Suspension Data Acquisition

[instefan]

Hi guys I am responsible for the DAQ of the Suspension in our car and I'd like some insight from other teams because it's our first go and we don't know exactly what we'll need.
So far we have decided to use 4 linear potentiometers for lift/dive measurement, one 3-axis accelerometer at the CG of our car or a 2-axis accelerometer and a gyro.
Could you tell me from your experiense will that be enough or we are going to need something else?

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ioannis Stefanoudakis
Suspension
Aristotle Racing Team

[exFSAE]

Well.. what are you trying to achieve?

[instefan]

Originally posted by exFSAE:
Well.. what are you trying to achieve?

So far all the data we used were from simulation. First of all, we would like to compare these data to some actual ones from our car to evaluate our current and future engineering choices.Also we want to be able to check our drivers' performance and get a more constructive feedback from the track

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ioannis Stefanoudakis
Suspension
Aristotle Racing Team

[Owen Thomas]

You need to have a discussion with your suspension guy and figure out what parameters you need to log in order to validate the simulations. Also brainstorm with him about what data will make it quicker or easier to dial in your suspension setups.

Just for suspension, I would suggest adding steering angle and throttle % at the least, since those two will help you get a better read on what's happening in the cockpit (driver reaction to oversteer, etc). This should very greatly help in determining grip balance, and speed up your suspension tuning. Tire temperature sensors can be useful to see if you're scrubbing your tires a whole lot or not using them effectively, but if it's your "first go" that's definitely not a high priority.

[Zac]

Another thing that should be part of those discussions is test procedure. I've seen a lot of teams dive straight into hot laps on a competition style course (within hours of getting their car together) and get themselves really confused with all the data they've just collected. Often your time can be better served doing less glamorous testing, like skidpads and slaloms.

[mech5496]

Wheelspeed and GPS worth considering too...

[rrobb]

Sounds like you need a more structured plan of attack.

What tests are you planning on running? The answer to this question will suggest what sensors you need.

Driver - steering and throttle

Shock tuning - damper pots

Tire - 6x IR temp sensors (12 is better but $$$)

Aero - pushrod strain

[Dylan Edmiston]

You need to talk to your system leaders and see what they need and make sure it's within your budget. While tire temps are good to have for some more "advanced" tuning, it's not needed.

I'd recommend picking up some books and reading what you can do with certain measurements and why or why not your team will need that data.

Chris Brown's "Making Sense of Squiggly Lines" is a fantastic introductory book into data analysis.

[jlangholzj]

if your looking into driver performance, there's a few things that are more beneficial than shock pots and even tire temps (if you don't have them all yet).

must haves:
-rpm
-wheel speed (front and rear)
-brake pressures
-steering angle
-lat accelerometer
-lon accelerometer
-throttle position


these are going to give you a great start on evaluating driver performance and car performance for that matter. You'll need at least a few of those to correlate to other data anyway.

[BillCobb]

I would recommend avoiding the 'pile-on' technique for loading up your Data Recorder. First of all, that's all you really want: a data recorder box, not a data 'processor'. Do that afterwards.

Evaluating 'driver performance' is a pretty nebulous undertaking, and by the way, your list of signals necessary to do that is still missing some key elements.

Instead, have your Team Leader decide what tests are necessary to compare the open loop performance of the current state of the car against its objectives. Gains and bandwidth and response time are the usual metrics of interest to a control system design project. (Yes chillins, a car is just another nonlinear control system). And, the metrics need to capture the nonlear traits successfully.

Then you run procedures on a skidpad or straightaway to measure the current state of the system using several procedures. You change system parameters (bars, springs, pressures, connection positions and brands of thisngs to see what contributes to the meeting of the engineering goals. The best driver for this is a robot, second best is a test driver who knows how to run tests and data acquistion systems or radio equipment. Someone who is strong, fast, consistent and reasonably educated in control engineering processes. Someone who can 'feel' subtle difference in car performance such that this can be related to the closed loop control problem solution that comes next.

I'm just not seeing this floating upward in the discusssion. If all you want to do is measure lap times, a stop watch is actually all you need. If the driver's shoes are too pretty, its probably not a good sign and a 'step' in the wrong direction.

[instefan]

Originally posted by jlangholzj:
if your looking into driver performance, there's a few things that are more beneficial than shock pots and even tire temps (if you don't have them all yet).

must haves:
-rpm
-wheel speed (front and rear)
-brake pressures
-steering angle
-lat accelerometer
-lon accelerometer
-throttle position


these are going to give you a great start on evaluating driver performance and car performance for that matter. You'll need at least a few of those to correlate to other data anyway.

Well all we have right now is rpm and throttle position so I guess you don't suggest the pots just for now with the accelerometers, wheel speed brake pressure and steering angle having higher priority.


Originally posted by Dylan Edmiston:
You need to talk to your system leaders and see what they need and make sure it's within your budget. While tire temps are good to have for some more "advanced" tuning, it's not needed.

I'd recommend picking up some books and reading what you can do with certain measurements and why or why not your team will need that data.

Chris Brown's "Making Sense of Squiggly Lines" is a fantastic introductory book into data analysis.

We don't know the budget yet, but we are always looking for the most we can get. Especially this year that we don't have a car to manufacture we'll spend all our funds for such equipment that will stay on the next cars.

[Will M]

I would agree strongly with BillCobb that how you test is much more important that what you measure.
So you should decide what you want to test (steady state cornering), figure how it can be tested (skidpad), figure out how to measure what you are testing (lap timer), and then develop a test plan (what variables you want to change).

But I think you are just looking for equipment so…
Steady state cornering -->skidpad-->lap timer* (or accelerometer)
Acceleration--> 0-75 meter event --> wheel speed on undriven wheel (or GPS)
Transients --> slaloms --> wheel speed on undriven wheel (or GPS)

For a first year car I really would recommend that you keep it simple and cheap.
You will be able to do a lot with just a good lap timer* and a single wheel speed sensor. I would also recommend a tire pyrometer, but it could be a $100 hand held unit.

Once you have some more experience and want to dig into what the tires, take a look at adding things like.
Tire temp sensor arrays, ‘shock pots’, GPS, accelerometers, gyrometers, four wheel speed sensors.
Just remember that collecting data is time consuming and analyzing it even more so, not even count the effort and cost of buying and integrating the sensors.
Properly understanding a handful of major variables and measurements is much more valuable than a weak understanding of a dozen variables and measurements.

Just my 2 cents,
William


* I would recommend against a stopwatch, the normal delay for a human using a stop watch is ~.25 sec and the difference between the skidpad times for the top 7 teams at MIS last year all fit in .25 sec

[murpia]

Hey, 2 threads!

http://fsae.com/eve/forums/a/t...20097151#82320097151 (http://fsae.com/eve/forums/a/tpc/f/125607348/m/21320987151?r=82320097151#82320097151)

As to a core sensor list, to understand your driver you need:
<UL TYPE=SQUARE><LI>Steering angle
<LI>Throttle position
<LI>Brake pressures (front & rear if you have a balance bar)
<LI>Gear
<LI>4x wheel speeds
<LI>Lat G
<LI>Long G
<LI>Engine rpm[/list]
You can't really ignore any of these or you'll have an incomplete picture.

If you want to understand your car add (in priority order):
<UL TYPE=SQUARE><LI>4x Suspension position
<LI>4x Suspension loads
<LI>Z-axis accelerations (hubs & chassis)
<LI>Tyre temps[/list]
A yaw gyro is nice for both. I've left out the data needed to understand your powertrain.
Regards, Ian