My First Boost Converter Design

[Jonathan]

Hi

I've been posting to this board for a while, and in the background I've been working on a boost converter design. My board is based upon the LT1618 chip, and is intended to run long series strings of LEDs from rechargeable batteries on a bicycle...I use it to boost 12V to about 24V for strings of 8 red LEDs.

Anyhow, the project has reached the stage of being publishable:

My Boost Converter

My tail light design

-Jon

 

[dat2zip]

Jonathan,

Nice work there. I've played with that chip too and I have found that the package is thermally lmiting.

In otherwords, it will overheat before you can reach the max output power the device can deliver. Try gooping some Artic Silver III or the new Artic Alumina?? since the latter is totally non-conductive under the IC before soldering the IC down. This help immensly on the micro-SOIC footprint. All the micro-SOIC (MSOIC, uMAX, etc...) have this characteristic.

Even the beloved LM2621 thermally shuts down before it reaches the SW_ILIM.

Gooping over the whole thing after it is solder also works. You might be able to get the 4W without encoutering the heat limits.

If you want to goop the top side, try a 50/50 mix of Artic Alumina?? and Epoxy.

I've gooped several converters now with 50/50 Artic Silver III and crossed my fingers. Seems to have worked and not shorted anything out.

 

[MrAl]

Hi Jonathan,

Looks like an interesting circuit there.
Did you do some efficiency measurements?

Good luck with your LED circuits,
Al

 

[**DONOTDELETE**]

Jon,

Glad to you see you getting your design up and running. Hope it works out well.

Anyways, thought i'd point out something in your circuit which could result in catastrophic failure of your LEDs.

On your output, I see you are monitoring both current and voltage. The problem is that you are monitoring current in the circuit before you are branching out to the parallel strings of LEDs (at least that is how it looks in your schematic.)

This is a classic example of how a SUPER CASCADE CATASTROPHIC FAILURE could occur.

If you happen to lose any one of your parallel strings of LEDs, your regulator will continue to pump out the set output current into the remaining strings of LEDs. Depending on your current levels, this may be too high for the remaining LEDs and kill yet another string. Then so on and so on . . .

Just thought I'd point this out as you may want to think of a circuit modification to account for this. You might even be able to monopolize on the voltage feedback input as a fault current input or something.

Hope this helps!!!!

Dan

 

[KenBar]

Space Cat
Nice pics on your Telsa! Ole Nikola Tesla would have been proud!

Give this a shot for your next project and let me know how many LED's it runs.
Free energy for your LED's\


ps..
Others who built it.

This is my next project...and then permanent new generation LS's.

 

[Jonathan]

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

On your output, I see you are monitoring both current and voltage. The problem is that you are monitoring current in the circuit before you are branching out to the parallel strings of LEDs (at least that is how it looks in your schematic.)

This is a classic example of how a SUPER CASCADE CATASTROPHIC FAILURE could occur.

If you happen to lose any one of your parallel strings of LEDs, your regulator will continue to pump out the set output current into the remaining strings of LEDs. Depending on your current levels, this may be too high for the remaining LEDs and kill yet another string. Then so on and so on . . .
<HR></BLOCKQUOTE>

Thanks for the suggestion. I thought about this exact issue, and made an intentional decision not to address it with a more complex circuit.

My reasoning is that this was intended as a small boost converter intended to operate at low power levels (1 to 2 watts) rather than a high power converter intended to operate an illumination array at full power all the time. I expected that the usual mode of operation would put only a single string of LEDs on the converter.

For the tail light, the LEDs are being run in parallel, and if a string of LEDs fails open circuit, then the other LEDs will carry correspondingly more power. However for the tail light the LEDs are underdriven in normal operation, and were a string to fail the LEDs would still be in their normal operating range.

-Jon

 

[dat2zip]

Jon,

I did an experment with a series string of white LEDs and the LM1618. I found the transistor switch element inside the IC turned on and off much slower than I had anticipated. With a standard LS the inductor flyback only reaches the LS voltage plus diode drop. So, rise/fall times are not as critical.

With 6 LEDs the flyback voltage will be 24 plus volts. I think I measured 5uS or slower for the switch time. I was rather disappointed since it reduced the efficiency really low compared to what the device could have achieved (like 50%). The other thing I notice was it pulsed only once in a while to maintain the output regulation and the switch voltage waveform was aweful. it slowely skewed up to peak and then slowely skewed down. So, the rise/fall times were very noticable.

I'd be really curious to see of the switching is improved with higher VIN voltages. I was running the series string off a 2AA setup. Just wondering if Vin was at 6V would the IC then be able to turn on/off the switch element transistor more faster making the IC more efficient.

-Wayne

 

[**DONOTDELETE**]

Right, if they are underdriven, then the remaining strings will be alright if one fails.

Actually two other things which could be added are:

1. A zener diode across the output to limit the output voltage . . . Oooppss . . . thats right, you're already monitoring output voltage with the voltage divider. Nevermind.

2. Possibly adding a diode in series with your input + voltage. This would be a simple reverse voltage protection (of course you have the nominal diode drop which increases your minimum operating voltage) If you reverse the input voltages, the FET will reverse conduct and blow your controller chip. (I know by LOTS of experience!!! Man is it a pain to solder those micro-sot packages)

Good luck!

Dan