Savel brain dump in English!

To light blue and white light-emitting diodes (LEDs) the voltage of 3V is required. By the way, white LED is the same as blue LED with additional luminophore inside. This luminophore is activated by blue light and lights up. Finally it lights as white.

What to do if you wish to connect a white or blue LED to a simple AA or AAA battery or even worse to the dead battery? It’s very simple. It’s enough to assemble a self generator on one transistor and the LED will light. That’s a real sample:

On the Internet this scheme is often called “Joule Thief”. Though the idea of self generation is well known and used in various strange devices.

What do you need for repeating the experiment? A simple transistor firstly. I used the transistor NPN 2N3904.  Secondly, you need 1K resistor  and  a small transformer wound on a ferrite ring. The ring can be very small, as device is not very powerful. My ring has 2 x 30 winds of thin copper wire.

Wind the transformer as follows: take 1m of  thin wire and fold it in half. So you have two wires. Wind them on the ring simultaneously. When the ring is wound, crop and tin each end of the wires. Connect the beginning of one wind to the end of the other wind and to the positive post of a battery. (Check with a tester or in any other way if the connection was made correctly. You can use coloured wires, it will be easier to recognize them).  The second wire of the transformer should be connected to the resistor, the third – to the collector. Connect Led’s anode (+) to the same collector and free end of the resistor to transistor base (to the central leg). Transistor emitter and LED’s cathode (-) should be connected to the negative post of the battery.

 Scheme is here:

Theoretical remarks (untested).

You can use PNP transistor, but it will be necessary to reverse emitter with collector, anode with cathode and to connect the power supply in opposite way. You can also use old Russian germanium transistors. Then the device will work more stable at low voltage. Such device could be powered from “primary power sources” as lemons or apples with two different wires inside… 🙂

I adapted the collimator of the higher quality. The voltage 250 mA is enough to light a match. And there is no need to hold it long.

The collimator was tore out from a copier. It’s quite good, that you can precisely focus the laser beam by turning a screw. You minimize the power and focus the laser beam till minmum. It it impossible to notice that the laser beam is minimized when everything around is light due to full power.

Video: DVD with New Collimator (Xvid, ~1Mb).

I tried to turn “a laser arrow” to the sky, but the beam gone into nonexistence. But it was perfectly visible on all the nearest objects. I tried to make some photos, but they came out so so…

After digital “processing “:

Youtube version:

Older lazeriniai postai.

At last I finished (2008.02) my torchlight project. I was looking quite a long time for good LED driver. When I found driver chip, I’ve lost all my white LEDs. So I used white LEDs from bad USB web cameras’. There are 4 LEDs in one web camera. I dissasembled several ones and collected 12 (in last edition 16) bright white LEDs.

baltų šviesos diodų projektą.

Here is last PCB test before finishing assembly:

I compared light output using simple cheap incandescent bulb, but powered from one Liion battery. Bulb is 2.4V and 0.75A. As it was powered from 4…3.7V source, output power is much more. And it was very hot- plastic details near lamp melted.

“Light power” comparison:

Conclusions:

After comparing both types of the lamp I can definetely tell, that LEDs are better in case of energy use. But only if LEDs are connected using special driver chip. Cheap China torches are made without any driver and are not very good for advanced use. Looking from spent money, the situation may be different. LED driver chips and custom hardware is expensive. Another advantage using drivers chips is that light power is constant during all liftime of battery.

Continue reading →

This is not up to date post (post date 2008.02)

I rescanned DVB-T channels using areal antena and found some new stations. As my antenna is quite big I am receiving TV stations from some near cities. I find new transponder at 778MHz. Maybe it is TEO? During tests all channels were uncoded. Now I need card to view.
Some changes in free to air channels. Also, picture size is smaller. Size is reduced from 720×576 to 544×576. But bandwidth is same, so overall quality is same. Also, less CPU power is needed to decode TV programs. Maybe size is changed because cheap TV tuners are with very “thin” CPU.

The are only 3 channels in 786Mhz…

Screenshots:

More cropped pictures: Continue reading →

Maybe you noticed, that new home cinema systems, LCD/Plasma screens, receivers have very small amplifiers. Physically small. But the power output is quite big. And there is no big, heavy heatsinks on these devices. And no big transformers. At a first glance I thought that all this output power is represented in “Chinese wats”. But in this case, the wats are real. And these devices are using “digital” D-class amplifiers.

What a hell is D-Class amplifier? It is old-new invention- digital amplifier. “Digital” means only one bit 🙂 The analog output is made using pulse width modulation (PWM) output. Some sound is converted to digital form using CPU ar special chips and this PWM sygnal is fed to powerfull MOSFETS. Real DAC conversion is made on small inductors near the output.

Here is image of one such amp:

I removed small heatsink from the top of chips. There is no need for big heatsinks as efficiency of such amp is about 90% (acording datasheet).

Here is closeup of the PCB. Note the size of SMD resistors and compare to the size of the chips. Chip is TAS5111 and TAS5112 made by Texas Instruments.

TAS5111 is buwoofer driver. Output of the chip is 70W RMS (@ 0,2% THD+N), 95dB dynamic range. And all this for single polarity 30V PSU (3.3V for logic). Other chip, TAS5112 is stereo amp (two channels) 2 x 50W (@ 0,2% THD+N).

The principle of operation is very simple. PWM sygnal from sound processor is fed to the chip. The chip is driving output mosfets in H-Bridge connection. As there are two PWM channels, the output is connected in full bridge. The chip is full of various protection circuits. All DAC is made on quite small 10uH coils (blue cylinders on PCB).

All amp is 6+1. Total 370W RMS from board sized like a CD box. Heat sink need to sink only 40W of heat during max load!

This particular board is damaged, I didn’t examined what is bad is this PCB, I picked up this PCB for weblog ilustration only.

This weblog entry is more interesting for Lithuanian readers. It is about our biggest internet provider and the problems related to DSL connection.

Nice title?

I am using cheapest DSL service at home. It is working only on evenings and weekdays. And speed is very slow- limited to 256kbits. But the cost is very low (check TEO webpages for rates).

My telephone line is quite old, and cable from crossbox somewhere in the begining of the street (about 1km from my home) is damaged. There is so called “semigrounded” pair. Sometimes I can hear AM radio in my phone… But I don’t use my phone for voice calls, I use it for my DSL connection. Few times I asked telephone company to repair, but due to private property problems they didn’t managed to swith pairs in the cable to select better wires.

So I didn’t cared about bad insternet connection. I blamed bad, old cable. Until one morning. I started my computer and noticed, that there is internet (in the morning). But must not have any internet at this time. I checked the internet by downloading some bigger file from rapidshare… wow! The speed was incredible and now lags and no “web page not found, no DNS” errors!

After some time I’ve made this screenshot, the speed increased upto 347 kilobytes per second (2.8 Mbit).

I started all internet leaching software I have. Bittorent and eMule speed was about 90…100 kilobytes per second. (Bittorent log show 1120 kB/s maximum working with eMule at same time). I started these, connection hungry, programs just to test if there is some session limitation. Normaly, bittorent and emule opened too many sesions and DLS connected freezed. Now, no problems at all!

And DNS server is working without any problems. In regular way, sometimes it lost packets and didn’t replay to DNS requests…

So I’ve made decision, that speed limitation software used in ISP TEO.LT is buggy. Also, that I have limitations in session number and this limitation is not stated in my contract!

Same effect I can achieve if I select very expensive “full speed” option in my DSL settings. I receive up to 5Mbit/s speed and no lags!
And don’t blame bad cable.

🙂

WRT54G or WRT54GL is well known Linksys/Cisco Systems wireless router. It is famous for its third party firmware. But be aware, new model sold in shop are neutered, the ROM and RAM size are reduced. Look for version with “L” index. It is rumours, that original router with new firmware became to good and was working better that “professional” models for Cisco company. So they neutered and removed some ROM and RAM chips. All these talks, that they reduced manufacture cost using smaller chips is nonsense. In our days, the price difference between 8 and 2Mbyte chips are very small. And I think is hard to get smaller chips 🙂

If you decided to get this router, it is worth to look for “L” version, unless you want to do tricks with firmware or you want to use your solder iron. If you select “L” version, more advanced options will be installed in “basic” router.

There is problem with some of these routers. Some time they hang up. One client was using “G” model with third party firmware, but he needed to reboot device from time to time. I offered him to test my personal router with “L” index, but it hang up too. At my home, this router was working fine. After more detailed investigation, I decided, that there is some problems in power supply and it’s filters. So I opened both devices and checked PCB.

Internal power suply is clasic step-down pulsed PSU. I was surprised, that manufacturer used very small and very suspicious capacitors- all caps were 25V x 220uF. Even in CPU core circuit (1.8V) and 3.3V power lines. I didn’t measured ESR, but I think the result could be very ugly.
So I decided to change this caps to bigger ones, from spare computer mainboards and VGA cards.

Here is WRT54G PCB before repair:

There is spare place for 5V regulator?

After:

Input cap is 25V x 470uF, output caps all 6.3V x 1500uF. One additional cap is soldered to empty PCB place. I detected 3.3V in that hole.

Two PCB boards for comarison: WRT54G (upper) and WRT54GL (lower). GL has 5V regulator and bigger memory (Flash ROM and RAM). I replaced all caps.

After this repair, both routers are working fine and no freezes.

Both model use third party, linux based, firmware.

In one company I was forced to install several copies of Norton Antivirus 2008. This was retail version of software in nice boxes, and I needed to enter all registration-configuration data in every computer. At a first glance, everything is very simple, just place CD, push few buttons… And I forgot that it IS Norton Antivirus. First, there were older versions of NAV in computers. Norton antivurus requested to reboot each computer twice for upgrade. And all this reboots require to enter administrator passwords…

Here is the story: go to computer, open nice box, take the CD and CD-key printed on envelope. Place CD into computer and wait for automagic install. No way! Place CD and wait for few minutes (tens of minutes- depending on the state of the computer) and then you receive message, that there is some problems with older version of NAV in this computer and need to uninstall software. Press fix-it. You jump to old version of NAV control panel- and there is RED warning that I need to dowload and install some update to remove that damn software. I killed that process and tried to uninstall software from add-remove-programs. No way, there are other users loged to this computer. And installer exited. I rebooted computer to eliminated logged users, hanguped software and other unknows reasons “not to install NAV”.
After reboot, install process continues. Then I need to register to some NAV-Symantec oline service. I don’t want, but until I didn’t register, I can’t continue configuring antivirus… But if you press “Next” ignoring red messages about invalid entries, this “registration” is skipped. After this, antivirus is loaded, and it began to worry that antvirus database is out of date. Dowloading of databases and updated took very logn time (it is not internet connection speed problem). And then… we need to restart windows.

Compared to old (7.x) version of AVG, NAV 2008 is ugly slow, time comsuming, CPU eating, silly designed and bad working software.

It took 50 minutes to install NAV2008 to old Celeron (1.6Ghz, 512Mb RAM) computer!

It is not worth such money…

I am using small Hobby drill (cheap dremmel tool) t make holes in PCB. Recently I broke several high speed steel drills. I decided to explore, why the drills are broken. After some investigation, I discovered that drill is moving a bit to side. The problem is in the drill grip. Sometimes, one element of the grip is “delayed” and the drill is not centered. When I use flexible steel drills, the drill is flexible enough to compensate the movements. But when I used high speed carbon steel for PCB, the brittle drill is damaged very fast.

I disassembled the drill. Here the image of the parts…

The drill is very simple: Chinese made electric motor, brass axle, one ball bearing and drill grip.