Line Driven LED + Waveforms

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,262
Location
New Jersey
Hello again everyone in LED land :)

I recently designed a line driven LED
circuit for running a single white LED
from a 120vac 60Hz line source without
a dropping resistor or transformer.
The total power consumption is about
1/3 of a watt.
The circuit with the results of a simulation
run are posted at: http://hometown.aol.com/xaxo/

Not shown is the pulse simulation that
would show the peak current through the
LED for a 170 volt fast rising wavefront
simulating circuit turn on when the ac
input wave is at a peak. The max surge
current through the LED during this pulse
test was about 40ma.

Keep in mind the actual circuit hasnt
been built yet so its still a little prelim.
The size of the fuse may change to 150ma
if that works out ok. If i can get some
MOV's to test out i might replace the 50uf
cap with a MOV instead.

Good luck with your LED circuits :)

--Al
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:
I recently designed a line driven LED
circuit for running a single white LED
from a 120vac 60Hz line source
<HR></BLOCKQUOTE>

When I made circuits like that I used a single diode antiparallel with the LED instead of the bridge rectifier. Of course you get much more flicker then.
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by PeLu:
When I made circuits like that I used a single diode antiparallel with the LED instead of the bridge rectifier. Of course you get much more flicker then.<HR></BLOCKQUOTE>

Yes, plus you need a capacitor twice the size
(1uf, 400v) in order to pump 20ma through
the LED because half the time the LED isnt
conducting :) That means the 500 ohm
resistor dissipates twice as much power too.
I thought the best utilization of parts
and ratings came from using a bridge
rectifier so i stuck with that.
Of course there are other drawbacks, such
as buying a good quality cap for the input.
It might turn out to be cheaper just to buy
a dc adapter and power the LED with that :)

--Al
 
This would be good for a night light or a background light, like maybe for behind a computer monitor.

Any thoughts on a multiple LED circuit design with one or two batteries?
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:
Yes, plus you need a capacitor twice the size
(1uf, 400v) in order to pump 20ma through
the LED
<HR></BLOCKQUOTE>

Maybe I had a too large capacitor at hand, cannot remember.
Anyway, I made several NiCd chargers like that. Built into a (larger) mains plug and charging 1-4 24V or 36V batteries. It had 3 different selectable charging currents (50-200mA).
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by jeff1500:

Any thoughts on a multiple LED circuit design
with one or two batteries?
<HR></BLOCKQUOTE>

Hi again Jeff,

Yes i think the "Brinkmann" circuit will
work for up to maybe 6 LED's wired in
parallel. The resistor R2 has to be lowered
in order to supply the extra current.
The coil L1 has to be a bit bigger though.
You can probably get away with a hand wound
toroid with 10 to 20 turns.
I did a simulation on that circuit with
1 to 6 white LEDs in parallel, but i put
one 10 ohm resistor in series with each
LED. Also, i used a Schottky diode on the
output with a 5uf output filter capacitor.
I found full brightness to be easier to
attain with the diode and output cap.

The simulation showed that if you can get one
LED to conduct 20ma ( by adjusting R2 ) then
you can get 6 LED's to conduct 20ma by
dividing the value of R2 by 6 and connecting
5 more LEDs in parallel (with a series 10 ohm
resistor for each LED).
The transistors used were 2N4403 PNP and
2N2222A NPN.
The only thing i couldnt determine was
whether or not the 2N2222A transistor
would get too hot or not. It looked ok
with 6 LEDs and R2/6, but i would have to
build the circuit to find out for sure.
Should work fine for 4 LEDs, with the value
of R6 for one led divide by 4.
Replacing the 2N2222A transistor with a
2N3055 general purpose audio transistor
and the circuit doesnt work at all.
I suspect its because the transistor gain
and switching speed arent enough.
Because of this, i would think other high
powered general purpose transistors wont work
either (because of low gain), unless you
get a special type that has high gain, fast
switching speed, and fairly low Vsat.
Zetex makes some some good transistors, but i
havent been able to get ahold of any yet.

Regulation isnt all that good, so brightness
diminishes as battery voltage falls.

I ran a simulation on another circuit which is
very similar but has feedback. The only drawback
is that it takes 2 more transistors to provide
the regulation circuitry, and a few more resistors
and one more capacitor. Im not really sure
the extra complexity is worth it.
I'll have to build this circuit up to make sure
it behaves like the simulation did also.

Can you get LM339 ic's? These chips can be
quite useful in these kinds of apps.
They are dirt cheap too :)
The data sheet says they will work down to
2 volts also.
Im thinking of making a circuit up using
one of these too, with super regulation
and multiple leds etc.
Zetex has some specialized chips out just for white leds,
but again i would have to get ahold of some before
i could use them. The other drawback is the
best one only comes in surface mount package.

I did get to try out a MAX756 ic chip, and its
pretty good for 2 cell apps, but is quite a pain
in one cell apps even though the data sheet says
it works down to 0.7 volts. With only one AA cell,
you can only really drive 1 LED because of the
very unusual way the chip works. Yeah it 'works'
down to 0.7 volts or so, but at hardly any
output current.
For 2 cell apps though, you could probably drive
20 leds or something. Once you get about
2 volts input, the chip runs smoothly.
Below 1.2 the output current gets very
skimpy. Unfortunately, there is also no
apparent way to fool the chip with a
pre-regulator/booster either, as the same pin
that is used for the power supply input is
also the voltage feedback pin! It bootstraps
the MOSFET drive voltage for 'normal'
mosfet drive once the output voltage ramps up
to par. The only trouble is, if you have a
significant load on the output, it never
ramps up to par :) so you never see
the benefit of the bootstraped drive voltage.
In any case, for one cell apps i would stay
away from this chip except for driving only one led.
Even then, i would go with the MAX757, which allows
for output voltage adjustment with two resistors.
(Oh yeah, the MAX756 has two preset output settings:
either 3.3 volts or 5.0 volts, so you also cant
fool it into current feedback mode of operation
either :-( Instead, you have to settle for setting
it up at 5 volts and use a current limiting resistor.)

The nice thing about discreet transistors is you can
use them down to about 1 volt any way you care too :)

Summing up:

If your in a hurry for a simple circuit to drive 4 leds
then use the brinkmann circuit with resistors and Schottky
etc as above (two cell app) and home wound coil.

If you can wait a little and you dont mind using
4 transistors i'll be working on the regulator version,
which provides 20ma to 4 LED's (probably 6) from an
input of 2 to 3 volts or so (probably regulates down
to 1.5 volts input and can take up to 4.5 volts input).

Good luck with your LED circuits :)

--Al
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by PeLu:

Anyway, I made several NiCd chargers like that. Built into a (larger) mains plug and charging 1-4 24V or 36V batteries. It had 3 different selectable charging currents (50-200mA).
<HR></BLOCKQUOTE>

I wish i thought of that :)
I think ill try that too...

--Al

ps. Oh one thing i did do sorta like that:
i used a 40 watt light bulb in series with
a single diode to make a half wave 12volt
emergency car battery charger one time :)
Although now i realize that the battery
gets a nice bang of high current when the
charger is first turned on as the bulb
is heating up :) I wouldnt do a small
battery this way :)
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:
Replacing the 2N2222A transistor with a
2N3055 general purpose audio transistor
and the circuit doesnt work at all.
I suspect its because the transistor gain
and switching speed arent enough.
<HR></BLOCKQUOTE>

I have made my last circuit with a 2N3055 (called BD130 by Siemens) in the seventies, but as much as I remember, it's beta is pretty low.
(It was a step up regulator for an 8W fluorecent tube, these things were not available at this time).
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:
I wish i thought of that <HR></BLOCKQUOTE>

Actually all these tiny rechargeable flashlights (available since the late 50ies) which you put into the mains socket work with capacitive chargers.

My problem was that I had to charge several of these large batteries at a time and this would need a pretty large unit. As I charged them pretty slow, overcharging was less of a problem (at least I thought so 17 years ago). I just checked them for temperature by hand. Of course was this unit very far from beeing save, if a cable breaks, you have full mains voltage on your batteries. I also included a tiny neon bulb for checking 'polarity', that means that the batteries always must be at the 'low' end of mains voltage.

You can make easily such a small flashlight out of 3 NiCd cells ( I would prefer them for this purpose) one or more LEDs and the simple charging current. I used to use this NiCd memory backup batteries from PCs (now out of style), they had some 70-100mAh and were available for free when I soldered them out of defect PC mainboards. They whole 3.6V battery had about half the size of a 123 cell. I did not get much runtime, but as it is so easy to just plug it in for recharge, this was almost never a problem.
The whole unit fits easily into a normal plug.

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>
i used a 40 watt light bulb in series with
a single diode to make a half wave 12volt
emergency car battery charger one time :)
Although now i realize that the battery
gets a nice bang of high current when the
charger is first turned on as the bulb
is heating up
<HR></BLOCKQUOTE>

Actually this was common practice in the 50ies and 60ies. And some people think that this high current spike does something good to the battery. Selenium rectifiers could take this high peak easily.
The very first article published about recharging Leclanché cells looked similar to this (Netherlands, late 1940ies).
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:
The simulation showed that if you can get one LED to conduct 20ma ( by adjusting R2 ) then you can get 6 LED's to conduct 20ma by dividing the value of R2 by 6 and connecting 5 more LEDs in parallel (with a series 10 ohm resistor for each LED).<HR></BLOCKQUOTE>

That's really good information. Should make for a good project this winter. The regulator circuit sounds good too.

I think a lot of people have been wanting to build things like this but it's been hard to figure out how. Buying something from a factory certainly has it's place, but building things yourself adds a whole new dimension.
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by jeff1500:

I think a lot of people have been wanting to build things like this but it's been hard to figure out how. Buying something from a factory certainly has it's place, but building things yourself adds a whole new dimension.
<HR></BLOCKQUOTE>


Im glad you can use this, it's kind of
interesting to check this stuff out.
Also, if anyone is interested i did
a small compact LED flashlight and
i wanted pushbutton/on/off and intensity
control, so i used a cheap DIP switch
to switch on/off (along with a small
momentary contact push button) and
to switch between light output intensities
(just switches in diff value resistors).
The dip switches available almost everywhere
are very small so they fit into even small
flashlights. An 8 position dip switch
has 8 tiny switches all rated at about
100ma each and fits into a small space.

Also, im working with someone else who deals
a lot with white LED's and we are in the
process of designing a new kind of LED
flashlight that is solar rechargable and
has a touch on/off/dimmer switch built in.
It's kind of neat because the harder you
touch the contacts, the brighter the light
output becomes. Is it the first LED
flashlight with a touch contact dimmer
switch? :)
If anyone is interested, let me know.
I cant say exactly when these will be
available just yet though.

Take care for now,
--Al
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by jeff1500:
MrAl's circuit as I understand it

Here's a link to a diagram I just drew with the ideas we've been discussing.
<HR></BLOCKQUOTE>

That's a nice drawing Jeff :)

I should just point out the following notes:

R2 might have to be scaled for the type of NPN transistor used ie low gain means lower R2,
but then after that the value is determined using the "divide R2 by the number of LEDs"
rule.
This means if 4k works for one LED (for 20ma) then 2k for two LEDs, 1k for 4 LEDs.
For example, 2.8k worked for one LED in
one circuit with a 2N4401 NPN transistor.

The wire gauge used for the coil can probably be as low as #32, which works out to
a series resistance about 0.15 ohms, which is very insignificant in this low current
circuit.

C2=2uf per LED seems ok, the current doesnt have to be perfectly smooth. This means 2uf for 1 LED, 4uf for 2 LEDs, 8uf for 4 LEDs.

ADDED LATER:
I just looked up that NTE11 transistor, and
i can see its a high current device.
I cant say that the values of R2 we are using
will work very well with that transistor.
If you can find me a "2N" part number for
that device i might be able to find values
that will work with that transistor.
I used a 2N4403 or a 2N2222 transistor in
all the circuit simulations.
END

ADDED EVEN LATER:
I tested another NPN with that circuit.
It's a 3 amp device and has a dc gain around
350. This transistor required R2=8k for
one LED.
Also, the input voltage plays a big
part in the selection of R2 also.
For 1 cell apps, a lower R2 value will
have to be used then for 2 cell apps.
Since the coil L1 inductance plays a part in
this also, i would suggest the following:
(using one 10 ohm resistor in series with
each LED)
1. Select Q2, wind L1, and select the number
of cells (or input voltage) and make R2=20k.
2. Hook up the circuit with 1 LED, then
measure the current through the LED with
a dc meter.
3. Lower R2 untill you get 20ma through the
LED.
4. Now to add LED's, add one LED and lower
R2 to 1/2 its value obtained above. Measure
the current to make sure its 20ma through
one LED.
5. To add 2 more LED's (total 4) lower
R2 again to 1/2 its value from #4 above.
Add the LED's and measure again the current
through one LED and verify its still 20ma.

Also, it looks like as low as 1uf per LED
will work out ok too, so 4uf for 4 LED's
(or higher is ok too).

One more thing, since there is a 10 ohm
resistor in series with each LED, you can
measure the dc voltage across the 10 ohm
resistor to determine the current. Simply
divide the voltage reading by 10.
This means each LED's 10 ohm resistor should
have close to 0.200 volts dc across it.
Any inexpensive digital or analog meter
should work ok for this.
END ADDED #2.

Good luck with your LED circuits,
--Al
 
I have an NTE11, ECG128P, and a 2N2222A. All three of them work in the one transistor circuit. The NTE11 got a little roasted the other day when I hooked something up inside out, but it still seems to work okay.

I have some values. I don't really know what they mean:

NTE11:
Vcbo 40v, Vceo 20v, Vfbo 7v, Ic 5A, Pd 750 m, hFE 230 min, fT 150MHz

ECG128P:
Vcbo 100v, Vceo 80v, Vebo 7v, IC 1A, PD 1W, FT 100 MHz, Hfe 100 min

2N2222A: hFE 132

I've also go a 5k trim pot for testing R2. Sounds like time for a trip to the store to find a few more pieces.

I keep looking at my TV remote control. It's a thin ergonomic shape that fits nicely in one hand. It holds two AA cells and has lots of room for a circuit board. I wonder if I can clone it for a package?
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by jeff1500:

I have some values. I don't really know what they mean:
<HR></BLOCKQUOTE>

This link gives more specs: http://www.nteinc.com/Web_pgs/device_list.html

From what i see there, it looks like the
NTE11 transistor is a good transistor
for this kind of circuit :)
From what i saw of the 2N3055 transistor,
the gain is low and so is the frequency,
so i would rather use another type, unless
i didnt have anything else laying around.
It will oscillate but with a much lower
value of R2 and only at a lower frequency.
After all, its original app was linear audio,
not switching. The NTE11 is about 10 times
better except of course the collector current
rating is lower (but still plenty high
enough for this app).

Oh by the way, five of the most important
transistor characteristics:
HFE=dc current gain
Vceo=max voltage across collector emitter
IC=max collector current
FT=transition frequency
Vsat=collector emitter saturation voltage

HFE is the current gain in a dc circuit.
For example, if you have 2ma going into the
base and 200ma flowing through the collector,
you have an HFE of 100 because 2ma times 100
equals 200ma.

Vceo is just the breakdown voltage of the
transistor across collector and emitter.
If you exceed this voltage, the transistor
will break down and be permanantly damaged.

IC is just the maximum current that can
flow through the collector before the
transistor breaks down. Sometimes there
are two ratings for this:
continuous current and max current.
The continuous current rating is just that,
while the max current is the max for a pulse
of current.

FT is just the ac frequency at which the transistor gain goes down to 1. Since
the gain goes down as frequency goes up,
at a high enough frequency the transistor
isnt useable any more because you cant get
any gain out of it, and thats what makes the
transistor useful in the first place:)

Vsat is just the voltage drop across the
collector and emitter when the transistor
is turned fully 'on'. There is usually a
collector current spec with this (like
200ma or something) so you know what current
is flowing through the collector at the time
the Vsat is spec'd. If you go above this
current, the vsat goes up too.
The Vsat spec is important for high efficiency and for app's that have to
work at very low voltage (like 1.5v).
The lower this spec, the better the transistor, but keep in mind most NPN
transistors like we are using all have
Vsat below 0.4 volts for currents around
150ma or so.

Oh as far as the parts go, i would say if
you like building these circuits up then
you should probably get more parts to
try out. You can get packs of 50 resistors
with an assortment of values. Do you order
on the web too?

Well good luck with your LED circuits,
and ill be posting more circuits on my
home page now and then too, along with
some waveforms.

--Al
 
I got a few more parts today. I got NTE 11 and 12 for NPN and PNP respectively. They come in complementary pairs.

We've been talking about 10k for R1. All the 10k packages were gone in the store. I got 11k instead. I wonder what will happen? It's kind of awkward to get to that particular store so I took what I could get.

I hope to have time to put it together in the next few days. I'll report results be they good or bad. Thanks for all the help.
 
Geeze Jeff, if your having that much trouble
getting parts i'll send you a selection of resistors free -- no prob. I could aford
a couple of the most common values :)
How much would it be to mail a small
envelope of a few parts or so from the USA?
1/4 watt resistors dont weigh much he he.

BTW, i think i found another improvement
to the circuit-- adding one more
resistor and a small capacitor seems to
boost efficiency by about 10%. I'll have
to do more testing. Looks good so far
though. I'll be posting on my home page.

Maybe someday this forum will come up with
its own picture uploading capability for
schematics and stuff.
I kind of like the ascii schematics too but
they are a real pain to create.

--Al
 
http://edusite10.tripod.com/led2/brink/bboard.html

Here's a breadboard layout sketch I made at lunchtime today. Ya, resistors, there's a radio shack at the mall. They've got lots of 'em.

Actually I think I'm ready to go with the parts I've got. I've got a 5k trim pot, some 11k and some 1k resistors. I'll hook them all together and then gradually lower the value to see what happens.

The hardest part about the project is trying to decide if I should do this or do all my chores instead.
 
I decided to do both my chores and work on the multi-led circuit.

I put it together this morning but it's not working yet. I'm using this circuit:
http://edusite10.tripod.com/led2/brink/multi_led.htm

Here's a photo of the test setup. http://edusite10.tripod.com/led2/brink/test.html

I got some 10k resistors for R1. I tried it with one led with R2=3.5k without D1,C2, or R3 but still no voltage gain. I just measure 3v at the led. I tried various R2 values between 1k and 10k with one and two leds.

I'm suspicious about the C1 capacitor I'm using. It's 680 pico Farads but it has a very high voltage rating of 1000 volts. It's a ceramic disk type about 1/4 inch in diameter. Is this a problem? Radio Shack has some 1/8" diam capacitors about 420 pico Farads. I can't remember the voltage rating but it's much lower. Would that be worth trying? Is there a test I can do with the 680 pF capacitor and a multi-meter?
 
I got it working, but now I'm full of questions again.

1. I had ecb backwards on the PNP. The drawing on the box is from the bottom looking up. I read it as if it were from the top looking down. Ok, I can live this way.

2. It's running now with 2 leds. Delta v for each led is 3.45 volts, delta v for R3 is .27 volts. That's with R2 = 4k ohms, R1 = 10k ohms. So that's 27 mA going through each led.
Q1, pnp, NTE12, measured hFE=362
Q2, npn, NTE11, measured hFE=374

3. Here's the thing. It's hard to start it up. I've got R2 set up as four 1k resistors in series. When I first hook up power, the led(s) glow dimly, no voltage gain. If I jump the R2 series for an instant to short it to 2k, it lights up to full brightness and then runs well on R2 = 4k. It also starts if I pulse the positive battery connection kind of like striking a match. (It starts like an old car I used to have.) How do I fix this? I think all the connections are sound. Is it that big voltage rating on C1?

4. Each plain led draws 40 mA when connected directly to a 3.6 volt Nicad phone battery.

5. I like the circuit. Just gotta figure out why I've got to kick it to get it started.

6. http://www.e-insite.net/ednmag/index.asp?layout=article&articleid=CA90758&industryid= 2281

This is the circuit that uses two NPNs. The article says it also provides voltage regulation.

I think all I have to do is get a few more resistors and then I can build and test this one too. What resistor values would be appropriate for multiple leds? The pattern must be similar.
 
Top