I feel a bit better today, so decided to build USBasp – an USB port ISP programmer for Atmel product family. I used STK 200/300 dongle before, but it works from parallel port only.

Lets start: the schematics is from http://www.fischl.de/usbasp/ . You can find some schematics and pre-routed PCB’s there.

What I need is USB B type or MiniUSB connector, SMD parts, small footprint, single – sided PCB. The one I liked is by Thomas Pfeifer, http://thomaspfeifer.net/atmel_usb_programmer.htm. The problem is I found no files for Eagle there and had to re-create the PCB.

Additional features are:

• RC filters on MOSI, MISO and SCK lines to suppress ringing, as suggested in AVR Freaks forum;
• Signal from oscillator (XTAL1) is routed to the pin 3 on IDC connector via jumper – a handy feature to debrick the MCU if you accidentally fused external oscillator;
• Inductor on the power line.

The PDF files with the layout are here, an A6 – size mirrored image for direct printing on OHP is here.

Added on 2008-10-01:


First, it works. Both connected directly or via USB hub.
Second: don’t create anything while you are ill. Actually, pin number 5 on miniUSB connector MUST be grounded, while pin number 4 can be grounded or left floating. I’ve used i jumper wire to solve the issue.

Sometimes i’m asked to “cure” one or another product, protected by registration or network license manager, like FlexLM. And you know, a lot of keygens for these product found in the Internet, do not work. The usual error is “Internal Error #{1..6} – Please be sure the app is running and on the license screen!”. What’s wrong, why the keygens from the glorious crack groups do not work any more?

Search on the Internet revealed almost nothing: from “the kyegen does not work” to “it works perfectly”, some says one must use genuine windows the keygen to work (wow…).

Let’s start a little investigation.

Some keygens are packed with some kind of packer, like UPX. I’m too lazy to extract them.
So i’ve found an unpacked version of keygen for AutoCAD 2008, created an evaluation installation of AutoCAD 2008 from Autodesk Revit Architecture 2008 (i have a license for it to run on the network, by the way) and started my explorations.

A usual PE executable consists of some sections: header, data, imported functions, static variables and text resourses. Text resourses are usually located near the end of the file, and it is the very first place to look for an information.
The very first string’s i have noticed, at the very beginning of text resourses were “acad.exe” and “~de81d5.tmp“. But wait, when running AutoCAD, i have 4 files with constant names in “%TEMP%/AdskCleanup.0001.dir.0000″. Their names are “PfdRun.pfd“, “~de0acb.tmp“, “~df294b.tmp” and “~efe2.tmp“. Only two of them has names with 7 symbols (like in keygen’s “~de81d5.tmp”). The *.tmp files seems like PE executables.
A quick edit in the keygen to match the actual names and …

The mysterious error is gone, you can paste a request code in the keygen.

And …. “REQUEST CODE ERROR!”

Let’s summarize:

• every product line Autodesk ships uses unique set of registration keys. This also means that AutoCAD 2008 sold as separate product will have different encryption keys that the AutoCAD2008, sold as “Revit Architecture suite 2008″;
• the keygen searches for predefined file in “%TEMP%/AdskCleanup.0001.dir.0000″, extracts encryptions keys (?) and generates authentification code, depending on your request;
• if the keygen does not find a predefined file in the predefined place, a famous “Internal Error #…” will be generated;
• as keygen works with predefined location of encryption keys, it is not capable of generating authentification code for different products

So, what’s to do if You got the famous “Internal Error #..”?
Buy the product. Or search for another keygen.

As, I’ve mentioned earlier, we had a pleasant trip through Italy the last summer. I have recorded the whole trip to my GPS unit using i-GO navigation program. Seems like 10 second recording intervals are fine and the track is quite precise and smooth.
Unfortunately, during upgrade from WM5 to WM6 I had to reinstall i-Go and all tracks just dissapeared from track list in i-GO. Stupid me, i HAD to save ‘igo.db’ file somewhere.
I found no way to put them back, so let the fun begin.
A short glimpse into one of the files and..

...
$GPGSA,A,3,04,20,13,23,31,11,,,,,,,2.0,1.2,1.6*33
$GPRMC,061120.287,A,5414.2734,N,02337.5314,E,54.08,261.55,290707,,,A*50
$GPGGA,061121.000,5414.2726,N,02337.5281,E,1,06,1.2,152.2,M,31.0,M,,0000*54
...

Hey, do You think the same way? Yes, it’s NMEA data in plain text.

Ok, the Tool – GPSbabel is ready. It can read and write a lot of GPS data formats. Let’s try:

$ sudo apt-get install gpsbabel

, a handy script ‘nmea2gpx’ from here
and let the fun begin:

$ for i in `ls *.bin`; do ./nmea2gpx $i; done

The conversion is smooth. You can discard files files up to about 400B – they usually contains only empty GPX structure.

But the resulting data set is HUGE: initial data set is 182MB, the resulting GPX files uses MB.
It’s fine if you need very detailed track info, but two problems arise:

• due to the nature of GPS it’s inaccurate, if you use the receiver in dense urban areas
• you do not need this amount of information to create daily track or a track for the whole journey

And again, let’s turn back to GPSbabel. It has a lot of filters to improve overall usability of saved data:

$ gpsbabel -i gpx -f 2007-08-07.gpx -x simplify,crosstract,error=0.002k -o gpx -F 2007-08-07-e2m.gpx
$ gpsbabel -i gpx -f 2007-08-07.gpx -x simplify,crosstract,error=0.005k -o gpx -F 2007-08-07-e5m.gpx
$ gpsbabel -i gpx -f 2007-08-07.gpx -x simplify,crosstract,error=0.01k -o gpx -F 2007-08-07-e10m.gpx
$ gpsbabel -i gpx -f 2007-08-07.gpx -x simplify,crosstract,error=0.05k -o gpx -F 2007-08-07-e50m.gpx
$ gpsbabel -i gpx -f 2007-08-07.gpx -x simplify,crosstract,error=0.1k -o gpx -F 2007-08-07-e100m.gpx
$ gpsbabel -i gpx -f 2007-08-07.gpx -x discard,hdop=10 -o gpx -F 2007-08-07-hdop10.gpx
$ gpsbabel -i gpx -f 2007-08-07.gpx -x discard,hdop=6 -o gpx -F 2007-08-07-hdop6.gpx

 $ ls -al *08-07*.gpx
-rw-r--r-- 1 ejs ejs 5321544 2008-02-24 20:41 2007-08-07.gpx
-rw-r--r-- 1 ejs ejs  343808 2008-02-24 21:00 2007-08-07-e10m.gpx
-rw-r--r-- 1 ejs ejs  909181 2008-02-24 20:56 2007-08-07-e2m.gpx
-rw-r--r-- 1 ejs ejs  541929 2008-02-24 20:59 2007-08-07-e5m.gpx
-rw-r--r-- 1 ejs ejs  117973 2008-02-24 21:31 2007-08-07-e50m.gpx
-rw-r--r-- 1 ejs ejs   69973 2008-02-24 21:38 2007-08-07-e100m.gpx
-rw-r--r-- 1 ejs ejs 5072358 2008-02-24 20:52 2007-08-07-hdop10.gpx
-rw-r--r-- 1 ejs ejs 4983846 2008-02-24 20:52 2007-08-07-hdop6.gpx

Summarize the above:

• reducing maximum allowable Horizontal Dilution of Precision (HDOP) makes almost no change to data size
• using ‘crosstrack’ filter and removing points maximum 50 meters away from track, reduced dataset size 10 times. Using 100 meters error reduces dataset almost 75 times, but you can see in the image and here, the data it is still usable.

The perfect and free tool to overlay the GPX track onto the globe is GPS visualizer . Use it.

The proposed workflow is:

1. download tracks the tracks you need to the computer
2. convert them to GPX
3. merge day track to single file
4. apply ‘crosstalk’ filter with maximum error of 100 – 200 meters to the daily track file. Your goal is to get 150-500kB file, depending on the road network and conditions.
5. merge daily track files to single trip file
6. apply ‘crosstalk’ filter with maximum error of 1 – 2 kilometers to the trip file, depending on the road network and conditions.
7. register at Google Maps to get your API key, as it’s stated in GPS visualizer FAQ
8. proceed your files with GPS visualizer
9. donate
10. enter your Google Maps API key into every saved HTML file from GPSvisualizer.com. You HAVE saved the files, have You?
11. arrange the presentation of your wanderings
12. enjoy

$ gpsbabel -i gpx -f 2007-07-31.gpx -i gpx -f 2007-08-01.gpx -i gpx -f 2007-08-02.gpx \
 -i gpx -f 2007-08-03.gpx -i gpx -f 2007-08-04.gpx -i gpx -f 2007-08-05.gpx \
 -i gpx -f 2007-08-06.gpx -i gpx -f 2007-08-07.gpx -i gpx -f 2007-08-08.gpx \
 -i gpx -f 2007-08-09.gpx -i gpx -f 2007-08-10.gpx -i gpx -f 2007-08-11.gpx \
 -i gpx -f 2007-08-13.gpx -i gpx -f 2007-08-14.gpx -i gpx -f 2007-08-15.gpx -o gpx -F full.gpx

$ gpsbabel -i gpx -f full.gpx -x simplify,crosstract,error=0.5k -o gpx -F full-e500m.gpx
$ gpsbabel -i gpx -f full.gpx -x simplify,crosstract,error=1k -o gpx -F full-e1km.gpx
$ gpsbabel -i gpx -f full.gpx -x simplify,crosstract,error=2k -o gpx -F full-e2km.gpx
$ls -al

lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-07-31.gpx -> ../daily/2007-07-31/2007-07-31.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-08-01.gpx -> ../daily/2007-08-01/2007-08-01.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-08-02.gpx -> ../daily/2007-08-02/2007-08-02.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-08-03.gpx -> ../daily/2007-08-03/2007-08-03.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-08-04.gpx -> ../daily/2007-08-04/2007-08-04.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-08-05.gpx -> ../daily/2007-08-05/2007-08-05.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-08-06.gpx -> ../daily/2007-08-06/2007-08-06.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:38 2007-08-07.gpx -> ../daily/2007-08-07/2007-08-07.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:39 2007-08-08.gpx -> ../daily/2007-08-08/2007-08-08.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:39 2007-08-09.gpx -> ../daily/2007-08-09/2007-08-09.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:39 2007-08-10.gpx -> ../daily/2007-08-10/2007-08-10.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:39 2007-08-11.gpx -> ../daily/2007-08-11/2007-08-11.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:39 2007-08-13.gpx -> ../daily/2007-08-13/2007-08-13.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:39 2007-08-14.gpx -> ../daily/2007-08-14/2007-08-14.gpx
lrwxrwxrwx 1 ejs ejs        34 2008-02-26 02:39 2007-08-15.gpx -> ../daily/2007-08-15/2007-08-15.gpx
-rw-r--r-- 1 ejs ejs    186733 2008-02-26 02:59 full-e1km.gpx
-rw-r--r-- 1 ejs ejs    111056 2008-02-26 03:01 full-e2km.gpx
-rw-r--r-- 1 ejs ejs    329869 2008-02-26 02:53 full-e500m.gpx
-rw-r--r-- 1 ejs ejs 107624591 2008-02-26 02:45 full.gpx

by, the way, while processing the last two gpsbabel commands and running Xorg , KDE and Iceape as usual:

top - 02:58:54 up  6:52,  6 users,  load average: 2.66, 1.93, 1.19
Tasks: 121 total,   4 running, 117 sleeping,   0 stopped,   0 zombie
Cpu(s): 99.3%us,  0.7%sy,  0.0%ni,  0.0%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:    516620k total,   508624k used,     7996k free,     2004k buffers
Swap:  1052216k total,   157092k used,   895124k free,    93384k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 6173 ejs       25   0 86368  82m  840 R 48.9 16.3   2:59.09 gpsbabel
 6175 ejs       25   0 86372  82m  840 R 48.9 16.3   1:28.35 gpsbabel

The resulting GPS data with 1km error is presented here.

The unit served as almost perfect GPS navigator for our summer trip though the Italy. One regular February evening i took my Asus PDA and…
Yep, a small problem. And a serious one.

But… there’s a country named China.
I like chinese. Really, dude. The price for 639′s LCD screen here in Lithuania is around 120€. You can grab one at E-bay.com for around $130 – it’s about 90€ in today’s exchange rates. If you look into ebay.co.uk, the same LCD goes for £40 – it’s 60€. Shipping – add 25€ more.

So, prepare 85€, wait several days and you can start the EXCHANGE.

what’s inside the package?	invoice, screen, tools, cleaning cloth, protective film	ah yes, i had to take the screen out
take out protective plugs and unscrew 5 screws with the tool provided. Separate the bottom shell.	CAUTION: do not push the LCD now!	remove the keypad plate
put the top part of the screen, widen the plastic shell and the screen will just pop in	and yes, the front side of MoBo. Note the empty U7 place near the SD card. What it is for?	and now, the connectors: keypad on the top, speaker in the middle and the LCD on the bottom
there’s almost no space left around and the LCD connector is hard to put in place.	attach GPS antenna connector. Be carefull, it’s hardly repairable.	Wake up, darling

And the protective film. Well, i should have it fitted before i’ve started to assemble the PDA…

Sometimes i need to check the ESR values for second-hand or fresh electrolutis capacitors. Although ESR meters are available, it’s hard for me to spend a lot of bucks on one of it. So: the schematics is from ludens.cl, the case is from local supplier (a low-cost analog multimeter from China).
OpAmp, resistors and SMD capacitors are mounted. Black stripes at the bottom are milimeter marks on the metal ruler.

Stupid me. Used DIY regulated transformes to start my Alfa Romeo today and hooked on the wrong connector. Alfa started. The insulation on the wire made a blue smoke, but … still working

Anyway, got 6mm² wire for replacement.
It’s time to replace car battery too, i think. Or check the generator.

As promised earlier, i started the temperture controll project.
The last Atmel AVR in DIP package i had went to some ‘black box’. Have no sources for firmware, so i don’t want to touch it now.

It happens i’ve got Atmega8 in TQFP package. The only problem is the connection of TQFP to veroboard is either complicated or expensive. I’ve decided to create my own modular and testing board for Atmega. It has: Atmega8, place for crystal, power supply, ISP connector, place for EEPROM – currently 24c16; RS-232 tranceiver, reset button; every pin of the MCU is routed to the connector. The board is 80x50mm – a quater of standart Euro PCB.

Here it comes. Before sanding, some tracks corrected, some still not. Hope the routing is correct

The process is simple:

1. create the board using Eagle editor;
2. print it mirrored on an OHP tranparency. I use quite old Epson Stylus Color 440 printer, set media to “Photo Quality Ink Jet Paper” and print at 720dpi. Setting of ‘OHP transparency’ gives horizontal lines, which may be quite ok for OHP, but mkes no sense for contact exposure using light box;
3. expose the UV-sensitive PCB. My result if ~12 minutes, while there are no signs of exposue on the board;
4. remove exposed areas using 10% solution of liquid glass at temperature of ~50 deg Celsium. Make sure every exposed area is clean from residues of UV resin;
5. etch the board in FeCl3. Heat it to ~45 deg Celsium for better results. I use indirect heating: a tin can with water placed on gas stove and PVC can with etcher inside

What i learned at last for today:

• keep an eye on drill diameters: usually they are too big for some parts;
• drill of 0.6mm is fine for vias. Just make sure diameter is at least 1.2mm
• keep spacing around vias bigger (about 0.025″) for it will merge with other tracks otherwise;
• get a better printer. Ink-jet is not suitable for fine PCBs
• keep min spacing of 0.025″ or tracks will merge
• also: tracks of 0.01″ tend to dissolve, if etching takes longer. Use wider tracks – 0.012″ in the case a lot of copper is to be removed
• i need a better light box: the lamps are too close to the PCB and the light field is not uniform in current setup

So, inputs and cotroll board is on the way.

This is because the boiler behaves strangely. Although the it’s full of hot water, cold water is going out.

So, the pile of details gave me: MCU Atmega8, by Atmel; some LM335, temperature sensors by National.com. A pair of CD4051 will allow to use up to 16 analog inputs to the Atmel MCU.

Also, schematics at
http://www.geocities.com/vk3em/avr-tempc/avr-temp.htm. It has some weird details, anyway i will try to master some kind of temperature sensor for the boiler. Just to make sure the will be enough hot water when i’m going to the shower.
Brr…

Quite interesting article on image processing. The MatLAB source, presented on the site, need some minor enhancments. Authors may contact me, if they want .
Attention: the program is really slow. I used ‘direct solver’, and it took almost 10 minutes to proceed the image.