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.

TPS922052: Strange delay when moving from off to low duty cycle

Part Number: TPS922052

Tool/software:

Configuration is as follows:

- Bridgelux 18v/500mA COB
- PWM @ 1kHz, 9-bit resolution
- Analog @ 20kHz, 8-bit resolution

Using PWM mode, I can get the LED to dim down to duty cycle 2/512. If I move from duty cycle 0 to 2, the LED turns on instantly.

Using analog mode, I can get the LED to dim down to duty cycle 1/256. If I move from duty cycle 0 to 1, the LED turns on instantly.

Using hybrid mode, I can get the LED to dim down to duty cycle 2/512. If I move from duty cycle 0 to 2, the LED turns on instantly.

Of course, I'm selfish and I want the LED to dim even more, so I want to use dynamic mode. I've tried a bunch of different methods, and in this example, I can get the LED to come on when analog duty is 20/256 and PWM is 2/512. Here's where the strange part comes in: the LED takes a few hundred ms to actually come on and then it fades in quickly, it's a very strange (and 100% reproducible delay).

This is the case all the way up to analog duty 20/256 and PWM duty 9/512, where the LED comes on instantly.

I've tried changing PWM frequency, altering the ratio of PWM/analog, and nothing seems to change the behavior. I've tried different initialization sequences for the chip, but even leaving that out completely and just sending in PWM behaves the same, which makes sense given the startup sequence is in the microseconds and what I'm observing is in the hundreds of milliseconds.

So basically–I can get the dimming ratio I'm looking for, but there's a weird delay when going from dark to dim.

Any ideas would be much appreciated!

  • Our expert will give you reply next week.

  • Dear Sir,

    Thanks for reaching out!

    For dynamic mode, do you refer to flexible dimming or hybrid dimming? And could you pls kindly confirm the test condition when you see strange delay? From my understanding, when the issue happen, the initial state is that the duty of ADIM is 20/256 (20kHz) and the duty of PWM is 2/512 (1kHz)? And then you will increase the duty cycle of PWM signal to 9/512? And what is the sequence between VIN power on, ADIM signal active and PWM active? It will be really appreciated if you would share me a figure to better show this scenario. I will first try to reproduce it on EVM board.

    Of course, I'm selfish and I want the LED to dim even more, so I want to use dynamic mode. I've tried a bunch of different methods, and in this example, I can get the LED to come on when analog duty is 20/256 and PWM is 2/512. Here's where the strange part comes in: the LED takes a few hundred ms to actually come on and then it fades in quickly, it's a very strange (and 100% reproducible delay).

    Thank you!

    BRs

    Lucia

  • Hi, thanks for your reply!

    I apologize, I want to use Flexible Dimming, that's what I meant by "dynamic mode".

    VIN power is on for a long time before the test starts, so it's not a startup issue with VIN.

    1. VIN is on, ADIM and PWM are both low

    2 Send 20kHz 20/256 PWM into ADIM, and 1kHz 2/512 PWM into PWM, and the LED should turn on immediately, but it takes 200-300 ms for it to come on.

    If I use higher duty cycles, the LED comes on instantly (ADIM 20/256 and PWM 9/512).

    Please note my output capacitor is a bit higher than eval board.

  • Dear Elan,

    I will check your question and get back to you before end of this week. Thank you!

    BRs

    Lucia

  • Dear Elan,

    Thanks for your feedback. I would like to confirm with you another thing about the PWM signal. For example, you said that "when you send 20kHz 20/256 PWM into ADIM, and 1kHz 2/512 PWM into PWM, the LED should turn on immediately, but it takes 200-300 ms for it to come on." I am not sure after you set 2/512 PWM, the PWM signal will always stay as 2/512 and you need to wait 200-300ms for the LED to turn on? Or the PWM signal will increase slowly so that when the LED turn on, the duty cycle of PWM signal at that time is increased and not be 2/512?

    The reason I am asking is because the initial enable PWM pulse lasting more than 5 μs is required at EN/PWM input pin to enable the device. However, with 1kHz PWM signal and when the duty cycle is pretty low, it might not meet 5us requirement.

    Thank you!

    BRs

    Lucia

  • I am not sure after you set 2/512 PWM, the PWM signal will always stay as 2/512 and you need to wait 200-300ms for the LED to turn on?

    Yes, exactly.

    If I calculate the pulse time, that is ~3.9us, so you're right, that might be the issue. However, I have tried a few initialization sequences and I cannot get it to make a difference. Please let me know if you see an issue with this sequence:

    - Vin stable for a long time (seconds)

    - Set PWM frequency to 10kHz for ADIM.

    - ADIM low, PWM high for 1ms

    - PWM duty 0/512, ADIM 4/256 for 2 ms (1.56% duty, much longer than 4 periods)

    - Set desired combination, e.g. as above 20kHz 20/256 into ADIM, and 1kHz 2/512 PWM 

    Does that sound correct?

  • Hi Elan,

    Our expert will give you feedback later.

  • Hi Elan,

    Thanks for your feedback. Just want to double confirm, so this issue only happen with flexible dimming, am I correct? If you use only PWM dimming (always pull ADIM high) and set EN/PWM as 1kHz 2/512 after the dimming mode selected and dimming mode start, will you also see the several hundred ms delay issue?  

    Thank you!

    BRs

    Lucia