[POTM] dAISy - A Simple AIS Receiver

[chicken 630]

Hmm, signal strength is easy, I already have that information. Probably would need to be calibrated depending on the antenna, but it's a solution that avoids a GPS module and packet inspection. Not that they are expensive (starting at $15) or that power hungry (~ 50 mA), but certainly saves time in code development and system complexity.

[cowlum 1]

One consideration with strength of signal is I beleive there is Class-A and Class-B transievers. Class A being far more powerful so Class B may appear distant in comparison.

[PaulTech 2]

http://www.bosunsmate.org/ais/ has some info on the NMEA packet structure.

As for the sailboat, checking every 5 minutes for 1-2minutes for a new vessel within a packet may be all that is required.

I can't test this yet as I'm still building mine, had to buy a hot air rework station, solder paste etc as the filter component changes are way too small for a soldering iron. Anyway it is about time I moved into the 21st century.

[chicken 630]

Complete AIS protocol documentation can be found here: http://catb.org/gpsd/AIVDM.html

 

@@PaulTech my cheap hot air station saved my butt more than once, best investment ever.

 

In other news, as posted on other threads, I found a case for dAISy USB. It required redoing the board layout, but I really like the new form factor.

 

I also learned, that the LDO I use (MIC5205) does not like MLCCs as output capacitors. I had to go back to tantalum caps to get rid of a 200mVpp oscillation.

 

And I finally got a "real" AIS receiver off eBay. Turns out dAISy still has a long way to go to compete with these. But then, this particular model costs $450 new ($400 without GPS) and is about 7x the size (volume) of dAISy.

[PaulTech 2]

interesting that the commercial model doesn't support NMEA according to their site.

In what way does the dAISy have to catch up?

[chicken 630]

Not sure what NMEA they refer to. It sends AIVDM messages over old-fashioned serial, plus addtional messages about its own position.

 

Compared to dAiSy it seems more sensitive. With the same antenna, about 5 times more messages are received and range is about double. The SR162 is much better at receiving targets that are not in direct sight, e.g. behind hills.

[PaulTech 2]

double the range is aout a 6db better sensitivity, so a MMIC at the antenna end or FET amp would easily get this result. The MMIC have the advantage of being very easy to work with. Even if the was a bit of filtering before the amp it should still get about an extra 10db improvement.

[chicken 630]

Thank you for your input @@PaulTech. My current suspicion is, that noise and/or selectivity play the bigger role than sensitivity:

• @@gmtii's station typically covers around 50 NM, and up to 500 NM when weather cooperates
• when adding a 162MHz, 14MHz bandpass filter, local reception improves (at the loss of fewer outliers beyond 100 NM, which I attribute to the 1.9dB insertion loss)
• even for ships clearly within range of dAISy, the professional AIS receiver is much better at picking up the (2.5x longer) static and voyage related data messages

I already work on adding a bandpass filter to the PCB, so maybe I should just add an amplifier and see what happens. Any pointers to a suitable MMIC, and/or articles for beginners about the topic?

[PaulTech 2]

Try the minicircuits site for MMICs as well, the filter looks like an easy solution as it would stop a lot of interference that the onboard filter just isn't designed to deal with.

http://www.minicircuits.com/products/Amplifiers_what_is_new.shtml

you really don't need much bandwidth so a MMIC that goes to 1GHz would work. I still have lots of the first gen MMICs as I had to buy a whole strip as they were special order way back. Even these old MMICs are very very handy, you can mount at antennas very easily. Just check the noise figure, as some can be a bit noisey and when you amplify the signal by 15db ,you also amplify the noise by 15db + amplifier noise.

[PaulTech 2]

just had another quick read thru the Si4463 doc, and noticed the channel filter bandwidth is configurable from 850KHz to 1.5KHz wide. Too narrow or too wide will cause problems too.

[chicken 630]

I set RX BW to 25kHz (see WDS screenshot in first post). I experimented with lower values (18.5 and 15kHz) but only noticed negative effects.

 

That being said, with my home being out of AIS range I didn't do thorough and systematic testing.

[Lgbeno 189]

I agree, I think that a LNA MMIC could make a big difference in range. Avago also has some good options. Mini circuits is cool though because you can get connecterized modules and plug them right in for testing.

[PaulTech 2]

MMICs are super easy to use at 162MHz, they're designeg to have 50ohm in &out, they just need a little power. Deadbug on a tiny bit of PCB.

the channel spacing for AIS is 25KHz and so the filter could be set a little big bigger 30-40KHz, not to much or more noise will get in , but a little wider to allow for any transmitters which either be drifting or have the modulation set a little too high. As you noticed too narrow doesn't work too well.

[chicken 630]

the channel spacing for AIS is 25KHz and so the filter could be set a little big bigger 30-40KHz, not to much or more noise will get in , but a little wider to allow for any transmitters which either be drifting or have the modulation set a little too high. As you noticed too narrow doesn't work too well.

This made sense to me, and I ran a few test today with 30 and 40kHz. The results are, that 25kHz seems to be the sweet spot.

 

I also found an app note about IR calibration for the 446x (AN790 http://www.silabs.com/Support%20Documents/TechnicalDocs/AN790.pdf), and it turns out I did it wrong. From my unscientific tests I think doing it right reduced the error rate a bit and extended range.

 

I also no longer put the MSP430 into low power mode, as I observed a minor (<10mV) ripple on the power rail when the CPU woke up. Again, I imagine a slight increase in successfully received messages.

 

I will push these two changes to Github shortly.

 

The MMIC is next in line, but needs a bit more preparation.. probably won't happen before thanksgiving.

 

Thanks all for your input.

[chicken 630]

I built up and tested the new revision with MMIC amplifier and bandpass filter (direct coupled resonator bandpass filter, ~6 MHz bandwidth). Still not up there with the professional receiver, but definitely an improvement. Thanks again to everyone for your input!

 

I put the amplifier after the bandpass filter:

 

But thinking about it now, putting the amplifier before the bandpass might have been better. As it is now, I amplify noise introduced by the bandpass and might also lose some signals to the attenuation of the bandpass (about -5dB according to LTSpice).

 

Is this correct, or is the impact of the order negligible?