This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3433/3434 is not working as expected

Other Parts Discussed in Thread: LM5002, TPS92640

I have implemented circuit of AN-1937 LM3433 10A to 40A LED Driver Evaluation Board with PIC24 in 3.3 supply mode but there is some audible noise from MOSFET (might be some switching noise) & MOSFET's are getting heated. Even i am not getting expected light output on CBT 140. When i have disconnected load the LM3433 gets burned. Please provide me any solution.

  • Are you using the actual EVM, or is it a different PCB/layout with the EVM circuit? Is the audible noise during PWM dimming? The EVM was designed for basically room temp evaluation to save cost, in real operation the FETs will likely need heatsinks at those current levels. In either case yes you will likely damage the circuit if the output is opened during operation with those kinds of currents. D3 was added to offer some level of protection, but we have found it is not really enough at those current levels.
  • Thank you for replying, I have copied schematic of Demo board (Except power supply, Snubber ckt & other omitted components such as 0 ohm resistors & DNS components). I took necessary steps to make bigger planes with large no of vias to dissipate heat.

    I have not connected Q5 Q6 (PWM mosfet) but still getting noise. 

    After my initial trial i had tried again & again & found that Gate driver is getting heated, according to me the real problem is switching frequency, on Low side i am getting pulses at 1.5 MHz & 70 % Duty cycle. So i have decreased frequency to 300 KHz & doing trials on it. 

    Please provide your valuable suggestion regarding above modification.

  • I see. Yes, 1.5MHz is way too fast for these current levels. The AC losses will make everything way too hot very fast, especially the FETs and gate drivers. 300kHz is much more reasonable. In fact the lower you go the cooler things will be.

  • Thank you for your reply.

    I have soldered new LM3433, TC4424 (U3 & U4), SiR798dp (Q1 to Q4) on my board for 300KHz switching frequency & 3.3 Volt for digital section & -12 volt for Analog section. After power on i was getting Light output in pulses of some milliseconds for every 5-6 Seconds & LM3433 was getting heated (consuming 800 mA from 3.3 supply).  3433 does not get heated when supplying only -12 volt supply.

    Please check for Driver Mosfet (TC4424 - SiR798dp) combination & provide your valuable feedback regarding above problem.

    Is there any other reliable alternative for this application from TI or any other manufacturer ?

  • It sounds like something is either damaged and/or the part is not soldered down correctly. I only mention the second one because it took me some time to get decent at replacing the LLP package. I always check for shorts between pins and do a substrate diode check on each pin to ensure it is soldered down. If it isn't soldered correctly the end result is still sometimes something gets damaged. I would also make sure you power sequence if you are not using the LM5002 circuit (it has a UVLO set so that it will not provide power to VIN until VEE has at least -9V across it. Alternatively the circuit consisting of Q10 and Q11 on the EVM provides a UVLO function. Basically if you already have VIN powered up there is no UVLO on VEE so it will try to switch with much to low of an input voltage which can cause damage as it will be out of regulation.

    The TPS92640 is another alternative that can be used, but you may need to be careful with FET selection as it has lower gate drive capability.

  • After soldering i had checked for shortings but everything was good. I have Q10, Q11 circuit on my PCB but i didn't used it as i was not aware about required power sequence. So first i have to turn on -12 volt supply & after 1-2 secs i have to turn on digital supply ?

    Any other precautions to be taken?

    Is it a reliable IC for medical devices ?

  • To avoid operation out of regulation you just need to make sure the |VEE| voltage is the LED forward voltage plus 3V headroom before switching begins. The 3.3V rail can be powered before the -12V rail if the EN is held low until the -12V rail is powered up, otherwise yes, you need to power up the -12V rail first. This is what that circuit does for you, but you can accomplish the same thing in multiple ways. There doesn't need to be 1 to 2 seconds unless the -12V comes up very slow. Yes it is reliable and is used in many different applications due to its common anode ground capability. The only other precautions I would take is to make sure you use the current limiting resistors on EN, DIM, and ADJ if there is any chance of spikes or signals on these that may exceed VIN. You should also always use the 100ohm resistor before C2 (on the EVM) going to VIN regardless of the source for input filtering.