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.

TPS92661-Q1: Circuit Design Considerations

Part Number: TPS92661-Q1

Hi Folks and Authors,

I have gotten a project recently that consists of a night vision system for some traffic applications. My client wants to use a camera to get violation data at night and for this purpose, we have used high power IR LEDs. We used twelve LEDs(SFH4715AS) that are best for my client's application. We have made some experiments and the camera gets sufficient images from violations. Until now we have used DC power supply to get images and now we want to use a proper LED driver. My client wants that all other LEDs work normally if any problem with one LED other. In that case, I have a couple of research about the LED drivers with 12 channels at that high current(~1.5A), and then I find out that your chip(TPS92661-Q1). Here my questions,

1.) Below I share my planned system, is it OK for our application?

2.) Is this chip proper for my client request?
3.) If it is not proper, what are you suggest for us?

4.) Do we want to use also "SST-10-IRD-B50-T850" LED? Is there any change?

  • Mustafa,

    There are some assumptions here that are not correct.  First off, LED drivers are current sources.  They don't regulate voltage like a DC power supply.  You need a 1.5A constant current source (LED Driver) to regulate current through the string of 12 LEDs (that roughly have a 40V forward voltage at that drive current according to your specs). 

    I can't comment on your LED choice, but I can comment on your converter topology.  A pure step-up (boost) converter will not work well with a matrix device like the TPS92661.  The matrix device has switches in parallel with each LED.  When the switch is off, the LED conducts the current.  When the switch is on, the switch conducts the current and LED is not illuminated.  This is done at a frequency above human eye detection (~200Hz or greater...programmable) and the duty cycle controls the average brightness.  When the switch turns on the voltage quickly changes across the output capacitor.  This dv/dt will cause an overshoot current that flows through any of the LEDs in the string that are not currently bypassed (i.e. switch open).  If there is a relatively large output capacitor like necessary on a boost converter that has discontinuous output current, this will damage the LEDs.  

    So, the solution is to create a two stage power supply with a boost voltage regulator that steps up the battery voltage to a voltage greater than the LED string forward voltage and then use a high bandwidth buck current regulator to regulate current with minimal output capacitance.  

    One other note, 1.5A on 12 LEDs will be far too much power for these switches when they are all turned on.  One single switch can tolerate 1.5A (metallization, bond wires, etc), but you must take into account the thermal rise for 12 switches.  Even 1.0A through 12 switches is extremely difficult as the maximum total resistance at high temp is 4 ohms.  Therefore that's 4W of power on that device when all switches are on continuously.  Needless to say, that requires metal clad PCB and very careful thermal management.

    In your case, I would make the change to the 2 stage power supply (look at ti.com for this architecture in the system block diagrams) and I would use 2 TPS92661 devices and split the load (6 on each) with the other 6 switches externally shorted (not used).  Then you may have some success with these difficult specs.

    -JP

  • Hi JP,

    Thanks for your quick reply. I appreciate your suggestions and I will take into account your suggestions for my design. I have some questions, too.

    1.) There is no answer about my client want that is "My client wants that all other LEDs work normally if any problem with one LED other", is this chip okay for this request?

    2.) What are your suggestions for a two-stage power supply? What are your suggestions? I will research if you share some chip links or forum threads?

    3.) Is there a need for use of metal clad PCB, if we make all of your suggestions as a two-stage power supply and two TPS92661-Q1?

  • Mustafa,

    Answers:

    1. Yes there is fault protection and if an open occurs, the switch will detect the overvoltage and close to protect itself, thereby re-establishing the current flow through the rest of the string.  This re-regulation occurs immediately since the converter is a current source (not a voltage source).

    2. This device was designed for automotive headlights, which have similar requirements.  Please look at this page on ti.com and you will find many resources on this topic. https://www.ti.com/solution/automotive-headlight

    3. I can't do your thermal design for you.  However, even with the configuration I gave you...the worst case power dissipation on chip is 1.5A^2*2ohm = 4.5W!  and the typical power dissipation at room temperature would be roughly 2.25W.  So, absolutely you would need incredibly good thermal handling which could only be obtained through a metal clad type design.

    Honestly, this is a very difficult design and you will need good power electronics knowledge and excellent thermal management to achieve it even using my suggestions.  If you want to make the thermals a little easier, you could spread it into 4 TPS92661 devices with 3 LEDs per device.  Then your worst case is 2.25W and your typical is 1.125W roughly.

    -JP

  • Thanks for all JP. I appreciate your answers and kind suggestions. I am very happy to say that that was a satisfactory conversation for my trouble.