I got a bunch of BC337-40 transistors; I don’t know why since I already have hundreds of BC338s and I haven’t really used a lot of them in years (see note). I have been using the BC337-25s because I felt that it was more representative of what the average transistor is going to be when the average experimenter gets hold of a transistor. The difference is in the current gain: the -25s have current gains centered around 250, the -40s are centered around 400. If I pick a 2N4401 or a PN2222A, it is going to be closer to the BC337-25 than the -40, so comparisons are going to be more equal, without a big advantage for the BC337-25. If you want to see the actual specifications, do a search for the two words BC337 datasheet or PN2222A datasheet etc.
I reckoned the BC337-40 was going to be a hotter performing transistor in a JT. So I heated up the soldering iron and put one together. For the core I used a 1/2 inch core from All Electronics, cat. no. TOR-23 (no longer sold). I bifilar wound about 13 turns of 24AWG solid insulated telephone wire, and measured the inductance of each at 83 microhenrys. I used a 1k resistor and a white LED. At 1.5V supply V, it drew 140 mA, and the frequency was 43 kHz. In comparison, the -25 usually draws about 100 mA at this voltage, so as I said, this -40 is hotter performing.
I didn’t put a 1 ohm resistor in series with the LED, so I couldn’t measure the LED current, but it was very bright, and from past experience I guesstimated that the current was well above 20 milliamps, so I decided that I had to reduce it some. I put a 100k (logarithmic audio) pot in series with the 1k, and powered it back up. I adjusted the pot until the supply current was under 100 mA, and powered it down and measured the total resistance at a little under 5k. I powered it back up, and adjusted the pot to its maximum resistance and the LED went dark. As I turned the pot down, the LED started to light and the frequency was over 180 kHz. As I further reduced the pot, the LED brightness jumped, and the frequency jumped, too – there were some kind of mode changes as I adjusted it. Further reduction brought the supply current up to about 80 mA, where the frequency was 60kHz and the LED was bright.
Conclusion If the BC337-40 is used for a JT, the 1k resistor will give greater than average LED and supply current. The resistor should be somewhere around 4000 to 5000 ohms, with 4.7k seeming to be a good choice. I’ll have to try a few more to confirm these results, but I can say that the results are right about where they would be expected to be, all things considered. And it confirms my ‘theory’ that the BC337-40 is a hot performer.
I measured a couple BC337-40s on the cheapo DMM’s hFE range. The first one measured over 600, the second one measured over 650. That’s very high, but the measurement is taken at low current. It’s not representative of the gain of the transistor in the JT because the current is much greater. I measured a few BC337-25s and they were about 350 to 400. In contrast, a few PN2222As measured around 200.
Note: A long time ago I ordered a hundred of the BC337s from Futurlec, and they sent me the BC337-25. I didn’t know which ones they would send because they don’t say in their list. Later when I got the BC338s, they were genuine Philips, and not marked with a dash suffix. Theoretically I should have received an even distribution of all three gain ranges, -16, -25 and -40, in the mix. But since they were from an eBay seller, I could not be certain they were evenly distributed. I do know that the BC338s perform as expected in a JT circuit.
Back to experimenting…