300w dimmable PFC LED driver (buck/boost)

Discussion in 'The Projects Forum' started by electricnut, Feb 11, 2016.

  1. electricnut

    Thread Starter New Member

    Feb 9, 2016
    5
    0
    Hi folks,

    I am building a dimmable 300w PFC LED driver using TI's TPS92075 and their design spreadsheet with the buck/boost topology.
    It takes 120v, dimmed with a regular dimmer, and is supposed output a calibrated max of 2.4A @ ~95vdc for a string of 3 100w, 30-32v LED chips. (=<240w; these are Chinese LEDs, so it is recommended not to run them at full power)
    I sized all my components for at least 300w, hoping to reduce heat and consequently the need for fan cooling.

    However, when powering it up, outputting ~80v, 2a, my efficiency is roughly the same as it would be if I used a linear regulator. Consequently the MOSFET and the inductor get very hot after only a short time. Also, to get 2a, I have to reduce the resistance of the current sense resistor, presumably because the energy needed to get the full 2.4a is getting wasted in inefficiencies. Also, if I tweak the current up much higher than 2a, my MOSFET short-circuits. I have no clue why!
    Everything else functions great, the TPS92075 is truly an amazing chip!

    Drive MOSFET: IPA60R460CE
    Inductor: PM2120-221K-RC (MOUSER link)
    Diode: BYV29FX-600
    Schematic: attached

    I currently have the TPS92075 in a breadboard, so I can fine tune the angle sense, as well as add overheat/overvoltage protection. The switching circuit, as well as the AC filter capacitors I have soldered, so my breadboard is not causing added resistance problems there. The TPS92075 is getting powered by an external isolated bench supply at the moment. There is a solid ground connection between the two, and noise is not enough to cause problems. I have monitored the MOSFET's gate to make sure it is getting a clean signal with enough voltage, and everything looks very good through the oscilloscope; the MOSFET is getting at least 10v and a nearly perfect cut-on to the gate.

    I'm thinking that maybe I somehow sized the inductor wrong. Though that doesn't explain why the MOSFET blows when I go much over 2a. Unfortunately, I don't have a /10 probe for my oscilloscope, so the highest voltage I can measure with it is at best 80v. Thusly, I can't measure to see if the drain voltage is going too high or too low. Though I highly doubt the MOSFET's 600v top limit is getting exceeded.

    Thanks guys, any help/ideas will be greatly appreciated.
     
  2. RichardO

    Well-Known Member

    May 4, 2013
    1,235
    384
    Looking at the inductor specification my guess is that the 5 amp maximum is not enough. When stepping the voltage up, the inductor current must be more than the output current. Often, the inductor current is 4 times the output current.

    The inefficiency and MOS-FET failure is likely caused by the inductor saturating. Saturation causes the inductance to go down which increases current.
     
  3. electricnut

    Thread Starter New Member

    Feb 9, 2016
    5
    0
    The inductor I got is rated at roughly twice the output of my project. But I wonder if maybe the 7 amp peak current isn't overloading the inductor? I couldn't find anything about the peak current in the inductor's data sheet, however, the TPS95075's design spreadsheet suggested an inductor with an average current of 2.4A, so I doubled that hoping to reduce heat. However, at your suggestion, I will try something in the ~8-10A range. My only question concerning that is: what happens to the efficiency if the inductor is over rated?

    The TPS95075 doesn't use a fixed R/C block with PWM as the power adjustment. Rather, it uses the current resistor feedback to tell when the inductor is charged, than immediately switches off the MOS-FET. The output power is determined by an R/C block that adjusts only the off time; the inductor is charged the same amount of time regardless of whether the power is at 10% or 100%, the main thing that changes is the delay between charging cycles. So then, the inductor would have to be saturating at a low power just as much as at 100%, if I'm not mistaken? That's why I don't understand how the MOS-FET could blow at 90%, and not 20%.
    Regarding the saturating issue, does this mean the TPS95075's design spreadsheet is suggesting the wrong uH value and I really need a larger inductor, or is this also related to the inductor being underrated for the current draw?

    I'm sorry for the dumb questions; I have no education on this subject, so I'm extra thankful for people like you who are willing to help. Thanks!
     
  4. RichardO

    Well-Known Member

    May 4, 2013
    1,235
    384
    The inductor get bigger, heavier and more expensive but the performance improves. A new thought: double check that your inductor is rated for the operating frequency fo the converter. The inductor core material determines this.

    I agree. That is a stumper.

    I am still guessing that the inductor is saturating. Too large an inductance will reduce full load current and too small an inductance will tend to reduce efficiency.

    Hey, I have let my share of smoke out of circuits. That's how you learn. I just hope I can reduce the smoke you make and speed up your learning at the same time. :D
     
  5. slackguy

    Member

    Feb 11, 2016
    29
    1
    i guess check the current draw on te LED

    for one they last decades if driven lower brightness (incandescent - a century or more is possible), but weeks or years if driven near limits

    another is forrest mims says LED have huge current inrush and low voltage at start-up, and then pick up voltage and drop the current to normal levels

    and your LC frequency? that changes things as well (rms volts and current and also current direction if not ac blocked). that's a b i can't look at diag right now.

    end of story i would not use RLC for small projects because alternatives are far easier to plan and troubleshoot
    , if nothing more, education on alternatives can't hurt

    you could use spike limiters, v regulators, diodes, etc to protect any valued parts. slap in resistors to test values running at a much lower level.
     
  6. slackguy

    Member

    Feb 11, 2016
    29
    1
    a regular dimmer? you mean household dimmer? those are like resistors and run very hot when dimming an incandescent.

    i could be wrong but you should be operating at "typical" not maximum ratings. 2.4A sounds like allot for a LED driving chip - like it's max. you can't run something at it's max unless it's cooled and it's unstable (ie, circuit fluctuations uncontrolled will cause issues)

    then there is the LED ratings: you might test them to see they are doing what the box says.

    maybe the chip is not meant to drive that much

    as far as efficiency i cant look at it close now. my guess is the unit looses efficiency when operated closer to it's limits due to heating and other things - but not familiar with it to say so.
     
  7. electricnut

    Thread Starter New Member

    Feb 9, 2016
    5
    0
    I did some research on that topic, and couldn't find any concrete answers. Most documents refer to a SMPS transformer rather than a buck/boost power supply. Most favor the torroidal inductors (which is what I am currently using) due to the slightly higher inefficiencies of ferrite inductors or others. As far as core material goes, I couldn't get far—neither the inductor I bought, nor the inductors I am now looking at getting mention anything about what the core is made of in their datasheets. :(

    Thanks for the tip. I am now looking at these inductors:
    mouser.com/ds/2/3/AIRD03-38423.pdf (220uH)
    mouser.com/ds/2/54/140_series-776825.pdf
    (220uH, not sure about this one; the datasheet calls it a choke.)
    Both are ferrite inductors, which were slightly less efficient than torroidal inductors in my research... but at this point I'm willing to try anything.
    Any thoughts?

    You've been a great help, thanks!
     
  8. electricnut

    Thread Starter New Member

    Feb 9, 2016
    5
    0
    For the present difficulty I'm having, the load (LEDs in this case) are not the source of problems. For the sake of simplicity, let's just pretend it is an indestructible resistive load that consumes 3A @ 96V.

    This is a constant current power supply, so that is not an issue.

    I will look in to this; though with the way the TPS95075 charges the inductor and adjusts output power, I can't tune it to a specific frequency.

    Due to the nature of the PFC power supply, it would wreak havoc if the R/C part of the circuit were tuned to resonate at any particular frequency; since it's frequency varies depending on the AC wave and dimmer position.

    I am aware of this. However, I don't know what is causing my MOS-FET to blow; be it over voltage, reverse voltage, spike current, gate over voltage/current, etc... it could be lots of things, but I haven't been able to pin-point it, and I hate blowing up my MOS-FETs trying to find out!

    Yes, a regular (modern) household dimmer, though ultimately this project will be used with a "professional" dimmer/lighting controller.
    Regular household dimmers do PWM on each AC wave; basically cutting out a section (starting from the top) of the AC wave based on the position of the dimmer knob.
    Back in the day, resistive light dimmers were used, but they required massive cooling making them unpopular. They were replaced by the modern light dimmer most of us see nowadays, which are way more efficient, using the method mentioned above. Most of the heat generated by them is caused by the voltage/inefficiency of the TRIAC.

    This is somewhat irrelevant to the problem I'm having, but the LED moduals (COB) are rated 32-34V (typ.) @ 3000mA.
    I intend to run three of them in series, at a maximum of 2400mA to reduce heat and increase life. Especially considering that these are genuine Chinese LEDs.;)

    I have tested them at their rated current for long periods of time, and they perform as expected.

    The TPS95075 requires an external driver; the chip was designed only to drive the gate of a MOS-FET or related load. Thus, unless I'm severely mistaken, it should be able to control a power supply of any size.

    Yes, as RichardO has suggested, this might be my problem with the inductor. In any event, I'm running something over the limit to cause the problems I'm having!
     
Loading...