2011 Jul 05 AC128 Germaniums
I was looking on eBay for some germanium transistors that might have a better chance at being used for a Joule Thief. I blogged my Germanium Supercharged Joule Thief a few days ago and said I came to the conclusion that germaniums were a poor choice for a SJT. I still think the germanium transistors that are least uncommonly available are not a good choice. But I should have said that the common germaniums are typically made for small signal use, and not for higher current use.
Further explanation
When I did some comparisons I found that the germaniums were constrained a lot more than silicons, due to their inherent limitations. If I put a germanium power transistor, even a lower power one like the AD162, into a JT, it would not work at frequencies above the audio band, roughly 10kHz. So power germaniums had a severe frequency restriction (a SJT likes to run at 100kHz or more). If I used the 2N404, as I did in the blog, I found they were capable of handling the 100 plus kHz SJT frequency, but they were pushed to their limit regarding current and hence power output. I got around that by putting two in parallel. So in other words, I had used a small, low power germanium transistor, and a big, high power germanium transistor, but not the ones in the middle, the medium power germanium transistors.
This led me to search for germaniums that were somewhere between small signal and power, that would handle a half amp of current, and be able to easily go as high as 200 kHz. I thought that 2N1038 might be a good choice. But I’m not sure because I see in the datasheet that its max freq is only 225 kHz.
I started looking for germanium transistors on eBay, to see what was available. I saw that the AC128 was shown in the picture as having a square metal sleeve that is used to hold it to a heatsink, implying that this transistor can handle higher than average current. I looked in a Mullard datasheet and found that this transistor can handle an amp, and has max freq of 1.5 MHz, much higher than the 2N1038, and that’s really what I want. I found a pair in this auction, but this link may be expired by the time this is read. Just search eBay for AC128. My only complaint is that the bloody things cost an arm and a leg (two for $6.66 plus $6 shipping), and they are being mailed from Bulgaria, about the worst place from which to receive something.
I finally found one seller who was located here in the U.S., sold them for cheaper, $2.00 apiece, and would give a break on shipping for multiple purchases. For now, it looks like four should be enough to do experimenting; I can order more later. When I receive them later, I’ll update this.
Update Jul 12
I finally received the four AC128s that I ordered. Even though the case is different than the 2N404, the pinout is identical: E – B – C looking at the base. The C lead is shorter, but that doesn’t mean anything after the leads get trimmed during use. I measured the gain, and three had gains of over 110, but one had a gain of over 200. It’s hard to tell what the gain is because the heat from your fingers changes the gain several points. Even at 200, the gain is still below that typically found in silicon transistors, which typically have a gain of over 200. We have to keep in mind that the germanium’s leakage current is much higher, too.
I soldered one into my Germanium Supercharged Joule Thief circuit that I’ve been using. This uses a silicon diode and the capacitor is 1000 pF. At 1.5V supply, the LED current was less than a milliamp, about 0.9. I reduced the supply voltage to 0.5V, and the LED wouldn’t light. I increased it to 0.7V, and the LED lit dimly, and the LED current was less than 0.1mA. I was getting much better performance from the original 2N404.
I increased the supply V back to 1.5V. I tried adding a 470 pF capacitor, and the LED current went up above 1 mA. I tried adding several more capacitors, and found that it liked a 2200 pF in parallel with the 1000, for a total of 3200 pF or 3.2 nF. With this combo, I was getting an LED current of about 8.8 mA, and a supply current of about 30 mA, which was just about the same as the original 2N404. The frequency was about 90kHz. The supply V situation was the same: nothing at 0.5V and dim at 0.7V.
These results were somewhat disappointing. I was hoping that the AC128’s higher current capability would make it easier to get 10 or more milliamps LED current at 1.5V. I hoped that at lower voltages it might do better, too. In both cases it wasn’t any better than the 2N404. I did find that it had difficulty at the frequencies above 100kHz, and the added capacitor solved that.
I believe that this confirms my earlier conclusion that germanium transistors do not offer anything more than silicon when they are used in my Supercharged JT circuit, and at low supply voltages do not do better than silicon.
Back to experimenting…