• State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 61 subscribers
  • Views 732 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

lm3429 design questions

Ronald Jansen
Intellectual 520 points
Other Parts Discussed in Thread: LM3429

Hello All,


I am working on a buck-boost design with the LM3429, for which I have some questions.

The design will have specifications:

input voltage: 20-28V

Output voltage: 20V

Output current: 5A

The LEDs will be connected between Vin and Vout (so not ground referenced). That means that the boost output voltage can be up to 48V.

Question 1:

Is it feasible design the supply on IMS?  The metal substrate will only allow me to use one layer. That means that I cannot use a ground plane below the circuit or cross tracks by using other layers.

Now if I need to make a few crossings I can still use 0R resistors as bridges. But it cannot be too many. Another concern is EMC. I do have a large chunk of aluminimium below the circuit, but I can only bind this to a potential by grounding the whole housing. Also Aluminium probably does not conduct as well as copper. As the switching frequency is around 500Khz and the switch node voltage will have a delta of up to 48V I am not sure if emission can be within limits.

Question 2:

would it be wise to choose a 1206 package for RCS? The average power dissipation would be about 350mw, so the dissipation is not a problem, but the peak switch current will be 11A. This means that the current density will be rather large and the pad sizes are relatively small. On the other hand if I would use a 2512 package the loop may be larger. In any case it would use more area.

Thanks in advance for your help.

Ronald

  • 0
    TI__Expert 3495 points
    Ronald:
    1. Yes, it is possible to design this power supply on IMS. The layout is going to be more difficult since you can use one layer board. But it can be done. 500KHz switching frequency is a liitle bit too high. The power efficency is going to be low with fsw 500KHz. Suggest lower switching frequency to 300KHz.
    2. LM3429 uses high side current sense. The CS resistor is in series with the LED string. The current on the CS resistor is the same as the LED current. You can use 0.5W 1206 size resitor if the power dissipation is 350mW. The current limit resistor carries the switch current. You can connect the IS pin to the drain of the MOSFET and you can reduce the value of the current limit resistor because the RDSON of the MOSFET becomes part of the current limit resistor.
    David Zhang
  • 0 in reply to David Zhang
    Intellectual 520 points
    Hi David,

    Thanks for your help.
    Is the RDSOn of the FET reliable enough to use it as a current sense? I have seen it being done a couple of times, but I always thought that production tolerances and temperature differences would affect RDSOn a great deal.

    I also thought that the switching frequency is rather high.
    I used a Webench design as a starting point and I though that a number of things are at least not ideal. 550Khz is what Webench chose for the medium efficiency grade. It is presented as the best trade-off between size, costs and efficiency. I thought that there must be a better solution, so I set it one step closer to "highest efficiency". In that case Webench chooses 327KHz, but I found that the calculated efficiency does only increase about 0.5%. I am not sure if this reflects reality, but indeed I would prefer 327Khz. Only I will need more capacitance and a better (or larger) inductor).

    About the Cin. Webench suggests a 20milliohm electrolytic capacitor. I prefer ceramic, and the datasheet prefers ceramic as well. I an not sure why Webench chose elektrolytic. Of course it does have a higher capacity, but at the calculated ripple current the ESR alone will give a ripple 20-100mV.
    It seems that 100uF ceramic would give about the same result as 300uF electrolytic. Do you agree?
  • 0 in reply to Ronald Jansen
    TI__Expert 3495 points
    Ronald:
    The RDSON o fthe MOSFET varies with the temperature. You can not use it for current sense. But you can use it as part of the current limit resistor to limit the maximum siwtching current. It is a protection function.
    I agree with you. Ceramic caps work better for ripple current filtering than electrolytic caps. Some people like to use electrolytic caps for lower cost.
    David Zhang
  • 0 in reply to David Zhang
    Intellectual 520 points

    So how do you set the maximum peak mosfet current? Usually I would set the the limit to not exceed the indutor saturation current, but rather a current that will match the maximum power demand. If I use the RDSon instead of external resistor that would mean that I would need to choose a mosfet with a higher RDSon? This could be interesting. I now chose the best trade-of between RDSon and gate charge. Its RDSon is around 4 mOhm. If I would choose a mOhm that might give a lower gate charge and therefore lower swithing losses.

    What about this IS current sense pin? The datasheet says it is robust enough to handle the drain voltage including a certain amount of ringing. Is it really safe to assume that the ringing will remain within the pins input range, or should I rather add a clamping diode?

    About the input buffer. As I will be using ceramic capacitors I want the buffer to be as small as possible while still being effective. What I still do not understand. The datasheet tells how to calculate the input voltage ripple, but I though that we would want to reduce the input current ripple. Using the the voltage ripple instead assumes that the impedance of the power supply is much higher than the buffers impedance. This kind of depends on the power supply, which is generally supplied by customers. Is there a rule of thumb for a realistic voltage ripple and reflected input current ripple? I would usually include an extra inductor before Cin, but an inductor rated for 5A will be rather large. So it would be good if I can safely omit it.

  • 0 in reply to Ronald Jansen
    TI__Expert 3495 points
    Ronald:
    I said using the RDSON of the MOSFET as part of the current limit resistor. For example, if you designed with 14mOhm current limit resistor, you can use 10 mOhm current limit resistor plus the 4mOhm of RDSON of the MOSFET. The power loss on the current limit resistor will be reduced by 30%.
    The maximum voltage rating of the IS pin is 76V. If the peak of the ring voltage is under 76V, it is safe.
    For the buck boost converter, the input current ripple is the same as the inductor current ripple. It has nothing to do with the AC/DC power supply. The data sheet shows how to calculate inductor current ripple. The input voltage ripple equals input current ripple times the ESR of the ceramic capacitor. The ceramic capacitors, there are ripple current ratings. Please make sure the ceramic caps are rated to carry the ripple current.
    David Zhang
  • 0 in reply to David Zhang
    Intellectual 520 points

    Aha, I seem to have missed that. That explains a lot. Still I think it might make sense to choose a MOSFET with higher RSDon and trade off some of the switching loss to RDSon loss. There has to be some loss in a sense mechanism anyway. Also Generally a MOSFET has better heatsinking capabilities than resistors. The only thing is that steeper switching slopes (necessary to minimize switching loss) in turn might increase the EMI. In an case I can add a source resistor to get the required sense voltage. That is a good point.

    I am not sure I agree with you that the AC/DC power supply is not a factor. When handling the ripple current the capacitors are charging/discharging. This in turn will generate a ripple voltage. The datasheet explicitly mentions the ripple voltage as a one of the inputs for calculating Cin.  If due to the ripple voltage the supply voltage would be higher than the Cin voltage the ripple current will not be supplied by the capacitors, but by the power supply instead. This is probably what you do not want. I guess that in practical situations the inductance of the cables will provide for enough impedance to filter the ripple current. But I am not sure if I can always count on that.

    For the ceramic capacitor the ripple current rating is no problem. Their ESR is around 5mr and I will need to put quite a few in parallel to get to the required capacity. The dissipation will be neglible.