Building your own GPS data acquisition system

I know I've mentioned it here, before, but we have a parts list to make your own data box, capable of taking GPS and accelerometer/Gyro data.

The software intended is Race Chrono- http://www.racechrono.com/ which does a good job at plotting the GPS data really well, and a pretty good job at the rest of the data. Not the easiest thing to work with, but all things considered, I like it.

So the box consists of some basic things- microcontroller, a/d input, GPS+ antenna, 3 axis accelerometer, 2 axis gyros, SD card slot, and Bluetooth (optional). You also need a box and power supply....

Here's the list of parts so far:

From Parallax- Propeller Prototype board with USB- $29.99; 8 channel, 12 bit a/d- $5.22 (this one has limited range and power, which is ok for the first 8 channels); SD card adaptor- $14.99, and box $19.99.

From Sparkfun- Bluetooth module with board- $34.95; Venus GPS + SMA antenna- $62.90; integrated 3 axis accelerometer + 2 axis gyro- $44.95.

When you add it all up, we are a little over $200.

Somethings to work out- the Bluetooth module board is actually assembled from a $14.99 board, and we are not sure if we have to have the module, or if there are other choices. So that could save $20, but to be honest, you don't have to have bluetooth- it's really nice to have if you want to stream the data to a handheld. The first goal is to put all the data on an SD card.

Also, if you don't think you need the gyros, then that can save quite a bit, if you can put a simple x,y,z accelerometer on the board- which is only $10. But the gyros can be quite useful, in that they will tell you car angle, and then be able to correct the accelerometer data to be correct.

now, you don't HAVE to have the accelerometers. But if you try to take GPS data and calculate it, you'll find that it's not that good of data. While accurate to small distances, and capable of 10 hz, the speed spec on all of them make calculated accleration due to noise at 10hz really bad. And I do mean ALL GPS units. I've looked into the ones that most GPS speed units use, and they all have the same spec. More important, there are no real 20hz systems out there. They can output data at 20hz, but most of them interpolate everything but the 5hz data. Nothing wrong with that, but there are only so many companies who make GPS units, and they publish all the capabilities. Even ones who have different data streams do tell you that the data isn't pefect. It's just there. So we think the 10hz Venus choice is good.

Now, we are going to check over the list to see if we can make any other changes.

Oh, and if you have a question about this micro vs. another very popular one.... Yes, this one is 50% more expensive, but $10 isn't a lot. And the main reason we are choosing this Parallax Propeller is that the guy building it is familiar with programming it (and really important, Parallax has software modules to borrow, one of which will work perfectly with the gyro data). If you can do the same with a different board- go for it. All they really do is take the inputs, and then put that in a format that a software package can use- as long as they are capable of the speed and inputs- the micro type isn't really relevant.

I plan on buying the hard ware pretty soon- we had a decent working prototype, but I can't get in touch with the guy who I loadned it out to- we know it works. It's a matter of builduing it, and putting it on my Alfa.

Eric

Great... now combine this with a LCD screen and a radar/laser detector and I am a buyer ;)

I actually did this once - I used at atmega128 processor, 4x20 LCD, SD card (microDOS), 3 axis accelerometer chip and a serial port (USB/BT was not that inexpensive or easy at the time).

I had managed to get as far as logging the accel data and NMEA records to the SD card, displaying lateral G's to the display was in the middle of coding up the routines that would calculate the triggers to do lap timing when I discovered Harry's Lap Timer app for the iPhone. Buying a well packaged, pretty and proven reasonably accurate solution for $11 really took the wind out of my sails.

All in all it was a fun idea (and project). I learned a lot about calculating position and error correcting from GPS but... but today - my new Droid X has better, faster and more accurate hardware. Hell, its better than the $30k professional data collection systems of just a few years ago. Tons of RAM space and real APIs with floating point sprintf so I don't need to spend a week laying out my board components, soldering or trying to render analog input values to numeric strings in a tiny address space.

If it is a fun exercise - go for it knowing that $200 could buy the solution in several different ways.

Money aside, this is a really cool project. I'd sure like to hear more as you progress through build and testing.

My own foray into the do-it-yourself DAQ world, circa 1999, is here:

DAQ ancient history

alfadriver's system is FAR more sophisticated!

mike wrote: Money aside, this is a really cool project. I'd sure like to hear more as you progress through build and testing. My own foray into the do-it-yourself DAQ world, circa 1999, is here: DAQ ancient history alfadriver's system is FAR more sophisticated!

Mike- I didn't realize that this was you! I had one, too- but it wasn't good at taking RPM data- Steck and I tried to work it out, and we had some plans. But never got to it. This one goes a little farther. (BTW, I used your panhard bar design from the TI for my car. have to still put it in)

GPS- honestly, I have thought about using a smart phone for this. But that's all I would use it for, so the $200 + extra monthy charges would have to be figured in. More importantly, though, was that I need a/f, engine speed, throttle position, and engine temp to log in my Alfa. That was the bare minimum requirement- since those are the basic inputs to the mechanical injection. The driving dynamics was just an addition. Not cheap, though. And the 10hz GPS data is a lot better than the 1hz phones have.

Eric

alfadriver wrote: GPS- honestly, I have thought about using a smart phone for this. But that's all I would use it for, so the $200 + extra monthy charges would have to be figured in. More importantly, though, was that I need a/f, engine speed, throttle position, and engine temp to log in my Alfa. That was the bare minimum requirement- since those are the basic inputs to the mechanical injection. The driving dynamics was just an addition. Not cheap, though. And the 10hz GPS data is a lot better than the 1hz phones have. Eric

You can find unlocked Droids on ebay for under $100... you don't need the phone part so do not hook it up to a carrier. Just install the apps you want. They make OBD software for it too. Once you have all that - you might not need to write any code at all - but if you do - the Android API is pretty robust and its in java so the development environment, compiler, tools and emulator are all free. They are well documented also. Depending on the model - you get Accel, GPS, BT, USB, data storage, real display and WiFi. Lots and lots of cool options to input/output data. Frankly - I am gassed by the march of affordable technology.

I'm not trying to push you one way or the other - this is just a subject that I have played and paid attention to for a while. I'm interested in whatever you come up with so post updates.

Giant Purple Snorklewacker wrote:

alfadriver wrote: GPS- honestly, I have thought about using a smart phone for this. But that's all I would use it for, so the $200 + extra monthy charges would have to be figured in. More importantly, though, was that I need a/f, engine speed, throttle position, and engine temp to log in my Alfa. That was the bare minimum requirement- since those are the basic inputs to the mechanical injection. The driving dynamics was just an addition. Not cheap, though. And the 10hz GPS data is a lot better than the 1hz phones have. Eric

You can find unlocked Droids on ebay for under $100... you don't need the phone part so do not hook it up to a carrier. Just install the apps you want. They make OBD software for it too. Once you have all that - you might not need to write any code at all - but if you do - the Android API is pretty robust and its in java so the development environment, compiler, tools and emulator are all free. They are well documented also. Depending on the model - you get Accel, GPS, BT, USB, data storage, real display and WiFi. Lots and lots of cool options to input/output data. Frankly - I am gassed by the march of affordable technology. I'm not trying to push you one way or the other - this is just a subject that I have played and paid attention to for a while. I'm interested in whatever you come up with so post updates.

I still need a/d, which was the original need of starting. I don't know of any smart phone that can take a signal off of my coil, read an a/f meter, the throttle position sensor, and an engine temp sensor. The driving aid stuff came later when we saw that a high performance GPS was available....

I know that good smart phones make excellent driving aids, for sure.

There was one like this built on the Parallax platform already. Great sounding solution in both cases.

http://www.parallax.com/DAQPacAutomotiveDataLogger/tabid/849/Default.aspx

I think the builder of this is in your area and working on getting into automotive engineering. Might be worth looking him up and chatting about his project, code, etc.

A quick calc on having 1 of the 4-layer DAQPac boards made by Express PCB is about $154 and then there is another $200 for the components. Costs will drop if more are produced, but for the hobbyist, this should be a good way to start.

The only real trick is the use of surface mount components which complicates things a bit when doing the assembly.

Also, not sure if you've seen this yet:

http://intrepidcs.com/sensorspy/index.html

Interesting solution, you'd still need a way to transmit the data via WiFi/BT, etc. Still, its better than all of the OBDII solutions on the market, which are too expensive and not at all useful for those of us with old cars, or a need to monitor data that the OBD system currently doesn't.

In reply to turboswede:

We've looked at that project, a lot. I'm sure my coder will talk more to him as time goes on. And if the board wouldn't be so expensive, that would be a great option.

The one other issue is assembly. I should do more study on how the MS guys do it, but the intention for our project was some simple surface/wire assembly, and let professionals do the hard part. For instance, the integrated accelerometer/gyro cost $44, but is made up of a $25 gyro and a $9 accelerometer. So we pay a $10 premium to make it easy to work with. Same goes for the GPS.

What I'm hoping for is that we assmeble this, and get some suggestions or have some help laying out the basic board, and maybe cut the costs then. But even with all the parts at ~$200, it's already cheap.

(BTW, apparently you can do surface mounting in an oven.....)

I'll check the other part this evening while my wife watches Idol....

turboswede wrote: Also, not sure if you've seen this yet: http://intrepidcs.com/sensorspy/index.html Interesting solution, you'd still need a way to transmit the data via WiFi/BT, etc. Still, its better than all of the OBDII solutions on the market, which are too expensive and not at all useful for those of us with old cars, or a need to monitor data that the OBD system currently doesn't.

In theory, that would be able to read what we are building. But I know the team at RaceChrono are in the middle of writing the RC code for android phones. Probably including using android phone hardware.

Pretty cool.

Our goal is for two kinds of file taking- one with a mobile device (PDA, phone, whatever) via bluetooth, and the other to an SD card for post processing. RC is designed to show a live display while it's recording data, as well as some extra stuff to define locations, tracks, etc.

alfadriver wrote: More important, there are no real 20hz systems out there. They can output data at 20hz, but most of them interpolate everything but the 5hz data.

I had to deal with some data a while back that was output from a 1hz GPS unit ....and sampled at 1kHz. That was an annoying day of explaining why "It doesn't work like that".

This project looks interesting though. I haven't bought a Parallax product in 10 years or so, but I will admit the propeller boards looks cool. I'm looking forward to seeing everything put together.

alfadriver wrote: In reply to turboswede: We've looked at that project, a lot. I'm sure my coder will talk more to him as time goes on. And if the board wouldn't be so expensive, that would be a great option. The one other issue is assembly. I should do more study on how the MS guys do it, but the intention for our project was some simple surface/wire assembly, and let professionals do the hard part. For instance, the integrated accelerometer/gyro cost $44, but is made up of a $25 gyro and a $9 accelerometer. So we pay a $10 premium to make it easy to work with. Same goes for the GPS. What I'm hoping for is that we assemble this, and get some suggestions or have some help laying out the basic board, and maybe cut the costs then. But even with all the parts at ~$200, it's already cheap. (BTW, apparently you can do surface mounting in an oven.....) I'll check the other part this evening while my wife watches Idol....

Actually, for just one board, its only $154. To produce 5 would be $214 total. The part cost for a single board is about $208.48 (not including shipping since the parts are spread across several vendors). 5 would be $1042.40, etc.

Realistically, one could take the design and break it out as most of the major assemblies are available as individual boards that can be connected together. GPS, accelerometers, Wireless, Thermocouple, Parallax controller, etc. Buying pre-assembled "kits" and connecting them with a simple Parallax board is within the realm of the average person and seems to fall into what you're trying to do.

One question that no one has answered so far in all my searching on DAQ solutions, is simply, how does one capture the timing from an AMB transponder system? That could be a nice feature to add to the data set you might combine with your video using Race Render

Yep, I've looked at SMT stuff using ovens (many use toaster ovens)

turboswede wrote: Actually, for just one board, its only $154. To produce 5 would be $214 total. The part cost for a single board is about $208.48 (not including shipping since the parts are spread across several vendors). 5 would be $1042.40, etc. Realistically, one could take the design and break it out as most of the major assemblies are available as individual boards that can be connected together. GPS, accelerometers, Wireless, Thermocouple, Parallax controller, etc. Buying pre-assembled "kits" and connecting them with a simple Parallax board is within the realm of the average person and seems to fall into what you're trying to do.

Really, it comes down to ease of assembly. Nothing against MS, but our idea is to assemble a set of pre-made stuff. When you price out each component, you get a strong idea of the assembly cost. It's a trade off, for sure. Lets say we go with even our simplified set up- the cost would probably drop somewhere around $50- which would pay for the custom board. But, to me, the ease of putting it together is worth that.

And I'm sure there are a lot out there who would not think twice about assembling a pile of small parts- to save some money. Which is fine.

One question that no one has answered so far in all my searching on DAQ solutions, is simply, how does one capture the timing from an AMB transponder system? That could be a nice feature to add to the data set you might combine with your video using Race Render Yep, I've looked at SMT stuff using ovens (many use toaster ovens)

Unless the AMB system is willing to broadcast the information, the best you could do is use the line trigger to define a lap. Or use that as a GPS point to do that- RC is capable of that.

As for RaceRender, I'm pretty sure it will work to combine the data. One of the guys on the Miata board does this - http://www.mh-motorsports.com/lap_timers.shtml, and they do mention putting the GPS and video data together. I want to do that, too. Especially with a track map.

I'll probably order the parts pretty soon, and take pictures of it all.

alfadriver wrote: Mike- I didn't realize that this was you! I had one, too- but it wasn't good at taking RPM data- Steck and I tried to work it out, and we had some plans. But never got to it. This one goes a little farther. (BTW, I used your panhard bar design from the TI for my car. have to still put it in) Eric

yep, still kickin'. My solution to the RPM problem was a massive kludge, but it worked. I bought a $20 Sunpro tachometer from Pep Boys, and tapped into the freq counter downstream of the frequency-to-voltage converter. I calibrated the signal voltage with my old Fluke 88 multimeter plus spark plug wire clamp, and then mapped the signal in the DAQ application running on the Palm IIIxe. Still, the 3 Hz sampling rate was very limiting. At least the channels were 16-bit (or were they? it's been a long time).

Panhard bar design - can't take full credit, the basic design was put together by Gary (my dad) years and years ago.

alfadriver wrote: I still need a/d, which was the original need of starting. I don't know of any smart phone that can take a signal off of my coil, read an a/f meter, the throttle position sensor, and an engine temp sensor. The driving aid stuff came later when we saw that a high performance GPS was available.... I know that good smart phones make excellent driving aids, for sure.

If you are grabbing that from OBD it will be in ASCII already so all you need is the usb cable or BT dongle.

Giant Purple Snorklewacker wrote: If you are grabbing that from OBD it will be in ASCII already so all you need is the usb cable or BT dongle.

Yea, I've had a tough time trying to find the OBD port on my mechanically injected '74 Alfa....

(but you are right)

alfadriver wrote:

Giant Purple Snorklewacker wrote: If you are grabbing that from OBD it will be in ASCII already so all you need is the usb cable or BT dongle.

Yea, I've had a tough time trying to find the OBD port on my mechanically injected '74 Alfa.... (but you are right)

Yeah, on a '74 you are going to have to be creative. My old personal favorite for A/D is the atmega128 in DIP form. A little Vreg., and wire the UART to a BT transmitter. I'm sure sparkfun probably have development board all pre-loaded with that for $30 or so. Then all you need is the free ATMEL gcc toolchain and a programmer ($15 or DIY for less).

I am looking for a reason to buy one of these... and a 74 Alfa would be made much easier to instrument with a PC.... how sweet is this (if not a little pricey):

http://www.gumstix.com/store/catalog/product_info.php?products_id=250

In reply to Giant Purple Snorklewacker:

that's kind of the point- how in expensive can we assemble hardware to get 4 channels of A/D that I need, incoluding engine speed.

IMHO, with the components that we have, it can be done a lot cheaper than that project- off of my list, all I would need is the $30 micro, the $10 a/d, the $14 SD card adaptor, and even the $34 Bluetooth adaptor. So for well under $100, I can have 8 12 bit channels plus RPM. And I can output the format in a form that I can use later.... Log to a handheld to review immediately, or use a computer to look at the SD card later.

The driving dynamics part was added when we saw that it can be added for a pretty reasonable outlay. And it was simple to add.

I had an a/d similar to what Mike above had, and it wasn't capable of reading RPM at 10hz, since all it could do was count switches during that time. So at typical firing speeds, you'd get low, medium, and high- and that was it. The alternative was using a tach....

As far as I can tell, even with an A/D, you need a processor to make some sense of it, and then format it. There are some that will process the info directly to a serial line, but when you add the cost... well.

thinking off the top of my head, I bet you can do a decent S/D card a/d for about $50. http://www.adafruit.com/index.php?main_page=product_info&cPath=17_21&products_id=243 + a cheap a/d board, and you could even afford a $15 bluetooth chip if you know how to wire it.

(BTW, I don't want a PC in my car, and I would rather write to an SD card- looks better for a box)

Take a look at the data sheet for the ATMEL atmega line of micros. They are the brains to do your signal processing for not a lot of money. This whole setup was roughly $55. 75% of that cost is because I used a dev board from sparkfun pre-built with a 28 pin DIP socket and clean power because I'm lazy. You could make your own board layout and make similar for under $20. The whole toolchain to program the processor is free (google GNU GCC ATMEL)

They have a whole line of processors that are powerful, flexible and super cheap.

In reply to Giant Purple Snorklewacker:

The cool thing about Race Chrono- as long as you format the information it uses right, you can make whatever you want to generate the data. MS, propeller, ardunio, aMTEl- whatever. One of the big reasons I like it so much.

BTW, do you have a link to teh ATMEL stuff?

And, remember, the main reason we are using propeller is that the guy doing the hard work knows it. We can save some money by using something else- sure. But in the scheme of things- adding up all the parts, well- it's only a few bucks to keep the free labor happy.

that would be different if i could find a processor with USB, A/D, and maybe even SD, and bluetooth already installed. But that's hard to find.

Atmel AVR linky

After you look there - look at their ZigBee stuff for adding wireless/BT modules.

FYI: The Arduno uses an Atmel AVR processor. The Propeller uses a Parallax processor and is good stuff other than the somewhat higher start-up costs.