Hey Anthony, I'm only scratching the surface here, of the GIT, but I imagine theres some interesting code in there. What a great 30th birthday present...! I can understand the considerations on the USB audio side of things, after Jim highlighted the problems with the USB, I felt a little despondent, and then thought "Well, what about chan_voter?" To that end, I purchased an RTCM and started playing with it. I still haven't powered it up - thanks Lithgow Leading Edge Electronics for NOT getting my parts in... eBay it is... I'm planning on using a Cubieboard (www.cubieboard.org) with this, faster CPU, more on chip USB and separate ethernet channel, lots more I/O, SATA, etc etc. Have to see how it goes, I'm taking it... On Sun, Jan 20, 2013 at 1:45 PM, Anthony Percy <anthcp@gmail.com> wrote:
Hi Shane, I have also uploaded the ARM Raspberry PI version of the allstar code as well which did include chan_voter...at https://github.com/anthcp/Arch-ARM-RAS-PI-Allstar I hope it helps. The compile process is the same as my previous email.
Just to to let you know, I found the sound on chan_usbradio with a CM108 usb fob was distorted for about 20% of the time when I did the *99 test tone on the PI. It could be acceptable on voice but I didn't test it any further.
I then tried to use chan_alsaradio but I think the CPU overhead was too high due to the downsampling from 44.1hz (?) to 8Khz. I have ordered a Griffin iMIC2 usb sound card which has hardware 8Khz sampling built in to see if it will work OK on the PI.( still waiting for it) If it does then it will mean that we will have to use the PI GPIO's for PTT and COR etc but that shouldn't be to hard.....(hopefully :-) )
Regards
Anthony, VK2ACP
Date: Fri, 18 Jan 2013 09:11:29 +1100 From: Shane Morris <edgecomberts@gmail.com> To: Eric Guth <eric@efratnetworks.com> Cc: app_rpt mailing list <app_rpt-users@ohnosec.org> Subject: Re: [App_rpt-users] Arch Linux version of allstar has been stable for 2 weeks now.... Message-ID: <CANZw+5eg7T3OmjMVJquwFhZLJ3-yWo9iY-hDeBtTzgJWYdeYdw@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1
Interesting!
Certainly smaller than my FitPC, but you knew that, cheaper too. Its a single core 1GHz x86, whereas my FitPC is a dual core 1GHz. I imagine its all passively cooled Eric?
I was planning to run Allstar on a single core 1GHz ARM at some point in the future, Anthony mentioned in his first post two weeks ago he hopes to port the software to RaspberryPi (not to bring up the debate again, I'm only mentioning it for completeness - my goal is to run chan_voter, which DOESN'T need USB) and I think it would be simple to port to other ARM platforms, for the sake of argument, the Beagle Board, running newer kernel revisions.
My FitPC does get quite warm Eric, even under "no load" conditions. I'm not entirely certain what it would do in 35+ degree Celsius heat in the middle of the Australian bush. I hope "work fine," not "explode"! So that might be a consideration for a totally passively cooled unit.
My unit is going to be used to conduct a RTCM network, using chan_voter. None of the USB interfaces will be used, say, for URIs or other sound fobs. It'll be connected to the remote stations by long shot wifi links (Ubiquiti Bullet2s and Loco2s over ethernet).
Eric, if you're going to use the USB ports on the 3350-AP, have a look at the structure of the USB hubs inside the computer. That'll tell you alot about whether you can use the sound fob with other devices, or you'll run into issues like the RPi. USB performance is critical, and I'm uncertain what the "on chip" USB ports would be like for such an integrated chipset.
I hope this helps - technically, it should run it, from a "raw performance" standpoint. However, the practicality might be a little different. I do hope it works - please let me know how you go, I'm rather interested! That is a cheaper computer than the FitPC!