2011 Jun 30 Germanium Supercharged JT
Note: links are broken until I replace them.
Peter asked me if my Supercharged Joule Thief would be a good circuit to use with AC128 germanium transistors. But I admitted that I had never tried using germanium transistors in a SJT. So I took a SJT that was already built with a silicon BC337 and changed it to work with the 2N404, a PNP germanium transistor from the 1950s. I had to reverse the polarity sensitive parts such as the LED, diode and ‘lytic cap.
Circuit
The coil I used was a half inch diameter high permeability ferrite core with both windings the same number of turns. One winding was 24 AWG solid insulated telephone wire, the other was stranded insulated wire, probably 26 AWG. I counted about 13 turns for one winding. Each winding measured 490 microhenrys.
The diode was a 1N4148 diode, nothing special, just a regular small signal diode. The resistor was a 1.2k 1/8 watt. The capacitor was a 560 pF ceramic disk. The bypass capacitor was a 10uF tantalum, but I replaced it with a 0.33 uF poly capacitor because it was non-polarized. The LED had a 1 ohm resistor in series with the cathode lead so that I could measure the LED current.
Performance at 1.5V
I connected it up to the 1.5V power supply. Initially with the 560 pF capacitor the supply current was about 25 mA and the LED current was 7.8 mA. This was below what is typical for the circuit, but the 2N404 was not designed to handle currents above 100 milliamps. The frequency was 225 kHz, which is about right for this SJT, but this was probably pushing the higher end for this germanium transistor which has a max frequency of only a few MHz.
I decided that it might help to increase the capacitor, so I put a 470pF across the 560, and the LED current went from less than 8 to more than 11 milliamps. I added more capacitance but the LED current started decreasing, so the total of 560 plus 470 pF, or right about 1000 pF is a good point for the capacitor. The frequency went down to 190 kHz, the supply current was about 30 mA and the LED current was 11.4 mA.
Performance below 1.5V
I now wanted to find out if the circuit would work at lower voltages. I dropped the supply voltage to 1/2V and the LED went out – no light at all. I increased the voltage up to 0.7 volts and the LED came back on but dimly. The circuit would not work below 0.7V, which seemed odd because the typical silicon SJT will start at 0.6V and a germanium should go even lower. At 0.7V, the LED current was only 0.1 mA, which is very dim, and really not useful.
Further changes
I put a Schottky diode in parallel with the 1N4148, and it increased the LED current from 11.4 to 14.4 mA at 1.5V supply. When I reduced the supply to 0.5V, the LED would light, but it was very dim and the current was less than 0.1 mA. I changed the diode from Schottky to germanium, and the results were the same. I then put a second 2N404 in parallel with the original, and the LED current went up to almost 18 milliamps at 1.5V supply. The supply current was a bit over 50 milliamps, which is just about typical for this circuit with a silicon BC337. However, when I changed to 0.5V supply, the LED current was sometimes 0.1, and sometimes 0.0 milliamps, and the LED was very dim.
I think there is more than one reason for this poor performance, but I haven’t confirmed them. One is that the germanium transistor has lower current gain than silicon, so it takes more current to get it running and keep it running. The silicon diode must have 0.6V to conduct enough current, so the supply voltage has to be at least 0.7V to keep it going. Another reason is that the frequency is higher than what the germanium is best operated at, so the performance suffers some and causes further strain on keeping the circuit working. Perhaps changing the winding ratios might help with this. And I forgot to mention that germanium transistors are much more sensitive to heat, and tend to go into thermal runaway when run at high currents – just another reason not to use them.
The performance of this SJT circuit with germanium transistors is lower than with a common silicon transistor, so I don’t think it is worth experimenting with germanium when it’s much easier and better to use silicon. Like, why bother to use a horse and buggy when a Volkswagen will do? I may think of some other minor tweaks to try, but I think that the poor performance and the rarity and high cost of germanium transistors make it a poor choice for use in this circuit. At two dollars for each of the two 2N404s, the price is 100 times the cost of a BC337, which is four cents.
I should add that last year I bought some AD161 and AD162 germanium power transistors and used them in conventional JT circuits. They cost twice as much as the common 2N3055 silicon power transistor. They have some limitations that I pointed out. I blogged them here, here. I also might add that the low voltage performance of germanium transistors can also be obtained by using FETs, as I have shown in my earlier blogs.
I just wanted to see what would happen if I changed the SJT back to a silicon transistor. I removed the 2N404 and soldered in a BC327-25. I also removed the germanium diode. This is probably the first time I have built a SJT using a PNP silicon transistor. With the supply set at 1.5V, the LED current was over 25 milliamps, and the supply current was about 60 milliamps. The frequency was 157 kHz. With the supply voltage at 0.7V, the LED current was 0.4 mA, more than four times that of the germaniums. At 0.5V supply, the circuit would run but not start and the LED current was about the same as the germaniums, about 0.1 mA. This just leads one to conclude that the germanium small signal transistors such as the 2N404 are a poor choice for the Supercharged Joule Thief.
Followup blog
I did some research on a better germanium transistor and purchased some AC128 transistors. I changed the transistor in the SJT that I used in the above blog to these AC128s and did some measurements. My blog on this is here.
Back to experimenting…