Electronics help please

[Boostinjdm]

I have a customer that seems to have too much time on his hands. He is regularly bringing me ideas he wants built. Many of these ideas are presented to me as a concept and I am left to work out the details. Some come as prototypes he has rigged up and I am asked to refine them. Quite a few ideas get shot down either for not being cost effective, or because somebody has beaten us to it. He does not seem to be interested in developing the ideas or becoming rich off of them. He strikes me as wanting to be an end user of the product.

I see his latest idea as having real potential. It could change the way many people operate, as well has creating some new businesses based around the data collected. I'm not going to explain exactly what it is yet. I have not found any evidence of anybody testing or selling a system that does what we discussed. So I don't want to let the cat out of the bag. I don't think the exact purpose is needed to get the help that I need.

I'll see if I can describe what I'm wanting to do as simply as possible.

I would like to take data collected from sensors and plot them on a map.
I believe that accelerometers would be best suited.
There will be anywhere from 2 to 10 of these mounted on a piece of equipment moving at a steady speed. I need to plot impacts to these sensors and their severity on a map for later reference.
The sensors need to be durable, weather proof, and fairly easily mounted.

I envision about 7 sensors mounted to the machine. The sensors are linked to a "brain box" by simple cables up to about 15' long. This brain box will have GPS capabilities and either store the data or transmit it to something like an IPAD or laptop. Bonus points if the individual sensors can be located with GPS (for more accurate plotting).

I know I'm out of my area of expertise. Just where do I begin?

[darryl]

I don't think the sensors will be a problem. They can probably be off-the-shelf parts, and may be available in mountable units that are sealed. If not, a suitable mount can be made and the parts wired and encapsulated in the mount. That length of cable would not likely be a problem unless you're needing to see very fast response times with little distortion of the waveforms.

Do the sensors need to be located via GPS, or can they be mapped to a representation of the machine on a computer screen? It almost seems to me that there's a complication here that really might not be required. In almost every development, the KISS principle applies. Keep it simple.

As I see it, you'll first need to know how fast the sensors must operate, as this will determine much of the circuitry and the mechanicals. Are the impacts going to be thuds and thumps, or sharp and short-duration pulses?

[fixerdave]

Well, you can get a cheap Android tablet from China that has a GPS and an accelerometer built into it, and wifi. For that matter, you could get one that has all that and a cell radio if you want it to call home. They would be as durable as the box you bought to put them in. Throw in a wireless access point so they can talk to each other and you've got your hardware. Beyond that, you just need software, somebody to write the app to get it all running together. One side note to this approach... getting it to all run from power-up unattended might be difficult. Probably not impossible, but something to consider... how will said app start?

Another route is to prototype in Arduino for the sensors... there are various ways they can talk with each other, depending on requirements like electrical noise immunity. All kinds of sensors are easy to hook up to them. You can even bluetooth to a tablet fairly easily. If you go this route, at least you have a mass-production option with cheaper hw... you can just use chips rather than the Arduino boards.

Or, you can skip the Arduino step and go straight to a bare microcontroller with sensors. Harder up-front design, but cheaper end product.

It all depends on how many you want to make and how much effort you want to put into development.

Oh, and as an aside:

If you want high-accuracy for GPS, you can get very fancy and expensive... high gain antennas using differential GPS can get surprisingly accurate. It just costs more.

"Simple cables" may not be so simple. If you are in an electrically noisy environment then there's more to it than just wiring the stuff together. There are various approaches and you will need to pick one.

Good luck with your project... hope this helps,

David...

[Boostinjdm]

I think you guys are on the right track. Let's see if I can try to clarify what I'm wanting to do with a similar concept.

Imagine a car traveling traveling at a fixed speed. A sensor mounted to each rear control arm. I want to plot on a map every pot hole this car drives over and how deep (roughly) it is. I then want to be able to give this map to somebody else to go fix the pot holes. The road has a heavy layer of leaves on it so neither the driver, nor the guy sent to do the repairs can see it without manually clearing the leaves. The guy sent to do the repairs also needs to know if the pot hole is too small to worry about, needs a shovel worth of filler, or needs to bring in a backhoe without ever seeing the hole.

The equipment should not be electrically noisy. Cost isn't a major consideration at this point. Lower is better. Somewhere between $1000 and $10,000 would be manageable if the system does what we want. Simple is good.

If we could rig something up in the few hundred dollar range to test the concept, that would help open up the funding for a more proper set up.

[PStechPaul]

Just having finished a homework "class" on surface measurement, what you propose is similar to a profilometer, which measures surface roughness of a machined part. but in this case it is the road. You could fairly easily rig up a travel indicator of sorts in the shock absorber, using the principles of an LVDT (Linear Variable Differential Transformer) which can provide a signal corresponding to the vertical movement of the wheel, and that can be plotted against the travel of the vehicle along the road to provide a profile. If this is in a vehicle that is specifically used to identify and quantify potholes, the operator could click a button when it was encountered so that the information immediately before and after the pothole can be flagged or stored separately for later analysis. This might also trigger a camera in the rear of the vehicle to get an image of the pothole that was just encountered, and there could even be one mounted under the vehicle with appropriate lighting taking video clips of the pothole as the vehicle passes over it. Assuming the front wheel hits it first, that could be a signal to start the video clip to get an image from various angles.

[Boostinjdm]

A travel indicator will be unreliable in my application, but the shock from hitting the pothole can be depended on. The operator is paid minimum wage and cannot be relied on to do anything other than drive. Video is not going to work because the "potholes" are hidden by "leaves". This is not a dedicated vehicle, but one that is driven down this road anyways and has had this "pothole locating system" added with minimal modification.

[topct]

It will soon be that time of year here when there will be thousands of pot hole detectors driving around on leaf covered streets. Simply listening to the answering machine at the street dept will give exact locations and severity. Locating the holes is not the problem. Getting the street dept to fix them is.

[J Tiers]

The data may have to be "reduced" by processing later to identify the start and end of each pothole. That will assess the size, at last in that direction of travel. Correlating left and right would locate huge ones, trenches, etc.

The location will not be to "inches", but may be pretty good, GPS security was rolled back a while back, increasing consumer GPS accuracy. I don't know if Homeland security has reversed that yet.

It is quite a "do-able" thing. the railroads use that sort of deal already for tracks.

[MrSleepy]

Piezo Discs : The type they use on electronic drum machines. Robust..Simple wiring requirements and easy to interface to mcus etc.

Rob

[browne92]

How accurate do you need your GPS points to be? I think the one in my phone is only good to about 30 feet. I routinely use one at work that measures "sub-meter", or within one meter. I understand there are "sub-decimeter" units, they tend go get spendy.

[jlevie]

Imagine a car traveling traveling at a fixed speed. A sensor mounted to each rear control arm. I want to plot on a map every pot hole this car drives over and how deep (roughly) it is. I then want to be able to give this map to somebody else to go fix the pot holes. The road has a heavy layer of leaves on it so neither the driver, nor the guy sent to do the repairs can see it without manually clearing the leaves. The guy sent to do the repairs also needs to know if the pot hole is too small to worry about, needs a shovel worth of filler, or needs to bring in a backhoe without ever seeing the hole.

For that I'd think you want a travel sensor on each shock, a three axis accelerometer, and GPS.

[fixerdave]

How accurate do you need your GPS points to be? I think the one in my phone is only good to about 30 feet. I routinely use one at work that measures "sub-meter", or within one meter. I understand there are "sub-decimeter" units, they tend go get spendy.

Yeah, you can get stupid accurate with differential GPS... it uses the same signals but processes them differently. It's been a while but getting to that accuracy used to take time, as in minutes per fix. Not something you could do while driving down the road. There are other ways, especially if the route is fixed, say plowing a field or something, so the GPS units can start at a known location and then calculate/deal with the error. It all depends on how much money you want to throw at it.

However, the OP does need to understand that programming will likely be the biggest cost, not hardware. The stuff he's talking about doing is pretty much off the shelf gear. It's the code that ties them together that will cost. The cheaper the hardware, the more coding time will be needed. The more polished the output, the more coding time as well. Then there's up-time... is it going to be an hours setup, run for a few hours, then spend a week processing the collected data? Not too hard. Is it always-on, processing data on the fly and automatically calling it in via cell, for years? A lot harder.

For example... Get 5 cheap tablets (one for GPS) and mounting boxes, turn them on, start the app, bolt in place, and then go for a drive. At the end of the drive, gather up the boxes, go home, download the data and start processing it, using the GPS clock to sync everything. Manually correct the GPS location on your map and send the results to the pothole crew. Lots of work each time, but not that hard to develop. Maybe that's okay if it's only a few times per year. But, if it's an always-on thing doing automated reports... the above hardware would be a pretty stupid way to go as it won't last a week. The hardware needed for alway-on will probably end up cheaper but the development time to make it all work will be substantially more. Same goes for the end report, if it's to be automatically generated instead of manual, then there will be a lot of up-front development time. Programming is expensive; it sucks a lot of time and people like getting paid. Finding the right person to do it isn't exactly easy either, especially if you're trying to keep a secret,

David...

[MrSleepy]

I bet you could almost do this with off the shelf components from racing orientated companies.
Aculux make this LASER sensor 3 - 8 ins detect at 1000 samples a second. (1 inch spacing resolution at 60mph approx)

http://www.aculux.com/pdf/Aculux%20Ride%20Height%20Sensor%20product%20sheet.pdf

A bar with an array of these sensors mounted underneath the car maybe (and debri protected).

McLaren Electronic Systems also provide a wide range of interface - datalogging -software (Matlab compliant) with graphing facilities (Their XP based Embedded PC Loggers and Data recorders look fit for purpose).

http://www.mclarenelectronics.com/Content/Catalogue/ProductCatalogue2014-rev.0013.pdf

Probably spendy...

Rob

[Mike Burch]

This is a bit like charting the seabed using only a lead line, as our ancestors did. The measurement is very accurate, but only for an area about 2" round, the size of the bottom of the lead. In your application, you'd be measuring only what the tyres touch, which is a small fraction of the road surface. The opportunity of missing significant potholes is not zero.
Uncorrected GPS is accurate to only about ten metres (33') at the 95% confidence level in the horizontal plane. Differential correction can get seriously expensive.

[Forestgnome]

Take a look at this site: https://www.sparkfun.com/

They have lots of sensors and "how to" info. You could link a GPS module to an Arduino that would also read the sensors. The Arduino could report to a data repository using wifi. They show how to do that as well.