High Intensity NIR LED Array

Hi All,

I am working on a project (high intensity NIR LED Array for Photobiomodulation).

So far this is how far I have gotten. I have some questions about the power supply and whether I have designed it right or wrong.

So the requirements

1) A (largely) portable device
2) Fairly long life.
3) A simple circuit design
4) 8 LED strings with 10 LED on each string.

I have decided to use the following LED for my application - Rohm SIM-040ST for their high (comparatively) power output of 20mW/sr over a narrow viewing angle.

For the LED Driver IC I have selected the TLC5917 from Texas Instruments. First the questions then the designed values

1) This LED Driver IC is supposed to connected to a microcontroller to operate at some frequency. I would prefer it to be continuously on. Can I then connect the LE(ED1) pin to a voltage high and the OE(ED2) to ground to have "always on" operation?. I guess SD1 should also be held high then? From the datasheet LE(ED1) and OE(ED2) pins are described in this way

LE(ED1)

-Data strobe input: Serial data is transferred to the respective latch when LE(ED1) is high. The data is latched LE(ED1) 4 I when LE(ED1) goes low. Also, a -control signal input for an Error Detection Mode and Current Adjust Mode (see Timing Diagram). LE(ED1) has an internal pulldown

OE(ED2)

-Output enable. When OE(ED2) is active (low), the output drivers are enabled; when OE(ED2) is high, all output drivers are turned OFF (blanked). Also, a --control signal input for OE(ED2) 13 I an Error Detection Mode and Current Adjust Mode (see Figure 11). OE(ED2) has an internal pullup.

Now about the design. Going by the design guidelines on Page 21.

If all the outputs are driven high (default value) then the current multiplier (CM) is 1 and programmable voltage gain (VG) is .992 (on page 21).
I want 100mA along each LED String thus

Rext = (1.25*15)/(100e-3)
= 188 Ohms

Which then gives an Iref of

Iref = 1.25/188
= 6.6mA

Additionally as each LED requires 1.7V across it each string would take 17V and thus allowing for a bit of headroom I should use a VLED of 18V.

Considering now power calculations (Page 25).

What is the junction to ambient temperature and how do I make sense of the last table on Page 4?
How do I determine what supply voltage I need? Like what type of battery do I need to power the IC. Also I imagine that VLED is separate from VDD? In that case how do I determine what type of supply I need for VLED and VDD?

Im guessing I will need a heatsink for this structure as the power dissaption is running into 7-9W.

I think your choice of IC is wrong.

With 8 strings of LEDs you need 800mA of current, total. The IC can only handle so much heat. 800mA and with only 1V across the IC is 800mW of heat.

The LED: The voltage drop could be as high as 2.5V, typically 1.7V and as low as (does not say)
If you get all the LEDs at the same time they will probably have about the same voltage drop but this is not guaranteed.
The LED voltage changes with temperature! (10 LED in a string, -25C to +85C will cause the voltage to change 2V) so (+25C to 85C = 1V change)
There are some tricks on the PCB layout to get heat out of the LED.

Hi,

Thanks for the input. Why do you believe the IC choice is wrong? From the datasheet it says it can drive upto 8 strings with 100mA per string. As far as I can see I have chosen all the currents and voltages to be within the maximum specs specified by the datasheet of the IC. Where did you get the 1V across the IC from?
I was going to attach a heatsink to the circuit to draw away heat.

What sort of IC would you recommend?

ElectronicsStartup said:

Where did you get the 1V across the IC from?

Click to expand...

I just made up a number. (actually I found it on the data sheet, Item 7.3 IO ) How many volts are you planning on putting across the IC?
From the graph below a typical part will need 0.5 to 1V to regulate at 100mA.

Hi Again,

Thanks for the reply.
As I said in my original post one of the questions I had was "How do I select the Power Supply Voltage?" Thus I am not totally sure what voltage I want across the IC as of yet.
Can you provide some guidance?

You should allow for 2.5 V for each LED, as that is the maximum Vf quoted in the data sheet. On top of that you need to allow around 1 V for the IC, because it won't regulate to 100 mA at anything less. That gives you 26 V for 10 LEDs. Depending on what power supply you chose, you might need to allow for that to change a bit, but for now I'll assume that you can get 26 V exactly.

The typical value for the LED voltage is 1.7 V, so the voltage drop across the IC would end up as 26 - (10 * 1.7 ) = 9 V. With a total of 800 mA, you would get 7.2 W dissipated in the IC. The IC is quoted as having a thermal resistance of 31.8 °C/W, so you would get a temperature rise of 229 °C which would take you way beyond the maximum junction temperature.

You could reduce the voltage, and aim for the typical Vf for the LED, with a small margin on top. That would give you say 1.9 V per LED allowed for, 1.7 V typical. The voltage supply would then be 20 V, with a 3 V drop across the IC. Power would be 2.4 W, and temperature rise would be 76 °C, which would be on the hot side but probably OK if you are always keeping the electronics somewhere near room temperature. That runs the risk that it you get some or all of your LEDs that have a higher voltage, and so end up dimmer. It probably wouldn't be acceptable in production, unless the LEDs were tested first.

The other way to do it is to have some type of switch-mode LED controller. That would handle the possible voltage variation efficiently.

Thanks for the answer. I guess my mistake was assuming I could be bang on typical instead of leaving room for the maximum.

ElectronicsStartup said:

I guess my mistake was assuming I could be bang on typical instead of leaving room for the maximum.

Click to expand...

If you are making only a hand full you can build it any way you want as long as it works.
When I make 100,000 I need to plan on getting some not typical parts.
Seveal times I found that some one big got all the typical parts, before me. An example was +/-5% capacitors (big high current high voltage capacitors) but there were no +/-2.5% capacitors delivered to me. So my parts were -5 to -2.5 and +2.5 to +5%. So I had to sort my caps into two piles (plus and minus). For LEDs we some times request to get each shipment labeled with the Vf= . When all the parts are 2.0 to 2.1V for a month and 1.7 to 1.6 for the next month I can handle that. I can not parallel a 1.7 and a 2.1V LEDs.

ElectronicsStartup said:

1) This LED Driver IC is supposed to connected to a microcontroller to operate at some frequency. I would prefer it to be continuously on. Can I then connect the LE(ED1) pin to a voltage high and the OE(ED2) to ground to have "always on" operation?. I guess SD1 should also be held high then? From the datasheet LE(ED1) and OE(ED2) pins are described in this way

Click to expand...

I don't know if you saw this, but there is just a Appnote describing how to run the part without a micro-controller. See appnote SLVA346–July 2009.

Here is a Snippet from the app note.


According to the data sheet, there are two different voltages to worry about and that have different specs. One is VDD spec, and one for LED voltage. See table in data sheet.
min max
VDD Supply voltage 0 V 7 V
VO Supply voltage to output pins OUT0–OUT7 20 V

Perhaps you can cascade the current sinks to give yourself the needed headroom. Almost sounds like you be better off design your own current sink, (lots of circuits on the web).
There was another thread on here by Evil Genius where he design his own, but his might be current source rather than sink, I can't recall

If you rather go with a chip solution, then cascading would be a possible solution, maybe 4 strings per chip. There are much smarter people than me on this forum that may have better solutions. Good luck

----------------------------------edit--------------------------------------

Gadzooks, after I went through the trouble of getting just the right amount of spaces in the table I created, it went to crap after posting... Just see data sheet.

RON Simpson.

Yes I am only making a handful right now. Really only need one prototype for personal use.

As an alternative solution - Can't I just get a power supply that converts AC to say 18VDC and use this voltage as the voltage drop over 10 series LED and have multiple of these in parallel? In this case the most critical thing is the power and current that the power supply can provide. This should be easier right?

You could get a 24V supply that will current limit at 800mA. Then the voltage will drop back to (18V?) or what ever it takes to hold 800mA.

Here is part of a schematic I built a while back. This is 5VDC to LEDs. So I am boosting 5 volts to some voltage. (maybe 10V)
Note "Vout" and "N$13" are connected together.
There are 8 strings of 4 LEDs.
I used a resistor in each string. If one of the strings pulls too much current then the voltage across that resistor goes up and reduced the voltage on that string. It helps share current.
At the bottom you will see a line called "feedback". This goes to the PWM that is making the voltage. So Vout goes up until Fout=0.5 volts. Then Vout holds at that voltage, keeping the current constant.
I could have used 8 resistors to average the voltage across R1-R8 and fed that back to the PWM. (not much room on the board) So I just watched the current in one string and hoped the others are the same.

You could use one of the cheap step up regulators from Ebay with Ron's design (Post #12) They can be modified to provide a constant current source like this.
As you would have 8 chains of 10 LEDs the regulator would be providing about 19.25 volts (18 for the LED chain and 1,25 across the current sense resistor.) As you require 100 mA through each chain the resistors (R1 to R8) would need to be 12.5 ohms. (12 ohm resistors would be OK) The feedback wire from Ron's circuit would connect to the bottom of R2 in my circuit. (R3 and R4 would not be used.) The step up regulator could be fed with any voltage between about 9 volts and 17 volts so a 12 volt battery would be a suitable power source.

Les.

I agree with Les,

Getting from the power line to some regulated voltage is something better left to experts. You can pay money and get some thing already made.

Then use a boost up or a buck down to regulate current in to the LEDs.
In the link Les Jones gave, the second schematic, is what I did (close). The schematic shows 300mA so you could use three if these. I think each will drive 3 strings of LEDs just fine.

I used a little FAN5333.

Hi Ron,
I've not seen that chip before. I like the low reference voltage so less power is wasted in the sense resistor. The ready built units from ebay are cheaper than I could buy the switching regulator chip. I have to admit that the added zenner diode was fitted AFTER one of the LEDs in the chain went open circuit and the regulator attempted to supply the set current into an open circuit destroying the output capacitor. (I was surprised that the regulator chip was not also destroyed.)
ElectronicsStartup, I agree with Ron that if you are running it from the mains that you should buy the low voltage power supply. (I thought when you said portable that you wanted it powered from batteries.

Les.