1
\$\begingroup\$

I'm trying to build a 1A LED buck basic buck led driver

It works well, current in leds depends of pwm duty until 1A and more, but when I watch signals with an oscilloscope, there are noise or oscillations on Rsense. See a screen shot of signals here: signals scope In yellow, the pwm signal and green, Vense.

Why so much oscillations around Rsense? How can I clean up Vsense in order to use it for the Ileds current regulation?

\$\endgroup\$
0
\$\begingroup\$

This is why I prefer to use a LED driver chip with on board MOSFET.

I am currently doing a 1A driver for 16 white 2.85V LEDs and a 48V Vin.
Using a TI LM3414HV. I can set the switching frequency between 250 Khz and 1 Mhz.

This circuit is almost 98% efficient using a 250 Khz switching frequency and a 1.5 mOhm 33 µH inductor which keeps it in continuous conduction mode. I could use a less expensive 110 mOhm inductor and lose 2% efficiency using a switching frequency of 400 Khz.

Very simple, low parts count, and minimal real estate. And I do not have to manage the PWM.

enter image description here
Source: TI WebBench



UPDATE

Could you send a picture of the inductor L1

I have not finalized the BOM. I have a footprint on the PCB that will accommodate over 50 BOMs. I have designs for 500mA, 750mA, and 1A with efficiencies ranging from 94% to 98% all using one PCB layout.

The easiest way to make component selection is to use TI's WebBench app. The you go to the BOM and pop up a list of alternate parts for L1. You can easily see many trade offs between cost, footprint size, height, and DCR. After selecting a new inductor it will recalculate the optimum switching frequency, efficiency and thermal characteristics. It is a great learning tool. When you change a part value or target efficiency you can find exactly what parameters are affected.

The LM3414HV datasheet has a good 1A design and an eval board for it.

The one inductor for the above circuit was the $2 (qty 100) Bourns PQ2614BLA-330K
But with the low DCR the winding is very heavy and the inductor is too big at 1.098" L x 0.780" W (27.90mm x 19.80mm)
enter image description here

Looking at my latest BOM this 68µH, 140 mOhm, Panasonic ELL-8TP680MB 0.315" L x 0.315" W (8.00mm x 8.00mm) Height: 0.197" (5.00mm) is a contender.

enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ Yes, this is a good solution too. Could you send a picture of the inductor L1 or the reference model used? \$\endgroup\$ – Teddol May 7 '18 at 2:46
  • \$\begingroup\$ see my update.. \$\endgroup\$ – Misunderstood May 7 '18 at 4:06
  • \$\begingroup\$ Good information! I will see Ti website and tools. Thanks! \$\endgroup\$ – Teddol May 7 '18 at 10:08
2
\$\begingroup\$

There will always be some noise at switching transitions, particularly those on which inductors are switched off. Proper attention to grounding, where the loop currents cross the ground net, etc, can minimize the transients.

However, the real problem is that the current sense is in the wrong place. At best it will only tell you the current during the time the switch is on. If you are trying to regulate the LED brightness, then you need to know the average current, not the peak current at one point in the waveform.

This is a case where controlled open-loop can be a good scheme. From knowing the input voltage, switch on time, inductance, and LED forward voltage, you can compute the average current as a function of duty cycle. You can have the micro that produces the PWM pulses perform that calculation, then update the PWM duty cycle accordingly. Since your load is fairly fixed and won't change quickly, these calculations don't need to be performed every pulse.

Depending on the processor and what exactly can vary in your application, you may be able to use a lookup table to get the PWM duty cycle for a particular desired brightness, given a measured input voltage.

I have done exactly that in a high volume battery operated LED driving product on a small PIC 16. The average LED current stayed within 10% of the desired value over the whole range of possible battery voltages. There was no current feedback, only occasional measurement of the battery voltage. Note that 10% brightness is very difficult for humans to perceive, even in a side by side comparison.

\$\endgroup\$
  • \$\begingroup\$ Thanks Olin for these explainations. The pwm is performed by an Arduino board so i can setup up a math function as current = x*Duty or other. I will try and will see linearity of leds current. I would like to know why my self comes wery hot. It's an 470µH 2A ? If I replace it by a 470µH 10A (thicker wire), does it reduce the heat ? \$\endgroup\$ – Teddol Apr 15 '18 at 14:11
0
\$\begingroup\$

The spikes are so high because you don't switch the FET fast enough, it can be seen in the pictures. I don't think Arduino has otputs capable to drive a FET load which is highly capacitive. This is also the source of heating. Search for a proper FET driver or use many parallel outputs together to increase the current capability. Still, you must expect some spikes, use an RC filter to cut them before the Arduino input.

You can know the otput current by reading only the curent sense, it's double than the average current through the current sense resistor

\$\endgroup\$
  • 1
    \$\begingroup\$ I will try to use a NE555 to drive the FET with a 12V pwm output. \$\endgroup\$ – Teddol Apr 17 '18 at 13:49

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.