LDO Driver - IRC 1205 Idea

[HarryN]

Hi

I have an idea which is sort of sacriligious, but I thought I would ask for some input before I try it.

I have seen an IRC 1205 LDO datasheet, which is nominally rated for up to 300ma and available in a variety of output voltages.

So here is the idea
- 2 x 123 cells (nominal 4 - 6.5V during lifetime.
- IRC 1205 LDO chip set for 3.6 V
- No caps on Vin or Vout, just the chip
- Direct feed into a 1 watt Luxeon, nominal 3.1 - 3.4 Vf

It seems like since the Vout from the 1205 will sag from 3.6 to nominally 3.1 - 3.3 volts as it tries to feed the 300 - 400ma, it will reach some kind of natural (but perhaps not entirely predictable) equilibrium with the Vf curve of the LED.

Since there is plenty of current available from 2x123, a cap on the Vin seems extraneous - ideas ?

Since Vout is not that critical, then a cap on Vout seems extraneous - ideas ?

The 1 watt LED will be most likely under driven, but even if it is running at 200 - 500ma, I am ok with that.

If this works, then I could use this to make a single chip (no extras) essentially "semi smart" resistor and skip a PCB altogether.

I am expecting efficiency would be 50 %, maybe 60 if I am lucky.

Now we are down to 50 cents for a simple, small regulator, which would seem to be at least as efficient, and more stable than a simple resistor.

Comments ?

 

[MrAl]

Hello there,

Where is the data sheet, ill take a look too.

Take care,
Al

 

[pbarrette]

Hi MrAl,

I think it's International Rectifier.

pb

 

[MrAl]

Hi again,

Thanks for looking that up pb.

This idea not only looks like it might work
(with adjustable version),
we might be able to turn it into a near zero
volts dropout voltage regulator.
It's getting late here ... I'll have more
details tomorrow.

Take care,
Al

 

[MrAl]

Hello again,

Ok, it's tomorrow :-)

Here's the circuit:

http://hometown.aol.com/xaxo/page9.html

This has been simulated but not actually breadboarded yet.
The dropout appears to be almost zero!
Can't substitute the zetex transistor unless it's another
very low sat high gain type.
I'll start another topic so you can still talk about
the original idea here.

Take care,
Al

 

[HarryN]

Hi, I think I did not make my point very well.

The idea is that we mostly have the more complex circuits which are very efficient, and at the low end, a simple resistor.

I am looking for a single component to replace the resistor but which is a little more efficient, does the regulation a little better, in a specific 2 x 123 configuration.

If I properly understand the resistor situation, power consumed in a 1 watt application is

.385A x .385A x 10 ohm = 1.5 watt loss in resistor, and 1 watt used in the LED. (for a nominal 6 V in)

Also, the light output drops off with battery voltage.

My question really is, can a simple device like an Int. Rect 1205 LDO get this efficiency up to 50 or 60 % (with no external parts at all ) and maintain a flatter light output than a simple resistor.

If so, then this would be a reasonable alternative to the people that run DD or resistor based flashlights due to either cost or just not able to build circuits into their lights.

 

[MrAl]

Hi again Harry,

Sorry about that! In my over-zealousness to spin off a new
design i never got back to the question at hand :-)

It looks like the max input voltage for a 3.3v LED at
350ma will be about 4.8 volts (approximate) in order
to keep the chip out of thermal shutdown.
This might be pretty good though for an Li-ion rechargeable
or three batteries in series!
Current limit might go a little high, but might still
work in the long run.
A bench test would be in order...any takers?

Take care,
Al

 

[Doug Owen]

I'm cool to try to build this (and especially Al's 717 circuit).

Anybody know how to get the parts? IMO we're looking for both the fixed voltage and the adjustable versions?

Doug Owen

 

[MrAl]

Hi again Doug,

I havent checked DigiKey yet for the parts, i just assumed
someone out there had them :-) I've gotten samples from
IRF before so they must have a way.

Well, apparently there are two ideas that stem from
Harry's one idea. Listing all three:

1. A fixed voltage regulator with the hopes that
the current limit will be just right for a lux 1 watter.
2. An adjustable version that can be adjusted to fit
the approximate voltage of the lux 1 watter with two
external resistors, or with only 1 potentiometer.
3. An adjustable version with the external FMMT717 Zetex
transistor to reduce dropout voltage.

Perhaps for testing an adjustable version can be aquired
and used to test all three (or more pending) ideas.
It can be set for 3.6v to simulate the fixed 3.6v version
(just an idea; if it works non-adj version would have to
be aquired anyway).

#1 above was Harry's original interesting idea.
#2 is just in case #1 doesnt work exactly as is, maybe
adjusting the voltage alone will make it work good
enough.
#3 is my idea, mainly to get the dropout voltage down
to near zero (approx 50mv). This of course isnt
as simple as #1 and #2 though, and brings in more parts.

I'll take a look on the web for the IRF parts.

Take care for now,
Al

 

[MrAl]

In addition to the above post:

Hello again,

Yeah, Digikey has them.

which brings me to idea #4 :-)

#4:
Purchase the 3.3v version and
put a resistor in series with the ground terminal
of the IRU1205 to drop a little (tiny) voltage
when the chip is regulating. This will increase
the output voltage by a tiny amount (if needed)
to bring the full load voltage up to 3.4v or 3.5v
or whatever is needed by the particular Luxeon.
Very roughly, if the pnp pass transistor requires
1/100 base drive while the output is 350ma that would
mean about 3.5ma would flow through the ground terminal.
This would mean the resistor value would be about
285 ohms per volt. Since we might need about
0.1 to 0.2 volt increase over the 3.3v fixed point
this would mean a resistor value of between about 28
and 56 ohms. The power dis would be nil, so it
shouldnt reduce efficiency to do this. Chip resistor ok
too. A single 100 ohm pot may add brightness adjustment.

Drawbacks:
There will be some variation in brightness with
temperature variation, but possibly not that much
for a hand held flashlight.
Would have to be tested in an oven.


Take care,
Al

 

[Doug Owen]

OK, so I ordered ten. Now to cut out tea long enough to solder 'em up.

One thing sure, to get popular as an alternative to DD someone's gonna have to do something about the package....

OTOH, it might be neat to glue the little rascal down to the heatsink on the LED with thermal epoxy and protect the LED at the same time.....

Anyway, it's in Digikey's experienced hands, soon to be in my eager ones.

Doug Owen

 

[HarryN]

Hi Doug, Thank you very much for stepping into this. I was about to order and assemble some, but I cannot really test, so the next stop would have been your door step with the parts.

My very crude implementation idea for the unskilled hobbiest (me):
- Start with a piece of Al sheet as a PCB / heat sink
- Sand, clean, oxidize / seal in hot boiling water and +12V for 15 - 20 minutes (sort of like the last step in anodizing)
- Use the Al sheet mounting the LED and 1205 (using the oxide to isolate the LED heat slug)
- Solder paste all connections
- Thermally bond the 1205 to the heat sink Al sheet

Vout - direct connect to the LED
Vin - direct connect to the 2x123 out
Enable - Vin
Grd
Cbyp - either let float, or solder paste / glue directly to the Al. The oxide will form a small, un predictable, but leaky cap - maybe enough

I was going to use a 1205 - 36CL (3.6 V version). That way, since it sags about 300 - 400 mv at 300ma, it would cover a pretty wide range of Vf for 1 watt Luxeons.

Thanks for doing this. I was actually just expecting that someone had done this before in another thread.

 

[MrAl]

Hello again,

Wow, Doug, that's great! What voltage types
did you order?
Cant wait to see some results, especially Harry's
original idea!

Harry:
Great idea! Cant wait to see some results :-)


Take care,
Al

 

[Doug Owen]

I ordered the adjustable one since it can be set up as any voltage needed. This will allow us to try all the options, I think.

If not, Digikey will may end up with (yet) more of my money a bit sooner than otherwise.

Doug Owen

 

[MrAl]

Hello again Doug,

Wow that's great. Cant wait to see some results with
the output set to maybe 3.6v with two resistors and
no other parts. If the current limit holds maybe we
can short the pin to ground (as in your other post which
im getting to next).
I've read something like 300ma to 420ma current limit
so lets keep our fingers crossed :-)
If it turns out low, we can always increase it.
Most people wouldnt mind 420ma either i think.

Take care and THANKS MUCH for trying this,
Al
ps I almost ordered a couple myself :-)