1
\$\begingroup\$

I'm trying to make a PWM'd floodlight with 6 Cree CXA 2540Ns. My setup is a simple 555 timer set up for PWM, driving a 3904, which in turn drives an FDA59N30 N-channel MOSFET, properly cooled to handle roughly 500 watts of power safely. This is going into a vehicle, so I needed to boost the vehicle's 12 to 36 V as required by the 2540Ns. I did this with a cheap boost converter purchased at Banggood. Here is the link to that particular module:

https://www.banggood.com/1500W-30A-DC-DC-Boost-Converter-Step-Up-Power-Supply-Module-Constant-Current-p-1087084.html?gmcCountry=CA&currency=CAD&createTmp=1&utm_source=googleshopping&utm_medium=cpc_elc&utm_content=frank&utm_campaign=pla-mix-ca-pc-0630&gclid=EAIaIQobChMItJexw7S53gIVz2B-Ch3RBQ-6EAkYBiABEgL6FfD_BwE&cur_warehouse=CN\

After hooking everything up as per my schematic, I get no illumination no matter the position of the pot controlling the PWM! The terminals at the boost converter's output measures 36 V DC, and the Vds on the FET measures 16 Vpk-pk, with 0 V base. The Vgs measures 12 V pk-pk with 0 V base, as expected.

U1 simply represents my PWM 555 circuit. It works as expected when driving small logic-level LED's and tranistors. The transistor is switching the vehicle's battery voltage to drive the FET, which is switching the LEDs at the boosted 36 V. The reason for the two different ground symbols is merely to highlight the fact that the LEDs do not return to the vehicle's ground directly, but rather back to the boost converter's negative terminal.

schematic

simulate this circuit – Schematic created using CircuitLab

Thank you for any illumination you might be able to provide!

To jonk: 4.5 kHz, 77-100 nC, 3590-4670 pF. All values copied directly from the datasheet, except the PWM frequency. I kept that a few orders of magnitude lower than the 150 kHz frequency of the boost converter, to help prevent issues with it's regulation.

To Felthry: I have run these LEDs directly from both the boost converter and my benchtop power supplies at 36 V. My DMM and benchtop power supplies all concurred that the current was 1.1-1.2 A at that voltage. This is in agreement with the datasheet. I also have no desire to continually run these fairly expensive arrays at their maximum forward voltages! And yes, I know that they're grossly over-rated. But I had them, and a I'm a sucker for safety factor anyway, heh.

To pipe: I have not! There's no "ground" labelled as such, just a negative terminal at the output. But I will try this!

To Janka: I have PWM'd this before, with another module. It delivered flickery results as you swept up and down the duty cycle. I assumed that this was because it was designed for motor control, and any flickering in the output would be smoothed out by the motor's inertia. However, with LEDs, it was very apparent. So I know I can PWM this boost converter, but I made an assumption that a cheap H-bridge would not be balanced well and some duty cycles would produce flickers. Thus my move to a single FET.

\$\endgroup\$
  • \$\begingroup\$ Have you connected the DC-DC boost output ground to the PWM controller ground? \$\endgroup\$ – pipe Nov 4 '18 at 0:06
  • \$\begingroup\$ One thing that comes to mind: your FET doesn't need to be able to handle 500W of power; it's not a problem but you're using a higher-power device than you need to. Also, 36V isn't going to be enough to drive these LEDs at their full current; the rated maximum forward voltage is 42V at 1.1A, which is only about half the rated forward current. \$\endgroup\$ – Hearth Nov 4 '18 at 0:10
  • 1
    \$\begingroup\$ If you switched the output of the converter instead, it's not going to work because neither the + nor the - output of the converter are GND. It's current-controlled. Your PWM circuit does not share a closed circuit with the output. You could use an optocouple and some more mumbo jumbo but buying the right boost converter is the way to go. \$\endgroup\$ – Janka Nov 4 '18 at 1:31
  • 1
    \$\begingroup\$ You show two different ground symbols and neither of the two power supplies has its negative connection shown. If both of those ground nodes and the negatives of both power supplies and the ground node of the 555 circuit are mutually connected, your circuit should work. Is this in fact how you have everything connected? \$\endgroup\$ – ajb Nov 4 '18 at 3:40
  • 1
    \$\begingroup\$ You really need to implement some form of current control (even a simple series resistor for each LED. The way you have it the current will increase as the LED modules heat up and give you quite large variations in brightness for a fixed input voltage. Setting a current limit on your boost converter is not adequate since in this case you would have to set it to 12A ….loss of a module or a mistake in wiring could then critically overload the other LED modules. \$\endgroup\$ – Jack Creasey Nov 4 '18 at 15:34
1
\$\begingroup\$

There are multiple problems with the schematic you show.

  1. If you connect multiple LED modules in parallel you will have no idea of how much current flows in each module. The Vf for the modules will vary and more variation will occur due to thermal effects. You should drive each LED module with its own FET, and you should implement some form of current control for each LED module.
  2. You have selected a FET with a very large gate capacitance, and this form a low pass filter with your R1. This is the reason you are getting almost no drive with your current schema.
  3. You need to run the Boost converter at an output voltage higher than the maximum expected by the LED module and drop that extra voltage using the FET. This will allow current control for each module.

I'd suggest that something along these lines may be viable to drive each module:

schematic

simulate this circuit – Schematic created using CircuitLab

Notice here that D2 is NOT a Zener, but a TL431C shunt regulator. This limits the maximum current through the LED module to 1.66A (I think anything above this might be very ambitious) using M1 as the series pass element.

M2 provides a simple interface back to your PWM generator.

Note: there is no need for the PWM frame rate to be 3.5kHz, it would be better set at something around 1kHz.

The Boost DC-DC can be set to any voltage above the maximum that the Cree module might need, but will result in more heat being dissipated in the M1.

\$\endgroup\$
  • \$\begingroup\$ You have to be careful, as the supply he choose is already constant-current, apparently. Not sure what will happen if the supply current setting is set higher than the current limiter of your circuit. Probably the boost supply output voltage will raise to its max (90V specified), and the FET will have to handle a huge voltage difference and will burn. But the supply specifications are so sparse and inconsistent that it's hard to tell. \$\endgroup\$ – dim Nov 6 '18 at 9:05
  • \$\begingroup\$ @dim He would set the Boost supply current limit to 12A or above for the LED modules being used. The output voltage pf the DC-DC would NEVER go beyond the voltage set ...in this case I said 40V so your scenario is completely false. \$\endgroup\$ – Jack Creasey Nov 6 '18 at 16:50
  • \$\begingroup\$ Right, I didn't realize there was both a voltage and a current setting on this supply. But in this case, the additional current limiter is redundant. \$\endgroup\$ – dim Nov 6 '18 at 17:24
0
\$\begingroup\$

If all LEDs will get the same current, use two 4.5 Amp constant current boost drivers with PWM.

Or boost the 12V up to 36V and use an (or two) LM3409 buck CC driver.

Then use an LM3466 with each LED to balance the current.


BTW, you are using an old CoB. The CXA have been replaced with the CXB one year ago.

There are better (higher efficacy) CoBs than the CXB from Bridgelux and Citizen. I always sort LEDs by lm/W when selecting. Higher efficacy equates to more light, less heat.

If each CoB needs to be controlled individually then use a simple CC boost LED driver with integrated FET and PWM for each CoB.

\$\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.