Looking fro high speed LED driver

Hi Pavel,

Which application do you need for such a LED driver? We have many LED drivers that could satisfy the above requirements, so we need to confirm the following spec:

automotive or non-auto?  

What is Vin range? Now I know Vout is about 45V.

Do you need only 1 string? What's the Imax of this LED string?

BR, Gracie

Hi Gracie,

First, thanks for feedback.

Application - open space optical communication, so driver should NOT have PWM control, but should feature control input that switch ON/OFF all LEDs.

Vin: it depends on number of LEDs (non-critical parameter). The number of LEDs can vary from 8 to 12. Given that Vf about 3.3V, for 8 LEDs Vin should be 28 ... 32V, for 12 LEDs - 40 ... 45V.

Concerning strings - perhaps 2 strings, e.g. 2 x 8 LEDs.

Imax - 20 ... 30 mA (depends on LED).

Sincerely

Pavel.

Hi Pavel,

May I understand what your Vin range is?

BR,

Steven

Hi Steven,

Sorry, I misunderstood Gracie's request.

Well, the Vin is supplied by a USB powerbank ... so 5V.

Hi Steven,

Thanks. Yes, your understanding is correct. The device you propose TPS92641 features PWM controller (as well as all other LED driver that I saw until now). For my concept I need LED driver without PWM ... but I guess such device doesn't exist. So I'll boost 5V to some level (e.g. 40V), as you suggest, and then use opamps, MOSFET., etc., i.e. "not-integrated" solution.

Sincerely,

Pavel

Hi Pavel,

For the "LED driver without PWM", do you mean a non-switching-type LED driver like TPS92612?

BR,

Steven

Hi Steven,

Yes. ON/OFF is finction of control signal because it is trasmitting data, so NO any PWM activity during phase ON and OFF.

Sincerely,

Pavel

Hi Pavel,

I am not sure if we are understanding the same thing. For switching-type regulator, the current is always on (but with some ripple) unless you turn off the IC. Here is an article that may give you some ideas on how to achieve your target: Common LED Functions and LED Driver Design Considerations

BR,

Steven

Hi Steven,

Seems I misspoke in my previous post.

Without a detailed review of 2 datasheets (i.e. TPS92641 and TPS 92612) I guess the TPS 92612 seems a better fit for what I'm looking for.

In TPS 92612, the PWM is an external input, which can turn on/off the LED array.

On the other hand, at TPS92641, the term PWM is rather related to an internal 1MHz oscillator where the ON/OFF ratio is controlled in a certain way, but what is essential, there is no input where I could apply the data signal.

So indeed, it is interesting to simulate the TPS 92612 in order to check if it is able to drive a series of diodes with a certain datarate (e.g. 200kb/s). It remains to be seen, is it possible to have TPS model 92612 in order to un a simulation and LTSpice.

Sincerely,

Pavel.

Hi Pavel,

Steven is out of office for a biz trip and will reply to you when he comes back, thanks for your patience.

BR

Patrick

Thanks Patrick!

Hi Pavel,

I am currently out of office. I will reply to you next Monday when I am back to office. Thanks for your understanding.

BR,

Steven

Hi Steven,

Thanks, no problem.

Sincerely,

Pavel.

Hi Pavel,

For TPS92641, you can apply the data signal to SDIM pin to use the shunt dimming. The TPS92641 is just an example here to illustrate the shunt dimming feature which can achieve high on/off speed. The output current spec of TPS92641 is over-spec for your requirement.

For TPS92612, I am not sure if its output voltage spec can meet your requirement. Also, I am not sure if the on/off speed of TPS92612 can meet your requirement. If you are interested in it, I will assign this thread to the corresponding engineer to help you.  

BR,

Steven

Hi Steven,

Not sure that TPS92641 could fit my concept.

Well ... there is Shunt NFET that redirect current on itself switching OFF the LEDs (BTW when the LEDs are OFF, the current is consummed nevertheless). But when the LEDs are ON, the current flowing through the LEDs manifests "the ripples" (i.e. not constant) ... because whole this stuff at left (i.e. Ron, Con, etc.) continue to charge/discharge the coil, isn't it ?

On the other hand, I noticed this sentence in the TPS92612 datasheet: "It is a simple, yet elegant solution to deliver constant current for a single LED string".

Combining it with a single control input (i.e. PWM) gives me some confidence that this controller is capable of doing the job. Why you aren't sure that it can't meet my requirement ?

Sincerely,

Pavel.

P.S. What about simulating TI controllers in LTSpice ... or the only solution is to install TI version of Spice ?

Hi Pavel,

Pavel Yermolenko said:

But when the LEDs are ON, the current flowing through the LEDs manifests "the ripples" (i.e. not constant) ... because whole this stuff at left (i.e. Ron, Con, etc.) continue to charge/discharge the coil, isn't it ?

Alright. I think I got your point. You need a linear LED driver for your application. Am I Correct?

Pavel Yermolenko said:

Combining it with a single control input (i.e. PWM) gives me some confidence that this controller is capable of doing the job. Why you aren't sure that it can't meet my requirement ?

1. For the input voltage range, it can only accept 4.5 V – 40 V (lower than the 45V as you mentioned)
2. For the on/off speed, please refer to Section 6.6 Timing Requirements from the datasheet. I am not sure if this could meet your 200kHz on/off requirement.

If you are interested in TPS92612, I will help assign the corresponding engineer to help you.

Pavel Yermolenko said:

P.S. What about simulating TI controllers in LTSpice ... or the only solution is to install TI version of Spice ?

This is another topic that is out of my knowledge range. You can raise a new E2E thread regarding this topic and other engineer will help you.

BR,

Steven

Hi Steven,

Steven Li said:

Alright. I think I got your point. You need a linear LED driver for your application. Am I Correct?

Yes, I don't know how to qualify the type of driver I'm looking for, but briefly... when it's ON, it must be really ON, ie without ripples, when it's OFF it must be really OFF , that is, there is no current through the LEDs.

Steven Li said:

For the input voltage range, it can only accept 4.5 V – 40 V (lower than the 45V as you mentioned)

Well... that's not really a bad thing because I'd probably opt for two rows of 8 LEDs in series. Since the Vf of these LEDs is around 3.3V, it might work.

Steven Li said:

For the on/off speed, please refer to Section 6.6 Timing Requirements from the datasheet.

Frankly, I didn't quite understand the meaning of the values in the table in section 6.6. For example, the minimum value for t(PWM_delay_rising) is declared to be 10us. What does this mean exactly? If I apply PWM with t(PWM_delay_rising) = 1us, what will happen? Will the device explode? Or does it just mean that the internal logic driving the NMOS, the source of which is connected to the OUT pin, has a limited slew rate?

Steven Li said:

This is another topic that is out of my knowledge range.

Ok, can you just confirm that the macromodel of TPS92612 is fairly faithful and I can trust simulations?

Steven Li said:

If you are interested in TPS92612, I will help assign the corresponding engineer to help you.

Yes, please.

Sincerely,

Pavel.

Indeed, TPS92612 doesn't match speed requirements.

I've simulated the usecase with PSpice for TI and found that maximum clock at PWM about 10kHz (at 20 kHz it's already quite bad)

Hi Pavel,

I also realize that 100kHz 50% PWM may be enough for your 200kb/sec requirement as there is one on and one off in one cycle in 100kHz 50% PWM.

BR,

Steven

Steven Li said:

You can refer to Figure 16. 100 kHz 50% External FET PWM Dimming VO = 42 V (12 LEDs); VIN = 48 V from LM3409 datasheet to check how fast shunt dimming can do.

Hi Steven,

Thanks.

But as can be seen in Fig. 16, when the LED is ON, the current has fairly large ripples that invalidate this device for my app.

Are there any LED drivers that avoid this shortcoming ?

Sincerely,

Pavel

Hi Pavel,

Pavel Yermolenko said:

But as can be seen in Fig. 16, when the LED is ON, the current has fairly large ripples that invalidate this device for my app.

There are methods to make the ripple as small as possible, such as enlarging the switching frequency or enlarging the inductor value.

Pavel Yermolenko said:

Are there any LED drivers that avoid this shortcoming ?

Sorry. The shunt dimming is currently the only method I know that can fit your 200kb/sec requirement.

BR,

Steven

Hi Steven,

Ok, perhaps by tuning this stuff we can reduce ripples, but classical method (without LED driver) avoids completely this drawback ... like this one:

Here is what this gives:

Well it's less integrated then when using driver (need to add boost converter) ... but finally it does what I want.

Sincerely,

Pavel.

Sorry, there are problems on TI webserver. Here is schematic:

Hi Pavel,

Got it. Thanks for sharing this info. Please let me know if any further support is needed.

Best Regards,

Steven

Hi Steven,

No, we can close this case.

Thanks.

Sincerely

Pavel

Hi Pavel,

You are welcome. I will close this thread.

BR,

Steven