Joule Thief

[Farmhand]

I ordered 50 pieces of adapter boards for SMD parts, up to 8 pin devices with two different sizes one side of the adapter pcb has
a small 8 pad mount and the other side has a slightly larger one, they should fit my needs I think.

With the 1400's parts pin 1 and pin 2 can be soldered together on the left side and the Lx and ground pin are "one pad apart" on the other side which makes soldering a bit easier.

The hookup is fairly simple + rail to coil to Lx pin, diode from Lx pin to + of output capacitor, gnd to circuit gnd, pin 1 and pin 2
together and connected to the output side of the diode. no contact pin 3 is isolated, it could be folded up out of the way maybe.

Also the enable pin 1 "CE" can be left floating as it has an internal pull up. Only pin 2 the output voltage monitor pin actually needs to be connected to the output side of the diode.

Sharp solder point, lots of magnification, thin solder and maybe some solder paste helped me a lot.

I quickly became a fan of surface mounting in general where structural strength is not needed.

.

[MarkE]

For fairly low cost SMT soldering Hako has a bunch of product.  I like the OKI/Met-Cal gear better.  If you can spring for an MFR1110 and three or four tips you will have a nice set-up for casual work.  For more serious work, you want a preheater, a hot air gun and solder paste.

[Dave45]

https://www.youtube.com/watch?v=jhVCY1yq-xc

[d3x0r]

I'm really liking these NTE172A aka 2N5306 darlington with 7k HFE. (MPSA18 work about the same; tested somewhat success)


(ya I know, it's not really much like a joule thief anymore; I'm calling it a ka4ep torch.. I should probably just start its own thread; and so I did http://www.overunity.com/14794/ka4ep-torch-kacher-torch/ )


I'm waiting on this test to finish to see how long 2 series AAA batteries last... going on 26 hours now; down to 2.55V. 
Joule thieves were nice to start with, but they draw a lot of current for not much gain; though I did find a high inductance third winding as a pickup made lighting lots of LEDs easy... since no matter how many you have in series you still only need 20mA... voltage is cheap, it was getting current that was the issue...
( http://www.allaboutbatteries.com/Energy-tables.html *sigh* this says an alkaline AAA yas 1150 mAh; which is 76 hours or like 3 days ... or 1.41Wh, of which I've used 0.49 in 26 hours... so this 3Wh supercap should last for days... need to figure how to use more power with this )
10mA for 500nS at every 12us(83Khz)... 24% duty cycle.. 2.5mA across the load :/ that's not very good power usage at all. well.. I guess... it's 25x the voltage... so can't be too bad (the 500ns isn't constant it's just the width of the bellcurve generously)
-----------


The capacitor that's used kinda matters...
The load that's used kinda matters....


Wish my scope had an integral (sum under the curve) function... It looks to me that 3 of these are lower current than the lower left, even though the lower left peaks out lower... Need a slightly bigger 350V cap (40uF 400V foil wound cap is bad)
Ceramic caps of various voltages also have basically a constant current draw.


Was puzzled by the continuous current draw (lower left image in composite) for a while because the only place power is connected is on the one coil... so why should it be a constant draw... then I realized the cap has an ESR, which the battery apparently can't quite keep up with... actually it's probably more of an capacitive impedance issue... that it takes X power to charge the cap at Y voltage so it can only take so much current in at a time... so I experimented with caps, and only found this one that goes down to 0 current.
I had added CSR's in various places and finally found that I am only returning 2mA for a short time in an decreasing slope (2mA max, decreasing over the span of T2 in the second image).  And since I had a constant draw of 15mA that was basically nothing  (it does help though, and is more than 0).


(yellow trace in composite is in same location; green trace in composite is on the collector, and it's biased from the 2.55V, yellow resistor is 1Ohm so mV=mA.  Lower right image green is moved to top end of secondary)

---
Edit: also the lower right ended up missing the bounce-curve, but that's cause I had a scope probe connected where I did... adding a few hundred pF of capacitance to ground does the same effect... but reduces the output a lot... have to tinker with that some.

[MarkE]

If your scope exports .csv data then you can use a spreadsheet program to derive almost anything that you think of calculating.