What is the mystery behind "white light led light from 1.5 Volt AA battery"

On Friday, November 14, 2014 4:58:46 AM UTC, daku...@gmail.com wrote:
> On Thursday, November 13, 2014 7:02:01 AM UTC-5, Ian Malcolm wrote:
> > dakupoto@gmail.com wrote in
> > news:74eddd6b-53e9-4bac-b29d-5cc40e3544a1@googlegroups.com: 
> > 
> > > Could some electronics guru please shed some light 
> > > on this ? Several Web sites(please check list below)
> > > have schematics and flashy color photographs of 
> > > single white led light powered off a single 1.5 Volt 
> > > AA battery.
> > > http://cappels.org/dproj/ledpage/leddrv.htm
> > > http://www.electroschematics.com/6195/high-efficiency-led-with-1-5-volt
> > > s/ 
> > > dinspirations.com/appliedcontent/Projects/Proj-WhiteLEDDrvr/Proj-WhiteL
> > > EDDrvr.html As most of these are very simple, I have simulated 
> > > a number of them with SPICE, but none of them 
> > > produce the advertised results. What is the mystery ?
> > > Some even have supposed oscilloscope traces !! Any 
> > > thoughts hints would be helpful. Thanks in advance.
> > 
> > This class of single transistor blocking oscillator boost converter 
> > designed to drive a LED from a single cell is colloquially known as a 
> > "Joule Thief".   Searching for that term with the keywords 'SPICE 
> > simulation' will find many simulations of varying degrees of accuracy.  
> > The basic circuit is pretty forgiving, and can easily be got working with 
> > junk-box parts which accounts for its popularity.
> > 
> > If the simulation doesn't model non-ideal behaviour of key components to 
> > a reasonable level of accuracy, it will *NOT* produce realistic results.
> > Have you accounted for the cell's internal resistance, the resistance of 
> > the transformer primary, transformer core saturation, the LED's junction 
> > capacitance and the transistor's gain falling at high collector currents?
> > 
> > Post your simulations (preferably in LTSPICE format) and the observed 
> > results you are trying to match up with and chances are, someone will 
> > point out what you've missed.
> > 
> > -- 
> > Ian Malcolm.   London, ENGLAND.  (NEWSGROUP REPLY PREFERRED) 
> > ianm[at]the[dash]malcolms[dot]freeserve[dot]co[dot]uk 
> > [at]=@, [dash]=- & [dot]=. *Warning* HTML & >32K emails --> NUL
> 
> Yes, I agree totally that unless the non-linearities in the key inductor 
> behavior are modeled correctly, the 
> SPICE simulation would work. As for 
> the transistor, I have used the standard
> SPICE model for 2N4401. The diode 
> has not yet been inserted. I was just
> trying to see if the oscillations could 
> be simulated first. Unfortunately, I 
> use HSpice at work, and Ngspice at 
> home(the simulations were done in my
> spare time), both of which use text
> based source files as input, as opposed
> to LTSpice's GUI based format.

LT does the textfile format too


NT

On Friday, November 14, 2014 10:56:07 PM UTC-5, meow...@care2.com wrote:
> On Friday, November 14, 2014 4:58:46 AM UTC, daku...@gmail.com wrote:
> > On Thursday, November 13, 2014 7:02:01 AM UTC-5, Ian Malcolm wrote:
> > > dakupoto@gmail.com wrote in
> > > news:74eddd6b-53e9-4bac-b29d-5cc40e3544a1@googlegroups.com:=20
> > >=20
> > > > Could some electronics guru please shed some light=20
> > > > on this ? Several Web sites(please check list below)
> > > > have schematics and flashy color photographs of=20
> > > > single white led light powered off a single 1.5 Volt=20
> > > > AA battery.
> > > > http://cappels.org/dproj/ledpage/leddrv.htm
> > > > http://www.electroschematics.com/6195/high-efficiency-led-with-1-5-=
volt
> > > > s/=20
> > > > dinspirations.com/appliedcontent/Projects/Proj-WhiteLEDDrvr/Proj-Wh=
iteL
> > > > EDDrvr.html As most of these are very simple, I have simulated=20
> > > > a number of them with SPICE, but none of them=20
> > > > produce the advertised results. What is the mystery ?
> > > > Some even have supposed oscilloscope traces !! Any=20
> > > > thoughts hints would be helpful. Thanks in advance.
> > >=20
> > > This class of single transistor blocking oscillator boost converter=
=20
> > > designed to drive a LED from a single cell is colloquially known as a=
=20
> > > "Joule Thief".   Searching for that term with the keywords 'SPICE=20
> > > simulation' will find many simulations of varying degrees of accuracy=
. =20
> > > The basic circuit is pretty forgiving, and can easily be got working =
with=20
> > > junk-box parts which accounts for its popularity.
> > >=20
> > > If the simulation doesn't model non-ideal behaviour of key components=
 to=20
> > > a reasonable level of accuracy, it will *NOT* produce realistic resul=
ts.
> > > Have you accounted for the cell's internal resistance, the resistance=
 of=20
> > > the transformer primary, transformer core saturation, the LED's junct=
ion=20
> > > capacitance and the transistor's gain falling at high collector curre=
nts?
> > >=20
> > > Post your simulations (preferably in LTSPICE format) and the observed=
=20
> > > results you are trying to match up with and chances are, someone will=
=20
> > > point out what you've missed.
> > >=20
> > > --=20
> > > Ian Malcolm.   London, ENGLAND.  (NEWSGROUP REPLY PREFERRED)=20
> > > ianm[at]the[dash]malcolms[dot]freeserve[dot]co[dot]uk=20
> > > [at]=3D@, [dash]=3D- & [dot]=3D. *Warning* HTML & >32K emails --> NUL
> >=20
> > Yes, I agree totally that unless the non-linearities in the key inducto=
r=20
> > behavior are modeled correctly, the=20
> > SPICE simulation would work. As for=20
> > the transistor, I have used the standard
> > SPICE model for 2N4401. The diode=20
> > has not yet been inserted. I was just
> > trying to see if the oscillations could=20
> > be simulated first. Unfortunately, I=20
> > use HSpice at work, and Ngspice at=20
> > home(the simulations were done in my
> > spare time), both of which use text
> > based source files as input, as opposed
> > to LTSpice's GUI based format.
>=20
> LT does the textfile format too
>=20
>=20
> NT

Well,
The source file is below. It runs on Ngspice-26.
The inductor is non-ideal, and the model for white light LED is also includ=
ed. As shown below, I have tried to measure current and voltage, and hence =
the three current measurement
nodes. The circuit is a minor variation of one
I saw at one of the Web sites I referred to, in=20
my initial post.

.PARAMETERS PI=3D3.14 PERMFS=3D0.000001256 RELPERM=3D500
.PARAMETERS NUMT=3D50 RAD=3D0.005 LEN=3D0.018825 CAPVALNOM=3D0.00001

.MODEL 2N4401 NPN(IS=3D1.75E-12 ISE=3D5.92E-14 ISC=3D9.42E-14 XTI=3D3.00 BF=
=3D3.03E2 BR=3D1.00E-2 IKF=3D2.11E-1 IKR=3D1.00 XTB=3D1.5 VAF=3D3.60E2 VAR=
=3D1.64E1 VJE=3D3.00E-1 VJC=3D3.00E-1 RE=3D1.00E-2 RC=3D1.07 RB=3D8.63E1 RB=
M=3D1.00E-2 IRB=3D9.62E-3 CJE=3D2.64E-11 CJC=3D1.37E-11 XCJC=3D1.00 FC=3D5.=
00E-1 NF=3D1.10 NR=3D1.71 NE=3D1.26 NC=3D1.00 MJE=3D4.09E-1 MJC=3D4.89E-1 T=
F=3D5.16E-10 TR=3D233.7E-9 PTF=3D0 ITF=3D5.09E-1 VTF=3D1.09E5 XTF=3D1.64 EG=
=3D1.11 KF=3D1E-9 AF=3D1)

.MODEL NSPW500BS D(Is=3D0.27n Rs=3D5.65 N=3D6.79 Cjo=3D42p Iave=3D30m Vpk=
=3D5)

.SUBCKT NIINDUCTOR 1 2
* 1 IN
* 2 OUT
C0 1 2 0.75nF
L0 1 3 L=3D'RELPERM*PERMFS*NUMT*NUMT*PI*RAD*RAD/LEN'
R0 3 2 1.5
.ENDS

D0 9 0 NSPW500BS
XL0 3 4 NIINDUCTOR
XL1 3 9 NIINDUCTOR
Q0 6 7 0 2N4401
R0 1 2 1
R1 4 5 1
R2 7 8 0.1

VP 1 0 DC 1.5 AC 0.0
VTST1 2 3 DC 0.0 AC 0.0
VTST2 5 6 DC 0.0 AC 0.0
VTST3 8 9 DC 0.0 AC 0.0

.OPTIONS METHOD=3DGEAR NOPAGE
.IC V(2)=3D0.0 V(3)=3D0.0 V(4)=3D0.0 V(5)=3D0.0
+ V(6)=3D0.0 V(7)=3D0.0 V(8)=3D0.0 V(9)=3D0.0
.TRAN 20.0ms 2000.0ms 5.0ms UIC
.PRINT TRAN V(6)
.END

On Friday, November 14, 2014 11:31:40 PM UTC-5, Jasen Betts wrote:
> On 2014-11-14, dakupoto@gmail.com <dakupoto@gmail.com> wrote:
> 
> > home(the simulations were done in my
> > spare time), both of which use text
> > based source files as input, as opposed
> > to LTSpice's GUI based format.
> 
> unfortunately (for you) LTSpice has become the lingua franca 
> for sharing simulations in this newsgroup, mainly because the
> price is right *1, it'll run on anything *2 and it's easier 
> to use than ms-paint *3
> 
> The save format is an ASCII *4 netlist with some embedded GUI 
> layout information, it can run other ASCII netlists.
> 
> *1 Free download
> 
> *2 Well, anyting with an x86 cpu - it runs OK in wine, and
> whatever it is that recent macs use to run windows apps
> 
> *3 if your goal is drawing electronic schematics
> 
> *4 If you tell it to use 'u' for micro, there's a checkbox 
> somewhere,else Windows-1252.
> 
> -- 
> umop apisdn

Ngspice is completely free, and has its own
users' and developers forums on SourceForge.
Though initially developed for Unix-like and
Linux platforms, precompiled binaries are 
available for Windows -- no issues at all. 
No need to install Wine, Linux/Unix is its 
native environment. Output files can be 
plotted with Excel or OpenOffice or Gnuplot.
The text based input makes it very easy to 
include special models -- about 2 years ago,
I had to do a lot of work with the sub-micron
MOSFET models from UC Berkeley(BSIM 4.6) --
all I had to do was have an .INCLUDE statement
in my source files.