Thread: How do I...

You are reading my mind. That is what I am thinking of doing. I am going to build it out of mdf first. And then transfer it to aluminum. I was in the one piece mode. I can use muggy weld on the joints. If you have never heard of that do a google on it. I used the real low temp stuff to fill holes. Works great.

Originally Posted by Dan2008

Don't bend it, machine it to fit.

I'm making my stuff from 2" billet aluminum and modeling it using, are you ready for this, cabinet/furniture design software! The software creates all the g-code output for the CNC so all I have to do is create a correct model.

I have the models machined in MDF until I get them perfect then commit them to billet. After the finished product is done I polish it to a mirror finish.

Here are some pics of the model for one of the taillight bezels. It has a step pocket on the front to hold the lens and a single pocket on the back for the wiring ti the LEDs. The front perimeter is also rounded-over.

Hi Dan,
Just reading through these notes, I know this is not releated to the forum, but just a quick question, what software are you using "Cabinet/Furniture design software"?
The reason I ask, is I built my own CNC machine that accepts G-Code.
I have used many CAD/CAM programs to produce models, but what I find is that these programs are "way over the top", I just wanted something basic and straight forward.

Many Thanks
Baz.

Better stick with the CAD/CAM software...the software I use outputs a CNC file only for a specific cabinet parts CNC machine.

Cabinet/furniture design software gets pretty expensive pretty fast. Most is in the $20,000/year range and the output is limited to a specific $250,000 CNC.

Another limiting factor is that most require you to be a licensed cabinet or furniture maker before they will license their software to you. The learning curve is also so steep that many require you to take a training course before you actually get the software. Training can run anywhere from $750 to well over $3000 depending on the software.

I'm thinking of rebuilding my dash, and making it symmetrical for better sound and imaging. I want to lose the factory cluster and was thinking of doing something like the guy with the chevy truck above.

Following this thread (and others today) I wish Babelfish had a translator for "Techie."

What I was wondering, tho, was if the output from the sensors is linear or not?
If the range is say 50-250ohms, (thus, a range of 200 ohms, is 100 ohms the center, or does it change as it approaches the top or bottom of the range?)

Thanks.

Jay

Depends on the sensor. That's a question for the manufacturer.

This thread has been a great help many years later. Like Dan, my vehicle (85 Honda motorcycle) has no OBD stuff. I have a shop manual and have values for 4 different coolant temps for the stock sender, as well as a range for empty and full on the fuel sender.

I made a calculator for what was discussed here in a form design program (I don't know how to use Excel), so like the cabinet CAD software I can't pass it along.

I made two versions- first with 12V as the input, and a second with 5V as the input.

The fuel sensor was easiest since it is linear. I was able to use a simple math formula to fill in what the values should be for 1/4, 1/2, and 3/4.

On a related note, I know the Arduino has a feature where you can input values and it will create 1023 data points (I think I am saying this correctly). Does the FB software do this as well? This would help to avoid having the "needle" jump from F to 3/4, and so on.

The coolant temp gauge was another issue. The 4 points ranged from 140F to 248F. When I looked at the differences, it was not linear (I haven't cracked the math for exponential senders). I'm thinking a GM 5V sender with a wider value range would work, or what I may do is add a Dallas 18B20 5V digital temp sensor (or equivalent) via a brass pipe plug.

I'll attach my 5V page for reference.

I use Excel for that, because it is really good at what it does. If I am not near a computer my TI-89 does the same thing just as well. No need to do fancy math unless you want to keep your statistic skills sharp.

For the temp sensor this is what I got:



So the equation of y = -53.367x + 365.42 is 99.79% accurate over that range which is pretty darned good for a simple linear equation. You will note that X is in volts, and Y is in temperature. So you can take any reading from the FB, put it in that equation, and out pops the exact temperature it is reading right then +/- 0.21%.

Now this is what I got for the fuel sender:



You say it is linear, but I say not. It looked "droopy" to me, which is a sign that a simple polynomial can be used. Instead of manually doing the taylor series, Excel does it with the click of a button. Linearly, the equation is y = -47.054x + 119.83 which is off by just under 3%. The polynomial equation of y = -13.093x2 - 12.346x + 104.34 is dead on with an Rsquared of 99.95%.

So if you read in 0.75v from the FB, it will be (-13.093 * (0.75^2)) - (12.346 * 0.75) + 104.34 = 87.716 which is 87% full and you would have a "needle" or whatever style show 87%. It would not jump between.

Here's the workbook too: http://www.fusioncontrolcentre.com//...sfer/bike.xlsx

Thanks for the math assist. I dreaded Algebra in school & it's always been a struggle for me. Looks like Excel is another thing to learn.