My LM3410x based Boost converters: Spark driver! PART 1

[erdabyz]

You don't need superduper capacitors for those drivers, just what the manufacturer recomends. And you don't want to use tantalum capacitors or things like that, the high switching frequency of these ultrasmall converters makes them like ceramic ones better. But you need good ceramic capacitors, X5R at least.

Input capacitors are also not really a "neccesity" but just a "slight improvement" in battery operated well designed circuits with no other integrated circuits laying around. Just pick a datasheet from Linear technology of a dedicated LED driver intended for flashlights (like LTC3490) and see how they say that input cap is just optional. Well, better if it's there, for sure.

You have all the host's case connected to ground provding shielding for EMI and all those things.

I have made quite a bunch of switching drivers to the date with various IC's, and tested some of them also with o-scope and i've never needed output caps higher than 10uF (most comon value I use is 4.7uF) to achieve millivolt-microamp ripples in the output with devices operating at >= 1Mhz.

The most important part of making a swithching driver is picking the correct inductor according to the voltage difference between input and output and the current requirements, giving special attention to the current startup transient which the inductor must withstand in order to start the driver.
We have massive shielded ferrite core inductors rated at up to 7 Amps that fit perfectly inside an aixiz module, so size is not an excuse.
Also PCB design is critical.

 

[geologic]

You don't need superduper capacitors for those drivers, just what the manufacturer recomends. And you don't want to use tantalum capacitors or things like that, the high switching frequency of these ultrasmall converters makes them like ceramic ones better. But you need good ceramic capacitors, X5R at least.

Input capacitors are also not really a "neccesity" but just a "slight improvement" in battery operated well designed circuits with no other integrated circuits laying around. Just pick a datasheet from Linear technology of a dedicated LED driver intended for flashlights (like LTC3490) and see how they say that input cap is just optional. Well, better if it's there, for sure.

You have all the host's case connected to ground provding shielding for EMI and all those things.

I have made quite a bunch of switching drivers to the date with various IC's, and tested some of them also with o-scope and i've never needed output caps higher than 10uF (most comon value I use is 4.7uF) to achieve millivolt-microamp ripples in the output with devices operating at >= 1Mhz.

The most important part of making a swithching driver is picking the correct inductor according to the voltage difference between input and output and the current requirements, giving special attention to the current startup transient which the inductor must withstand in order to start the driver.
We have massive shielded ferrite core inductors rated at up to 7 Amps that fit perfectly inside an aixiz module, so size is not an excuse.
Also PCB design is critical.

I set R1 to 1ohm to drive about 190mA, and the inductors are the same as suggested in the LM3410 datasheet (rated 3A), so i think inductors are not the problem.
The only component that is a little different is the schottky diode, i'm using one with Vf=0,5V (instead of Vf=0,36V of the LM3410 datasheet).

Do you think it can be the cause?

 

[geologic]

PCB design which is not optimal shouldn't cause it to blow. It will affect efficiency and regulation though. Just make sure you read the datasheet through and through so you dont misinerpret the layout.

If you blew a capacitor, was it a tantalum on reverse polarity?

BTW, I found that in sepic config, the lm3410 can't seem to handle the same currents it can in boost mode. The datasheet doesn't give many specifics or examples as to what it is capable of as a sepic converter.

Do you know the current limit in SEPIC mode? I'm using it to drive 190mA, but i would like to reach about 1,5A.

 

[erdabyz]

I set R1 to 1ohm to drive about 190mA, and the inductors are the same as suggested in the LM3410 datasheet (rated 3A), so i think inductors are not the problem.
The only component that is a little different is the schottky diode, i'm using one with Vf=0,5V (instead of Vf=0,36V of the LM3410 datasheet).

Do you think it can be the cause?

I actually haven't read the whole thread, just a bunch of posts, so don't really know what happens to your driver. Having a schottky with higher Vf will just compromise efficiency a bit but shoudn't affect output stability, as long as it can withstand the currents involved it should be fine. We use schottky diodes instead of silicon ones not because of the lower forward voltage but because of they faster recovery and switching times, and that "speed" is intrinsic to metal-semiconductor junctions.
It would affect stability if your diode were too slow but i don't think that should be case.

 

[geologic]

I actually haven't read the whole thread, just a bunch of posts, so don't really know what happens to your driver. Having a schottky with higher Vf will just compromise efficiency a bit but shoudn't affect output stability, as long as it can withstand the currents involved it should be fine. We use schottky diodes instead of silicon ones not because of the lower forward voltage but because of they faster recovery and switching times, and that "speed" is intrinsic to metal-semiconductor junctions.
It would affect stability if your diode were too slow but i don't think that should be case.

Well, before using my last LM3410 on a new PCB, i tried to "downgrade" my SEPIC to a boost design. I remove L2 and shorted C3, put a bunch of nichia 5mm leds to get 9V @ 200mA as a load, powered with 3 AA batteries and... IT WORKS!

So, my design is ok, all components seems ok, the problem must be in C3 or L2... Since L2 is the same as L1 and i think i saw smoke from C3, the problem must be C3... I'm using MLCC X5R 2200nF, 25V from Taiyo Yuden

Shoud i use a big one?

 

[pocketfluff]

2.2uf should be fine for C3 (thats your coupling capacitor right?)
im looking at your sepic board you posted up. Is it just me or do BOTH inductors have a connection to pin 1 (switch pin). Might just be the small size playing tricks on my eyes. Should be fine since it worked for boost and the rest of the sepic board looks ok.

And if you're after 1.5A from the sepic, remember that its not quite as efficient as boost and that you will likely only get that output with when the board is stepping the voltage down.

Erdabyz is right in saying that 10uf of ceramic is adequate. But sepic is a little different. The output cap is not just to lower ripple, but also provides the output current on one part of the switching cycle. So higher currents needs more capacitance. National has an app note on the sepic converter where you can calculate values for all of this.

 

[geologic]

2.2uf should be fine for C3 (thats your coupling capacitor right?)
im looking at your sepic board you posted up. Is it just me or do BOTH inductors have a connection to pin 1 (switch pin). Might just be the small size playing tricks on my eyes. Should be fine since it worked for boost and the rest of the sepic board looks ok.

And if you're after 1.5A from the sepic, remember that its not quite as efficient as boost and that you will likely only get that output with when the board is stepping the voltage down.

Erdabyz is right in saying that 10uf of ceramic is adequate. But sepic is a little different. The output cap is not just to lower ripple, but also provides the output current on one part of the switching cycle. So higher currents needs more capacitance. National has an app note on the sepic converter where you can calculate values for all of this.

Yes, i'm using 2.2uf as coupling capacitor.

About the drawing, just L1 is connected to SW (pin1). L2 is connected to GND (pin2). After "downgrading" the circuit, i soldered C3 and L2 again and the circuit stoped working. I removed L2 and shorted C3 and it worked again... (so f**** frustrating). Boost is ok, SEPIC is not. I don't understand why this happens, since i'm using the components showed on the LM3410 datasheet. C3 it is not the same, but it has the same specs... And i'm sure it is not a drawing error (just 2 components are different from the boost design).

I want to use it with 3 AA batteries, so almost the time it will be stepping down the output to 3V, so i think it can handle 1,5A. I choosed this IC because it can operate from 2,7 to 5V and can handle 2.8A (?).

I'm completely lost... :yabbem: I'm going to check the app note to see if i can figure wat is wrong.

 

[erdabyz]

Yes, i'm using 2.2uf as coupling capacitor.

About the drawing, just L1 is connected to SW (pin1). L2 is connected to GND (pin2). After "downgrading" the circuit, i soldered C3 and L2 again and the circuit stoped working. I removed L2 and shorted C3 and it worked again... (so f**** frustrating). Boost is ok, SEPIC is not. I don't understand why this happens, since i'm using the components showed on the LM3410 datasheet. C3 it is not the same, but it has the same specs... And i'm sure it is not a drawing error (just 2 components are different from the boost design).

I want to use it with 3 AA batteries, so almost the time it will be stepping down the output to 3V, so i think it can handle 1,5A. I choosed this IC because it can operate from 2,7 to 5V and can handle 2.8A (?).

I'm completely lost... :yabbem: I'm going to check the app note to see if i can figure wat is wrong.

Well, it has a switch rated for 2.8Amps, it doesn't mean it can source 2.8Amps of continuous current. Currents switched in the coil are bigger than average current delievered to the load (at least in boost mode). Having a 2.8A switch means that it sould source up to 1A or so. My highest current design with LM3410x delievered 650mA in boost mode (limited by resistor).

 

[geologic]

Well, it has a switch rated for 2.8Amps, it doesn't mean it can source 2.8Amps of continuous current. Currents switched in the coil are bigger than average current delievered to the load (at least in boost mode). Having a 2.8A switch means that it sould source up to 1A or so. My highest current design with LM3410x delievered 650mA in boost mode (limited by resistor).

I read somewhere that it can delivery up to 1,5A continuous, and little more pulsed, i'll do some tests if i can put it to work in SEPIC... :undecided:

I realize that you build several LM3410 circuits, any in SEPIC mode? What components did you use?

 

[pocketfluff]

Are you certain that your test with the sepic didn't damage the IC? a I've found that when it doesnt turn on, theres a decent chance ive killed it (aside from reverse battery). Measure the FB voltage in boost mode to check on its regulation.

Don't worry too much about your selection of components at your currents atm.

When you resoldered L2 and C3 did you use a new part just to make sure?

I drove 800mA at 4v from a 5v source using the sepic. 20 or 30 uf at the output. I think 4.7 for the coupling.

 

[geologic]

Are you certain that your test with the sepic didn't damage the IC? a I've found that when it doesnt turn on, theres a decent chance ive killed it (aside from reverse battery). Measure the FB voltage in boost mode to check on its regulation.

Don't worry too much about your selection of components at your currents atm.

When you resoldered L2 and C3 did you use a new part just to make sure?

I drove 800mA at 4v from a 5v source using the sepic. 20 or 30 uf at the output. I think 4.7 for the coupling.

Yes, in boost mode FB voltage always around 190mV, so the IC is alive.
I found a apps note at http://www.national.com/an/AN/AN-1484.pdf about SEPIC design. They say something about the coupling capacitor:

"The SEPIC capacitor must be rated for a large RMS current relative to the output power"

With their equation, i calculate Ics(rms)=1,45A for my design, so there is any chance that my C3 is not rated for that current? How the hell i find Ics on the capacitor datasheet? I'm using TAIYO-YUDEN TMK212BJ225KG-T...

 

[pocketfluff]

Dang I was thinking your IC was funky, but its alive and kicking.

I wouldn't worry about the cap. X5R and X7R can handle quite massive currents through their little bodies. I believe its typically over 10A. You mentioned it was 2200nf. Try a diff cap in case its 2200pf or something silly like that

 

[geologic]

Dang I was thinking your IC was funky, but its alive and kicking.

I wouldn't worry about the cap. X5R and X7R can handle quite massive currents through their little bodies. I believe its typically over 10A. You mentioned it was 2200nf. Try a diff cap in case its 2200pf or something silly like that

Well, i changed C3 for a new one 10microF (the only value i had in SMD) and the SEPIC is working!!! VFB is around 190mV and output 3.4V @ 190mA (R1=1ohm). Nice!

Next step, i changed R1 to 0,22ohm to get the output up to 860ma. Powered on and... the led starts flashing... (i build my first strobe ) This time no smoke, just regular flash around 4 times / sec.

So, i realize:

1) My old C3 was a crap and was the problem of the SEPIC not working
2) 10microF is not a good value for C3, at least for high current

I'll try another values for C3, maybe near 1microF... I don't have any here, so i have to buy some.

 

[pocketfluff]

Wait why are you reducing the value of C3 from 10uf to 1uf. And try to put more capacitance for the output as it supplies power AND filters the output. If that doesn't work you could always try laying multiple caps in parallel to get your desired value. I'm not sure if that made a difference for me but its worth a try.

Good luck!

 

[geologic]

Wait why are you reducing the value of C3 from 10uf to 1uf. And try to put more capacitance for the output as it supplies power AND filters the output. If that doesn't work you could always try laying multiple caps in parallel to get your desired value. I'm not sure if that made a difference for me but its worth a try.

Good luck!

I was trying to get a new value for C3 as close as the LM3410 datasheet (original value is 2,2microF), but i'll try your suggestion first, since i have lots of 10microF, i can parallel them to make 20microF or 30microF...

I'll be back...

 

[pocketfluff]

Bear in mind that the design example they give is for 350mA. The coupling and output capsneed to be higher for your uses.