• State Resolved
  • Locked Locked
  • Replies 14 replies
  • Subscribers 63 subscribers
  • Views 4230 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Infrared LED buck driver for high output current stability

Mackenzie Wilson
Prodigy 100 points
Other Parts Discussed in Thread: TPS92513, LM3414, LM3406, LM3409, TPS92640

Hi there,

I am looking for a step-down (buck) LED driver to drive up to 4 alternating infrared LEDs, either all with the same driver or with up to 4 different drivers and one output each (only one LED will ever be on at any given time).  My current LED is LED Engin’s LZ1 infrared LEDs at 740-940nm, 2-4V and 1A, and I will be using about a 12V supply to power these LEDs.  For simplicity, and due to my inexperience, I am looking specifically for an evaluation board for an LED driver that suits my project.  I looked on a third-party distributor’s site and found the TPS92513, LM3414, LM3409 and LM3406 (and their respective evaluation modules) to be promising.  I was wondering if you could make a suggestion between those four products for my purposes, or others of a similar price range (since this site has more selection).

My main project need is impeccable current accuracy and stability due to precise, medical applications (i.e. very low output current ripple and predictable current output true to its settings), as well as true LED “colour” or (more accurately) wavelength.  The board also must have PWM dimming that can be controlled externally from a pin or something similar.  Size and efficiency are not main concerns right now, and I would prefer sticking to a true buck topology.  Additionally, I don’t want to be blasting the LEDs at their max current rating of 1A, so it would be ideal to either use a driver with a slightly lower rating (still >600mA, for example), or that also has analog dimming so that I can dial back the output current that PWM uses in its duty cycle.  Either way, the idea is to set the max current output ahead of time and leave it; I won’t be changing this during operation.  Beyond that, I also have some questions:

1) The LM3406 evaluation board has four current output settings.  How do these work? Are they selections made when ordering, or is there an electrical component change that must be made to the board? If I am only interested in using one setting can I order it with that setting to avoid having to make changes?  

2) Is there a technical/performance difference between “LM” models and “TPS” products?

3) For the TPS92513 and LM3414, the first page of the datasheet boasts a “current accuracy” percentage. Is this regarding the regular fluctuations or ripple, or is it something else? For the TPS92513, it also mentions a 20mA current ripple, so what does the 5% represent?

4) Where can I find information pertaining to the size of the output current ripple and current stability, for models that don’t give percentages?

5) The LM3409 evaluation board’s PWM has a much higher ratio than the other products (10000:1). Is this better or worse for current stability and LED “colour” or wavelength? On this model, can analog and PWM be used simultaneously?

6) In contrast, the TPS92513 has only a 100:1 PWM dimming ratio. How will this much lower ratio affect its performance, current stability and output wavelength while dimming?

7) Do you have any other suggested models that would be better suited, perhaps one that can handle multiple outputs with the above regulations for each output?

8) Lastly, if using a single driver with just one output, would it be appropriate for my requirements to put a power transistor on the LED side so that I can electronically switch between LEDs on the same output channel? Would this damage the LEDs?

Thank you very much for your help!!

~Mackenzie

Hi there,
I am looking for a step-down (buck) LED driver to drive up to 4 alternating infrared LEDs, either all with the same driver or with up to 4 different drivers and one output each (only one LED will ever be on at any given time).  My current LED is LED Engin’s LZ1 infrared LEDs at 740-940nm, 2-4V and 1A, and I will be using about a 12V supply to power these LEDs.  For simplicity, I am looking specifically for an evaluation board for an LED driver that suits my project.  I looked on a third-party distributor’s site and found the TPS92513, LM3414, LM3409 and LM3406 (and their respective evaluation modules) to be promising.  I was wondering if you could make a suggestion between those four products for my purposes, or others of a similar price range (since this site has more selection).
My main project need is impeccable current accuracy and stability due to precise, medical applications (i.e. very low output current ripple and predictable current output true to its settings), as well as true LED “colour” or (more accurately) wavelength.  The board also must have PWM dimming that can be controlled externally from a pin or something similar.  Size and efficiency are not main concerns right now, and I would prefer sticking to a true buck topology.  Additionally, I don’t want to be blasting the LEDs at their max current rating of 1A, so it would be ideal to either use a driver with a slightly lower rating (still >600mA, for example), or that also has analog dimming so that I can dial back the output current that PWM uses in its duty cycle.  Either way, the idea is to set the max current output once ahead of time and leave it; I won’t be changing this during operation.  Beyond that, I also have some questions:

-          The LM3406 evaluation board has four current output settings.  How do these work? Are they selections made when ordering, or is there an electrical component change that must be made to the board? If I am only interested in using one setting can I order it with that setting to avoid having to make changes?  

-          Is there a technical/performance difference between “LM” models and “TPS” products?

-          For the TPS92513 and LM3414, the first page of the datasheet boasts a “current accuracy” percentage. Is this regarding the regular fluctuations or ripple, or is it something else? For the TPS92513, it also mentions a 20mA current ripple, so what does the 5% represent?

-          Where can I find information pertaining to the size of the output current ripple and current stability, for models that don’t give percentages?

-          The LM3409 evaluation board’s PWM has a much higher ratio than the other products (10000:1). Is this better or worse for current stability and LED “colour” or wavelength? On this model, can analog and PWM be used simultaneously?

-          In contrast, the TPS92513 has only a 100:1 PWM dimming ratio. How will this much lower ratio affect its performance, current stability and output wavelength while dimming?

-          Do you have any other suggested models that would be better suited, perhaps one that can handle multiple outputs with the above regulations for each output?

-          Lastly, if using a single driver with just one output, would it be appropriate for my requirements to put a power transistor on the LED side so that I can electronically switch between LEDs on the same output channel? Would this damage the LEDs?

Thank you very much for your help.
  • 0
    TI__Mastermind 25500 points

    To try and answer your questions:

    1. The LM3406 EVM uses a shunt to select which current sense resistor(s) on the board are in circuit. To change the output current you just change the shunt position.

    2. Not really.

    3. A lot of devices are specd differently so you need to know what to look for. The TPS92513 for example says +/-5%. This has nothing to do with ripple current, it is the current regulation accuracy. If you set the output current to 1A it could vary over temperature and from circuit to circuit between 0.95A and 1.05A. It also depends on what current sense voltage you use for devices with analog current adjust. In the case of the TPS92513 that accuracy is at the full 200mV current sense voltage. The number comes from the +/-10mV of offset in the error amplifier. So if you reduce the current sense voltage to 100mV the part becomes a +/-10% current accuracy. The LED ripple current all depends on the inductor value (inductor ripple current) and the output capacitor value if one is used.

    4. If no % is specified look for a current sense voltage range. For example the TPS92513 specifies a typical of 200mV but a min of 191mV and a max of 210mV. That is where the +/-10mV offset spec comes from. Or if that is not specified usually the offset itself will be specified somewhere so you can calculate the accuracy give the current sense voltage you choose to use.

    5. The dimming ratio has nothing to do with the current regulation while on unless your pulse widths get so small it is no longer dimming linearly.

    6. It depends on the PWM dimming frequency. The TPS92513 is a current mode regulator which is good for accuracy, particularly over input voltage variations, but it make its response time much slower since you need an output capacitor and the control loop takes time to respond. That is why it isn't practical to use with dimming frequencies over a few hundred Hz. The LM3409 is hysteretic so it needs no output capacitor and it is very fast. So it will have a much better dimming ratio even at much higher PWM frequencies.

    7. See below.

    8. Yes, you can switch in/out LEDs as long as it is during the OFF time during PWM dimming or while the device is not switching otherwise (disabled for example).

    It looks like the LM3414 is probably the best fit from what I can tell given it has high accuracy, a simple BOM, and it has fast PWM dimming (much faster than the 513, not quite as fast as the 3409). But to be sure I would need to know what PWM frequency you are using and what dimming ratio you need. If you need very fast PWM dimming say in the 10s of kHz you may need to shunt FET dim.

  • 0 in reply to Clinton Jensen
    Prodigy 100 points
    Hi again,

    Thank you for the very prompt and detailed response. With regards to Q/A #8, about using power transistors:
    1) What would happen if I did switch during an "on" phase, or while the device itself was switching?
    2) The two power transistors i have used in the past in different situations are the TIP120 and TIP125. Would these be good choices for my current use?
    3) Would you happen to have any reference circuits for power switching with the LEDs I am using? Or, can you simply tell me whether NPN or PNP is better in this case?

    With regards to your suggestion of LM3414:
    4) Would you say the LM3414 is the most accurate (in terms of current) out of the lot?
    5) I have not decided on a PWM frequency or dimming ratio; this project is highly exploratory, and I am willing to try out different specs and see what works for me. However, seeing as my goal is overall light stability and ideally lots of dimming flexibility, I think a fairly high PWM frequency and decent contrast ratio (i.e. 1000:1 or higher) would be necessary (correct me if any of my assumptions are wrong!). For example, I need to be able to dim the LED to less than 25% of its max intensity using PWM.

    Sorry for the vague requirements as it comes to PWM - as mentioned, my main requirement is current stability and accuracy, and I don't think I will need anything extreme for dimming such as shunt FET.

    Thanks again, I look forward to your response!

    ~Mackenzie
  • 0 in reply to Mackenzie Wilson
    TI__Mastermind 25500 points

    Hello Mackenzie,

    1. Depends on the dynamics, but it will likely cause current spikes and ringing since the internal compensation would be trying to adjust and it may not be fast enough.

    2. Those may work fine, but I would recommend a FET rather than a BJT. You could get a much smaller package most likely and they are faster.

    3. Not that I'm aware of. You would just have a FET in series with each and turn on the one you want. You could use an NFET or a PFET, it depends on what you are driving it with. But given the LEDs are referenced to Vin a PFET may be easier to implement.

    4. They are all pretty accurate, but it is one of the better ones. The most accurate may be the TPS92640, it only has 600uV of offset. But it may be a bit overkill for what you are doing, it requires a lot more components.

    5. The maximum contrast ratio you can get depends entirely on the dimming frequency. The lower the frequency the higher the contrast ratio you can get. If you just want to eliminate visible flicker and need a very high ratio I would stick to something in the 200Hz to 300Hz range, 250Hz is commonly used and you should be able to get 1000:1 or better in this range without trouble.

  • 0 in reply to Clinton Jensen
    Prodigy 100 points
    Thank you very much, this is very helpful and exactly what I needed. The TPS92640 requires higher voltage than I need, but it is great to know the LM3414 is suitable for me!

    Have a good day.

    ~Mackenzie
  • 0 in reply to Clinton Jensen
    Prodigy 100 points
    Hi again, I have one quick additional question.

    Would the Evaluation Boards for either LM3414 or LM3409 come with pins/functionality that would allow me to directly measure output current with an oscilloscope? If there is no direct hook-up intended for this, what would be the best other way that you recommend to use an oscilloscope to measure current output on these eval boards?

    Thanks again.
    ~Mackenzie
  • 0 in reply to Mackenzie Wilson
    TI__Expert 3950 points
    Hi Mackenzie,

    I guess you may want to measure the current from the wire connecting the driver and LED with a current probe.
    Those EVMs do not have test points for direct measure output current.

    Thank you!
  • 0 in reply to Issac Hsu
    Prodigy 100 points
    Thanks Issac.

    Sorry for yet another post. I ordered both the LM3409 and LM3414 and am now working with both of them. In a preliminary check that everything was working, I noticed that the LM3409 evaluation board will not light up my LED. I'm running it from a 12V supply, to power a single 1A, 2.5V infrared LED. The same set up works with the LM3414, so there is nothing wrong with the LED or power supply. I have quadruple checked the wire connections and they are all correct (i.e. Power supply + to Vin, Led + to LED+ turret, etc). I originally tried this with the shunt for Jumper 1 in position 1 (shorting Vin and EN) as the datasheet suggested, and additionally tried to run the LED with the shunt in all other positions (including off) just in case, but no luck.

    I am wondering if there is something I am doing wrong, or if I missed something in the datasheet, or if maybe this set up is problematic for the LM3409 board. Why can't it light up my LED?

    Thank you very much for all your help.
    ~Mackenzie
  • 0 in reply to Issac Hsu
    Prodigy 100 points
    Hi there, I am still wondering why my LM3409 board will not work, even with power lines and LED connections hooked up properly.

    Thanks.
  • 0 in reply to Mackenzie Wilson
    TI__Expert 3950 points
    Hi Mackenzie,

    I will direct you to Clint for technical details.

    Thanks.
  • 0 in reply to Mackenzie Wilson
    TI__Mastermind 25500 points
    Is the IADJ potentiometer turned all the way down? Try turning the potentiometer up if so. If that doesn't work it would be really hard for me to say without looking at the board in the lab.
  • 0 in reply to Clinton Jensen
    Prodigy 100 points

    I believe it is turned all the way up. In either case, I have tested it while turned all the way to both sides, just in case. Is the 12V supply sufficient? I have looked at a few simple connections using a multimeter, and found that there is a very tiny voltage drop between LED+ and LED-, about 0.6V, which is not enough to light my LED. The board's datasheet, however, says that it is rated for up to 15V output. 

    Is it possible that there is a defect in the board...?

    Hope this helps. Thank you.

  • 0 in reply to Mackenzie Wilson
    TI__Mastermind 25500 points
    Can you probe the CSN/FET node to see if the part is switching? If there is voltage across the LED it must be running some current. I would also check the voltage at the VADJ test point. It's possible there is a defect in the board, but the most common culprit is a potentiometer. I'm wondering if IADJ is just being held low by a bad pot. If so you can remove it and just drive the VADJ test point with a voltage either from a supply or derived from a resistor divider.
  • 0 in reply to Clinton Jensen
    Prodigy 100 points
    There is 12V from the CSN pin on the IC to ground. There is no voltage drop from the VADJ turret to ground.

    I will try driving the VADJ turret with a voltage - will this override the potentiometer or do I have to remove the pot first?
  • 0 in reply to Mackenzie Wilson
    TI__Mastermind 25500 points
    Usually you could put the pot at full resistance and that voltage would just override it. But in this case it looks like the pot might be stuck at low resistance so you may want to remove it so you don't overtax your supply. But you can measure the resistance first and if it is high enough that it will not draw significant current you can leave it on there.