This is a continuation of my earlier Questions about Joule Thief.. blog. Paul, who is from the U.K., said he is using the BCY51, BFY51 and 2N1711 transistors, and wanted to know if I thought they were suitable for a Joule Thief.
The information I could get online about the BCY51 was scanty, but what little I saw seemed to indicate that it was made for high radio frequency use and cannot handle the JT current of over 100mA, therefore I would say that it is not suitable for a Joule Thief, however poorly it may seem to work. According to the datasheet, the BFY51 is a general purpose transistor that can handle a full amp. The Philips datasheet did not show any graphs. The hFE at 10V and 150 mA is given as 40 minimum. That’s much lower than the typical 150 or so current gain of a PN2222A or 2N4401. The Vce(sat) at Ib=15 mA, Ic=150 mA is 0.350 V maximum. That is higher (not as good as) the PN2222A, which is 0.3VDC. Because it can handle the current but has the lower gain and higher Vce(sat), I would put this on the lower end of suitability for a JT. It is going to need a lower resistance than the 1000 ohms of the typical JT.
The Last transistor Paul used was the 2N1711, and I have a few of those. According to the ST datasheet, the 2N1711’s Vce(sat) at Ic=150mA, Ic=15mA is typically 0.5 V, maximum 1.5V. Oof! That last figure is the same as the supply voltage!! The JT transistor must conduct well over 100 mA in order to charge up the coil during its on time. Even if the 2N1711 is at the typical 0.5V, one third of the supply voltage will be wasted across the transistor as heat. I took one 2N1711 and soldered it into a Joule Thief which already had a BC337-25 in it. The BC337-25 was drawing 88 ma from the 1.5V supply, which is what I would expect from a Joule Thief putting out nearly 20 mA to the LED. When I put the 2N1711 into this circuit, the supply current dropped down to 64 mA. To get the supply current up to 86 mA, I had to drop the resistor from 1k down to 470 ohms. The lower current gain of the 2N1711 is what made the difference. I don’t have a current sensing resistor in series with the LED, but from the many other JTs that I’ve experimented with, I already can predict that the LED current will be only a dozen or so milliamps instead of the 18 to 20 it was getting from the BC337.
To me the choice is obvious. The 2N1711 and other transistors he has experimented with are packaged in a metal can, which costs many times as much as a BC337-25 in a plastic package. The metal package is not needed; the heat dissipated by the transistor (when it’s the proper transistor with a low Vce(sat)) is minimal and it won’t get warm. And the BC337-25 makes a very good choice for a JT transistor. The BC337-25 is available from Farnell for the miserly sum of 2.50 to 3 Pounds for 100. The 2N1711 will cost you 40 Pounds for a hundred. Ouch!
If you want to see an excellent JT transistor, try the ZTX1048A, from Zetex in the U.K. The gain at a half amp is typically 450, and the Vce(sat) is an incredibly low 0.027 volt! These are more expensive, but can drive several LEDs. Here’s a link to the Farnell catalog.
I used a lot of PN2222A transistors in my Blue Blinkies for Xmas decorations; they’re good for experimenting and available at Radio Shack stores. I noticed that they don’t seem to be in Farnell’s catalog. The 2N4401 is another one available at Radio Shack stores. The BC337-25 and BC337-40 are great for JTs – probably the best buy for the money at 5 cents apiece. The -40 will act like a racehorse in a JT and you have to say “Whoa, Nellie!” by increasing the resistance to 3.3k or more, lest the LED get too bright and burn out. The complimentary (opposite polarity PNP) to it is the BC327-25. The overseas company futurlec.com has components in small quantities and the shipment I got a few weeks ago took a couple weeks to arrive, but then some delay might have been because of the Xmas holidays. However I did have a problem (see my other blog), and they finally rectified it by sending me replacements which arrived a few days ago.
Paul asked about
The transistors that Paul said he has, for their day, were very good. But they were the ‘jack of all trades’ type which didn’t have any outstanding attributes. They were designed with few compromises. The transistors of today, such as the ZTX1048A, are more specialized, and have a few outstanding attributes, such as the Vce(sat), and high gain. But in order to get those, they had to sacrifice some other place. One sacrifice is the ZTX can only handle 17.5v maximum, less than half of what those earlier transistors’ maximum. There are other sacrifices, too. The laws of physics can’t be ignored, and the transistors have to obey just like everything else. One thing that has improved is the manufacturing process, so that newer transistors don’t have to have the greater margins of safety that older ones had. For instance, the equipment lets the designers control the process to closer tolerances, which allows them to make the transistors so that the junctions are closer together, thus giving higher gain. And the same applies to the chemistry and temperatures. One thing that bothers me, though, is that Fairchild or Philips may sell a transistor to us for five cents and still make money, but the ZTXs sell for nearly twenty times that much, yet they’re all made from the same silicon and similar processes. I guess it’s a question of supply and demand.
One important point. You don’t need a ZTX1048A for a Joule Thief with a single LED. This transistor is so powerful that you want to use it for a large number of regular 5mm LEDs or for a 1 to 3 watt LED. I tell others to use the BC337-25 because it matches up well with a single 5mm LED. If you use its higher gain version, the BC337-40, you will get more than the required 20 mA for full brightness with 1 LED; instead you can use it to power two or three 5mm LEDs. The ZTX can power many more than that, so don’t buy them if all you want is to power 1 or 2 LEDs.
You don’t need to order a minimum at some stores. I got some of my transistors from eBay in smaller quantities. But I have also had some bad experiences with eBay sellers. Almost all of the commodity priced components there are new and may even come on tape, so you know they’re not used. But some vendors sell certain things, especially LEDs, at good prices, and they don’t tell you (or may not know themselves) that these are factory rejects with flaws such as LEDs with air bubbles in them. For experimenting, this isn’t so much of a problem, but if you’re building something like a flashlight, you want a tight beam of light, not something that’s diffused by air bubbles.