Theo Verelst Local Diary Page 46


I've ditched the usual header for the moment, I think it doesn't help much anyhow.
This page is copyrighted by me, and may be read and transfered by any means only as a whole and including the references to me. I guess thats normal.

Home (alt)   Tripod Home   Local Diary list   Tripod Diary List   Previous Diary

This page is under contruction, so check back later, too.

  Fri Mar 25, 13:52, 2008


 

Software radio, what´s that

Well in essence a radio or TV tuner of which major parts work in software instead of in hardware electronics, altough of course the antenna and some important electronics like mixers and AD converters are still essential, but the signal detection and a part of the filtering is done in software, and even in programmable logic.

Because computers are getting so much faster and AD converters up to high sampling rates so much better and cheaper, it can pay to do software decoding of even live Analog to Digital converted signals, using other methods than are available in must ´hardware´ radio´s electronic signal path. Not necessarily better, but for instance for HAMs it is certainly interesting to change a radios´ reception method and experiment with various signal processing possiblities, and some signal processing possibilities which are very hard tomake in hardware or even impossible, like those requiring long delays and for instance FFT based filtering and decoding loops.

A little ssb radio experiment for little radio amatures ? Ehm, no.

This example is made by somebody actually decoding High Definition Television with a software radio setup!:



(I brighened the image, and shrunk it)

It comes from the gnuradio site, where a wholly open source software radio setup is described even with hardware with all the parts and pcbs available.

It is possible to internet-buy the hardware required for course tuning, AD conversion and USB interfacing from Ettus , see GettingStarted. It isn´t that expensive for what it is, an probably the whole is terribly cheap, but getting started with the hardware would cost apart from a PC about $700 and then some. Otherwise I´d probably try quickly to get acces to this and use the hardware also for oscilloscope and spectrum analyser (that´s electronicist equipment)

Would that work? Well, I´d as it is probably need the Universal Software Radio Peripheral to find out, but if I take the descriptions and independent user reports I´d say it works. I think back about my idea of using 100 meter of coax cable (See local dairy page 25) to make an interesting analog electronics string simulation (For which I made advanced real time simulation software as musical instrument simulation)  and honestly it would be cool to get satellite reception via a self controllable tuner which could also get hd channels (I use a computer for receiving mainly BBC, but it doesn´t get HD channels) and of course if I´d have a licence, creating sat (HAM) signals would be cool, or just taking two communicating discs and making them link with low power.

I wonder what the actual quality of the decoded signal would be. I as a teenager had built a found car radio wreck into an enclosure after fixing and upgrading it, which acted as a tuner for years, and I was very interested in signal quality of expensive tuners which were available at the time. I don´t know what the sensitivity and distortion figures of the software approach can be, but I guess having digital delay lines and computation facilities makes things interesting.

I installed the gnuradio package from Redhat Fedora ( version 3.1.1-4.fc9, 64 bits ) which works for simple examples like the testtone, I´m not sure it is the best way to dig in some /var/share tree to find the examples, but it worked, I got audio as supposed, and even from the right DA converter and setting the sampling frequency was OK, too.

The FFT revisited

Made with Free and Open Source Software analog synthesizer simulation, and LADSPA plugins on Linux (no windows was used at all for this music) this piece of music (not crap) which is just testing some possibilities features analog synth sounds with effects with actual synth sounds, not some prefab samples. Remember Emerson Lake and Palmer doing Peter Gun ? That´s much better, but this is fun, too.

 smalltheme.mp3          (1.47 MB, 47 seconds)
 smalltheme.wav          (8.1 MB, same music, 44.1 kHz 16 bits CD quality uncompressed wav file)
 smalltheme_jam.mp3  (1.44 MB, the same music but ´treated´ with Jamin, in mp3)

The last version has been fed through Jamin, the 32 band equalizer program for Linux, with mulitband compression and loudifier. Doesn´t sound better in my opinion, but some people like it.

Some work

I did some work in the household or ´klusjes´. Which is dutch for chores but then in a bit condenscending fellow sense and possibly with the connotation of being unofficial which is totally irrelevant in this case, like computer guys repairing their friends´ PCs  at work, or when I used to work at university with electronicists when ´jobs´ would be done at work or otherwise which could also be paid activities.

If I´d pick up a magazine (which I do at times) and lets say build an aquarium thermostat for the accurate heating of tropical fish, and go out an buy some electronics parts to do that for aunty X, that would not be a klusje, but if I´d be a plummer ´helping´ as unce Y to get some friends´ toilet fixed up, and get a proverbial bottle of wine as reward, it is possible that actually is a ´klus´. Enfin, who cares, I just remember someone telling me when I was asked for another university job at the time that it was supposed to be considered normal that the economy is ´greased´ with blackmail money. Huh?! I thought at the time...

Well I built some stuff I also built long ago myself, and some new things:


A carpet and vertical sunshields



Another attempt to build some bed (I built my own when I was in my own place which worked good for years). It´s not stable enough, so it needs more woodwork to be used good.



What´s all this ? That´s 4 Texas Instrument wireless connected boards with MPS430 microprocessors on them all connected up, basically to report battery voltage and temperature to the main station. The red boards thast is, two are on the computer on the right with USB connector board which supplies 3 volt to them, and also have programmers for the two wire interface, and both have serial connections, too, which I can read/write from tcl scripts.

Public use HD

some Chicago HD company:

  (great)






Adobe HD example:



An example froma Cinelerra HD demo source:



Some images grom my own HD footage (mind the motion effects and the straight character):



Original format from a filmish saturated sun bathed scene:



Filmed in Amsterdam:



From the zoo Artis:







Near the sea:



Scheveningen, from lower bandwidth encoded version:




A recent snowfall filmed from behind glass:



Yeah yeah, progress.

A place where I used to go as a teenager to sail in HD (screendump from mplayer):



Would you know where that is ? This is a "dezoom":








To get an idea of the results of my  HD material being crossconverted on the internet, see

   http://www.vimeo.com/955386

My place on the vimeo site, where anyone can upload 500 MegaByte for free in the form of video and 720p HD video (like mp4 mov etc.).


 

I uploaded 4 minutes of a selection of some mostly recent scenes coded in 720p mp4 at 13Megabit/Sec from the mp4 Cinelerra edit result, where there is no special editing, no sound, and almost no image enhancement or filtering at all, but some scenes were transcoded via h264 .

Before that however, I transcoded some scenes using the block motion detecting h264 format and mp4, which I have started to do without intermediate files to prevent incredible disk use. Normally using ffmpeg and the x264 program one needs a yuv file and gets a h264 file of considerable size, depending on the encoding quality. I've started using q=1, which means a very large file where the program tries t very acurately find the information present in the recording. The result I then pipe into ffmpeg to make decent mp4 from it, where use is made of the information without overdoing it.

The commands I use simultaneously (for repeating this you could use 3 seperate xwindows with bash shells, starting with the x264 program, and then the ffmpegs):

x264 -q 1 --interlace -o /tmp/u.264  /tmp/t.y4m 1440x1080 &
ffmpeg -i /Server/My\ Book/Tmp/garden1001.m2t -ss 44 -t 30 -pix_fmt yuv420p -f yuv4mpegpipe - > /tmp/t.y4m
ffmpeg -aspect 16:9 -i /tmp/u.264 -aspect 16:9 -b 90M -deinterlace ~theo/Videos/garden1_3.mp4

x264 -q 1 --interlace -o /tmp/u.264  /tmp/t.y4m 1440x1080 &
ffmpeg -i ~theo/Video/badkuip1.mpg -pix_fmt yuv420p -ss 20 -t 60 -f yuv4mpegpipe - > /tmp/t.y4m
ffmpeg -aspect 16:9 -i /tmp/u.264 -i ~theo/Video/badkuip1_au.mp3 -aspect 16:9 -b 90M -deinterlace -b 3M -aspect 16:9 -s 1280x720 /home/theo/My\ Book/Tmp/bakkuip1htp1.mp4

Of course this in more in the line of Pro Things than your average uncle the hobby filmer, but it can be understood what this does. The first command can be issued after having created two unix (I don't think this can work under windows) fifo pipes by 'mkfifo /tmp/t.y4m' and the other tmp pipe, and will take a pixel aspect ratio 4:3 uncompressed 1440x1080 pixels uncompressed yuv file as input, presuming it is interlaced (I don't know if it can detect it is bottom field first) and output a huge high quality h264 file at it's output pipe.

The first ffmpeg command creates the yuv file from the source 25 (twenty five) megabit per second mpeg2 material starting at a certain relative time in the file and in this case for 30 seconds.

The second ffmpeg command takes the h264 stream and creates a 90 Mb/s target bandwidth mp4 (mpeg4 for linux, playable as quicktime mp4) with the correct 16:9 format and deinterlaced, using the motion information fro the high Q h264 file, so it does a pretty good job at it. In this case the source file comes from the server machine (another computer) over the 100 BaseT network, and from an external 320GB disc on that server, and the result is written to the local disc of the dual core 3 GHz pentium machine. At the moment the two cores aren't busy all at the same time much, a little over 50% utilisation is achieved, probably because by far the most workload is the h264 coding, the two ffmpegs work relatively slow to keep up with the about 1 to 1.5 frames per second coming through the pipe. They would normally be about realtime, but the h264 process us the bottleneck, and runs on only one core. Still the 30 seconds fragments of transcoded HD video can be waited for, it takes a bit over 10 minutes. It is also possible to make 720p (actual progressive) in a similar way, from the h264, which is what I did for the above vimeo edit.




Les Miserables

The AES trip of the month was to a Rotterdam theatre where Les Miserables is being performed, to have a look at the audio setup in the theater which makes the performance audible and the musical work.







So what audio system would they use, or is that classified? Well, of course a sizeable amplifier array, and two hanging curved speaker arrays from the ceiling, two pilars on the side of the stage covered by cloth. And then over 80 or even more wireless microphones appearently in mpg format at some point mixed into a big digital mixer where the idea is to have only one AD/DA stage overall. Then surround speakers with lexicon PCM91s for effects, and a whole lot of screens and controles to make sure the channels can be controlled and opened in time and set so that the point of feedback is approached but not too close. Of course with so many portable (hair) microphones that isn´t easy.

A demonstration with 1 microphone on stage worn in the hair of the speaker was convincing that a degree of realism was achieved. I´d say the sound was fairly dead-ish but on the other hand pretty good, even pleasant, power enough, and no obvious added (negative) effects, and good dynamic range, though maybe a bit forced on the loud side, and the low (including transients) was quite filtered, I´d say.

If I recall correctly the speaker has made it clear there were no special feedback killers or more than the (computer) mixers´ processing, which I think is a good idea. I have experience with amplification of course, going back to primary and highschool, and I agree that as straight as possible signal path and control over the feedback with well chosen speaker placement and acoustics is preferable and more pleasant to listen to, certainly in the long term.

Theo, you run two heavy FFT based Jamin programs in a high quality 24 bit signal path live and you don´t want to use that? Euh, mostly: no I would not. It sucks too much.

HD computer graphics animations




See the Tcl wiki for a page on this subject.

The animation script is there, too, but not all the header files which are chunks from the NVidia example .pyg file, where the tcl script puts transformation commands in between. That´s not great, but it does work, and the whole think isn´t perfectly great in every way, but certainly good, and if you want grand looking HD tiff frames this works well.

I don´t speak gelato text file lingo good enough to be sure about the function of push/pop-transform and the generalities of transforms being applied to lights, scene and camera. I can search it out and look it up I guess, because it would be great to transform the scene, or parts of it, and make the light reflections change in an animation. I in this example have a turning spotlight, which already looks great, and the whole thing stays natural.





James Bryant from Analog Devices gave a lecture, well euhm, seminar like (but not for university) at Spoerle in Houten (middle of Holland) about an important subject: OpAmps:





(more pictures  1   2   3 )





From above (the picture with the actual boards and just above: the 4 nodes on one screen), the Texas Instruments wireless connection boards were presented in a Silica day near Utrecht:



I got another wireless connection kit like I had already bought from Ti for free.






This image may look familiar, I photographed the same place some pages ago, but this was with another camera.


Ultimate Rock Bands

These people (minus the first frontman who died already quite some time ago) were probably close to the ultimate rockband when I was in highschool.








I recall well the pounding basses and heavy rythms (where the word applies) and extremely musical songs coming from the 500 Watts disco system which luckily played also this music and Rolling Stones and Kiss and what have we. Absolutly marvelous, can hardly be played loud enough on a good or decent sound system. (Maybe a decompressed version Theo? Sure, probably even better, but impossible with the given video quality.)


There is a site with videos (basically bootlegs) from concerts, and I just looked at ´shotdown in flames´. Great band.


Not a rock band but good music which I knew as a kid: Cornelis Vreeswijk



There´s a documentary on the ´missed program´ site of the dutch television, which I was eventually able to see (and copy for personal use) on linux.

The EE's scheduler Execute Function

What does that mean? Well, EE means Electrical Engineer, scheduler is the thing in your computer (in this case) which gives little pieces of time to various programs (processes) that run on it, like your web browser, java applets, the operating system updater, etc., and the execute or fire function is some algorithm put in computer code which decides which process is going to run and for how long.

So presume we are a traveling salesman for a minute, we would have a schedule which is simple enough when it has been decided on, we drive to the next city, do our business, and go again to the next city, until were idle again, or plan the next move.

If however we are a local manager we will want to keep some track of all the employees which we in turn give attention to based on some decision making criterion, like who asks the hardest for it or where things are most wrong, or maybe following a fixed schedule.

If the entity is a computer, and the attention or schedule deciding factor is therefore part of a computer program, we could say that we describe the possible paths for the entity to go as a graph with connected nodes, and the going from one of the nodes of the graph to another as a decision which is based on a 'execute' or ' fire' rule. These aren't perfect comparisons, but lets say the computer has two programs which it can divide it's attention to, the schedule part of the kernel of the operating system (already at least since UNIX in the 70s) would at some point decide to fire up as it were either of the two programs and at some point to bud in again to let the other program run a while, in this case we'd be talking (hopefully) about time slices of the order of milliseconds, so the computer appears to carry out tasks in parallel.

An electrical engineer can assign many tasks to him (or her) self because the field of EE is still large and contains many possibilities for working on new and interesting subjects, and so the choice to use one of those possibilities to work on a subject is usually a choice out of many. Working on several tasks is of course the same way as with a computer possible in a time dividing way.

The interesting part is that an EE will chose for a number of tasks in the way which may correspond to his interests and desires. Like Steve Jobs wants a small solid state and energy safe sound player device for when he goes biking in the woods, and so he invents an ipod. Or, the director of Hewlett Packard (I know CEO) really hates the guts of some nazi competition guys somewhere and so decides to totally beat them by making computers which are much faster and cheaper then theirs.

More limitedly: Joe let's call him Frizzle has always been a HAM and would really like to get this one connection with an over the ocean radio buddy 3 dB more noise free and works in every area where he thinks this could work.

More everyday-ish: Mindy Ehe just loves it to play with Ohm's Law and became a network cable installer for fun.

Well, every year Bill Gates issues a great plan the campagne written down by the infamous giga hard plan bureau and it's iron guidelines, and all EEs follow this to the letter to become eternally saved from pergatory. Oh no, wrong story line. Sorry, one should never speak the truth in this microbrave world. Lets go save some woods in LA or so.

Anyhow, especially in electrical engineering, and I can say so, I am EE, following ones' interest should be and often is normal and a reasonable walk of life, including the consideration of personal honour, feelings of a moral kind, promoting normalities and good citizenship, and being good leaders.

This should work fine in the free world, but where are people still free, and unscutinized and let in peace with their own ideas, when Big Money and Big American Pie is at stake and the whole of the world of electronics. Well ....

And then the satan (who ??!!?!, well, you know..) comes by, wearing a suit, and starts managing the corporate world, and wallstreet if he can handle it, until all conservative (weak, unbrave ?!) souls are convinced that a big planbureau is absolutely necessary to survive, and of course allows the boys helping just a tiny bit to have fun while they try to save their soul. Get the picture?