MADJAK's NA8 - 200+HP N/A

[zossy1]

Great testing opportunity in August - MX5 Nationals at Wakefield! Come down with the rest of the QLDers and give it a try!

[madjak]

Great testing opportunity in August - MX5 Nationals at Wakefield! Come down with the rest of the QLDers and give it a try!

I'd love to attend, and I did seriously think about it, but unfortunately this year isn't very practical. It's a long way from Perth and trailering the car across the country with the family would probably mean I'd have to drop one of my local events. My focus at the moment is building the car up to compete in the National Hillclimb Champs later in the year.

Maybe next year is a possibility though.

At the moment I'm adding a full cage, changing the rear diff and soon to be adding some data tracking. My next event is the start of August and from then on to the national hillclimb champs it's fairly hectic.

[madjak]

I just ordered a RaceCapturePro data logger. I couldn't justify the price of the expensive digital dashes and in some ways this has much more capabilities to monitor and control whatever I need in the car. You can even get a telemetry option to upload the logged data to the web reatime.



I can hook this up to the CAN interface on my Haltech and datalog all the ECU data. I can then display it on an Android tablet on the dash. The unit itself has a 50Hz GPS, G-force sensors and heaps of additional analogue and digital in and outs. This means you get very accurate lap timing based on GPS (within 0.1 of a second) as well as split times and predictive lap times displayed on the tablet. It also has a scripting language so you can program external devices like a 3 stage shift light based on RPM and warning lights based on sensors exceeding thresholds. It has heaps of additional inputs so I'm planning on making some cheap ride height sensors so I can test my aero modifications and possibly use it to drive a DRS system on my rear wing.

I've ordered another 100PSI pressure sensor ($20) for my fuel system and also a couple of 1000PSI brake pressure sensors ($30) so I can see how much the fluctuating vacuum affects my brakes. I'm currently using the same 100PSI pressure sensor in my oil lines and it's working great.

I'll update once it's in my car and working... but I'm looking forward to having a play.

[Magpie]

Watching with interest!

[madjak]

As part the testing my rear wing I want to measure and log ride height on both the front and rear of the car. I can calculate the amount of downforce / lift different aero additions make to the car This way I can see the effect of varying angle of attacks and see when the wing starts to generate max downforce or stalls.

I will be able to log the sensors via the 5v analogue inputs on the RaceCapturePro. You can buy ride height sensors second hand for around $75 but they look a bit clunky and im not sure on their accuracy. The better ones are at least $150. Time for DIY!

Here is my design:


These will connect to a vertical pushrod connected to the suspension. I should be able to make these for $20-30 and they should be very accurate.

[Cus]

It's things like this that make me want a CNC and a printer.

I spy "input voltage 12V" in the corner, implying you're going to use the pots to limit the signal voltage. Please don't!

Pull the guts out of a lighter-socket phone charger ($10 at jaycar), it will output a nicely regulated 5V to use as your input voltage, that way if the pot fails/gets water in it/takes up meth the worst case scenario is 5V into the data logger, instead of 12V.

The downside is a little bit less resolution in the logging, the upside is if a sensor fails, it won't take the logger with it.

Then print a new case for the 5V power supply as well, because 3D printer

[madjak]

Thanks for the post... its exactly what I was worried about. I had planned on feeding the pot from a 12v source via the ecu to stop any fluctuations from the battery. Since its a 270 degree pot and I'm only using 80degrees it should give me a nice signal between 0 and 4-5 volts.

If I use a 5v source then output would only be 0-2 volts... which means the sensitivity is way down. The logger can handle intermittent voltage spikes up to 400v so it may be able to handle 12v... I'll email the guys from autosports labs and find out. Alternatively can you get something that will cap voltage at 5v? Some sort of voltage regulator that wont affect the 0 - 5v range.

[Magpie]

To calibrate the sensor just get some bags of cement and as you put them in the boot take a voltage. For the front it will be a bit more difficult but the same process can be employed, just finding a place the put the weights.

Another way is if you know your spring rates. However don't forget to double the spring rate when calculating. If you have 2 x 6kg springs, for each 1mm of compression there is 12kg of force.

Also make sure you do coast down testing to establish a baseline for drag. As you change the angle of attack downforce will increase but so will drag (generally speaking). Each time you change the angle do some more coast down tests. Coast down testing http://www.optimumg.com/docs/Aero_BertaReport.pdf.

This is one way of calculating frontal area http://ecomodder.com/forum/showthread.php/diy-frontal-area-measurement-462.html

5VDC power supplies http://oceancontrols.com.au/PSM-2531.html they also have 12VDC supplies as well.

[madjak]

Custom full cage is now in...




My goal with this cage was a CAMS complient cage that offered a bit more side and top protection than the usual half cage. The wonkey stock dash bar has been removed and replaced with a straight bar with custom steering bracket. This required to side dash mount points to be removed to fit. Also the side bars slope down for easier access but still offer goid protection. Ideally these would sit further out in the door but I didn't want to modify the door cover.

Dash is already trimmed... just need to paint it now.

[greenMachine]

Great minds ...!

I am currently scheming out exactly the same setup, sort of. All off-the-shelf gear though.

First though, a couple of comments. Your ECU or datalogger should give you a regulated 5v source, so that should eliminate the need for an additional component (aka failure point). Also, a key determinant of lift (downforce) and drag - and therefore L/D - is the wing's local AoA. Knowing this will enable you to focus your tuning on a much narrower band of AoA and speed up the generation of useful information (as opposed to interesting data). I am supposing you have lift and drag curves for your airfoil?

My approach to AoA is along the lines of tuft testing the region immediately in front of the wing leading edge (the only point where it matters in this context). It is likely to vary across the span due to hardtop effects.

I have a string pot which I am re-purposing (more later on the second sensor) to measure suspension travel. This will enable the sensor to be located somewhere relatively convenient (I hope) because it does not need a fixed rod to the actuating arm. The actuating arm is to be located mid point of the ARBs, to neutralise as far as practicable roll effects. I am not (yet) worrying about calibration, as the important part is relative movements - does a change in aero increase ARB deflection, no change, or decrease deflection? The front/rear aero balance I am planning on adjusting as a function of car balance, to reduce understeer for example, can I add front downforce (ie what worked to increase downforce that I am not now doing) or do I have to take some rear downforce off? In practice I expect that it will be a case of maximising front DF, and adjusting rear DF to balance the car, but who knows how it actually work out?

Now, that second sensor ... I was planning to buy another one off fleabay, but you seem to have identified a source that might also have string pots as well as the rotary ones - care to share, by PM if appropriate?

A final thought re the 5v source. My string pot will output a voltage proportional to the inputs, up to 32v iirc. If you datalog voltage, you could do a correction factor to eliminate voltage variations in the data. You have maths channels? Personally, I would not die in a ditch over it - the voltage readings should not vary much at speed, and if you casually monitor voltage between events or sessions/races you should pick up any variations due to system deterioration (assuming no operation of high draw electrical equipment, and an alternator in good nick).

Magpie's suggestion on drag is interesting, if that is a big concern for you as a subject matter. Again, at this stage I am not interested in drag per se, only car balance and car speed. Anecdote time: last outing I had massive understeer, and I was hitting Vmax (rev limiter) about 150m later than the previous outing there. This was because of high AoA that day, much higher than previously. Clearly the high AoA was generating high levels of drag, however my lap times were faster ... I was losing a little time in the later top speed, but the DF was giving me more car speed elsewhere. Sadly, that session I lost an endplate, so it was game over for the day, and no experimentation with AoA. If you are datalogging speed, you can also deduce relative drag from graphing speed on different laps (setups) on the same section of track, preferably on a calm or windless day.

[Cus]

First, I know, this post got way out of hand, I haven't been sleeping very well and woke up at 3am. This is what happens at 3am.

Your ECU or datalogger should give you a regulated 5v source, so that should eliminate the need for an additional component (aka failure point).
[..snip..]
A final thought re the 5v source. My string pot will output a voltage proportional to the inputs, up to 32v iirc. If you datalog voltage, you could do a correction factor to eliminate voltage variations in the data.

If it does give you a 5V output, use it - it'll give you the most accurate readings from the sensors, whatever they are - with the limited ADC knowledge I have, if the 'power' voltage varies at the same rate as the 'input' voltage due to headlights or whatever, it won't see as much of a difference. If you happen to be using the power supply magpie linked (which looks really good, but way overkill for the application at hand) then you'll start logging voltage difference in the car, as well as ride height changes.

If I use a 5v source then output would only be 0-2 volts... which means the sensitivity is way down. The logger can handle intermittent voltage spikes up to 400v so it may be able to handle 12v... I'll email the guys from autosports labs and find out. Alternatively can you get something that will cap voltage at 5v? Some sort of voltage regulator that wont affect the 0 - 5v range.

"Voltage spikes" could be reverse voltage, they're way easier to filter out than forward spikes.

Ask them the resolution or number of bits of the ADC (analog/digital converter) they're using, that determines the actual accuracy - eg; the arduino I'm using in my project has a 10 bit ADC and there's just over 400 steps between 0-2V (0.005V steps) which is still a lot of accuracy. ... i think. prepare for maths.

100mm lever will make a circle that is 628.32mm in circumfrence.
one degree of movement is 1.74mm, 139.62mm of total travel over the 80 degrees. (156.6mm for 90 deg)

At this point, I actually plugged a pot into the arduino and looked at the numbers instead of using "I think" statements.

Using 90 degrees of rotation on a 50k pot, I got these numbers out of the ADC:
Max: 893 (4.3646V)
Min: 540 (2.3693V)
Range: 353 (1.7253V)

The arduino gives me a number between 0 and 1024 that corresponds to 0 - 5V - the numbers in brackets are from a giant lookup table I have in a spreadsheet leftover from "the other project" - I used 90 degrees because I could line it up on the keyboard and be accurate, and the maths are slightly easier - I also didn't start at an end-point, because science.

I re-did my measurements a couple of times from different positions on the pot

Halfway: 520
Turn -90: 880 -> 360 "range"
Turn +90: 166 -> 354 "range"

Now, first thing to note is that the arduino's output wobbles by 2-3 in each direction, lots of reasons for this, all of them boring, but, it happens. I wrote some code that takes a sample every few ms, averages it with the previous 8 samples, and got the wobble down to +/- 1.

Anyway, so over the 90 degress of operation, we have 353 available steps meaning we have 3.9 steps per 1 degrees of rotation, 1 degree of movement is 1.74mm travel at the end of the lever, so we'll have an effective accuracy of 0.44mm!

From here the location of the sensor on the arm and some of magpie's suspension maths will give you excellent numbers to work with.

Actually, I'm not sure why Magpie suggests to double the numbers when working out the force - If you make one of these for each corner of the car, you're measuring the force on that corner - in that case the location of the end-point of your lever starts to affect accuracy more - measuring at the spring, you'd have (on 6Kg/mm springs) 2.64Kg/mm accuracy... that actually seems way less accurate now.

However, if it's a 12-bit ADC, there are 4096 steps in total.
1412 range; 15.6 steps/deg; 0.11mm/step; 660gm/mm accuracy

If it's a 15-bit ADC you're working with 32,768 steps.
11296 range; 125.51 steps/deg; 0.01mm/step; 80gm/mm accuracy.

So, it all depends on the resolution of the ADC being used. If it's a 10-bit ADC like I tested with, you'll log a change of less than half a millimeter, which is fairly accurate. Every extra bit of resolution doubles the number of samples per degree of rotation on the pot.

Alternatively can you get something that will cap voltage at 5v? Some sort of voltage regulator that wont affect the 0 - 5v range.

"I don't think so" - Most likely you can, but we're adding circuitry far more complex than we need, unless you're racing for big prize money, go with the above ;) even at 12 bits you're starting to get into logging the roughness of the track surface.

[Cus]

Update:

http://www.autosportlabs.com/racecapturepro-2/#specs

12-bit ADC. Does provide a 5V reference. You'll log with 0.11mm accuracy. *nods*

[Magpie]

I suggest double the spring rates if you are only using a sensor in the middle of the car. If you have a sensor on each side then no need.

[Cus]

That makes a lot of sense Magpie! I just decided "attached to suspension" translated to "making four sensors"

[Zabinskii]

do you have any specs on the motor? bore size, comp ratio?