Nice mount for home made lights.

[Humbug]

On 08/05/06 at 22:27:19 HughMann somehow managed to type:

>
> Random Data Wrote:
> >
> > The K2 looks to be an uprated III, or a pair of them in parallel on
> > one
> > dye. I'm giving serious thought to ordering a couple, since a pair
> > of them
> > would kick serious **** and be easily runnable from the 12V source
> > I've
> > got. A reasonably efficient driver would be easy enough to make as
> > well.
> > Dave Hughes )
>
> Yes it looks as if "practical" LED lighting might be taking another
> step forward.
> Noticed on the Lumileds site that there is no holder for the optics
> yet. Its either home brew or wait. Scroll down to the bottom of the
> page for info.
>
> Yes I am sure that between the crew on a.b. there is the potential for
> a great driver unit design. The Sillycon Chip mag effort being
> concluded this month is just over the top. A great exercise in SMT
> soldering and PIC programming etc but for the average person who
> clears the kitchen table to do some soldering its pretty daunting. I
> question the need for some of the "bells and whistles" in their
> design also. Some of the "concepts" may be pinchable thou.

.... I've pinched their PWM and output concepts and triplicated 'em
so I've got one PIC driving three PWM channels.

I've had a look at their code and decided that trying to work out how
someone elses assembler works (even WITH comments) makes my head hurt.
I've just started again, in C this time, using MPLab and CC5X. I'll use
their code as "guidance", etc. but I intend pinching nothing from it
unlike the hardware design from which I've pinch almost everything...

A very first cut of the hardware circuit is at http://tinyurl.com/j9w9l
It's not been prototyped or anything - it's just been drawn ("xv" and
some clip art). I haven't even checked it for glaring errors yet. The
code will be posted in a while where "while" means some unknown time in
the future...


>
> This might get somewhere as long as the nasties over on
> aus.electronics dont invade us again and have a flame war on our bit

I think I'd rather have a h*lm*t war. Hmmmmm better yet, neither of 'em



--

Humbug
Today is Setting Orange, the 57th day of Discord in the YOLD 3172

 

[HughMann]

SNIP
A very first cut of the hardware circuit is at http://tinyurl.com/j9w9l
It's not been prototyped or anything - it's just been drawn ("xv" and
some clip art). I haven't even checked it for glaring errors yet. The
code will be posted in a while where "while" means some unknown time in
the future...

Hey, thats looking good already. (dont know if its wrong thou)

Are you looking at selling the programmed PIC's or going to leave us thrash around by giving us the code and let us stuff it up.

Being far from expert in this stuff I suppose that everything will be done via the Power/Mode switch. Will that include a flashing or steady headlight, one headlight or 2, flashing or steady tail. Dont think that there is a need for dimming do you?

Where are the MOSFETS available from? Farnell should have them but they end up expensive for just a few bits. Might have to wait for next Sillycon Chip mag.

As with most electronics projects these days this will involve at least 10 different shops to get all the stuff together. Was so easy in old days of Radio & Hobies Mag when there were half a dozen shops selling real kits and mountains of good things to put in the shed.

Better go and practice with the etch resist pen. Or are we going to have a PCB design Too

Cheers

Hugh

 

[suzyj]

Humbug wrote:

> .... I've pinched their PWM and output concepts
> and triplicated 'em so I've got one PIC driving three
> PWM channels.

The output rectifier you show is pretty inefficient. You'll lose about 1.4V across it, which is pretty significant when you connect it to a 3.6V LED.

If you bung some small MOSFETs across the diodes, you can do a synchronous rectifier, which is staggeringly efficient. Then you only need itsy bitsy diodes and FETs as well,as they don't dissipate much. Only caveat is that the smoothing caps after them are no longer useful, as synchronous rectifiers conduct both ways when they're on, and simply discharge the caps backto the source. Shouldn't be a problem as long as the frequency is high enough for flicker not to be noticed, and the LEDsdon'tmindbeing switchedon and off.

Have a look at http://www.littlefishbicycles.com/synch.gif for an example...

Cheers,

Suzy

 

[TimC]

On 2006-05-10, suzyj (aka Bruce)
was almost, but not quite, entirely unlike tea:
> If you bung some small MOSFETs across the diodes, you can do a
> synchronous rectifier, which is staggeringly efficient. Then you only
> need itsy bitsy diodes and FETs as well,as they don't dissipate much.
> Only caveat is that the smoothing caps after them are no longer useful,
> as synchronous rectifiers conduct both ways when they're on, and simply
> discharge the caps backto the source. Shouldn't be a problem as long
> as the frequency is high enough for flicker not to be noticed, and the
> LEDsdon'tmindbeing switchedon and off.
>
> Have a look at http://www.littlefishbicycles.com/synch.gif for an
> example...

Also:
http://groups.google.com/group/aus.bicycle/msg/c8734fd7ece1c485?&hl=en

Alas, I googled "***** Hunt" LVR, but came up with nothing at all.

--
TimC
Anyone seeking the "Relativistic Quantum Mechanics" soft option
course, may wish to leave now. -- Intro lecture to RQM

 

[Humbug]

On 10/05/06 at 21:59:03 HughMann somehow managed to type:

>
> Humbug Wrote:
> >
> > SNIP
> > A very first cut of the hardware circuit is at
> > http://tinyurl.com/j9w9l
> > It's not been prototyped or anything - it's just been drawn ("xv"
> > and some clip art). I haven't even checked it for glaring errors
> > yet. The code will be posted in a while where "while" means some
> > unknown time in
> > the future...
> >
>
> Hey, thats looking good already. (dont know if its wrong thou)
>
> Are you looking at selling the programmed PIC's or going to leave us

If/when I get it working as I currently want I'll be able to programme
PIC's for beer or something of similar trading value... Send me a
PIC and I'll programme and test it and send it back I reckon would be
the way to go.

Perhaps when I get the whole shooting match working OK I'll look at
selling made up units. You supply the Luxeons and housings 'n stuff.

> thrash around by giving us the code and let us stuff it up.

The code will be available for all and sundry to fix/improve/etc.

>
> Being far from expert in this stuff I suppose that everything will be

You and me both - I'm just lashing some ideas from SC and manufacturers
data sheets together...

> done via the Power/Mode switch. Will that include a flashing or
> steady headlight, one headlight or 2, flashing or steady tail. Dont
> think that there is a need for dimming do you?

It'll be pretty simple with no dimming.

Connect a power source and the PIC initializes itself and makes sure
everything is OFF except for raising one pin that turns a standard
Vistalight/Cateye[1] blinky on. If the battery voltage is too low it'll
give a couple of short bursts of blinky blinking and then turn
EVERYTHING off.

A long press - longer than say, two seconds, will turn on the 1W red
and the 3W white Luxeons. If there are already lights on then a long
press will turn off everything except the blinky. If the battery
voltage falls below say, 10V then everything is turned off after a
couple of bursts of blinky blinking.

Short presses, while there are lights on, change which headlights are
on and go :-

a - Tail light and 3W head light ON.
b - Tail light and 5W head light ON.
c - Tail light and both headlights ON.
goto a

Short presses while only the blinky is on will do nothing.

Pressing the "mode/reset" switch won't interfere with the PWM duty
cycle of lights aready running except what's to be turned off in which
case the relevant PWM channel is turned off. If an OFF light is to be
turned ON its associated PWM channel is turned on and its PWM duty
cycle is set at some arbitary (low and therefore dim) value and normal
processing can sort it out.

Pressing the "reset" switch will re-init the whole shooting match and
leave you with just the blinky on. Unless battery voltage is too low in
which case nothing gets turned on except for a short burst of blinky to
let you know what the problem is.


>
> Where are the MOSFETS available from? Farnell should have them but
> they end up expensive for just a few bits. Might have to wait for
> next Sillycon Chip mag.

I'm buggered if I know where I'll get the bits from but I'll get 'em
(or equivs) from somewhere...

>
> As with most electronics projects these days this will involve at
> least 10 different shops to get all the stuff together. Was so easy
> in old days of Radio & Hobies Mag when there were half a dozen shops
> selling real kits and mountains of good things to put in the shed.

BRING BACK Waltham Trading.......!!!!!

>
> Better go and practice with the etch resist pen. Or are we going to
> have a PCB design Too

That was the reason for my question for Suzyj. If I can come up with a
workable design I'll make the artwork available - hell I'll make the
whole design available for whoever wants to have a lash at it.

There's a LONG way to go yet. I need to "breadboard" it (
http://en.wikipedia.org/wiki/Image:Breadboard.JPG ) and get a CRO
around some of the pins and fix the code to get what I want. Then I'll
build it "ugly" style (http://www.qrp.pops.net/ugly.htm ) or on a bit
of vero board and give it a good real(ish) world test. When that's all
going nicely I'll start messing with PCB's 'n stuff.


[1] The "blinky" isn't on the circuit yet. I intend to just raise a pin
which'll turn on a 3V supply to a standard, off-the-shelf blinky. ie
Replace the blinky batteries with a 3V connection to the controller
board.


--

Humbug
Today is Setting Orange, the 57th day of Discord in the YOLD 3172

 

[Humbug]

On 10/05/06 at 22:57:37 suzyj somehow managed to type:

>
> Humbug wrote:
>
> > .... I've pinched their PWM and output concepts
> > and triplicated 'em so I've got one PIC driving three
> > PWM channels.
>
> The output rectifier you show is pretty inefficient. You'll lose
> about 1.4V across it, which is pretty significant when you connect it
> to a 3.6V LED.
>
> If you bung some small MOSFETs across the diodes, you can do a
> synchronous rectifier, which is staggeringly efficient.

Thank you - it's as good as done... Just that saving there will
probably be nearly enough to run the blinky which isn't on the piccy
yet.

<snip>

> problem as long as the frequency is high enough for flicker not to be
> noticed, and the LEDsdon'tmindbeing switchedon and off.

I was thinking 4.88kHz or 1.22kHz as the PWM frequency. The SC design
runs at 7.8 or 13.0kHz but I don't _think_ I have those options.

>
> Have a look at http://www.littlefishbicycles.com/synch.gif for an
> example...

Ta for all that - it looks I've got some more reading/learning to
do...

--

Humbug
Today is Setting Orange, the 57th day of Discord in the YOLD 3172

 

[Humbug]

On 10/05/06 at 23:52:25 TimC somehow managed to type:

<snip>

>
> Also:
> http://groups.google.com/group/aus.bicycle/msg/c8734fd7ece1c485?&hl=en
>
> Alas, I googled "***** Hunt" LVR, but came up with nothing at all.

From memory the ***** Hunt LVR isn't any good for LEDs - it's designed
to be used with normal filament bulbs and to wring the last erg out of
a battery. A trip to the BikeCurrent mailing list archives would
probably turn up a heap more in time.


--

Humbug
Today is Setting Orange, the 57th day of Discord in the YOLD 3172

 

[Random Data]

On Thu, 11 May 2006 00:27:01 +1000, Humbug wrote:


> The first trick is finding a MOSFET that turns fully ON when the gate
> is raised to logic levels, most of 'em require more than that. I just
> picked on the ones spec'd in the SC design. The second trick will be
> finding a supplier.

Jaycar have BUZ71s (Cat ZT2225) for $3.75 ea. At around 1A the Drain
Source resistance is well under 0.5 ohm at 3V. That's probably
sufficiently efficient for this application. According to the basic data
I've got here most stuff turns fully on at 5V. That's logic level isn't it?


DSE have BUK456 or equivalent for $4.97 (Z1853). They have an RDS of 0.02
ohm, and will handle 60A claimed. For an extra dollar these will be a
touch more efficient, though at ~1A there's not that much in it (half a
volt or so).

--
Dave Hughes | [email protected]
"Hello. My name is Inigo Montoya. You killed my father. Prepare to die."

 

[slaw]

I recently bought a few bits from http://www.futurlec.com.au/ including some mosfets for 66c each. I was building a soft start circuit for my halogens. It cost me less than $9 including postage, for enough bits for 5 circuits. They were a bit slow with the order claiming they had trouble with my card number and it ended up being shipped from Thailand, but for the price, I didn't mind.


There is a guy in Brisbane who is making up 3x3W lights. He's importing the Luxeons and drivers in bulk lots. Details here: http://www.mtbdirt.com.au/modules.php?name=Forums&file=viewtopic&t=12873. Maybe his light and Humbug's driver could make a could cost-effictive solution.

 

[Humbug]

On 11/05/06 at 01:36:15 Random Data somehow managed to type:

<snip>

> data I've got here most stuff turns fully on at 5V. That's logic
> level isn't it?

It _should_ be but it's likely to be closer to 3V.
>
>
> DSE have BUK456 or equivalent for $4.97 (Z1853). They have an RDS of
> 0.02 ohm, and will handle 60A claimed. For an extra dollar these will
> be a touch more efficient, though at ~1A there's not that much in it
> (half a volt or so).

Hmmm the data sheets for the BUK456 reckon Rds(on) is nearer 5 ohms
unless I've totally read it all wrong. The data sheet can be got at
http://www.datasheetcatalog.com/datasheets_pdf/B/U/K/4/BUK456.shtml .

There are a couple of Fairchild equivs to the SC specified part though
- FDN359AN[1] and FQP50N06L[2] which is listed as a direct equiv[3]. I
reckon I'll go with FQP50N06L's. Not even Mouser or Small Parts carry
the SC specified part so it'll have to be an equiv. No problem -
another member of the radio club I belong to is going to the US for a
week pretty soon so I'll just add what I want to his growing
Mouser/Small Parts order...


The only circuit problems I have left are the transformer design
(HELP!!!) and the rectifiers. I'm not really all that keen on simple
bridge jobbies - getting rid of those losses can only be a good thing.
Suzyj's idea of doing a synchronous a rectifier appeals but in my mind
it raises a question.....

In the output waveform, isn't there going to be bits where the LED is
probably going to be off or close to it ? Seeing as the LED isn't
going to be driven to a reasonable level 100% of the time isn't there
going to be a perceived loss of brightness ? A solution would be
smoothing the output to (almost) pure, constant DC but with smoothing
caps out of the question, how?

On the code front I've just about got the power on, "reset" switch and
"mode/power" switch stuff happening (in an on hand PIC16F84A). The next
bit requires MUCH reading, learning and inwardly digesting all sorts of
stuff about registers and timers and other nasties and things...



[1] http://www.fairchildsemi.com/ds/FD/FDN359AN.pdf
[2] http://www.fairchildsemi.com/ds/FQ/FQP50N06L.pdf
[3] http://www.st.com/stonline/products/promlit/pdf/crmosfet-0202.pdf


--

Humbug

 

[Humbug]

On 11/05/06 at 10:23:46 slaw somehow managed to type:

>
> I recently bought a few bits from http://www.futurlec.com.au/

NBG if you want the stuff some time in the next month or so...:-( If
your order comes from outside Aust. they'll deny all responsibility and
tell you to call whatever place it's being dispatched from. CS is NOT
their strong point - they well and truly alienated me a good while ago.


--

Humbug

 

[suzyj]

Humbug wrote:

> The only circuit problems I have left are the transformer design
> (HELP!!!) and the rectifiers. I'm not really all that keen on simple
> bridge jobbies - getting rid of those losses can only be a good thing.
> Suzyj's idea of doing a synchronous a rectifier appeals but in my mind
> it raises a question.....

If the battery voltage will always be greater than the LED voltage, then you don't need to use a transformer at all. Instead you can design a simple buck regulator, which uses a simple inductor, FET, and (usually) a schottky diode.

Linear Tech have a good intro to switching regulators, called "switching regulators for poets".

http://tinyurl.com/o9kcp

They also have a really cool program for designing all manner of power supplies (and other stuff), called switcher CAD. The particularly neat thing is that it includes a powerful implementation of spice, complete with a GUI interface. I use it for all sorts of stuff, from designing audio power amps to dynamo lights to microwave (gotta do some work occasionally).

http://tinyurl.com/qfa8n

> In the output waveform, isn't there going to be bits where the LED is
> probably going to be off or close to it ? Seeing as the LED isn't
> going to be driven to a reasonable level 100% of the time isn't there
> going to be a perceived loss of brightness ? A solution would be
> smoothing the output to (almost) pure, constant DC but with smoothing
> caps out of the question, how?

Yes, the average brightness will be less than the peak, but the LED is able to be driven to much higher pulsed currents than average currents anyway, so simply arrangethe peak current such that the average is what you want, and it all comes out in the wash. Your eyes will smooth it out as long as it's over 25Hz or so.

Cheers,

Suzy

 

[TimC]

On 2006-05-11, suzyj (aka Bruce)
was almost, but not quite, entirely unlike tea:
> with a GUI interface. I use it for all sorts of stuff, from designing
> audio power amps to dynamo lights to microwave (gotta do some work
> occasionally).

How long does it take to cook 2 minute noodles in front of one of your
transmitters?

> Yes, the average brightness will be less than the peak, but the LED is
> able to be driven to much higher pulsed currents than average currents
> anyway, so simply arrangethe peak current such that the average is what
> you want, and it all comes out in the wash. Your eyes will smooth it
> out as long as it's over 25Hz or so.

And your multimeter should give you the average (RMS) current if it's
close enough to a sine wave. So just measure the current, and that's
that.

--
TimC
Skywalker> You are either with me, or you are my enemy.
Only a Sith deals in absolutes. -- Obi Wan Kenobi on George Bush Jnr

 

[Humbug]

On 12/05/06 at 08:01:24 suzyj somehow managed to type:

<snip>

>
> If the battery voltage will always be greater than the LED voltage,
> then you don't need to use a transformer at all. Instead you can
> design a simple buck regulator, which uses a simple inductor, FET, and
> (usually) a schottky diode.

Hmmm...reaches for ARRL handbook CD....
OK so if I make it look like
http://members.optushome.com.au/forsythm/BikeLights/output.jpg


>
> Linear Tech have a good intro to switching regulators, called
> "switching regulators for poets".
>
> http://tinyurl.com/o9kcp

Ahhh the ARRL handbook CD pointed me at that while I was looking for
stuff on Schottky diodes..

>
> They also have a really cool program for designing all manner of power
> supplies (and other stuff), called switcher CAD. The particularly
> neat thing is that it includes a powerful implementation of spice,
> complete with a GUI interface. I use it for all sorts of stuff, from
> designing audio power amps to dynamo lights to microwave (gotta do
> some work occasionally).

I only do this stuff for a hobby - sub mW low HF amateur radio mainly.
At work I try and teach long term unemployed how to get a job so I NEED
this stuff to calm down...

>
> http://tinyurl.com/qfa8n

Ooooh a shiny new toy - I'll have a play over the weekend between
playing radio.

Ta for all your input, you've been a load of help to this novice...


<snip>

--

Humbug

 

[suzyj]

Humbug wrote:

> OK so if I make it look like
> http://members.optushome.com.au/forsythm/BikeLights/output.jpg

I was thinking more along the linesof using a switching regulator IC to regulate the current to 700mA, then chopping that using the PIC. I just knocked up a Q&D 700mA current regulated buck regulator in LTspice:

http://www.littlefishbicycles.com/luxeon_buck.gif

Basically I took a standard buck regulator, and bunged an opamp in the feedback path so I could regulate output current. It should be relatively compact, and can be done using all SMD bits.

The PWM output from the PIC is inverted - when the PIC output is high, it switches off the converter.

Cheers,

Suzy

 

[Humbug]

Bet you all forgot about this thread, eh ?

Progress has been made in fits and starts (other stuff to do, etc.)

http://members.optushome.com.au/forsythm/BikeLights/

I've only used (so far) an el-cheapo Luxeon imitation to get it working
but it works as intended. I've used junk box parts apart from the white
LED so it's really only a first cut prototype but it does work as
expected. In the next week or so I'll borrow a CRO from a mate and get
the MOSFETs that I want and have at getting it happening properly. I
guess after that I'll have to bite the bullet and spend up big on
actual Luxeons.

The only real stumbling block, and it was a REAL STUMBLING BLOCK, that
I had was the code getting too big for the free C compiler I was using
which meant I had to rewrite the whole shooting match in PIC
assembler.....URK...:-( I s'pose I _could've_ forked out a BIG pile of
dollars for a non-free C compiler but...


--

Humbug
Today is Sweetmorn, the 35th day of Confusion in the YOLD 3172

 

[Kingsley]

On Fri, 30 Jun 2006 23:31:27 +1000, Humbug wrote:

> http://members.optushome.com.au/forsythm/BikeLights/
>
> I've only used (so far) an el-cheapo Luxeon imitation to get it working
> but it works as intended. [...]
>
> The only real stumbling block, and it was a REAL STUMBLING BLOCK, that
> I had was the code getting too big for the free C compiler I was using
> which meant I had to rewrite the whole shooting match in PIC
> assembler.....URK...:-(

What are you using the PIC for?

You've checked out http://www.gnupic.org/ ?
Not that I'd want to dissuade anyone from coding assembler,
it's becoming a lost art.

--
Kingsley Turner,
(mailto: [email protected])
http://MadDogsBreakfast.com/ABFAQ - news:aus.bicycle Frequenly Asked Questions

 

[cfsmtb]

Bet you all forgot about this thread, eh ?

Progress has been made in fits and starts (other stuff to do, etc.)

http://members.optushome.com.au/forsythm/BikeLights/

I'd persuade Bikesoiler to have a lookie at this..

 

[Humbug]

On 01/07/06 at 11:41:26 cfsmtb somehow managed to type:

>
> Humbug Wrote:
> > Bet you all forgot about this thread, eh ?
> >
> > Progress has been made in fits and starts (other stuff to do, etc.)
> >
> > http://members.optushome.com.au/forsythm/BikeLights/
> >
>
> I'd persuade Bikesoiler to have a lookie at this..

I've put the circuit there as well. Well, it's probably not THE circuit
but it's where I'm at ATM. It still needs a tidy up (but not anywhere
near as much as the code does) but I _think_ it's basically it. The
next step is to borrow a CRO and get it all "nice". After that I guess
I'll just have to make something a bit more permanent than a breadboard
and see how it all goes. Then after that I'll just have to build it up
in a box and test it on an actual bike...



--

Humbug
Today is Boomtime, the 36th day of Confusion in the YOLD 3172

 

[Humbug]

On 01/07/06 at 08:47:05 Kingsley somehow managed to type:

> On Fri, 30 Jun 2006 23:31:27 +1000, Humbug wrote:
>
> > http://members.optushome.com.au/forsythm/BikeLights/
> >
> > I've only used (so far) an el-cheapo Luxeon imitation to get it
> > working but it works as intended. [...]
> >
> > The only real stumbling block, and it was a REAL STUMBLING BLOCK,
> > that I had was the code getting too big for the free C compiler I
> > was using which meant I had to rewrite the whole shooting match in
> > PIC assembler.....URK...:-(
>
> What are you using the PIC for?

It's being used to manage a PWM drive for two or three Luxeons. A 3W
and a 5W headlight (either or both running) and a 1W tail light. If I
can lay my hands on a PIC16F777 I'll run 3 Luxeons but if I can only
lay my hands on a PIC16F877A it'll only be the two headlights that are
Luxeons with a normal incandescant, or maybe a slew of super bright
normal red LEDs, tail light. It'll also watch the battery and stop it
from getting below 1V / cell.

>
> You've checked out http://www.gnupic.org/ ?

All of the stuff I found could only _nearly_ do what I wanted which was
a bummer. I ended up going with Microchips MPLab IDE and the CC5X C
compiler from Knudsen Data which works nicely with MPLab.

> Not that I'd want to dissuade anyone from coding assembler,
> it's becoming a lost art.

PIC assembler isn't too bad - there's not that many instructions and
there's LOADS of help and doco on the Microchip web site. It does get
loooooong though...


--

Humbug
Today is Boomtime, the 36th day of Confusion in the YOLD 3172