High current switching regulator

fenix-store
Wilsonite

Wilsonite

Newly Enlightened
Joined
Feb 21, 2005
Messages
75
Location
Des Moines, Iowa
Hi Guys & Gals, Love ya love your show!

I am wondering if anyone can tell me if the switching regulators (more specifically step-down) can be modified for high current usage. (15-20W @3-4v) I didn't know you had to have an inductor on those little switchers, and I don't know enough about electronics to determine whether I can just slap the output into a suitably sized FET to scale the output? I assume not due to the inductor. Correct me if I am wrong, but it would be driving the FET linear right?
Actually, I think I just answered my own question there. /ubbthreads/images/graemlins/Christo_pull_hair.gif

/ubbthreads/images/graemlins/crazy.gif umm, yeah sure, that will be regulating the FET into full conduction... /ubbthreads/images/graemlins/ooo.gif grr...

Any thoughts/tips on building a high current model? Any sources to purchase them already brewed?

Thanks!

Wilsonite
 
Hi there,

I would think that if you provided your own output transistor
you must be able to switch more current. You'll have to
set up the drive such that it turns on and off properly.
I was thinking of trying this someday almost like that,
but for higher output VOLTAGE, using a higher voltage
rated transistor.
Of course the inductor would have to handle the much higher
current too, so you'd have to find an inductor that can
handle some 10 or 20 amps (or whatever your peak I will
be).
Also, your caps will have to be able to handle the high
peak current too, so you'll have to use a few in parallel
i bet.

It would not be driving the FET into linear mode, because the
output of switchers are switching to begin with.
You cant use the output of a linear regulator, but you can
use the output of a lower current switching regulator
like the National Semi. line of switching regulators.

What is the exact output you need, because computer supplies
are available at 3.3 volts (non-adjustable i guess) ?


Take care,
Al
 
Ok, lets see if my old BBS ascii skillz are still with me...
I was simply referring to hooking the gate of the FET to the output of the regulator, then realized that the regulator would be putting out ~3.3v to the gate, then depending on the FET driving it to saturation.

  +4.5-6v
   |              lux
   |             array
   +--------->|--+
   |                   |
+-+---+             |
|        | out+     /
| reg   +-------||S
|        | out-   |     N-chan FET
|        +--+     ||D
|        |    |       \
+-+---+    |        |
  |            +-----+    
  |            |
 gnd           gnd

Clear as mud?
(WHAT A PITA!!! this board doesn't like art /ubbthreads/images/graemlins/mad.gif removed all my spaces, had to use the good old alt255 with creative spacing, and it still doesn't come out right) /ubbthreads/images/graemlins/banghead.gif

I was looking for quick and easy, but I guess that won't work here. Anyone know if the SMPS controller will sink/source enough to drive a mosfet large enough to switch 6-7A @ ~3.0v? (to leave me some headroom)

My end result is to drive 4 3w stars (maybe overdrive) from a 4.5-6v source.

Thanks!
Wilsonite
 
Wilsonite,

Put your ascii-art inside <font class="small">Code:</font><hr /><pre> </pre><hr /> tags
 
Awesome! Thanks evan9162! /ubbthreads/images/graemlins/grin.gif

Forgot to mention above, as far as the pc supply goes, this is for a 4D Mag... Planning on replacing one of the cells with this regulator for 4.5v. Or if small enough, mount in tailcap electrically between the battery and the body for 6v.

Wilsonite
 
Hi - just some observations.
- 4 x 3W stars - Light overdrive approx 5 amps, hard overdrive maybe 8 amps (and likely short life, but real bright)
- Vf of the star will shift up toward 4 + V under overdrive conditions regardless of original Vf
- V bat will drop under high draw conditions.
- 3 x Alkaline D - certainly will drop to 4 V fast, and probably will underdrive your setup
- 4 x NiMH or NiCd will be approx 4.4 V under load, but could supply that kind of current flow even in AA size.
- 2 x Saft Li D cells would be around 7 V and impressive. Getting a little pricey at $ 15 / each, but they have the power.
- Most active buck setups need approx 1 V of headroom (V bat - Vf) to work well.

In Alkaline D at least, your Vbat will be so close to Vf that even a very small resistance (such as a stainless spring) will provide the necessary 1/2 ohm without electronics,

There are some current regulators prebuilt that can be purchased from cpf georges80, leddynamics, and wayne at the cpf section marked "Sandwich Shoppe", just in case you want to just build the light instead of the circuit.
 
Hi there,

Yeah that ascii art is a pain /ubbthreads/images/graemlins/smile.gif
Hope your not trying to use a linear regulator, unless
you dont mind having the MOSFET in the linear region
(which is entirely possible).

Take care,
Al
 
Thanks HarryN,
I have yet to own/test my first lux. they should be here in a few days :cool:
I was under the assumption that the vf was the "firing" voltage for the lux, and the max voltage would be within .3-.4v of that point.

Sorry guys, my experience with driving lamps at this time is exclusively in step-ups with mercury and HPS bulbs from 12v. and it's pretty limited at that... /ubbthreads/images/graemlins/crazy.gif What I know about electronics, I taught my self, and my knowledge is quite limited. /ubbthreads/images/graemlins/frown.gif I really don't understand the magnetics (inductors more specifically, transformers are ok) I get the basic concepts, but as far as "rolling my own" it's pretty much been the guess and check method. (thank god for my Tektronix 453!) [ancient o'scope]

I'm just paranoid...

as far as prebuilt models go, the ones that I am seeing (I've probably stared at them and missed em) are rated at 700ma or 1a, but not any higher. I don't have the room to put 4 regs in the head, nor the ca$h for them. Once again I am paranoid of smoking 4 stars in one brilliant, but short flash! /ubbthreads/images/graemlins/icon15.gif /ubbthreads/images/graemlins/grin.gif /ubbthreads/images/graemlins/icon15.gif

and unfortunately, I do not have a lab PS, I suppose I could put one together. The PS I do have is a 12-15V @ 50A...
Maybe I should get out of LED's and go back to hotwires. /ubbthreads/images/graemlins/blush.gif


Wilsonite
_____________________________________________
I don't know everything, just a little bit about it.
 
Mr. Al, I was trying to avoid running in linear mode, but after reading Jarhead's dissertation on PWM vs CC dimming, CC is looking more attractive. I just didn't want all that wasted heat. I suppose I could throw together something for linear operation, and scale it to my requirements, I may end up doing this anyway. Maybe it's just the coolness factor of PWM... /ubbthreads/images/graemlins/buttrock.gif

Wilsonite
 
You could build 4 high current, very low dropout linear regulators for less than $20 total. And with careful design, all 4 could fit together in a D-cell mag easily.
 
[ QUOTE ]
Wilsonite said:

Any thoughts/tips on building a high current model? Any sources to purchase them already brewed?


[/ QUOTE ]

Astec APC08F03 is along the lines of what you want. 3.3v out, 8 amp max, from a 4-6v input.

You should be able to get them from Digikey.
 
What type of batteries do you plan on using? Hauling 4+ amps out of two CR123s (or four NiMh AAs) will result in less than thirty minutes runtime. Depending on your design goals, increasing the input voltage or using 12Ah NiMh D cells might be worth considering. Four 12Ah D cells would be good for 2-3 hours, for instance.
 
enLIGHTenment, you hit the nail on the head! I am modding a 4D. I will be on alkies for a bit but plan on getting the 11-12Ah cells. I may build the reg into a D size form factor and run 3 "live" cells.

I still need to assemble the light, don't have the stars yet. I thought I was going to build one driver for each light, but don't think I have room for that. I have been thinking of doing a serial/parallel setup (kinda like the 5w dies) ... I dunno...

I am chomping at the bit to get my stars... this will be my first lux period. I was awestruck with them when HardOCP (pc hardware site) reviewed them back when they had the separate red die in reverse. I kinda lost interest for a very long time. A guy I used to work with had a Lion Heart, and a U2, so of course we started talking and he pointed me here. Finally, a place where I can hang out with fellow lightloons! -

Anyway, back on topic, I am trying to get ideas for how I want to set this up. As above, I thought I was going to run a switcher, but after reading Jarhead's post on CC vs PWM, I am thinking CC is the way to go. I was concerned by wasted heat/power in linear mode, but if I ser/par them, I should lower the wasted heat, but then will have to go to a higher volt (read exotic) power setup. I am trying to keep the power requirements as generic as possible, and, well, I want those 12Ah'ers /ubbthreads/images/graemlins/grin.gif

Wilsonite
 
[ QUOTE ]
Wilsonite said:
Anyway, back on topic, I am trying to get ideas for how I want to set this up. As above, I thought I was going to run a switcher, but after reading Jarhead's post on CC vs PWM, I am thinking CC is the way to go.


[/ QUOTE ]

Jarhead's data are only applicable to dimming, not to switching power supplies in general. For your purposes, a switching power supply is, by far, the best route.

A direct-drive series arrangement will severely underdrive the LEDs when using NiMh batteries. 2.4v Vf is well under the 3.03-3.90v that Luxeons like.

Linear regulation from 4.8v to 3.90v is, at theoretical maximum, 3.9/4.8=81% efficient. Real-world figures will be worse. In contrast, the Astec part I suggested can knock 5v down to 3.3v with 93% efficency. Of course, it will also maintain constant brightness until the input voltage drops below 4v.

Going to three NiMh cells would pretty much require parallel direct drive as there wouldn't be enough headroom to operate a regulator.
 
I would do it the other way - connect the four stars in series and run a step-up from 4.5-6V input. This will give you lower current but higher voltage on the output. Because the output voltage drop will be much greater than the schottky forward voltage the rectification losses will be relatively low. As I understand, you want to drive 3.3-3.6V luxeons, and that means the 1W ones? The inductor requirements should not be that awful... especially if you could do without regulation and let the output current drop as the cells are drained - should not be a big issue if you are using D-size cells.

Connecting LEDs in parallel is usually a bad idea unless they have been carefully hand-picked for matching forward voltage. If not, you will risk different current levels through the individual LEDs, some of them could be overdrived while others are underdrived. Connecting them in series on the other hand ensures that the same current flows through all LEDs.

BTW: I think that you could use the Zetex driver for this - the two-cell driver circuit described in their lighting handbook could be scaled to drive four LEDs from four cells. I would use a greater inductor value though (to keep operating frequency low).
 
[ QUOTE ]
operator_smooth said:
I would do it the other way - connect the four stars in series and run a step-up from 4.5-6V input. This will give you lower current but higher voltage on the output. Because the output voltage drop will be much greater than the schottky forward voltage the rectification losses will be relatively low. As I understand, you want to drive 3.3-3.6V luxeons, and that means the 1W ones? The inductor requirements should not be that awful... especially if you could do without regulation and let the output current drop as the cells are drained - should not be a big issue if you are using D-size cells.

Connecting LEDs in parallel is usually a bad idea unless they have been carefully hand-picked for matching forward voltage. If not, you will risk different current levels through the individual LEDs, some of them could be overdrived while others are underdrived. Connecting them in series on the other hand ensures that the same current flows through all LEDs.

BTW: I think that you could use the Zetex driver for this - the two-cell driver circuit described in their lighting handbook could be scaled to drive four LEDs from four cells. I would use a greater inductor value though (to keep operating frequency low).

[/ QUOTE ]

Ok - So let's assume you are going to run 3 x Lux IIIs at a reasonably high current.
Vf total = approx 3 x 4 V = 12 Vf total
I total = approx 1 amps (maybe 1.5 amps max)

Output to Lux series = approx 12V x 1 amps = 12 watts
10 % loss in efficiency boost converter = approx 15 watts

Georges80 has a converter like this off the shelf.

From each of the batteries (cells) (D cell NiMH)
Current = 5 watts / 1.1 V = approx 5 amps = viable, but need to be paying attention to the cell selection.

There is no advantage to staying with the 1 watt Luxeons over the Lux III - Lux III is basically a premium version of the Lux I with better color and electrical characteristics.
 
Ok, I am planning on running lux3's (just waiting for them to show up)

enLIGHTenment, I will not be implementing dimming at first but do plan to add either a half or a third power setting sometime. This is why I am stuck on the linear/pwm thing. The logic makes perfect sense. An led has a turn on time in the microseconds, where as in a incandescent the full bore current pulses don't matter. Then again, on the other hand, you are throwing away energy with the linear.

Operator_Smooth, If I decide to run them parallel, which I will probably do at the onset, I will put a tiny resistance ahead of each star, to try and minimize the uneven draw of each star. That is the reason I was thinking I was going to have an independent regulator for each star. All get the same current, and hopefully, all are happy. Not enough room for that though...

HarryN, I am a little leary of pulling 5A from the pack. I'm sure that will not be an issue with the 11-12Ah D's, but for the present, I will be running alkies. I am assuming your draw calculation is based on a 3 cell pack, I will be running 4 at first. If I do change it, I will take it to 3 and use the 4th slot for a D size regulator. /ubbthreads/images/graemlins/grin.gif Anyway, with 4 cells, this should place current draw, assuming loaded voltage of 1.1v/cell, at 4.4v/3.4A. (a bit more tolerable) I flat out need to experiment. The stars that are coming are "K"vf's. In the immediate, I will probably run them all parallel, with a resistor at each for a little drop. (paranoid, don't wanna smoke my first set, didn't have cash for spares) I have parts here that I can make a simple linear current regulator. I guess I never entertained the idea of running a series string, and stepping it up. I kinda like that.

However that leaves me with a question. Say that the 4 stars have current draws at 4v of 675, 700, 700, 725. Does this mean that all of the stars will be receiving 675 each due to the weak one? If this is true, can I place a moderate value resistor across it to bring the current across it, (not the emitter, but the emitter and resistor) to 700 to give the others 700ma each?

i.e.
supply - 675led+25r - 700led - 700led - 725led(@700) - supply

Thanks a million guys!

oops, had 625 in example rather than 675...
 
[ QUOTE ]
Wilsonite said:
I guess I never entertained the idea of running a series string, and stepping it up. I kinda like that.

However that leaves me with a question. Say that the 4 stars have current draws at 4v of 675, 700, 700, 725. Does this mean that all of the stars will be receiving 675 each due to the weak one? If this is true, can I place a moderate value resistor across it to bring the current across it, (not the emitter, but the emitter and resistor) to 700 to give the others 700ma each?

i.e.
supply - 625led+25r - 700led - 700led - 725led(@700) - supply

Thanks a million guys!

[/ QUOTE ]
No, that's not quite it. If you put the stars in series, all get the same
current, no matter what their individual forward voltages might be. No need
to equalize the LED currents or forward voltages at all. That's the beauty
of it. If you have a regulated power source @ 700mA, the regulator will apply
whatever voltage is needed to produce 700mA of current flow through the full
string of LEDs.

Each LED, in turn, will have a different voltage across it according to its Vf
characteristic -- whatever voltage it has when conducting 700mA. The unit from
your example that had a Vf of 4v at 625mA would simply have a higher voltage
across it than the others, e.g. 4.1v.

A drawback of boosting to drive a series string is the high switching
currents needed to do this. High currents mean either high losses, or BIG parts.

On the other hand, putting LEDs in parallel leads to current sharing problems,
which are worse if you intend to dim them (Vf curves may match at one operating
point, but not the 2nd).

Another choice, the most efficient electrically, is when the supply voltage
barely exceeds the Vf of the series string of LEDs. Some folks here convert 'D'
cell lights to use a stack of 'AA' NiMH cells in series using homemade AA-to-D
cell adapters, then use a downconverter ('buck') to drive series-wired LEDs.
Chief trade-off here is all those NiMH in series, which are a handful to recharge
& which must not be too deeply discharged, for fear of wrecking the batteries.

So, those are your options in a nutshell!

If you've got the room for a mongo converter, I'd be tempted to boost & drive
a series string of 4, or compromise & wire the LEDs in two paralled strings of
two series-wired LEDs. This would keep the battery pack uncomplicated, give
improved current-sharing between the LEDs (you could hand-pick the LEDs to match
the Vfs of the two series strings), and cut i^2*r losses in the converter as well
(compared to boosting into a 4-series string of LEDs).

Best,
Jamaica
 
You must log in or register to reply here.

Similar threads

S
New AAA Duracells...My Current Best AAA Battery
Replies
3
Views
1K
snakebite
snakebite
vicv
How does mosfet linear regulator work
2
Replies
24
Views
2K
aznsx
A
K
Circuit to distribute LED driver current
Replies
9
Views
1K
Dave_H
D
S
CR123 safe operating current
Replies
3
Views
1K
Kestrel
Kestrel
Poppy
high speed charger fail
2
Replies
25
Views
2K
hsa
H

Latest posts

Top