Two component LS current-mode booster!

[bikeNomad]

Well, after someone was complaining about my LT1308B circuit and its number of components, I started thinking about making a simpler, smaller, and cheaper circuit.

So I came up with a 2-component circuit.

My goals with this circuit:
<UL TYPE=SQUARE><LI> drive one or more Luxeons in boost mode
<LI> protect them from peak currents beyond their ratings
<LI> do not require trimming of output voltage or component changes for different/more luxeons
<LI> output at least 250mA average current
<LI> be really small
<LI> cost less than $3 in components in reasonable quantities
<LI> work down to below 1.8V
<LI> have good efficiency
<LI> possibly allow brightness control
[/list]

Well, I think I've done it, with a 2-component circuit. I'm ordering the IC right now, and will be trying it soon. It should work down to 1.2V.

This is just a teaser; the circuit will follow along with analysis as soon as I put one together.

Some things I realized:

<UL TYPE=SQUARE><LI> the Schottky diode eats efficiency and is only needed because of the output cap
<LI> the output cap is only needed to reduce output ripple
<LI> today's boost regulators already sense peak current through the inductor
<LI> the LED doesn't care if it gets something other than DC (if you're willing to work at a somewhat reduced average current, or if you don't care if peak currents go a bit above 350mA)
[/list]

My circuit puts out 350mA peak currents, with 270mA average with a 4.7uH inductor.
And in simulation gets better than 95% efficiency driving a single Luxeon from 2 alkaline AA cells.

Anyone guess how I did it?

 

[LiteFreek]

Hmmm, magic?

-the Local LiteFreek

 

[remuen]

bikeNomad, this looks very interesting! The only drawback in my eyes is that it works 'only' down to 1.2 volts input. But it won't be a real handicap if it will work very well down to this voltage.

When can we expect more informations?

 

[KenBar]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by bikeNomad:

Anyone guess how I did it?<HR></BLOCKQUOTE>

Not a flippin clue...and probably will never have one either....YET...

95%...is pretty impressive in anything. How much longer till you quit teasing?
What is max voltage it will run at?

I got tired of spending 25 bucks for chips and postage.. anything under 10 bucks sounds great!

 

[**DONOTDELETE**]

What is the SPICE model you are using for the Luxeon Star?

I am not able to get a model that behaves the way I think it should

Thanks!

 

[**DONOTDELETE**]

Are you using a 555 timer?

J/K

 

[bikeNomad]

No, no 555's here. Here are some more hints:

The way a typical current mode boost converter works is this: the switch closes, and the current in the inductor rises till it gets to a (usually fixed) value. Then the switch opens.

The inductor still wants to maintain the same current, though. So the voltage across it (which is in series with the input voltage) increases until it can produce the current that it was passing immediately prior to the switch opening.

This current then flows through the load for a while (the timing depends on the IC), and the whole cycle repeats.

What I figured out was that if you take a boost IC (like the LT1615, for instance) whose inductor current limit is set at 350mA
and set it up without a diode (the LS won't conduct much current below about 3V, after all) or output cap (since we don't care if the LS is getting DC), then you'll get what is more or less a pulse waveform across the LED.

This waveform will have an initial current peak at 350mA, and then the current will reduce linearly as the inductor discharges (higher inductance=less slope), until the switch closes again.

 

[**DONOTDELETE**]

The Zetex ZXSC300 spec sheet has a "maximum battery life solution" that claims 94% efficiency with four components, the chip, an inductor, a transistor, and a current sense resistor.

If you use a chip that has a built-in transistor and an internal resistor, then you're down to two components.

We played with this mode early on in the development of our Zetex project but were not impressed with the output level.

 

[bikeNomad]

You've got it: a chip with an internal current sense resistor and low Vce switch.

The efficiency is quite good, and I don't personally care about getting the full 350mA average current into the LS (especially when its efficiency is better when run cooler). I figure I can average 270mA on fresh batteries, dropping down smoothly (semi-regulated) to 40mA or so average (low duty cycle) at 1.2V input. Even at 1.2V input the efficiency is better than 75%.

This is more for people who need long battery life in a 2-cell light and don't mind trading a reduction in brightness for it.

If you want constant full brightness, use something like my LT1308B circuit, which maintains full brightness down to less than 1V input voltage.

 

[papasan]

i made a 3LED circular PCB for replacing a mag-light lamp, on this i used a small (1210) panasonic inductor...trying to over-drive it as much as possible i kept trying to up the output and it wouldn't budge...so i kept trying to increase the output until i burned my finger on the inductor...obviously i had reached the max of my inductor quite awhile back, that's why there was no increased output...anyhow, dunno what happens to an inductor that's running at peak too often, but heat is one of the things (which means that efficiencey is down)...

so you have an inductor and a PIC, i take it?...

 

[machspass]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR> so i kept trying to increase the output until i burned my finger on the inductor...obviously i had reached the max of my inductor quite awhile back, that's why there was no increased output...anyhow, dunno what happens to an inductor that's running at peak too often, but heat is one of the things (which means that efficiencey is down)... <HR></BLOCKQUOTE>

Papasan, I've looked at some of the inductors that Digi-Key offers, and I find that the SMA chip inductors have a considerably higher DC resistance than coil ones. In addition to being less efficient, I can see where a 1210 package inductor wouldn't be able to handle high currents. You might want to check some of the SMA power inductors made by J.W. Miller that Digi-Key carries. I have a couple on the way to me. I got the suggestion from the Zetex project discussed on here, as well as the Zetex data sheets. They suggested using an inductor with a 500 mohm DC resistance or less.

Can't wait until I get the parts and can begin playing!

-Jason

 

[bikeNomad]

I've used the Panasonic ELL6SH and ELL6RH inductors from Digi-Key; these are good for quite a bit of peak current and are only 6mm square. I'm using these in my LT1308B circuit and they work well.

There are also the Panasonic ELJ-PA 1210 inductors; a 4.7 uH one of these is good for a peak of 350mA and has a 0.340 ohm resistance. If you're using a current-limited circuit (like the LT1615) that can limit to 350mA you should be OK with these.

 

[papasan]

i've since re-designed it to take up to 1812 sized package...its bassically the whole bottom half of the circular PCB...it'll take about anything since it's mostly copper...i haven't tested anything out yet because i broke my carbide bits on the original ones i made...damn they're easy to snap...

 

[bikeNomad]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by papasan:
i haven't tested anything out yet because i broke my carbide bits on the original ones i made...damn they're easy to snap...<HR></BLOCKQUOTE>
That's one of the reasons I only do surface mount circuits any more...

 

[machspass]

Well, I got my parts from Digi-Key yesterday, and built up a Zetex310 circuit designed for maximum brightness that was in the data sheets. I have just started testing, and it looks very promising so far....

I ran this thing with a mostly dead battery for a few hours. It put out a half-way decent amount of light with a battery voltage of 0.8 volts. It's quite bright with a fresh battery.

Papasan:
I ordered some ferrite beads, along with the SMT inductors. I find winding my own inductors to be fun and much cheaper. I put about a dozen turns of #28 wire on a small toroid, and got an inductance of 130uH and a DC resistance of 250mohm. Of course, it helps to have a lab full of eqipment at my disposal to test this.

 

[papasan]

what are you driving with the 310 circuit?...if it's an LS check out the ZXCS300 thread, duggg says that his 2.5 turn hand-wound coil is best for efficiencey...i'm not sure how he came to this besides turning, testing, turning, testing, etc...i have almost no equipment, just a fluke that i had for testing power at work...and considering i need a new car and want to buy a house soon i'm not willing to put alot of money into getting scopes and variable constant power supplies and stuff...

so i'm not sure what the theories behind the hand-winding are...duggg brushed the surface but didn't go into much detail...he was saying that more turns reduce the efficiencey...from a higher DCR from more wire being used?...i would think that more turns means the coil would charge faster, but possibly lowers it's saturation?...if anyone wants to shed some light onto this it would be helpful for me in windign a coil for the max1674 circuit since the one i'm using now makes the circuit pulse when the batteries drop below .9V...

 

[machspass]

Papasan,
I am driving a single Nichia white LED with the circuit. I have read Dougg's web site on the Mag-Lite mod using the Zetex circuit. I'm basically using the circuit that's in the data sheet applications. The data sheet recommends a 100uH coil, so I took Dougg's advice from the Zetex web site on the particular ferrite beads he got from Digi-Key, and experimented with the windings. I have an LCR network analyzer at work, so I could easily check the inductance and DCR as I went along. I tried a couple of different ferrites with various number of windings.

I want to know too, just what the effects are on the performance using different inductor values. The one I wound that I'm using now is about 130uH, and DCR of approx 300mohm. The DCR does affect the efficiency of the circuit, but I read, either in the data sheet or Dougg's web site (I can't remember which), that a DCR of 500mohm or less is desirable. Even with as many as 12 windings of 28ga wire, the DCR wasn't above 500mohm. Torroids are extremely efficient inductors or transformers!

I ran a few quick bench tests to take some voltage and current measurements, and it looks really good so far. I can pass those numbers along to you if you like.

 

[**DONOTDELETE**]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by papasan:
Duggg says that his 2.5 turn hand-wound coil is best for efficiencey...i'm not sure how he came to this besides turning, testing, turning, testing, etc... i'm not sure what the theories behind the hand-winding are...duggg brushed the surface but didn't go into much detail...

The first step was finding a core material that had the right permeability. If the permeability was too low, a lot of turns were required. To fit a lot of turns, either the core had to be big (not desired) or the wire had to be thin, meaning high DCR and low efficiency. If the permeability was too high, the core saturated with too few turns, so its inductance wasn't high enough to provide the necessary 350mA average current.

After determining the best relative permeability was around 3000, we searched for the smallest core that would still provide the desired current. If the core was too small, it wouldn't hold the necessary number of turns, or really thin wire had to be used. Fortunately, the core was so small that only 1-4 turns would fit, so the wind-then-test process didn't take that long.

He was saying that more turns reduce the efficiencey...from a higher DCR from more wire being used?

That, and also the more turns, the more saturated the core is. The more saturated the core, the more it behaves like a resistor rather than an inductor.

...i would think that more turns means the coil would charge faster, but possibly lowers it's saturation?

Actually, the more turns, the higher the inductance, so the slower it charges. Inductance is proportional to the square of the number of turns.

Saturation is related to the permeability. The higher the number of turns, the more the core is driven into saturation, and the more the core behaves like a resistor rather than an inductor, thus lowering its efficiency.

If anyone wants to shed some light onto this it would be helpful for me in windign a coil for the max1674 circuit since the one i'm using now makes the circuit pulse when the batteries drop below .9V...

One inductor aspect we didn't have to worry about with the Zetex circuit was quality, which is how well the coil maintains its inductance as the current through it varies.

The 4-turn bead core inductor has very poor quality, because its inductance changes radically between the charge and discharge cycle.

Fortunately, the Zetex chip and circuit were stable enough to operate properly with such an inductor, and at 22 cents, the price certainly was right

You may not be as lucky with other chips and circuits. Since quality is generally related to the number of turns (the more the coil looks like an inductor, the better it behaves), you may have to accept higher DCR and thus lower efficiency.

Of course, at 0.9 volt, the chip itself may not work properly with any inductor. A Schottky rectifier, if not already used, may help stabilize the circuit at low voltages---at the expense of efficiency.
<HR></BLOCKQUOTE>

 

[**DONOTDELETE**]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by machspass:
I read, either in the data sheet or Dougg's web site (I can't remember which), that a DCR of 500mohm or less is desirable. <HR></BLOCKQUOTE>

Actually, you want far less than that, if possible. Good commercial 68uH inductors are on the order of 100 millohms. The 4-turn ZLT+ inductor requires only 6cm of #24AWG wire, which has a resistance of just 6 millohms, so inductor loss is virtually nil.