2011-12-09 Germanium Joule Thief

I have built quite a few germanium JTs, for the simple reason that they will start easily when the battery is below a half volt, and run the battery down to less than 0.2 volts while still putting out a small amount of light. Even after the conventional silicon transistor JT has run the battery down and stopped working, the germanium JT will continue to run off that battery for hours or even days.

In the past several years I bid on and won some lots of transistors, some having germaniums along with various other oddballs.  These and others I collected over the years have enabled me to build germanium Joule Thiefs that can literally suck the life out of a battery like a vampire drains blood from its victim – if you could see the electrons in a battery as red blood, the battery would be a pale and ghostly white after the germanium JT has finished with it.

In the past year I got some more germanium transistors from an eBay seller in Germany, and these were marked 2N404, the number that was as common in the 1960s as 2N3904 is today.  These work very well for sucking the last vestiges of juice from a ‘dead’ battery – I was pleasantly surprised at how well they work.  I used a high permeability ferrite toroid core with enough wire to give about 1 millihenry of inductance, usually about 20 or more turns.  For small cores this usually means the wire has to be thin, so I usually use about 24 inches or 60 cm of 30 AWG, three lengths wound trifilar, all at the same time.  I’m not trying to wind a super low resistance coil, I’m trying to keep the JT running at well below 50 kHz.  Two of the windings are connected in parallel for the primary.

Since most germanium transistors are PNP, including the 2N404, the positive battery line goes to common or ground, and the negative to the two windings, in other words the opposite of the normal connection for a silicon JT.  Also the LED’s cathode or flat spot must be connected to the collector.  I use the optional capacitor across the battery, and it is typically a 10 microfarad polarized electrolytic with the positive end to the positive of the battery, which is the JT’s common or ground.

I use a resistor that is lower than 1000 ohms or 1k ohms, although a 1k works okay.  I usually use a 470 ohm resistor to drain the battery a bit faster.  But this will cause the germanium JT to draw too much current if it is connected to a fresh battery, so i don’t connect it to fresh cells.

The LED is not critical; any color LED will do.  Typically I use the blue LED.  I bought a few dozen blue LEDs from a seller on eBay, and when I received them found that they (and other colors) had air bubbles in the lens.  I informed the seller, who sent me replacements, so I have even more blue LEDs.  I’m still trying to use up the various colors with the air bubbles.   I even have one germanium JT with an ultraviolet LED, which is supposed to be able to detect counterfeit bills.  I haven’t found any yet (thankfully..).

The picture shows a germanium Joule Thief made with a transistor from an old computer board, one of the many that were made surplus back in the early 1960s when the SSI chips (7400 series small scale integrated circuits) became available and computer (mainframe) makers replaced all of those old transistors.  The date code is 5926, or the 26th week of 1959, over 50 years ago.  But the transistor is still going strong, mainly because at that time they had learned enough about packaging transistors to hermetically seal the chip inside of a metal package.  The transistor is similar to the ones General Electric sold, but it has a house number, maybe from IBM.  We have to remember that germanium transistors were only capable of running at 1 to a few megahertz, so when silicon parts came along, the clock speeds could be dramatically increased.  The poor li’l germanium transistors were quickly abandoned for silicon.

The clothespin button cell holder allowed me to run those alkaline button cells down to nearly zero volts. One thing I’ve noticed is that the alkaline AA and AAA cells tend to leak, even if they are old but unused (I took some old ones out of a package recently and the ends were all furry from the leaked juice).  But I can’t remember ever seeing an alkaline button cell that has leaked.

Back to experimenting…