1
\$\begingroup\$

What I want:
Drive a 17V, 600mA COB LED from a 5V 3A USB powerbank.

Approach 1: Use a boost LED driver.
I have looked through Digikey as well as manufacturers' parametric searches to find a suitable boost LED driver. Unfortunately, the few parts that would work (e.g. the LT3477, almost 7€/piece) are either comparatively expensive, require lots of auxiliary parts or both.

Approach 2: Boost converter in series with a buck LED driver
If I were able to use a regular boost converter in series with a buck LED driver, the selection would not only be far greater but the parts would also be cheaper. Based on a forum post, connecting boost converters in series is only possible if they are of the isolated type (the boost converters I have looked at didn't mention isolation, so I reckon they are non-isolated). Does that limitation also apply to me, as I would connect a boost in series with a buck converter, not two boost converters (as buck/boost converters are very similar, I suspect it does)?

Question: Is approach 2 viable, or do I have to bite the bullet and go with approach 1? Do you have an alternative approach I haven't thought about?

\$\endgroup\$
  • 1
    \$\begingroup\$ What does for example TI webench recommend? I would aim for straight boost converter. \$\endgroup\$ – winny May 28 '18 at 11:34
  • \$\begingroup\$ @winny I have just (for the first time) tried webench, it gave me 11 results. The issue I have is that these very polished looking circuit are completely custom, making it hard for me to wrap my head around. Using a boost converter with internal switches would have made it easy for me to make changes. With the webench circuits, making changes feels like stepping through a minefield. Here's the simplest design \$\endgroup\$ – iMrFelix May 28 '18 at 12:26
  • \$\begingroup\$ Looks alright to me. What part scares you? BOM count? \$\endgroup\$ – winny May 28 '18 at 13:53
  • \$\begingroup\$ One thing I am unsure about is the sheer size of the ciruit. I know that component layout is critical for switching regulators, but the vast unused copper fills seem excessive. Couldn't I just shrink everything while keeping the general layout? Furthermore, I would like to incorporate PWM dimming. My understanding would tell me to replace the 28.7k resistor "RDIM" with the PWM signal (based on info from the LM3428 datasheet). On the other hand, if I select "PWM Dimming" in the "Simulation" window, it adds a PWM signal between GND and +5V Vin while keeping the "RDIM". \$\endgroup\$ – iMrFelix May 28 '18 at 14:41
  • \$\begingroup\$ Webench does SMD only with no heatsinks so you can easily do exactly as you suggest. See Andy’s answer below for a lower parts count solution of the same thing. \$\endgroup\$ – winny May 28 '18 at 15:59
2
\$\begingroup\$

Do you have an alternative approach I haven't thought about?

Find a generic boost converter that can reliably produce 17 volts at 600 mA then, rearrange the feedback loop with your LEDs as part of it: -

enter image description here

The picture above uses a 12.4 ohm resistor and a FB voltage of 250 mV in order to get 20 mA but there's little stopping you using a 1.24 ohm resistor and getting 100 mA.

If you pick a generic booster that has a 0.8 volt reference voltage for feedback and you want 600 mA to flow, the resistor would need to be 1.333 ohms.

\$\endgroup\$
  • \$\begingroup\$ This looks like a beautifully simple yet effective concept! Unfortunately, I have forgot to add my requirement for PWM dimming in the description! I do not see how I could implement that in such a design, but you are more than welcome to prove me wrong! :) \$\endgroup\$ – iMrFelix May 28 '18 at 18:28
  • \$\begingroup\$ You can inject a dc current via a resistor into the FB node to fool the converter into lowering drive levels to the LED string. \$\endgroup\$ – Andy aka May 28 '18 at 20:47
  • \$\begingroup\$ @Andy, English language does not seem to have the term, but are you a fellow species? Are you anoyed by PWM flicker as much as I am? \$\endgroup\$ – winny May 29 '18 at 6:52
  • \$\begingroup\$ @winny I’m not sure what you are driving at? \$\endgroup\$ – Andy aka May 29 '18 at 7:25
  • 1
    \$\begingroup\$ Sooo, injecting current to fool the converter is analog dimming, right? Instead of using PWM, I could then use a digital potentiometer controlled by an ATTiny. Am I right in my assumption that due to the analog dimming, flickering won't be an issue? \$\endgroup\$ – iMrFelix May 29 '18 at 17:21
1
\$\begingroup\$

Very simple and inexpensive design.

Integrated Silicon Solution Inc IS31LT3948
BOOST LED DRIVER WITH EXTERNAL NMOS

Input voltage range from 5V to 100V
Analog Dimming
External PWM dimming
Constant current output up to 2A
SOP-8 package

enter image description here

RC filter PWM dimming
enter image description here

enter image description here

\$\endgroup\$
-2
\$\begingroup\$

Yes, approach 2 is viable. You can use microcontroller to drive both boost and buck converter.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.