Joule Thief

[tgraca]

You may want to test different transistors. I have tested many, and I've found that MPSA18 and BC337-25 work well, and even the 2n2222a works well. In my testing the 2n3904 actually had the _highest_ start and stop threshold voltages in the basic JT test circuit I used. Depending on your circuit component values you may get different results. I tested only the "start oscillation" and "stop oscillation" voltages; other performance parameters like current drain and LED brightness may not follow the same pattern. But experiment, you never know unless you try.

Here's my "data table" giving the start and stop threshold voltages for various transistors.  These are the supply voltages. "Start" is the lowest voltage at which the JT oscillations will start and light an LED, and "Stop" is the voltage when the oscillations stop. The LED may go out at higher voltages, many JTs will continue to oscillate well after the LED is actually out.

I just got a bunch of 2222A's in... they finally became readily available and are now actually cheaper than the 3904's. I use the MPSA18's
if I am working with near half an amp and 3055's when I am working with big 12V batteries, but I have still managed to pop at least
one of my 3055's and I have 5 others from that experiment I haven't pulled to check yet... they may be toast too! lol

Thanks for posting that... it's very useful. -t

[MarkE]

I just got a bunch of 2222A's in... they finally became readily available and are now actually cheaper than the 3904's. I use the MPSA18's
if I am working with near half an amp and 3055's when I am working with big 12V batteries, but I have still managed to pop at least
one of my 3055's and I have 5 others from that experiment I haven't pulled to check yet... they may be toast too! lol

Thanks for posting that... it's very useful. -t

For work w/ 12V batteries, I would use a power N-MOSFET instead of a 2N3055.  There are many 20V - 50V N-MOSFETs that handle lots of current that can be had for less than $1. each in small quantities.

[tgraca]

For work w/ 12V batteries, I would use a power N-MOSFET instead of a 2N3055.  There are many 20V - 50V N-MOSFETs that handle lots of current that can be had for less than $1. each in small quantities.

I have some STP75NF75 TO-220's (75 Volt 80 Amp) that I am going to use to make joule thief battery chargers - 12v lead acid to ~ 9.5 V
Bedini lead alum batteries, but I want to build a joule thief battery charger from solar to the lead acid battery first... it's a fairly complex
project for me and involves running new wire for a pulsed DC lighting system in my home, so I haven't started yet.... I am still goofing
around with these really low powered projects, which is safer and a learning exercise. Also, I haven't worked with mosfets yet, so I will
probably do this with a 10 amp rectifier and the 3055's and save my experimentation with the mosfets for later... We'll see... It seems
there is never enough time in the day! lol

[TinselKoala]

My "current" favorite N-mosfet for low voltage applications is IRF3205. They are available for between $0.50 to $1.00 each.

Vdss 55 V
Id 110 A
Rds(on)  0.008 ohm
175 degrees C max operating temperature


There are a "lot" of fake 2n3055s on the market. Even sticking with name brands is no assurance that you are getting a real one. Check the internet for some "dissections" of fake and real ones, the fake ones have a tiny die inside the TO-3 case. Even good ones will vary in their ability to handle high voltage spikes, as I've found.

[allcanadian]

@TK

My "current" favorite N-mosfet for low voltage applications is IRF3205. They are available for between $0.50 to $1.00 each. Vdss 55 VId 110 ARds(on)  0.008 ohm175 degrees C max operating temperature

I like that series as well and after the ratings you have posted I tend look at the internal diode rise time in V/ns as well as the gate capacitance which can be an issue in regards to switching losses. There are so many things to consider and so many choices it is an ordeal in itself just ordering parts. I have a preference for the IRFP460 for general applications which has marginal ratings at best but has shown to be pretty robust in practice.


I have yet to find a perfect mosfet solution and a gain in optimal characteristics in one area is generally followed by sacrifices made in other area's. I'm always looking for better numbers if someone wants to throw some out.


I have lots of 2n3055s kicking around but don't use them anymore although they performed okay as a darlington pair in some cases. Many of the circuits I built at that time had ridiculous rise times inducing stray voltages in every conductor in sight and I could smoke the 3055's on command. Even worse they would not blow completely however the input theshold would change or the emitter/collector would start leaking like a sieve. In any case I never had much luck with them and found them problematic so I don't use them anymore.


AC