Constant current source

Go To Last Post
583 posts / 0 new

Pages

Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey

I'm making a circuit to drive some K2 LED's. They need 3.42V and about 0.5A. I would like to use LM317, but I can't figure out how to calculate the resistor values. Just using a resistor is IMHO waste of power. I was told to use a constant current source by the guy who sold me those K2 LEDs, but that controller he uses costs like 100$ a piece, and that is something I can't afford ;) So does anyone have a good formula I could use to calculate the values for those resistors or maybe a better, more efficient way to solve that problem. Current regulation (up to 1A) would be even better, but That would probably make my circuit a bit harder as big (1A tolerant) pots are hard to find here and cost a fortune.

Thanks in advance
Rain

Last Edited: Mon. Mar 5, 2018 - 10:38 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LM317 is one possibility, but a transistor can be set up as a current source too. Before starting to sketch: what supply-voltage is available ?

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Power supply comes from AA cells. I was thinking of 3 or 4, depending on the voltage drop of the driving circuit. (LM317 has a 1.2V drop, right?)

Rain

PS Heatsink is 1 A4 and 2-3mm thick sized aluminum sheet.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LM317 needs 1.2V between output and regulator-pin. But it needs a bit more to stay in regulation. I'll look that up.
I think a transistor-solution will be the most efficient. What NPN or PNP transistors have you lying in the "things"-box ? You'll need a power transistor and a smallish (like bc547)

Edit: @ 500 mA, the LM317 needs 1.6V to stay in regulation.

Do you have an adjustable low-drop regulator ?

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

Last Edited: Sun. Jul 22, 2007 - 08:07 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My parts box has nothing that can take 1A without overheating. The biggest I have is BC618 :P As I need to give a receipt for every part anyway I have to buy those things ;)

Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Do you have an adjustable low-drop regulator ?
And even with a LM1086 (Low Drop Reg) you'll need 1V across the regulator .... plus 1.2V.

I'll sketch something up

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You can also use an opamp as a current source, but I'm not aware of any opamps that can supply that much current.

As for your original question, The resistor value would be 1.25V (the fixed LM317 regulation voltage) /0.5A = 2.5ohms. So, you'd loose 1.25W at the resistor and, obviously, need a resistor that can dissipate that much heat. Also, the LM317 will also dissipate (at a minimum) the 1.6V Plons mentioned x 0.5A, or another 0.8W. You can reduce that with the low-drop regulator he mentioned.

I agree with Plons, since your application doesn't require precision (1% or less) regulation, and because you'll be battery powered, the transistor solution would be the most power efficient.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Do it seems, that I need another battery over there ;) ok I think I can take 5...

I would most appreciate if you can "whip something up" for me.

I found something with transistors. But it's not good, as it draws too much current itself and is not very accurate (I would like to get at least some accuracy).
(circuit can be found on the PDF):

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Kevin ! Agree with you.

OpAmps are great (also for current-source) but the lowest possible losses in linear regulation require the lowest supply-voltage: 3 * AA-cell

Rain, I made even a schematic ! I must have plenty of time :lol:

3 resistors 1R2 250mW are less expensive than a 0R39; so see what suits you best.

Same with the led D1: one small led may be more efficient (in cost or space) than 3 diodes.

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I once designed the attached circuit.
It delivers approx. 250mA regulated with a
temperature coefficient given by the lm334.

Perhaps changing the sense resistor R1 to 0.1 Ohm
may increas current to approx. 0.5 A (use a better transistor then).

But the circuit was critically with respect
to oscillation. In order not to kill
the expensive LEDs I tested alway first with a
dummy load constructed of some 1N4007 diodes in
series.

A further hint if you want to drive 2 diodes:
Its normally not considered a good practise to
put the LEDs in parallel, because current-share
might not be equal ! So either each diode gets
its own regulator, or you need current-equalisation resistors (they increase voltage drop).

Perhaps putting diodes in series and using a
special designed switching LED driver IC might
be the best !

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Whether you use a current source or just a resistor, the same amount of power will be wasted, so for efficiency it doesn't matter. You just share the dissipation over multiple components.

One advantage is, of course, the constant intensity (within limits).

You can use opamps with an extra driving transistor.

If you want better efficiency, you should use a switching regulator. There are a lot of (white ) LED drivers out there.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ossi's suggestion (greetings ! ... and yes, they were impressed ;) ) has a tendency to oscillate. Using a 5 pcs. 1N400x for test-purposes is a wise suggestion.

Jayjay: do you have a suggestion for such a smart led-driver ?

My setup : low cost, works fine with 3 * AA, and minimal amount of loss: 500 mW max with fresh batteries. A solution using a switching regulator is more probably efficient, but costs more.

I owe you some explanation for this circuit:

The voltage on the base of Q1 will be (more or less) fixed. The base-emittor of Q1 and Q2 take 1.4V. When Q2 is drawing current, the voltage across R2 will rise proportionally. @500 mA it will be 0.2V. If the voltage at the base of Q1 = 1.6V, the system is in regulation. Let's assume t5hat the Power-led wants to draw more current (f.i., it's getting warm). More current means more drop over R2, less voltage left to drive the 2 transistors, so less base-current for Q2.

If you want to protect the BD139 in case the Power-LED is not connected, add a resistor of 220R in the collector of Q1.

With fresh batteries, Ub will be 4.5V. The Power-led will take 3.4V, which leaves 1.1V for Q2 and R2. R2 will drop 200 mV @ 500 mA, so 900 mV over Q2.

The BD139 (Q2) needs Vce to be 200mV minimum @ Ic=500 mA to stay in regulation, so the lowest battery-voltage at which this circuit still regulates is 200mV for R2, 200 mV for Q2 and 3.4V for the Power-led: 3.8V total. Per AA-cel 1.26V. And at that voltage it's time for refreshment(s).

Some notes:
A led like D1 performs better than 3 diodes in series: it's more stable.
For R2, 3 small resistors in parallel is often easier than one resistor in the sub-one-ohm range

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just a quick selection from a search on Farnell's site :

LT1618 (step up)
LT3474 (buck)
LM3519MK-20 (step-up)
LT3465
MIC2293-15YML
and many more... ;)

Most cellphones today use white leds, so quite a number of devices have been developed for that market; unfortunately usually in super small SMT :(

Also, the relatively new market of high power leds has also led to development of switching LED drivers.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I use LM3404 for driving high power LEDs. It comes in SOIC and the datasheet seems pretty decent so I made my own PCB for it. Works just fine. I just like the higher efficiency of a switcher.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Nice 8)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I had a look at National's site ...... it's amazing how simple it is these days to "design" ... they do it for you ;)
... and, I am not complaining. It's great !

The LM3404 needs 6V as minimum input. So 5 AA's is a good idea.

Nice indeed Ezcomp.

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I once used the LT1618 to power a Luxeon 1W warm white LED.

MSOP10 isn't easy to prototype with :(

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

:)

I thought these led's always need a cooling-star to operate.
Is the MSOP10 on the rear-side?

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No, the chip is exactly in the center of the picture, above the black wire :D

Little black block at the red wire is the inductor; the green resistors are for current sensing.

These LEDs indeed need a heatsink, but I didn't care :P

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Joo ! That's small ..... you must have very good eyes ! :)

Just as a brain-exercise: how about this approach ?

Bedtime ....

Nard

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I made that circuit about 2 years ago; my eyes were a tad better back then :( The effects of aging :cry:

I used a small magnify glass, an indispensable tool when working with SMT :)

That circuit looks good, most cheaper ICs work that way.

There some project on the net that replaces the comparator with an Tiny15.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Uh oh! So many replies! Have to sleep less I guess...

Thanks for all the schematics and suggestions. I think the last schematic Nard posted is a bit of an overkill :P I think I'll take the one with 2 transistors, 3 parallel resistors and a red LED ;) Seems the most appropriate for me.

About LED drivers: Well I did a little search on them. Local stores only sell drivers for 16 LED's and with maximum supply of just a few mA. Now that's kinda pointless don't you think... And one for something as big as K2, the cost was bigger than what I payed for the LED itself (about 5€ per LED). As I think the end user is going to use NiMH or NiCD rechargeable I think I need some kind of alteration for the Nards schematic... You said, that 1.23V is too little for it, and that's about the voltage they give out when they are half full (or half empty?). 0.2V voltage drop is a bit too much to ask right? Then I think I'll go with 4 AA rechargeables giving me a total of 4*1.2=4.8V and 5.44V when they are fresh from the charger. Is it correct, that I have to change the resistors a bit or will it work without alterations?
Those LED's need very well regulated voltage, but the current can vary a few hundred mA's depending on the batteries status.

PS I think I'm going to make one for every LED as this seems to be a very low cost solution (on first sight and without searching what the parts actually cost ;)) I also assume, that only the second transistor needs a heatsink, right? and those 3 parallel LEDs should be a bit better, than 1206 SMD ;) others should be ok with any wattage. right?

Thanks again

Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hmmm. NiCd and NiMH instead of alkaline .... Yes, you'd need 4 cells then.

But, as you wrote: fresh charged NiCd/NiMH are 1.45V and rapidly decrease to 1.2V, where they hold on for a while, before going down ...
It's the variation in battery-voltage that makes the analog solution less attractive. With fresh cells, Q2 will dissipate 1.1 Watt. With 1.2V-cells: 600 mW.

Frankly, I think that with this pretty large variation in cell-voltage, a switched regulator will be the best alternative. Ezcomp's solution is very nice, as is Jayjay's.
My last schematic may look a bit overkilled, but is it ? I put in all components you'd need, and with a schottky instead of 1N4148, two more resistors can go out.
There is IMO just one objection for my switched variant: I didn't test it :lol:

1206 will do .... max dissipation in R5 ( 0R56 now) is 145 mW

For the Analog version: basically there is no need to change the component-values; it's a constant current source, and the current is (basically) independant of Battery-voltage and voltage across the Power-LED.

Nard

Edit: replaced schematic

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

Last Edited: Mon. Jul 23, 2007 - 12:53 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I can simulate the switching circuit, as at this very very moment I'm already simulating the simple single transistor current source in LTspice :D (autch, efficiency is terrible :P)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Great Jayjay ! Looking forward to the results.

The attached curve is something I wanted to share already on earlier occasions: a small led (in this case a modern 3mm red one) is a pretty good zener for low voltages: see the attached curve. The series resistor is 2k7. For Dean's ButtLoad, I also tested the temperature sensitivity of this light-emitting-zener: and that was quite good as well: better than a zener-diode. No, I don't have the results at hand.

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok thanks for all the suggestions.
I'm now really considering the OP-Amp version. But there is a problem, I don't really like the inductor over there ;) is it necessary? high current inductors cost a fortune here, and it's hard to make them myself (I don't have the know-how).
The cost of that op-amp version is going to be about 10$ (for 4 LED's). That doesn't seem that much... but the 2 transistor one is 4 times cheaper ;)

Is it so, that I can use any OP-AMP there? or do I have to use something specific (like the one in the schematic). Also the diode. Does it have to be that, or can I find something cheaper/similar to that?

(Client has a VERY small budget sadly).

Thanks
Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If inductors make you feel uncomfortable, you'd better stay away from the switching version :D

Switching regulators are quite difficult circuits, lots of things to consider.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No I don't have anything against inductors. I have used them before, but they cost a LOT. especially when they have to take as much as 500mA.
330uH 500mA cost: 2.53€. I need 4. That makes 10€. That's the amount I want to spend on the circuit itself ;)

Sorry for being so picky. As soon as I get back to civilization (I'm digging up the strawberries right now) I start testing the 2 transistor version.

Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It's not an OpAmp, but a comparator. LM393 is about the lowest-cost ;) I could find, and it's necessary that it works with input-voltages close to ground.

It's a switched constant current source, and I don't see a way (yet) to get rid of the inductor and schottky diode. Maybe one of our fellow members has an idea to solve that.
If you need to build 4 units, you could consider to make the inductors yourself. I used to avoid them as much as possible, but since I did some experiments with them (with a MC34063/163 and a Nixie PS), I am beginning to like them :)

I think it's a good plan to wait for jayjay's results with Spice.

Nard

Edit: the switched version I suggested is a free running one, so the inductor isn't critical (I think :lol: )
And Rain, I know you can improvise ;) and you're not picky : let's see where we can get with a low cost solution. Consider it a challenge. At least, that is what I do.

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'll see if I can find a good tutorial on how to make those damn little buggers ;) Maybe I too find, it is not that hard ;) As I also have space concerns I have to make them as small as possible (The schematic can be just 5mm high with the PCB :P)

And now, let us wait for jayjay ;)

Cheers,
Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
(The schematic can be just 5mm high with the PCB Razz)
That will be a problem I think. Even if we take an output-cap in SMT, 5 mm incl. PCB is not feasible

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

With a switching regulator you'll get triangular current through the inductor, which average is the load current. The peak current is higher, and the inductor must be able to handle this. It shouldn't saturate. It also should have low wire resistance to keep resistive losses down.

I wouldn't use a bipolar transistor, the saturation voltage is too high, and you'll lose a lot of power in there.

Suitable inductors are indeed rather expensive, as the diode.

Anyway, the original circuit didn't startup at 5V, the transistor dropped so much volts, that no switching was necessary to push 500mA through the LED :D

The LED I use drops ~3.5V, but I don't have a model for a BD140 :( So I used a 2N3904. With a slightly higher current sense resistor it started to oscillate :)
I used a BAT54 as reference (~220mV).

Still working on it, as I already expected it's all quite critical. For really good performance you need ceramic capacitors, nice inductors, a FET.

I you want this on the cheap, I think the discrete switching circuit isn't the right choice.

Buying a dedicated swithching regulator chip will likely save you a lot of headaches.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jayjay, the 2N3904 is not a good replacement for the BD140 IMO. The BD140 has a saturation-voltage of less than 200 mV @ 500 mA. Max current for the 2N3904 is less, and what's worse: it's NPN :)
The reason I used a BJT instead of a Pch Fet was the battery-voltage: at 4.5V it's hard to get the Fet full on.

I think it will be tough to meet all the requirements.

Rain, are you picking strawberries to make some money in your vacation ?

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Memory error :D I used the PNP version, 2N3906 or so :D

Anyway, still experimenting to come up with a nice circuit :P

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for all the suggestions.
I have big (in diameter) but low (in height) electrolytes, so that's no problem.

I'll see, if I can't get good results with just 2 transistors I think I have to consider the switching one with higher cost (I hope the client can take that).
Thank you jayjay for testing the circuit. I have never fully understood, how to use those simulation programs so I choose to stay away from those ;)

PS I'm not picking them anymore; I ate them before (when they were ready) and now I dug them out with a big shovel as we plan to fertilize the soil a bit and plant new plants next spring. The old ones didn't give enough berries anymore :( And that on/in (which is correct?) my summer cottage. Now I have to pick raspberries ;)

Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jayjay, that's better. At least a PNP. But still too small for the job. Can you find the BC640 in Spice? That's still smaller than the BD140, but a better replacement. Or try BD136 or BD138 .... but I guess you know those ...

I checked component prices, and so far it's only the inductor that's a problem, the rest is all quite low-cost (at least here in Cheesy Cheesy-land).
Rain, I think you need to do something on the specifications. But if you can live with the inefficiency of the analog solution, that's still an option. Low cost and low height.

Nard

Edited

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a very limited selection unfortunately, and I'm too lazy to find spice models :D :P

I'm now using a BCW68F.

If you need inductors, Coilcraft is very generous with samples ;) I got the DS3316 designer kit for free... And for another product I ordered about 50 pieces of the DS1810 series... Guess what, I also got these for free :shock: About ~25 quid worth of inductors, just for free :shock:

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the info JayJay. I will take a look.

Nard, can you tell me, how inefficient this analog circuit is? how much do I loose to that?

It seems that I really do have to talk to my client about the specifications. He wanted something as slim as possible. Maybe I'll make a small compartment for the batteries and electronics which is a bit thicker than the drawing area (maybe 10mm even). Sadly minimum I can get is 10mm total thickens as the Plexiglas and LED's take so much space themselves (3mm for Plexiglas and extra 5mm for the K2 LED.

PS I buy almost all my stuff from Elfa. www.elfa.se

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

From an earlier post:

Quote:
But, as you wrote: fresh charged NiCd/NiMH are 1.45V and rapidly decrease to 1.2V, where they hold on for a while, before going down ...
It's the variation in battery-voltage that makes the analog solution less attractive. With fresh cells, Q2 will dissipate 1.1 Watt. With 1.2V-cells: 600 mW.
It's hard to keep up with us, ain't it Rain ;)

I am having a look at some old MoBo's ..... maybe the switching PS's on them are built with usuable components for this purpose. Of course I realize that it will not solve *your* problem right now, but it may be a source for you in the future.

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here are the results of my simulation.

About ~78% efficiency at 5V. Works down to 4V where it stops switching, which is logical, voltage drop over transistor, inductor, current sense resistor is near the voltage drop of the LED :)

Also attached the ASC circuit file, so download LTspice and play with it. Really, there's no need to stay away from it ;)

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Uups. sry bout that. I knew I saw those numbers before!
but 300mA is a bit too much sadly :( I still have to use the switching one (oh goodie goodie).

Found this inductor: MSS1260-334KL 330 ±10%uH 0.630 Ohms(DCR) 4.0 MHz (SRF) 1.00 A (Isat) Coilcraft. This is good right?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

0.63 DCR is a tad on the high side, the less the better :)

Higher inductance values have higher DCR values, thinner wire needed to get all the coils on the bobbin. Lower inductance gives more ripple and higher peak current and higher operation frequency.

It's all a compromise.

100uH seems a nice compromise. Not too big, not too small.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

oh I can take smaller? that's good. I was searching for a 330 one. Good. Then MSS1260-104 will do the trick (DCR is just 190 mOhms ;)) SRF is 8MHz.

Downloaded the Spice too. I'll play with it for a while and see, what I can do with it.

Thanks
Rain

EDIT: Heh cool I'm starting to like that spice ;) nice graphs and this circuit seems to work just fine.
I think I can find the founds to make that switching one. I'll also see, if I can get samples from coilcraft. Any suggestions on what to get for future or something like that?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Very nice indeed. I am a bit old school I guess. Time for some change :!:

The switching PS on the MoBo is interesting too (I like to do some reverse engineering). They use Nch-FET's all over, of which the gate is driven with 12V. The used inductors you ask? 300 nH Impressive huh?
So it may be interesting to collect old PC's, as they are a great source for parts.

"Het enige dat u weggooit .... is de verpakking" :)
(in English: all you dispose is the packaging)

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Modern processors draw huge currents, 100A :shock:

A reasonably modern processor consumes around 90W at a Vcore of 0.8V or so :shock:

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Okay, an update

I built a switcher like I posted here, and the results are promising. Of course there is some parasitic oscillation, but that doesn't surprise me at all as it's built on a breadboard.
I use 5 * 1N4007 as dummy PowerLED, and a lab-supply as battery.
The current of 500 mA can be reached from 4.5V in to 15 V in. I wondered about the 4.5, as I expected it to perform better than that: and it's the dummy-load that's responsible: it takes 4.0V @ 500 mA. That's 0.6V more than the real PowerLED.

So it can be expected that the circuit will do fine from 4 Volt to 6 Volt, which are 4 empty NiCd/NiMH's resp. 4 fully charged ones.

It's coffeebreak now ... pictures and schematic update will follow soon.

It's fun ...

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

thanks a lot. The schematic will cost me exactly 10€ ;) 2.5 per LED. not bad actually... (of course I hoped for better, but what can you do ;). Redesigning the whole thing to drive 2 LEDs is i think stupid right now. I would have to use much more AA's then too and that's not smart either. Stepups won't work on that schematic (they only add more cost and inefficiency). I think I can find the 10€ to build it. Have to save some from the casing or something like that.

And Thanks Nard, for testing the schematic. Don't you have a similar LED at hand? Those are pretty cheap and cool little buggers ;) one can light a whole toilet/bathroom with just one of it (have to get a warm yellow one, bright "clean" white makes you feel, like you are on the surgery table, not on the can).

Cheers,
Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Your bathroom toilet will be dimly lit then :D

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Bathroom: if it's a "hard" job, my eyes are closed anyway :)

Time for an updated schematic; pls. read the notes, and the optimized values of some components.
I modified the dummy so it behaves now almost like the PowerLED.
By using the second half of the LM393 as driver, performance increased.

Pictures need some editing ...

Nard

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The pictures

Scope: upper trace 2V/div; comparator output == drive of sw. BJT
lower trace 20 mV/div; voltage over Rsense (= 0R15)
Picture taken with 330 uH, 5.76V as Ubatt, 500 mA outputcurrent, 3.5V dummy
Transistor = BC640
At 4.2V and lower, the current reduces; 400 mA @ 4.0V input ... not bad
(performance with BD140 was better: 4.05 V as minimum input voltage)

Did I mention that cooling is not necessary ? :)

Cheers guys !

Nard

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What's the efficiency?

Pages

Topic locked