[aprssig] Quick APRS hardware survey (APRS Hotspot)

[Scott Miller]

> Could you expand on this?  I did the calculations, and a reasonable SD card should have enough write cycles for several years.  Why would the system need to be read-only?

It's not really about the write endurance.  With a writeable file 
system, you always run the risk of corruption due to software problems, 
power failures, and so on.

All of the things I'm trying to do, you could do with a Raspberry Pi and 
a suitable add-on board to handle the radio interfaces and such.  It'd 
mean a lot less development effort, but also vastly greater system 
complexity overall.

The SR2's power consumption is a lot lower because it doesn't have all 
of the overhead of a bunch of dynamic RAM, a GPU, and a full-fledged 
operating system.  It boots in a fraction of a second, and can be on the 
network inside of 10 seconds.  The file system can be corrupted or 
totally wiped and it'll still boot and retain your settings.  Software 
updates are monolithic - when it's time to update it needs one file, 
about 2 MB max.

It's not prone to memory leaks because there's almost no dynamic memory 
allocation.  It won't ever get infected with a worm.  You don't have to 
worry about the processor load ever getting so high that it can't keep 
up with the radio channels because the resources for everything are 
basically reserved in advance.  There's also a hardware watchdog monitor 
that makes sure that if the software safety checks on things like 
transmitter timeouts fail to run for more than a quarter of a second, 
the PTTs are all disabled by an external circuit.

The down side is that it's a lot of work for *me*.  I've had to do stuff 
like write a super-lightweight telnet client and server from scratch.  
It has a screen-oriented text editor with syntax highlighting that 
occupies less than 5k of code space.  This is stuff you'd never have to 
worry about on a *nix box, because all of that code was written 30 or 40 
years ago.  I kind of enjoy that sort of programming, though.  Whether 
it'll turn out to be worth it commercially remains to be seen.

> FWIW, at Burlington Amateur Radio Club, we’ve had a couple of Pis at our remote location (one for APRS, one for D-Star) for about a year now, with few actual Pi-related problems.  We did have to replace a cheap SD card with a name-brand card, and we’re currently having trouble with a cheap Chinese USB-Serial interface.  In general, though, the Pi’s have been pretty solid.
Something like this is never going to replace single-board computers for 
everyone.  It's intended for places you don't want or need to put in 
that kind of effort.  It'll work as an appliance for common use cases, 
and I'm trying to make the BASIC interpreter powerful enough to easily 
handle most custom application requirements with small amounts of easy 
to learn code.  If for some reason you want to wait for a particular 
APRS station to come within a pre-defined area and then play a voice 
message on one radio, send an APRS message on another with some sensor 
readings, turn on a relay, send a text message via Twilio using their 
REST API, and establish an Echolink connection with a particular node, 
it'll do that.

You could do all of that on a Pi, of course.  But unless someone wants 
to develop and support a software distribution to do all of that, and 
the interface hardware to go with it, it's going to be a lot of effort 
to put it together.

Ideally you'd all buy a bazillion of these things so we could get the 
cost down to the Raspberry Pi range, but even if that doesn't happen I 
can still beat the price of a Pi and Rigblaster.

Scott
N1VG