ROV SONAR

[rossrov]

Been working on a primitive 2-dimensional SONAR that can be fitted to my as-yet-not-in-existence ROV to help with navigation and help in finding objects of interest. The plan is to show a small SONAR plot alongside the camera window. Move into camera visibility range with the SONAR, and then navigate/inspect visually via the camera.

The SONAR is designed around:
A modified (butchered) Eagle Magna 2 plus fishfinder and transducer
An EtherTen (Arduino Uno-like board that has built-in Ethernet) to get signals from the fish-finder and send collected echo data the surface
Netbook PC running the Google Chrome browser that can do graphics using HTML5 canvas
A stepper motor to "scan" the transducer over a given arc

Description of Operation:
Two inputs of the EtherTen are connected to the FF PCB via some resistors, a capacitor and a diode. One is an interrupt that lets the EtherTen's microcontroller know when a "ping" has been sent out by the FF. The other is one of the analog inputs that samples any echoes from the ping (reflections from objects). The EtherTen then sends data generated from the above process to the surface PC and displays in graphical form, much like a ship's RADAR or many commercial SONARs.

So far so good:
I was concerned that the A to D conversion time of the microcontroller may not be fast enough to capture echoes, but testing in a 2 metre tall/deep PVC pipe has given good results. What the results will be from open water and greater distance (weaker signals) remains to be seen. Calculating roughly, I think the conversion time gives about 80mm of distance resolution.

Copied then modified the code from http://startingelectronics.com/tutorial ... eb-server/ - very useful and well presented tutorial. Not the fastest way to send data, but I hope to be able to update a full 120 degree sweep once per second using this method

Display resolution at present is 16 arc segments by 64 range segments - mostly limited by the small amount of SRAM available in the EtherTen (same micro as Arduino Uno) and the way data is sent to the browser

Next job is to get some basic mechanical hardware built and do a salt-water test.

[sthone]

This sounds very interesting, well beyond my abilities but interesting none the less. Can't wait to see what happens.

I just took one of those finders off my boat this year too.

-Steve

[rossrov]

Thanks Steve. Easier ways to do it but I want to have all control, compass, depth, SONAR and video running over Ethernet. Been taking some pics and making notes so if it looks like working will post those and some video. Being an Electronics Tech and a boat nut makes the ROV "hobby" and this Forum a good fit for me.

Ross

[a_shorething]

I'm watching this thread with interest. Sounds very cool...

[rossrov]

Cheers a_shorething.

Did some testing yesterday. Rough setup at the moment but I'm pretty happy considering it's only first try outside of the drainpipe "test-tube". Really need a workshop with ocean access. Working on that.
I'm finding that the transducer beamwidth or "cone angle" seems too great. Either that or the display needs better gain control, which I plan to implement later anyway. The Fish-Finder manufacturers state cone angles but I tend to think now that the "beam" isn't actually cone shaped, but more of a flattened cone of elliptical cross-section. (We're getting into geeky waters now!) This makes sense when one thinks what a conventional fish-finder transducer needs to do. A wide swathe athwartships and narrow fore-and-aft. If this is the case then I have placed the transducer the wrong way with regards to the rotation mechanism. Next test will re-attach in housing 90 degrees and see if that reduces the azimuth beamwidth. Looking at the moulded (lens) shape of the transducer makes me hopeful this will improve things.



Some pics and rough video. A bit windy so had to jam the PC sun-shade (cardboard box) with PC in it under a table so wouldn't blow away. No stepper motor yet so used an old servo to rotate the transducer, but found the gears stripped. Connected the servo's pot to another A to D input on the Arduino and just turned the transducer shaft manually. Linear pot so angles accurate enough...The first few seconds of video shows a reflection which is probably the piling at the end of the pontoon 5-6 metres away. Then the owner of the pontoon puts the bucket in the water and it shows up closer in.

[Zaibach]

Cool beans, there are so few people out there messing with Sonar on a hobby level. Can't wait to see more of your project

[Oldsirhippy]

Your experiments are very interesting and I look forward to seeing your results.
I too am hoping to build a sonar system in time, when I have completed my ROV. To help with this I bought Mark Thompson's PDF book called 'Home built Sonar'. Whilst he uses a specific type of display, I hope to use the front end of his design and adapt it to feed an Arduino and LCD.
Mark designed a sonar array and I think an array narrows the sonar beam width, but I don't know why that is. The results he achieved enabled him to locate wrecks that had been lost, so his system is quite successful.
I sourced the sonar transducer, they cost about $9.50 each from a Chinese company. The transducers are used in ultra sonic cleaners, probably why they are low cost. Mark designed his array for 6 transducers.
The design uses low frequency transducers. Mark said that higher frequencies mean better resolution but more difficulty in the design.

[rossrov]

Thanks Zaibach. (edit: and SirHippy just saw your reply) Since last post made a new mounting so I could rotate the transducer. Also, the transducer now has a piece of expanded polystyrene behind it (what would be the top in normal fish-finder use) in an attempt to reduce any sensitivity from that direction.
With transducer rotated, I tested again at same location on the pontoon. The bucket showed up narrower than in the video, however there was a stronger reflection from the bottom. This makes sense with the transducer now rotated. The beam now looks to be wider vertically and narrower horizontally, as hoped.
On first re-testing, the transducer was about the same distance from the bottom as it was to the bucket. This cause me not inconsiderable alarm and some confusion at first, because on lifting the bucket out there was still a good echo the same distance out as the bucket. Moving the transducer assembly further into the water and so closer to the bottom separated the two echoes. Relief!
Noticed more "random" echoes than in the videoed test. I'm putting those down to small waves on the water's surface or reflections from the bottom. The transducer assembly I was holding in the water, not clamped with respect to the bottom, so movement would have been affecting things.
Probably any further testing will be with a stepper motor, not fingers, doing the scanning, deeper water, PC set up out of sun ideally inside a boat's cabin, fish-finder gain controls more easily accessible etc etc. Would like to try a transducer with a flat face, or file the current one flat to see if that narrows the vertical beamwidth so reducing reflections from the bottom or waves.

Pic transducer rotated:

[rossrov]

OldSirHippy: Thanks for info, especially the cleaner transducers. I had a quick look for the book and author. Could you post the link to where you purchased the book please? I'd be keen to do a similar setup with the phased array of transducers, but that would be a bigger project than what i can take on at the moment. Certainly I can see myself looking at doing that sort of thing in the future though.

Your mention of 6 transducers and finding shipwrecks makes me think that Mark's book is describing side-scan. Phased arrays are also used in forward-looking SONAR, recreational boating units are becoming available and much cheaper than anything previous, but I'd still place them in the "rich kid's" category. The way these work is simple enough in theory, but there is a fair bit of signal processing involved to make them work at all. Fast A to D converters (faster than those in an Arduino at least), digital signal processing.... Though you could do it without the DSP.

Because I'm thinking ROV use, radial scan is the choice, rather than side-scan, and I'd really like to be able to do continuous 360 scans. To keep this as compact and as simple as possible, and given that not alot of image detail is required, mechanical scanning using the one transducer or a small array looks like the go.

Arduino's A to D converter I roughly calculated to be able to give a minimum of 80mm distance resolution. Do not know whether this would be low enough/fast enough for adequate side-scan detail. Also, would be good if the EtherTen (same chip as Uno) had more SRAM

[Oldsirhippy]

Mark's web site is at http://www.mbtelectronics.com/side-scan-sonar.php

You're right it is a side scan sonar.

Do you think you could rotate the array in order to get a tighter beam?