overkill and 100x is silly.

griceiv

iTrader: (2)

I didnt say that low sample rates add noise to the signal. I said it adds noise to the derivative computation. I think the velocity data posted thus far supports this pretty clearly, but i dont mind if you disagree.

Iowa999

iTrader: (1)

I understood your point on low sampling rates adding "noise" to derivatives. Did you see where I mentioned that you only need to double the rate when working with derivatives, not raise it by an order of magnitude?

In any event, when we're talking about suspension movements, after the tires have killed all the real high frequencies, these are over-sampled data. Any formal analysis of the high-frequency stuff is going to tell you a lot more about the problems with linear potentiometers than about cars. That's my real worry. There's such a thing as too much data, IMO. You end up staring at some blip at 20 Hz when it has zero to do with grip or handling.

golgo13

iTrader: (134)

Great stuff here, curious how all this data is going to benefit the average gearhead (like me)?

EVOizmm

iTrader: (10)

I like your philosophy on this. unless you are spinning extremely consistent lap times try to leave setup alone.

Besides track laps, do you perform any other testing? skid pad or any other process to dial in damping?

Dallas J

iTrader: (1)

The nyquist rate is a very crude absolute minimum. You need a series of collection over time at a given rate to be able to reconstruct the frequency with 2x but when you have a rapidly changing rate where peaks are hit for tenths or hundredths of a second, you need to know the slope of the instant position change.

As Grice eluded to, you need instant derivative of each point if you want useful wheel speed information. An overtime average will give you just that, and average.

Iowa999

iTrader: (1)

OK, I give. If you want to look at stuff that lasts for a few milliseconds, I will not stand in your way.

Dallas J

iTrader: (1)

If you want to see peak events in the 10in/sec range, they yes you will need to be able to see tenths to hundredths.

I don't do any of this, don't have the capability on my car, but this is really counting dimes. Sampling in the 100hz range and dealing with the data is childs-play these days.

Iowa999

iTrader: (1)

Grrr.

Yes, playing with the data is child's play. But that's half the problem. You see people all hung up on some transient which quite often is an artifact of the potentiometer, instead a valid measure of the car, and wouldn't make any difference even if it were real.

But don't take this as me being against having more data. You just need the correct additional data. Wheel speeds, steering angle, throttle position, and brake-line pressure are what are needed. Not the ability to see a 10"/sec peak that lasted for 2 ms and is probably caused by slack in the ball-joints at the ends of the potentiometer.