A new light

"Jon Senior" <jon_AT_restlesslemon_DOTco_DOT_uk> wrote in message
news:[email hidden]...

Quoted message said:

Finally built it (Sort of). Uncased so not a long-term thing but it
works. Details (As always) can be found at
http://www.hoovesofdestiny.co.uk/led_light.html

Ooh, excitement. Any more details on your switch-mode chip (couldn't find
much on LM574 on the web or maplins - is this a typo?)

cheers,
clive

Clive George [email hidden] opined the following...

Quoted message said:

"Jon Senior" <jon_AT_restlesslemon_DOTco_DOT_uk> wrote in message
news:[email hidden]...

Quoted message said:

Finally built it (Sort of). Uncased so not a long-term thing but it
works. Details (As always) can be found at
http://www.hoovesofdestiny.co.uk/led_light.html

Ooh, excitement. Any more details on your switch-mode chip (couldn't find
much on LM574 on the web or maplins - is this a typo?)

It is indeed a typo. That should read LM2574 -ADJ. I'll update the site.
The chip requires 2 capacitors, a schottkey diode, an inductor and a
pair of resistors (To set the voltage).

I'll break out the multimeter at lunchtime and work out how efficient it
is.

Jon

In article <[email hidden]>, [email hidden]
says...

Quoted message said:

And whether would be worth spending the extra couple of quid on the LM2674
(the high efficiency one) :-)

Yes. Didn't notice that one!

My circuit pulls about 400mA @ 7.7v (Battery pack) and makes a high
pitched buzz (Very quiet, but definately there). The LED is current /
voltage limited to pull 250mA @ 3.44v so my basic understanding of
electronics suggests that the figure at the higher voltage should be
lower. Either the LED is somehow pulling 500mA at the reduced voltage
(Unlikely, I did some playing with a bench PSU before I built this) or
this is not very efficient.

Looks like tonight I'll be getting my head around the LM2674.

Jon

"Jon Senior" <jon_AT_restlesslemon_DOTco_DOT_uk> wrote in message
news:[email hidden]...

Quoted message said:

Clive George [email hidden] opined the following...

Quoted message said:

"Jon Senior" <jon_AT_restlesslemon_DOTco_DOT_uk> wrote in message
news:[email hidden]...

Quoted message said:

Finally built it (Sort of). Uncased so not a long-term thing but it
works. Details (As always) can be found at
http://www.hoovesofdestiny.co.uk/led_light.html

Ooh, excitement. Any more details on your switch-mode chip (couldn't

find

Quoted message said:

Quoted message said:

much on LM574 on the web or maplins - is this a typo?)

It is indeed a typo. That should read LM2574 -ADJ. I'll update the site.
The chip requires 2 capacitors, a schottkey diode, an inductor and a
pair of resistors (To set the voltage).

I'll break out the multimeter at lunchtime and work out how efficient it
is.

And whether would be worth spending the extra couple of quid on the LM2674
(the high efficiency one) :-)

cheers,
clive

"Jon Senior" <jon@restlesslemon_DOT_co_DOT_uk.remove> wrote in message
news:[email hidden]...

Quoted message said:

In article <[email hidden]>, [email hidden]
says...

Quoted message said:

And whether would be worth spending the extra couple of quid on the

LM2674

Quoted message said:

Quoted message said:

(the high efficiency one) :-)

Yes. Didn't notice that one!

My circuit pulls about 400mA @ 7.7v (Battery pack) and makes a high
pitched buzz (Very quiet, but definately there). The LED is current /
voltage limited to pull 250mA @ 3.44v so my basic understanding of
electronics suggests that the figure at the higher voltage should be
lower. Either the LED is somehow pulling 500mA at the reduced voltage
(Unlikely, I did some playing with a bench PSU before I built this) or
this is not very efficient.

Hmm - suggests that's about 30% efficient, which is an awful lot below what
it should be (NS say 70-80% or so - they specifically say it doesn't need a
heat sink, which if yours was losing what you say, it would.)

Is your input current accurate? How lumpy is it? ('scope).

If it is that inefficient, the chip should be getting _very_ hot - that
comes out as just over 2W heat loss (think of how hot a 2.4W halogen bulb
gets). Which suggests it's something else.

Quoted message said:

Looks like tonight I'll be getting my head around the LM2674.

Now it's built, I'd leave it - you're probably only losing 10-20% runtime
with the 2574.

cheers,
clive

disclaimer : I really have absolutely no idea what I'm talking about here
:-)

Jon Senior said:

My circuit pulls about 400mA @ 7.7v (Battery pack) and makes a high
pitched buzz (Very quiet, but definately there). The LED is current /
voltage limited to pull 250mA @ 3.44v so my basic understanding of
electronics suggests that the figure at the higher voltage should be
lower. Either the LED is somehow pulling 500mA at the reduced voltage
(Unlikely, I did some playing with a bench PSU before I built this) or
this is not very efficient.

There might be something simpler going on: LEDs are designed to be
driven with a regulated *current*, not a regulated *voltage*. The
relationship between voltage and current for LEDs is not guaranteed.
What this means is that even if your regulator puts out a constant
3.44V, the current the LED draws from that may vary from a few mA to
several amps depending on its temperature and which batch it was
manufactured in, amongst other things.

I'm fairly certain, although I'd have to check, that the forward voltage
of LEDs has a negative temperature coefficient, which is Very Bad in
this situation. As the LED heats up, it'll draw more current from your
voltage-regulated supply, and thus get hotter, so it'll draw more
current, and so on until something hits its limit, melts or catches fire.

It may therefore be that your LED is actually drawing somewhere around
an amp, which would explain the high draw from the battery pack.

HTH
Chris

Clive George [email hidden] opined the following...

Quoted message said:

Hmm - suggests that's about 30% efficient, which is an awful lot below what
it should be (NS say 70-80% or so - they specifically say it doesn't need a
heat sink, which if yours was losing what you say, it would.)

Doesn't seem right does it?

Quoted message said:

Is your input current accurate? How lumpy is it? ('scope).

[censored]. I'm not sure I can be bothered to learn how to use our
oscilloscope!

Quoted message said:

If it is that inefficient, the chip should be getting _very_ hot - that
comes out as just over 2W heat loss (think of how hot a 2.4W halogen bulb
gets). Which suggests it's something else.

Nah. Though it does make a noise which would be another way of disposing
of energy. Not sure where the noise is coming from, but my money is on
the inductor.

Quoted message said:

Now it's built, I'd leave it - you're probably only losing 10-20% runtime
with the 2574.

Ran it up tonight somewhere properly dark. Not quite as punchy as my old
10W halogen, but bloody close. Taking into account an estimated burn
time of around 10 hours I'm perfectly happy with it. Especially as I may
be able to power it from a S.O.N. dynohub which is a potential upgrade
for the bent.

Jon

Chris Jones [email hidden] opined the following...

Quoted message said:

There might be something simpler going on: LEDs are designed to be
driven with a regulated *current*, not a regulated *voltage*. The
relationship between voltage and current for LEDs is not guaranteed.
What this means is that even if your regulator puts out a constant
3.44V, the current the LED draws from that may vary from a few mA to
several amps depending on its temperature and which batch it was
manufactured in, amongst other things.

I'm fairly certain, although I'd have to check, that the forward voltage
of LEDs has a negative temperature coefficient, which is Very Bad in
this situation. As the LED heats up, it'll draw more current from your
voltage-regulated supply, and thus get hotter, so it'll draw more
current, and so on until something hits its limit, melts or catches fire.

It may therefore be that your LED is actually drawing somewhere around
an amp, which would explain the high draw from the battery pack.

An interesting idea. When I first played around with it, I used a bench
supply which incorporated a current regulator. I never ran it up for
very long so might have missed the phenomenom that you describe. That
said, the heatsink on this is quite something.

Regardless of this, I'll try metering the current draw at the LED
tomorrow to see how it compares. It's unlikely to be pulling more than
0.5A as at that point the chip would definately start to get warm.

More experiments / results to follow.

Jon

Jon Senior said:

Quoted message said:

If it is that inefficient, the chip should be getting _very_ hot - that
comes out as just over 2W heat loss (think of how hot a 2.4W halogen bulb
gets). Which suggests it's something else.

Nah. Though it does make a noise which would be another way of disposing
of energy. Not sure where the noise is coming from, but my money is on
the inductor.

My bet remains with the LED - I bet the poor beast is dissipating most
of the extra power, due to the current/voltage thing I mentioned
earlier. Email me for my top tips on converting a voltage regulator into
a current one, unless I've got the wrong end of the stick!

Other top suspects for inefficiency: the freewheel diode, if it's not a
high-speed Schottky one, will waste a lot. The electrolytic capacitors
in the circuit will be really struggling if they're not low-resistance
types designed for this sort of application, and they'll get hot and not
last very long. And finally the inductor - if it's too small, it'll be
saturating and/or the resistance of its winding will throw a lot of
power away.

The quick way to find out where the power is going is with a cautious
fingertip :-)

Chris

Chris Jones [email hidden] opined the following...

Quoted message said:

The quick way to find out where the power is going is with a cautious
fingertip :-)

Cautious experimentation earlier failed to find any hotspots. It's be
possible for the heatsink to dissapate a great deal more energy to no
noticeable effect as it's attached to a reasonably sized lump of
aluminium.

Jon

"Jon Senior" <jon_AT_restlesslemon_DOTco_DOT_uk> wrote in message
news:[email hidden]...

Quoted message said:

Nah. Though it does make a noise which would be another way of disposing
of energy. Not sure where the noise is coming from, but my money is on
the inductor.

Noise takes hardly any energy. Think how loud a little tranny radio can be -
and that's a watt or so.

cheers,
clive

Chris Jones said:

Other top suspects for inefficiency: the freewheel diode, if it's not a
high-speed Schottky one, will waste a lot.

That's usually the efficiency limiter with low voltage outputs - the
efficiency is limited by the rectifying diode voltage drop vs the LED
voltage drop. The only way round it is to go to synchronously switched
FET rectifiers but that's a lot more complex

Tony

On 12/10/04 11:54 pm, in article [email hidden],

Jon Senior' jon_AT_restlesslemon_DOTco_DOT_uk said:

Ran it up tonight somewhere properly dark. Not quite as punchy as my old
10W halogen, but bloody close. Taking into account an estimated burn
time of around 10 hours I'm perfectly happy with it. Especially as I may
be able to power it from a S.O.N. dynohub which is a potential upgrade
for the bent.

There seem to be a lot of dynohubs for sale now (two from Shimano). As these
can produce 3W, would it be possible to run a 3W Lumeon (and keep your
handlebars warm at the same time ;-) from that?

And any chance of a circuit diagram/parts list so that mere mortals who can
wield a soldering iron can have a go as well?

...d

David Martin [email hidden] opined the following...

Quoted message said:

There seem to be a lot of dynohubs for sale now (two from Shimano). As these
can produce 3W, would it be possible to run a 3W Lumeon (and keep your
handlebars warm at the same time ;-) from that?

Given the apparent efficiency of mine I doubt you'd get full whack out
of a 3W Luxeon but the 1W seems bright enough to full-darkness riding.

Quoted message said:

And any chance of a circuit diagram/parts list so that mere mortals who can
wield a soldering iron can have a go as well?

I'll knock one together later today.

Jon

In article <[email hidden]>, [email hidden]
says...

Quoted message said:

There might be something simpler going on: LEDs are designed to be
driven with a regulated *current*, not a regulated *voltage*. The
relationship between voltage and current for LEDs is not guaranteed.
What this means is that even if your regulator puts out a constant
3.44V, the current the LED draws from that may vary from a few mA to
several amps depending on its temperature and which batch it was
manufactured in, amongst other things.

I'm fairly certain, although I'd have to check, that the forward voltage
of LEDs has a negative temperature coefficient, which is Very Bad in
this situation. As the LED heats up, it'll draw more current from your
voltage-regulated supply, and thus get hotter, so it'll draw more
current, and so on until something hits its limit, melts or catches fire.

It may therefore be that your LED is actually drawing somewhere around
an amp, which would explain the high draw from the battery pack.

And the award for Bright Spark goes to Chris for correct guess. The LED
is pulling a quite healthy 630mA @ 3.44V which is around 300mA over the
recommende current and a good 130mA over the maximum pulsed current.
This might not be a good thing although it does explain why the light
was so bright!

Of course... noticing your email address should have been a dead
giveaway. Presumably you've been through all this before!

Jon

In article said:

On 12/10/04 11:54 pm, in article [email hidden],

Jon Senior' jon_AT_restlesslemon_DOTco_DOT_uk said:

Ran it up tonight somewhere properly dark. Not quite as punchy as my old
10W halogen, but bloody close. Taking into account an estimated burn
time of around 10 hours I'm perfectly happy with it. Especially as I may
be able to power it from a S.O.N. dynohub which is a potential upgrade
for the bent.

There seem to be a lot of dynohubs for sale now (two from Shimano). As these
can produce 3W, would it be possible to run a 3W Lumeon (and keep your
handlebars warm at the same time ;-) from that?

From comments on the bikecurrent mailing list, yes (with some rectification
between the alternator and LED), and the hub acts as close enough to a constant
current source to be safe with a 3W Luxeon. For more money, you can do it
rather more efficiently.

Alan Braggins said:

In article said:

There seem to be a lot of dynohubs for sale now (two from Shimano). As these
can produce 3W, would it be possible to run a 3W Lumeon (and keep your
handlebars warm at the same time ;-) from that?

From comments on the bikecurrent mailing list, yes (with some rectification
between the alternator and LED), and the hub acts as close enough to a constant
current source to be safe with a 3W Luxeon. For more money, you can do it
rather more efficiently.

Why rectify? That's going to drop some voltage and the LED will
rectify for itself, though I note that Luxeon's data sheet doesn't
list a permissible reverse voltage.

With a 'normal' dynamo, generating nominaly 6V at 0.5A, could you not
put two 1W luxeon in series across the dynamo and let everything sort
itself out - the LED can tolerate peak currents of 0.5A, and has a
nominal forward voltage of 3.45V.

If you wanted maximum light you could have four - two pairs each
lighting on the relevant half of the dynamo cycle. You could have one
pair in focussed white at teh front, and one pair in lambertian red at
the back.

(Incidentally, I note that farnell has wider stock than Maplin, and
also sells the collimating optic so you can put a 10 degree beam on
other emitters than the one in teh star/O, though the mounting would
be more tricky).

regards, Ian SMith
--
|\ /| no .sig
|o o|
|/ \|

"Ian Smith" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Alan Braggins said:

In article <[email hidden]>, David Martin

Quoted message said:

Quoted message said:

Quoted message said:

There seem to be a lot of dynohubs for sale now (two from Shimano). As

these

Quoted message said:

Quoted message said:

Quoted message said:

can produce 3W, would it be possible to run a 3W Lumeon (and keep your
handlebars warm at the same time ;-) from that?

From comments on the bikecurrent mailing list, yes (with some

rectification

Quoted message said:

Quoted message said:

between the alternator and LED), and the hub acts as close enough to a

constant

Quoted message said:

Quoted message said:

current source to be safe with a 3W Luxeon. For more money, you can do

it

Quoted message said:

Quoted message said:

rather more efficiently.

Why rectify? That's going to drop some voltage and the LED will
rectify for itself, though I note that Luxeon's data sheet doesn't
list a permissible reverse voltage.

For the 5W, it says > 5volts. It also says 'LEDs are not to be driven in
reverse bias', but I'm not entirely sure if that means you'll knacker them
if you feed them AC.

Quoted message said:

With a 'normal' dynamo, generating nominaly 6V at 0.5A, could you not
put two 1W luxeon in series across the dynamo and let everything sort
itself out - the LED can tolerate peak currents of 0.5A, and has a
nominal forward voltage of 3.45V.

If you wanted maximum light you could have four - two pairs each
lighting on the relevant half of the dynamo cycle. You could have one
pair in focussed white at teh front, and one pair in lambertian red at
the back.

Oy, stop nicking my ideas!

Actually I was still thinking of conventional red LEDs for the back, like I
have currently, and have all the white ones required at the front.

Question is, what will the dynamo (eg SON) produce as a peak current?
Somewhat more than 0.5A I'd guess, to make up for the troughs. Then, will
the luxeon handle this? (no worries about the rear leds, if you've got
sufficient). Would 3W ones be a better choice?

cheers,
clive

In article said:

"Ian Smith" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Alan Braggins said:

In article <[email hidden]>, David Martin

Quoted message said:

Quoted message said:

>
>There seem to be a lot of dynohubs for sale now (two from Shimano). As

these

Quoted message said:

Quoted message said:

>can produce 3W, would it be possible to run a 3W Lumeon (and keep your
>handlebars warm at the same time ;-) from that?

From comments on the bikecurrent mailing list, yes (with some

rectification

Quoted message said:

Quoted message said:

between the alternator and LED), and the hub acts as close enough to a

constant

Quoted message said:

Quoted message said:

current source to be safe with a 3W Luxeon. For more money, you can do

it

Quoted message said:

Quoted message said:

rather more efficiently.

Why rectify? That's going to drop some voltage and the LED will
rectify for itself, though I note that Luxeon's data sheet doesn't
list a permissible reverse voltage.

For the 5W, it says > 5volts. It also says 'LEDs are not to be driven in
reverse bias', but I'm not entirely sure if that means you'll knacker them
if you feed them AC.

Even if it doesn't knacker them, you won't get useful light out of them in
both directions, and "6V" dynamos can certainly produce more than 5V across
something not letting much current through. Using a second Luxeon for the
other direction might be efficient, but you will probably need a second mount
and reflector or lens.

(There was some talk of using FETs for active rectification to reduce the
voltage drop.)