Joule Thief - still not working....

[fungus]

On Jul 31, 12:44 pm, Jon Kirwan <j...@infinitefactors.org> wrote:

Are there any positive benefits to saturation in this case?  Perhaps.
If it is a very controlled amount.

My ham-fisted experiments seem to show that R1 can't
be used to control the output of a saturated circuit.

 

[Paul E. Schoen]

"Jon Kirwan" <jonk@infinitefactors.org> wrote in message
news:6ir675t6v3urp1i6ncn6ko3uip5tfeldet@4ax.com...

On Fri, 31 Jul 2009 14:38:58 -0700 (PDT), fungus
openglMYSOCKS@artlum.com> wrote:

On Jul 31, 12:44 pm, Jon Kirwan <j...@infinitefactors.org> wrote:

Are there any positive benefits to saturation in this case? Perhaps.
If it is a very controlled amount.

My ham-fisted experiments seem to show that R1 can't
be used to control the output of a saturated circuit.

R1 can't in that case. Did I seem to give a different impression?

Not in direct reply to this thread of the post, but I simulated the circuit
and I was able to add current regulation so that LED current stays within
20-27 mA from about 2 volts to 6 volts and efficiency is about 80%. With
some combinations of values (without regulation) I got efficiencies well
over 90%. But the output varies greatly with voltage, and with inductance
values. I found an 8:1 ratio worked pretty well. Of course real components
will not match the simulation.

Paul

Using LTSpice IV:

Version 4
SHEET 1 880 680
WIRE 80 0 -112 0
WIRE 208 0 80 0
WIRE 80 16 80 0
WIRE 208 16 208 0
WIRE 208 112 208 96
WIRE 368 112 208 112
WIRE -112 128 -112 0
WIRE 80 128 80 96
WIRE 208 176 208 112
WIRE 80 224 80 208
WIRE 80 224 0 224
WIRE 96 224 80 224
WIRE 144 224 96 224
WIRE 0 240 0 224
WIRE 96 256 96 224
WIRE 192 304 160 304
WIRE 288 304 272 304
WIRE 368 304 352 304
WIRE 496 304 368 304
WIRE 272 320 272 304
WIRE 496 320 496 304
WIRE -112 384 -112 208
WIRE 0 384 0 304
WIRE 0 384 -112 384
WIRE 96 384 96 352
WIRE 96 384 0 384
WIRE 208 384 208 272
WIRE 208 384 96 384
WIRE 272 384 208 384
WIRE 368 384 272 384
WIRE 496 384 368 384
WIRE 368 432 368 384
FLAG 368 432 0
SYMBOL npn 144 176 R0
WINDOW 0 33 47 Left 0
SYMATTR InstName Q1
SYMATTR Value 2N2219A
SYMBOL ind2 96 112 R180
WINDOW 0 36 80 Left 0
WINDOW 3 36 40 Left 0
SYMATTR InstName L1
SYMATTR Value 800ľ
SYMATTR Type ind
SYMBOL ind2 192 0 R0
SYMATTR InstName L2
SYMATTR Value 100ľ
SYMATTR Type ind
SYMBOL LED 352 240 R0
WINDOW 3 22 69 Left 0
SYMATTR InstName D1
SYMATTR Value AOT-2015
SYMBOL res 64 112 R0
SYMATTR InstName R1
SYMATTR Value 1k
SYMBOL voltage -112 112 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 3
SYMBOL LED 352 112 R0
SYMATTR InstName D2
SYMATTR Value AOT-2015
SYMBOL schottky 16 304 R180
WINDOW 0 24 72 Left 0
WINDOW 3 24 0 Left 0
SYMATTR InstName D3
SYMATTR Value 1N5818
SYMATTR Description Diode
SYMATTR Type diode
SYMBOL LED 352 176 R0
SYMATTR InstName D4
SYMATTR Value AOT-2015
SYMBOL res 352 288 R0
SYMATTR InstName R2
SYMATTR Value 30
SYMBOL npn 160 256 M0
WINDOW 0 33 47 Left 0
SYMATTR InstName Q2
SYMATTR Value 2N2219A
SYMBOL schottky 352 288 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName D5
SYMATTR Value 1N5818
SYMATTR Description Diode
SYMATTR Type diode
SYMBOL cap 256 320 R0
SYMATTR InstName C1
SYMATTR Value 100n
SYMBOL res 288 288 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 34 46 VTop 0
SYMATTR InstName R3
SYMATTR Value 1k
SYMBOL cap 480 320 R0
SYMATTR InstName C2
SYMATTR Value 10ľ
TEXT 280 72 Left 0 !K1 L1 L2 1
TEXT -66 360 Left 0 !.tran 10m startup

 

[fungus]

On Jul 31, 11:23 pm, Jon Kirwan <j...@infinitefactors.org> wrote:

In any case, you can try out the equations I discussed in selecting a
different core, too.  Pick another one (one you can get ahold of?) and
see where the numbers put you.  If they are about right (collector
winding in the area of more than 100uH, let's say), then you've got a
good shot at it.

I might pop into the local components store and see what they've got.
If I do I'll take my magnet along with me....

I don't think I'm up for ordering one from RS and paying $15 shipping
just as an experiment. Maybe if I end up buying some other stuff from
them I could slip one in.

 

[fungus]

On Aug 1, 12:21 am, Jon Kirwan <j...@infinitefactors.org> wrote:

On Fri, 31 Jul 2009 14:38:58 -0700 (PDT), fungus wrote:

R1 can't be used to control the output of a saturated circuit.

R1 can't in that case.  Did I seem to give a different impression?

I don't think you made a statement either way so I'm just making
it clear that there's a definite downside to saturation - loss of
control of the output voltage.

I agree that the best place to be is just below the saturation point.

 

[Jon Kirwan]

On Fri, 31 Jul 2009 17:20:22 -0700 (PDT), fungus
<openglMYSOCKS@artlum.com> wrote:

On Aug 1, 12:21 am, Jon Kirwan <j...@infinitefactors.org> wrote:
On Fri, 31 Jul 2009 14:38:58 -0700 (PDT), fungus wrote:

R1 can't be used to control the output of a saturated circuit.

R1 can't in that case.  Did I seem to give a different impression?

I don't think you made a statement either way so I'm just making
it clear that there's a definite downside to saturation - loss of
control of the output voltage.

I agree that the best place to be is just below the saturation point.

Or right at/just above, to shorten just slightly the time the BJT sits
at the tail end of its higher Vce-on with high collector currents.
This is something to play with, though. I suspect there is still more
yet to uncover on even this simple circuit.

Jon

 

[fungus]

On Aug 1, 2:11 am, "Paul E. Schoen" <p...@peschoen.com> wrote:

Not in direct reply to this thread of the post, but I simulated the circuit
and I was able to add current regulation so that LED current stays within
20-27 mA from about 2 volts to 6 volts and efficiency is about 80%.

I plugged those results into my graph:

http://www.artlum.com/jt/jt_outputs.gif

By extrapolated the line using my Winnie The Pooh ruler I
can see that crosses the y-axis of the graph at 16mA.

ie. with zero volts input you have about 16mA output.

Maybe you should patent this circuit...

Of course real components will not match the simulation.

Paul

 

[Jon Kirwan]

On Fri, 31 Jul 2009 18:00:38 -0700 (PDT), fungus
<openglMYSOCKS@artlum.com> wrote:

On Jul 31, 12:44 pm, Jon Kirwan <j...@infinitefactors.org> wrote:

Air core is acceptable and you've already got some thoughts on that
here.  If you have ready access to that CAT5 or CAT6 cabling, that may
work.  The downside there is that the wire itself is quite thick and
your transformer will be sizeable.  If you have fine-gauge magnet wire
(did I read that you said you had some?), you might haul out a long
length of it, fold it over, and try and impart a uniform number of
twists per foot/meter in it and use it as a bifilar pair.


Winding an air core would be quite quick compared to trying
to wind long pieces of wire onto a tiny ring (which is a real
pain in the ass btw.) so I might try the 4:1 experiment with
an air core.

I assume there has to be actual air there, not a stick, so whatever
I wind it onto has to be removed, right?

Nah. Use a piece of plastic or dry wood. (Wet? I don't know.) A
sewing thread spool?

Does the diameter of the core matter?

Yes.

It would be easy
to wind onto something with a large diameter (eg. a pill
bottle) then tie it with string to stop it falling apart when I
remove the bottle. Something long/thin like a pencil would
be much harder to do that.

See these, for fun:
http://microblog.routed.net/wp-content/uploads/2008/10/pancakewheel.pdf
http://www.qsl.net/in3otd/indcalc.html
http://www.phy.auckland.ac.nz/Staff/geb/Inductance%20Problem.pdf
http://www.moshier.net/coildoc.html

Plus there is this commentary from Alan Yates which may help. It
basically covers errors that come from having a short, stubby coil
because the assumptions made in deriving the equations no longer hold
true in that case:

"The other result is for an "long" single-layer solenoid, and ignores the
fringing fields that result from its finite length. The fringe field's
flux leakage means it doesn't cut all the turns so the self-inductance
measured is lower than calculated using this model. You might think of
it as an upper-bound on the inductance you are likely to get.

It is calculated using:

L = (u * n^2 * A) / l

Where:
u is the permeability of the core (k * u0 or free-space u0)
n is the turn count
A is the cross-sectional area of the coil (m^2)
l is the length of the coil (m)

u0 = 4*pi*1e-7 H/m

For small coils your measured values may be larger than the Wheeler
calculated value because of the thickness of the wire. The diameter
should actually be the average "current sheet" diameter, so you might
get a closer result if you add the wire thickness to the diameter (which
will increase the inductance calculated a bit).

The Wheeler formula is normally quite good, accurate to a few percent
for the average single-layer solenoid you'd use in RF applications. It
is specified as being accurate when l > 0.4 * d:

L = (d^2 * n^2) / (l + (0.45 * d))

Where L is in uH and the l and d in metres.

For shorter coils there is a series of length to diameter tables
published by Nagaoka which can be used, but I've never had to. One of
these days I'll improve the coil calulator with other formula."
-Mr Alan Yates-

 

[fungus]

On Jul 31, 12:44 pm, Jon Kirwan <j...@infinitefactors.org> wrote:

Air core is acceptable and you've already got some thoughts on that
here.  If you have ready access to that CAT5 or CAT6 cabling, that may
work.  The downside there is that the wire itself is quite thick and
your transformer will be sizeable.  If you have fine-gauge magnet wire
(did I read that you said you had some?), you might haul out a long
length of it, fold it over, and try and impart a uniform number of
twists per foot/meter in it and use it as a bifilar pair.

Winding an air core would be quite quick compared to trying
to wind long pieces of wire onto a tiny ring (which is a real
pain in the ass btw.) so I might try the 4:1 experiment with
an air core.

I assume there has to be actual air there, not a stick, so whatever
I wind it onto has to be removed, right?

Does the diameter of the core matter? It would be easy
to wind onto something with a large diameter (eg. a pill
bottle) then tie it with string to stop it falling apart when I
remove the bottle. Something long/thin like a pencil would
be much harder to do that.

 

[Michael A. Terrell]

fungus wrote:

On Jul 31, 11:23 pm, Jon Kirwan <j...@infinitefactors.org> wrote:

In any case, you can try out the equations I discussed in selecting a
different core, too. Pick another one (one you can get ahold of?) and
see where the numbers put you. If they are about right (collector
winding in the area of more than 100uH, let's say), then you've got a
good shot at it.


I might pop into the local components store and see what they've got.
If I do I'll take my magnet along with me....

I don't think I'm up for ordering one from RS and paying $15 shipping
just as an experiment. Maybe if I end up buying some other stuff from
them I could slip one in.

Look at the Amidon or Fair-rite websites for some information on the
different materials and the frequencies they are designed for.

https://www.amidoncorp.com/

https://www.amidoncorp.com/pages/specifications gives you an idea of
what materials they sell.


If you order anything, ask for a copy of their free brochure:

https://www.amidoncorp.com/items/55 or request a copy by following the
information on that page.

http://www.fair-rite.com/


--
You can't have a sense of humor, if you have no sense!

 

[Paul E. Schoen]

"fungus" <openglMYSOCKS@artlum.com> wrote in message
news:cb79617e-f274-4e94-bf75-486a9df14447@26g2000yqk.googlegroups.com...
On Aug 1, 2:11 am, "Paul E. Schoen" <p...@peschoen.com> wrote:

Not in direct reply to this thread of the post, but I simulated the
circuit
and I was able to add current regulation so that LED current stays
within
20-27 mA from about 2 volts to 6 volts and efficiency is about 80%.

I plugged those results into my graph:

http://www.artlum.com/jt/jt_outputs.gif

By extrapolated the line using my Winnie The Pooh ruler I
can see that crosses the y-axis of the graph at 16mA.

ie. with zero volts input you have about 16mA output.

Maybe you should patent this circuit...

Well, I never claimed it was linear. At 1 volt the LED current is 4.4 mA
Avg ( 20 mA RMS ), and 1.2 mA Avg ( 7.7 mA RMS ) with 0.7 V input. It stops
working at 0.5 V input.

32.5 mA at 8 volts and it stops oscillating at 10 volts. Even at 12 volts
the overcurrent protection turns off Q1 and there's just 89 mA in the LEDs
and the efficiency is still 88%.

Without the current limiter Q1 draws over 800 mA with 12 VDC applied so it
is toast. But at 2 VDC there is 25.7 mA, at 3 VDC it is 58 mA, and at 4 VDC
it's 99.5 mA. These values are just about 5 times your measured values, so
it shows that the experimental circuit is performing just about as the
simulation predicts, except the inductance values and other components are
certainly different.

In fact, If I use 20 uH for each coil, and an 8 kOhm resistor, I get almost
exactly the same readings you show on your chart for the megabead. But I
only get 23% efficiency at 3 volts, and 10% at 4 volts.

Of course at some point the inductor saturates and Q1 goes boom. The
simulation does not take that into account.

Maybe I should build this circuit. I have made an LED flashlight circuit
for about 40 watts at 12 VDC using a PIC and a boost converter. But this is
so much simpler!

Paul

 

[greg]

fungus wrote:

My ham-fisted experiments seem to show that R1 can't
be used to control the output of a saturated circuit.

That sounds right, because the end of the on-time
is being controlled by inductor saturation, rather
than the base current and transistor beta.

--
Greg

 

[greg]

fungus wrote:

I assume there has to be actual air there, not a stick, so whatever
I wind it onto has to be removed, right?

No, anything non-metallic and non-conductive --
wood, plastic, cardboard, etc. -- should make
very little difference.

Does the diameter of the core matter?

I would expect a large diameter and short length
to work better than the other way around, because
the total flux through the interior of a solenoid
for a given current is proportional to its cross-
sectional area and inversely proporptional to its
length.

--
Greg

 

[fungus]

On Aug 1, 9:13 am, "Paul E. Schoen" <p...@peschoen.com> wrote:

Not in direct reply to this thread of the post, but I simulated the
circuit
and I was able to add current regulation so that LED current stays
within
20-27 mA from about 2 volts to 6 volts and efficiency is about 80%.
I plugged those results into my graph:
http://www.artlum.com/jt/jt_outputs.gif
By extrapolated the line using my Winnie The Pooh ruler I
can see that crosses the y-axis of the graph at 16mA.
ie. with zero volts input you have about 16mA output.
Maybe you should patent this circuit...

Well, I never claimed it was linear.

I was just kidding...

I googled for LTspice to see what this "simulation" thing
is all about and had a look at your circuit - you added a
few extra components!

So what you've done is regulate the current better so the
output is a lot flatter as voltage drops then it suddenly falls
off at the end.

Maybe I should build this circuit. I have made an LED
flashlight circuit for about 40 watts at 12 VDC using a
PIC and a boost converter. But this is so much simpler!

I haven't got enough parts here to make one but if it works
as claimed then the extra efficiency could pay for the extra
ports quite quickly (with non-rechargeable batteries).

 

[fungus]

On Aug 2, 7:20 am, greg <g...@cosc.canterbury.ac.nz> wrote:

fungus wrote:
I assume there has to be actual air there, not a stick, so whatever
I wind it onto has to be removed, right?

No, anything non-metallic and non-conductive --
wood, plastic, cardboard, etc. -- should make
very little difference.

Ok, that makes life easier.

Does the diameter of the core matter?

I would expect a large diameter and short length
to work better than the other way around, because
the total flux through the interior of a solenoid
for a given current is proportional to its cross-
sectional area and inversely proporptional to its
length.

There seems to be disagreement over this. I'll
make a medium-size one when I get around to it.

 

[John Fields]

On Sun, 02 Aug 2009 17:20:07 +1200, greg <greg@cosc.canterbury.ac.nz>
wrote:

fungus wrote:

I assume there has to be actual air there, not a stick, so whatever
I wind it onto has to be removed, right?

No, anything non-metallic and non-conductive --
wood, plastic, cardboard, etc. -- should make
very little difference.

---
IME that's not true.

A couple of hundred years ago (it seems) I built a 2kW CW ultrasonic
system where I used a piece of PVC pipe as a former to wind the loading
coil for the transducer and when we fired it up and tuned the system to
resonance, the pipe melted.

"Dissipation factor", I later learned, was what had bitten me on the
ass, and I solved the problem by cannibalizing the ceramic (steatite)
core of an ARC5's loading coil and winding it to get the inductance we
needed for resonance at 40kHz.


JF

 

[Paul E. Schoen]

"fungus" <openglMYSOCKS@artlum.com> wrote in message
news:128b0b87-6c98-4e8d-8271-049d63d2e223@b15g2000yqd.googlegroups.com...
On Aug 1, 9:13 am, "Paul E. Schoen" <p...@peschoen.com> wrote:

Not in direct reply to this thread of the post, but I simulated the
circuit
and I was able to add current regulation so that LED current stays
within
20-27 mA from about 2 volts to 6 volts and efficiency is about 80%.
I plugged those results into my graph:
http://www.artlum.com/jt/jt_outputs.gif
By extrapolated the line using my Winnie The Pooh ruler I
can see that crosses the y-axis of the graph at 16mA.
ie. with zero volts input you have about 16mA output.
Maybe you should patent this circuit...

Well, I never claimed it was linear.

I was just kidding...

I googled for LTspice to see what this "simulation" thing
is all about and had a look at your circuit - you added a
few extra components!

So what you've done is regulate the current better so the
output is a lot flatter as voltage drops then it suddenly falls
off at the end.

Maybe I should build this circuit. I have made an LED
flashlight circuit for about 40 watts at 12 VDC using a
PIC and a boost converter. But this is so much simpler!

I haven't got enough parts here to make one but if it works
as claimed then the extra efficiency could pay for the extra
ports quite quickly (with non-rechargeable batteries).

------------------------------------------------------------

If you have installed LTspice then you should try their jig for LT1932
which is an LED driver that works down to 1.5 VDC and has built-in current
regulation. It runs at 1 MHz so the inductor can be tiny, and there are no
big spikes in the LED current as there are on the JouleThief without a
capacitor.

Also I tweaked the original circuit without regulation and achieved 90%
efficiency at 3 VDC and 87% at 2 VDC. The output current is roughly
proportional to the square of the input voltage, and the circuit works to
1.5 VDC. Some of the components might not be necessary, but I did find a
lot of variation with different BJTs. A 2N3904 seems to work best.

The Schottky diode D5 and filter capacitor on the output limit the high
current spikes that otherwise go through the LEDs. Another Schottky D7
limits the reverse voltage spikes. These spikes could damage the LEDs.

R1 does control the output current, and the small capacitor across it
greatly speeds up the transitions and increases efficiency. I added an
input bypass capacitor and realistic source resistance. If your voltage
source is not close to the circuit, a bypass is needed.

It seems to work OK with equal values (10 uH) of L1 and L2, but with L1=40
uH I was able to get it to work at 1 VDC input, and 69% efficiency with
about 3.4 mA output. The frequency is 800 kHz, which might be too high for
a successful deadbug prototype.

100 uH coils brings the frequency down to 135 kHz and about the same
efficiency, but less output. I boosted R1 and I ran some more simulations
with even better results, so I left it at that. It should be interesting to
build and test an actual circuit.

Paul

===================================================

Version 4
SHEET 1 880 680
WIRE -112 0 -224 0
WIRE 0 0 -112 0
WIRE 208 0 0 0
WIRE 208 16 208 0
WIRE 0 32 0 0
WIRE 208 112 208 96
WIRE 240 112 208 112
WIRE 368 112 304 112
WIRE 464 112 368 112
WIRE 496 112 464 112
WIRE -112 128 -112 0
WIRE -224 160 -224 0
WIRE 0 160 0 112
WIRE 80 160 0 160
WIRE 208 160 208 112
WIRE 288 160 208 160
WIRE 0 176 0 160
WIRE 80 176 80 160
WIRE 288 192 288 160
WIRE 368 208 368 176
WIRE 496 208 496 112
WIRE 208 224 208 160
WIRE 0 272 0 256
WIRE 80 272 80 240
WIRE 80 272 0 272
WIRE 144 272 80 272
WIRE 368 288 368 272
WIRE 0 304 0 272
WIRE -224 384 -224 224
WIRE -112 384 -112 208
WIRE -112 384 -224 384
WIRE 0 384 0 368
WIRE 0 384 -112 384
WIRE 208 384 208 320
WIRE 208 384 0 384
WIRE 288 384 288 256
WIRE 288 384 208 384
WIRE 368 384 368 352
WIRE 368 384 288 384
WIRE 496 384 496 272
WIRE 496 384 368 384
WIRE 368 432 368 384
FLAG 368 432 0
FLAG 464 112 Vout
SYMBOL npn 144 224 R0
WINDOW 0 43 29 Left 0
WINDOW 3 26 59 Left 0
SYMATTR InstName Q1
SYMATTR Value 2N3904
SYMBOL ind2 16 128 R180
WINDOW 0 36 80 Left 0
WINDOW 3 36 40 Left 0
SYMATTR InstName L1
SYMATTR Value 100ľ
SYMATTR Type ind
SYMBOL ind2 192 0 R0
WINDOW 3 37 69 Left 0
SYMATTR InstName L2
SYMATTR Value 100ľ
SYMATTR Type ind
SYMBOL LED 352 288 R0
WINDOW 3 22 69 Left 0
WINDOW 0 -28 23 Left 0
SYMATTR Value AOT-2015
SYMATTR InstName D1
SYMBOL res -16 160 R0
SYMATTR InstName R1
SYMATTR Value 7.5k
SYMBOL voltage -112 112 R0
WINDOW 123 0 0 Left 0
WINDOW 39 24 132 Left 0
SYMATTR SpiceLine Rser=.1
SYMATTR InstName V1
SYMATTR Value 1.5
SYMBOL LED 352 112 R0
WINDOW 0 -31 32 Left 0
SYMATTR InstName D2
SYMATTR Value AOT-2015
SYMBOL schottky 16 368 R180
WINDOW 0 24 72 Left 0
WINDOW 3 24 0 Left 0
SYMATTR InstName D3
SYMATTR Value 1N5818
SYMATTR Description Diode
SYMATTR Type diode
SYMBOL LED 352 208 R0
WINDOW 0 -29 19 Left 0
SYMATTR InstName D4
SYMATTR Value AOT-2015
SYMBOL schottky 240 128 R270
WINDOW 0 20 73 VTop 0
WINDOW 3 0 32 VBottom 0
SYMATTR InstName D5
SYMATTR Value 1N5818
SYMATTR Description Diode
SYMATTR Type diode
SYMBOL cap 480 208 R0
SYMATTR InstName C1
SYMATTR Value 1ľ
SYMBOL schottky 304 256 R180
WINDOW 0 24 72 Left 0
WINDOW 3 -20 -2 Left 0
SYMATTR InstName D7
SYMATTR Value 1N5818
SYMATTR Description Diode
SYMATTR Type diode
SYMBOL cap -240 160 R0
SYMATTR InstName C2
SYMATTR Value 100ľ
SYMBOL cap 64 176 R0
SYMATTR InstName C3
SYMATTR Value 33p
TEXT 24 24 Left 0 !K1 L1 L2 1
TEXT -24 416 Left 0 !.tran 10m startup
TEXT 264 8 Left 0 ;92% eff at 3 VDC, 110 kHz, 25 mA output
TEXT 264 32 Left 0 ;87% eff at 2 VDC, 127 kHz, 10.6 mA output
TEXT 264 -16 Left 0 ;92.7% eff at 4 VDC, 99 kHz, 44 mA output
TEXT 264 56 Left 0 ;86% eff at 1.5 VDC, 136 kHz, 5.4 mA output

 

[Jon Kirwan]

On Sun, 02 Aug 2009 18:15:08 -0500, John Fields
<jfields@austininstruments.com> wrote:

On Sun, 02 Aug 2009 17:20:07 +1200, greg <greg@cosc.canterbury.ac.nz
wrote:

fungus wrote:

I assume there has to be actual air there, not a stick, so whatever
I wind it onto has to be removed, right?

No, anything non-metallic and non-conductive --
wood, plastic, cardboard, etc. -- should make
very little difference.

---
IME that's not true.

A couple of hundred years ago (it seems) I built a 2kW CW ultrasonic
system where I used a piece of PVC pipe as a former to wind the loading
coil for the transducer and when we fired it up and tuned the system to
resonance, the pipe melted.

Interesting, and believable. Luckily... I think the 1/2 Watt system
will do about 4000 times less damage to the PVC. It just might
survive.

"Dissipation factor", I later learned, was what had bitten me on the
ass, and I solved the problem by cannibalizing the ceramic (steatite)
core of an ARC5's loading coil and winding it to get the inductance we
needed for resonance at 40kHz.

This is a fun addition. And hopefully the OP will enjoy playing
around with all this.

Jon

 

[fungus]

On Aug 3, 1:15 am, John Fields <jfie...@austininstruments.com> wrote:

A couple of hundred years ago (it seems) I built a 2kW CW ultrasonic
system where I used a piece of PVC pipe as a former to wind the loading
coil for the transducer and when we fired it up and tuned the system to
resonance, the pipe melted.

Ah, now I understand why the datasheets for ferrite
rings show performance graphs up to 500 degrees C ....

 

[bw]

"Paul E. Schoen" <paul@peschoen.com> wrote in message
news:%wqdm.353$Jg.293@nwrddc01.gnilink.net...

"fungus" <openglMYSOCKS@artlum.com> wrote in message
news:128b0b87-6c98-4e8d-8271-049d63d2e223@b15g2000yqd.googlegroups.com...
On Aug 1, 9:13 am, "Paul E. Schoen" <p...@peschoen.com> wrote:

Not in direct reply to this thread of the post, but I simulated the
circuit
and I was able to add current regulation so that LED current stays
within
20-27 mA from about 2 volts to 6 volts and efficiency is about 80%.
I plugged those results into my graph:
http://www.artlum.com/jt/jt_outputs.gif
By extrapolated the line using my Winnie The Pooh ruler I
can see that crosses the y-axis of the graph at 16mA.
ie. with zero volts input you have about 16mA output.
Maybe you should patent this circuit...

Well, I never claimed it was linear.

I was just kidding...

I googled for LTspice to see what this "simulation" thing
is all about and had a look at your circuit - you added a
few extra components!

So what you've done is regulate the current better so the
output is a lot flatter as voltage drops then it suddenly falls
off at the end.


Maybe I should build this circuit. I have made an LED
flashlight circuit for about 40 watts at 12 VDC using a
PIC and a boost converter. But this is so much simpler!

I haven't got enough parts here to make one but if it works
as claimed then the extra efficiency could pay for the extra
ports quite quickly (with non-rechargeable batteries).

------------------------------------------------------------

If you have installed LTspice then you should try their jig for LT1932
which is an LED driver that works down to 1.5 VDC and has built-in current
regulation. It runs at 1 MHz so the inductor can be tiny, and there are no
big spikes in the LED current as there are on the JouleThief without a
capacitor.

Also I tweaked the original circuit without regulation and achieved 90%
efficiency at 3 VDC and 87% at 2 VDC. The output current is roughly
proportional to the square of the input voltage, and the circuit works to
1.5 VDC. Some of the components might not be necessary, but I did find a
lot of variation with different BJTs. A 2N3904 seems to work best.

Nice analysis. I've tested many other types, "switchers" work best (2N1711
and 2N2019) on the original circuit, but with the R1 cap then almost any BJT
will work. There is more than hfe involved, transistor capacitance seems to
be significant, and frequency. I've never reached 50 percent efficiency,
though.

The Schottky diode D5 and filter capacitor on the output limit the high
current spikes that otherwise go through the LEDs. Another Schottky D7
limits the reverse voltage spikes. These spikes could damage the LEDs.

I've not tested these. I've not noticed LED damage after several days. I did
try hanging a 6 volt zener on the base, which does protect the BE junction
but cuts efficiency.

R1 does control the output current, and the small capacitor across it
greatly speeds up the transitions and increases efficiency. I added an
input bypass capacitor and realistic source resistance. If your voltage
source is not close to the circuit, a bypass is needed.

Yes, but it is not linear. Transistors are current devices.

It seems to work OK with equal values (10 uH) of L1 and L2, but with L1=40
uH I was able to get it to work at 1 VDC input, and 69% efficiency with
about 3.4 mA output. The frequency is 800 kHz, which might be too high for
a successful deadbug prototype.

100 uH coils brings the frequency down to 135 kHz and about the same
efficiency, but less output. I boosted R1 and I ran some more simulations
with even better results, so I left it at that. It should be interesting
to
build and test an actual circuit.

Paul

My coils seem to be around 40 to 50 uH. About 30 turns on a 1/4 inch diam
ferrite rod salvaged from an old TV. Symetric winding is not optimal, I
think there is improved output with 20 turns on the collector side and 10 on
the base side, but I've not checked the waveform. The original circuit with
a 3300 pF cap across R1 gives a fairly good waveform at around 100 kHz and
10 percent duty cycle. 1 1/2 volts input gives 12 volt peaks. Very
non-linear voltage response.

 

[bw]

"fungus" <openglMYSOCKS@artlum.com> wrote in message
news:b385ea09-e364-45b6-9eb6-646a57ac7ddf@o15g2000yqm.googlegroups.com...
On Aug 3, 1:15 am, John Fields <jfie...@austininstruments.com> wrote:

A couple of hundred years ago (it seems) I built a 2kW CW ultrasonic
system where I used a piece of PVC pipe as a former to wind the loading
coil for the transducer and when we fired it up and tuned the system to
resonance, the pipe melted.

Ah, now I understand why the datasheets for ferrite
rings show performance graphs up to 500 degrees C ....

I think your original tries were at too high frequency and you expected too
much.
The original circuit works well for what it is, a very simple, cheap LED
driver using a single 1.5 volt supply.