• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 62 subscribers
  • Views 695 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.

LM3409 Current Fall Time much longer than datasheet

Chris_P
Prodigy 110 points
Other Parts Discussed in Thread: LM3409HV, LM3409

Hi.

I'm using the LM3409HV evaluation board straight out the box.

The graphs in the LM3409 data sheet (figure 17) and the AN-1953 LM3409HV Evaluation Board User's manual (Figure 6- and Figure 7) showing LED current vs EN pin waveforms show the LED current falls to 0 in about 2 or 3 us max after the EN pin goes low.

On my scope I'm seeing a first order decay which takes about 100us to reach zero! (Time constant to decay 66% about 20us maybe)

Why the huge difference?

I have all the standard component values on the board - I have not modified the eval board in any way.

J1 is not fitted at all so it's in EN pin PWM control which I'm drive with a clean 3v3 waveform from a sig gen reference to the eval boards gnd.

I have limited Iadj with a fixed resistor divider external to limit LED current which works fine - this is fixed throughout test and does not vary though I have tried values from 100 to 500mA but doesn't affect time constant of LED decay current .

I'm observing the decay on the LED +ve drive o/p voltage and also by putting  very small (0.1ohm) high side sense resistor in line + amplifier to measure the actual current waveform.

I have 12 LEDs with Vfwd about 3.3V at 350mA.

The input voltage to the LM3409 is 48V.

I was wondering if this could be the main inductor time constant but it's the one on the eval board that presumably that the waveforms were taken with.  Why do I see such a long decay compared to the graphs?

I need fast decay time for the application in question.  I also need very high efficiency so ideally don't want to use the shunt method.

Thanks,

Chris

  • 0
    TI__Mastermind 25500 points

    Hi Chris,

    Please use a current probe to measure the actual LED current fall time to be sure first. Using the output voltage will not show you the output current, and due to the non-linear dynamic resistance characteristics of LEDs using a resistance in series with them is also suspect. You could measure a current across it simply due to the resistance of the LEDs changing. I'm not saying you do not have an issue, I'm just saying you need to be careful how you measure LED current. The only really reliable way is to use a current probe in a dynamic (PWM dimming) situation.

    Also, make sure you are not accidentally applying the signal to the VADJ point as that could have the a similar effect as you describe.

  • 0 in reply to Clinton Jensen
    Prodigy 110 points
    Thanks
    I got hold of a current probe (brand new and expensive) and the decay tail I was seeing doesn't show up.
    Not really sure why a diff amp and 0.1 ohm sense R should be an issue as that's how the part itself does it's sensing - maybe it was something to do with my grounding setup? Anyway, decay time looks good.
    Rise time on the other hand is awful..
    First there's a delay where there's not much rise at all (PGATE duty cycle is low here) then the current rises to the peak over 400us!
    This is at 1KHz. If I increase the frequency it does change but obviously the rise time itself eventually becomes larger than the PWM period.
    Any thoughts?
    (maybe this should be a new thread)
    By the way, rise and fall time look great in parallel shunt mode - only a couple of us rise/fall but adds nearly $0.50 to the BOM which kills the project and means looking at other drivers.
    (Jumper and PWM signals set and fed correctly to EN/PWM2 depending what mode)
    Help greatly appreciated.
  • 0 in reply to Chris_P
    Prodigy 110 points

    Sorry, I meant to say if I increase the PWM frequency, it does NOT change the rise time.

  • 0 in reply to Chris_P
    TI__Mastermind 25500 points

    Usually if there is a delay like that it is because the part has gone into low power shutdown mode (the EN pin is low too long). The easiest way to avoid this is to use the UVLO pin instead. Connect a small schottky diode to the UVLO pin, anode to the pin, and apply a PWM signal at the cathode. This will function the same as the EN pin except it will not ever enter low power shutdown during PWM dimming.