#oggstreamer – arrives in Spain ;)

The OggStreamer arrived in Spain and the people from the Medialab Prado filmed a “How-To” for it. Makes me smile 🙂

 

#oggstreamer – OggStreamer @ LiWoLi 2012 (Linux Week Linz)

I have the chance to give a short presentation of the OggStreamer at the Linux Week Linz on Friday 25.5. So if you are interested and want to drop by I’ll give a 20-minutes short presentation around 18h30 at the “Stadtwerkstatt Saal”.

It is a great event, offering a number of interesting topics – you can check out the LiWoLi 2012 Program here.

#oggstreamer – 24h Streaming Test

Some days ago I installed the first OggStreamer at the Remotestudio of the FRS – “Free Radio Salzkammergut” – the regional Radio. And I put the OggStreamer to stream for 24 hours non-stop – and it worked without interruption. So I think  the next Live-Emission from the Remotestudio can be done using this Setup rather than the temporary Laptop-Setup they have used before.

Another OggStreamer shipped to Switzerland yesterday and it should arrive there early next week – I am quite curious how their reaction on this device will be.

The next OggStreamers from the first batch are already in the making –  the list of customers receiving a device form the first batch is already defined. Currently I am evaluating possibilities to start another production run of maybe around 30 Devices.  But before thinking that far – The device first has to pass the necessary EMC Compliance Tests …

Another issue what bothers me a little is the delivery time of XPortPro’s – I ordered 4 pcs. of them in the beginning of march and they are expected to be delivered in CW 21 – I am anyhow busy till CW 21 so nothing to worry about, but nevertheless it is quite a bit of time for such a small number.

By the way thanks to David Schleef for putting me on his website and on Planet GNOME – This blog received quite a number of hits via their websites.

#oggstreamer – jlGuiApplet replaces ffmp3

Sorry to say – but I’ve dropped support for ffmp3 – the main reason for this is that flash only processes audio at 44100Hz so streaming ogg and listening with ffmp3 wasn’t possible at 48000Hz (it was just play-backed at 44100Hz). So I needed another solution for this. Its name jlGuiApplet which implements an Winamp-style Ogg/MP3-Player as a Java-Applet and supports the 48000Hz sampling-rate as well. This software is almost 5 years old but it does the job.

As you can see on the ssh-terminal running on the XportPro – streaming  to 1 client produces around 1% CPU-Load at 128kbps. Also note that we are using more than 8,5 Megabytes of RAM, which means that older XPortPros with just 8Megabytes in RAM total, won’t be able to run the full OggStreamer-Software including the builtin sighttpd streaming-Server.

#oggstreamer – New Streaming Server sighttpd

I was trying really hard to get the IceCast-Server working under the XportPro-uclinux buildsystem but after more than a week of work I ended up in desperation. And I dropped the Idea of running an instance of IceCast-Server on the Device itselfs. – Of course support for upstreaming to an IceCast-Server still remains. That’s the bad news. But the good one is that I found a sleek little Streaming-Server, that just does the job as it should. It only took a few hours and the sighttpd – Server by Conrad Parker was up and running. First tests show that this Server should be able to serve around 20 Listeners directly (for really high bandwith vorbis; tested with 350kbps).

Building sighttpd is a great choice for embedded systems, as it only depends on LibOgg and LibOggz, whereas IceCast (at least) needs LibOgg, LibVorbis, LibShout, LibXML2, LibXSLT. May be it was the number of dependencies that complicated my  attempts to get IceCast running. In fact it compiled just fine – but for some reason in this configuration pthread_create just didn’t work as it should… sighttpd also uses pthreads – but no problem with this configuration.

So let’s stick with sighttpd 🙂

 

#oggstreamer – THT components soldering

I just soldered the remaining THT components for 3 OggStreamers, and I want to share the following picture with you:

For those of you, who are unfamiliar with the terms THT and SMT. I want to briefly explain the difference: THT (through hole technology) are components that are mounted by sticking the lead through a drilled hole. Where as SMT (surface mount technology) components are placed on the surface of the PCB-Board. In this project the soldering of the SMT was done by a Pick’N’Place machine, and the THT components were soldered by hand (by me 🙂 )

 

#oggstreamer – Rev 1.0b Schematics

We were quite busy testing the prototype and fixing all the issues for our next Version Rev 1.0b – the schematics are now online as PDFs and after the PCB manufacturing proves alright, I will release the PCAD and Gerber Files as well. For now the PDFs of the adapted schematics:

oggstreamer_REV1_0b_digital_and_power

oggstreamer_REV1_0b_analog_and_vs1063

 

#oggstreamer – Having fun with non-intrusive debugging

Hi out there – I found the problem with the buzzing noise, there was a short on the VDD_Analog and RCAP Pins on the LQFP Package of the VS1063. Now the thing sounds quite decent. It still needs some fine tuning. Help for this is on the Way, I will meet an Audio Engineer probably next week.

Meanwhile I put together the device to see if there are any mechanical problems – but Verficiation in 3D (with SketchUp) paid off and no problems when assembling the device – well there are some minor issues with the housing but those are basically translation errors to the punching machine.

And to see how the VS1063 performs with various bitrate settings I added support for statistics to the STM8 Firmware. With the help of the STM Studio Tool which allows non-intrusive debugging it is now possible to monitor the bitrate and buffer status in realtime. The bitrate shows the average bitrate in the last 1/4 second – whereas the buffer status shows more or less random the current bufferstatus – because the update rate via the STM-Link is limited.

Run 1: CBR Mode with 128000 bps

This shows that we have some Peaks with around 170000 bps and average around 130000bps.

Run 2: Quality Mode with q=10

Bitrate varies between 250000 and 400000 bps

Run 3: Quality Mode with q=1

Bitrate varies between 0 and 135000 bps

#oggstreamer – first recording + NOISE

Well the OggStreamer is working … basically. I recorded today samples of classical music and it turned out that there is a problem with background noise. If you want to listen you can download the sample here

would have been to easy … i guess 😦

#oggstreamer – measuring the input amplifier in detail

The input amplifier is now designed to roughly cover a range from -20dB to +20dB, which translates to an amplitude amplification of x0.1 to x10

Today the frequency response of the ampilifier using a standard PC-Soundcard and the Shareware Version of the audioTester Software. This is far from an accurate measurement but it gives me quite a good hint that we are on the right way. I already noticed before that the output of my PC-Soundcard is slightly unbalanced and this can be seen again in this measurement. I also want to note that I measured the frequency response of the Soundcard only as well and used the result as calibration data for the actual measurement.

With +20dB gain the frequency response looks like this (with groundlift activated)

Without groundlift (here we short the input Capacitors on the Ground line C24 and C28)  The spike on the left is unclear why but maybe due to an error in calibration data.

Results: In +20dB gain Mode we see a drop in gain at around 16kHz. Groundlift does not really effect the frequency response in the audible spectrum.

With -20dB gain the frequency response looks like this (with groundlift activated):

And without groundlift:

Results: with -20dB gain we see an unbalanced Situation with groundlift activated. And we further see an steady increase in gain at around 6 to 7 kHz.

In overall the Results look acceptable – I also ran an online radio station through the input amplifier today and listened to music and the Music sounded nice – it sounded a tick more crispier when ran through the input amp.

Together with Michi we will use a professional audio analyzer to do the measurements in the next weeks