Design of series of 8 LED Driver circuit

SunnySkyguy said:

So far it appears to be a poor mismatch of wavelengths to the sensitivity of phototransistors , no daylight blocking or stray light blocking in the design and low efficiency in driving LEDs.

WHat is the purpose of your project? How will you control stray light and the threshold ? Is this for logic level out?

Too often young designers fail to grasp the importance of defining the requirements before implementation begins and the project fails in meaningful ways,

Click to expand...

i use fiber optic to couple between LED and PD in a custom made cast, so it is safer from ambient light; by decladding small potion of fiber optic and dip into different refractive index solution from 3.33 to 4.33 (i.e)8 channel then the output given to NI DAQ to plot the graph.

Yes, according to photodarlington poor wavelength adsorbance at 530nm; it is only available in my lab and can couple with fiber optic.

https://www.sciencedirect.com/science/article/pii/0925400593800213

treez said:

how about this....

https://www.linear.com/product/LT3465

..you can even simulate it on their free sim "Ltspice"

Click to expand...

treez said:

what about this, its an integrated Switch mode led driver, no design effort needed, just hook up...

https://www.linear.com/product/LTM8042

very efficient, simple

Click to expand...

which one is useful for my application ? can i use 8 LED in series

those can drive 8 leds in series yes, but I am not sure what operation you want from the leds.....do you want to be flashing them off in some form of ON/OFF keying comms scheme?..if so , what s the frequency of the flashing?
I thought you were doing a led lighting application, not comms or detection , whatever you are doing.

You will huge path spacing loss with gaps
Attenuation with frequency as not rated but with 10us step response time into 100 Ohms,
it will not perform well if you intend to operate at high freq. (1 MHz) or is this DC?

Perhaps you still need to define overall optical requirements.

Perhaps consider a linear CCD sensor with 10MHz https://www.engineerlive.com/content/compact-cmos-linear-image-sensors

Why does everybody say to use a voltage regulator when the LEDs "might be" 3.1V? The minimum and maximum voltages are not stated, they all might be 3.5V each.
The first post shows a perfectly good LM317 current regulator. The minimum input voltage to it is (8 x 3.5V) + 2.8V= 30.8V where the 2.8V is the minimum voltage needed by the current regulator. Good luck getting 30.8V from a little 9V battery that drops to 6V during its life.

A little 9V battery (6V?) cannot even power one LED and an LM317 current regulator if the LED is 3.5V and the battery is 6V.

treez said:

those can drive 8 leds in series yes, but I am not sure what operation you want from the leds.....do you want to be flashing them off in some form of ON/OFF keying comms scheme?..if so , what s the frequency of the flashing?
I thought you were doing a led lighting application, not comms or detection , whatever you are doing.

Click to expand...

for now constant LED Lightning is sufficient may be in future i may use PWM at 1KHz. Suggest the ic's for both application i will consult with my guide

Thanks in advance

If you put in series, you can still get a wide range in brightness at constant 20mA.
How will you calibrate this? with CUSTOM series resistors? for equal intensity at end of fiber?

for 8x3.1V or 8S you required 24.8V + current regulating voltage drop of 0.1 to 2.5V depending how it is done.
for 4S x2P you need 12.4 + . . .
for 2S4P you need 6.2V + . . . .
When 5mm LEDs which have 60mW rating and 16 Ohms ESR it is wise to add 5Ohms in series with each to prevent thermal runaway from mismatch unless you know the quality of your sources.

for 8P you need a 3.3V ultralow LDO regulator from a 3.7V Lipo which can drop to 3.4 before loss of LDO regulation. giving reasonable discharge time but requires low voltage sensor.

Forget LM317 too high Vin-Vout drop

Using Ohms Law try

Vf of Green LED= 2.8V + If*16Ω =3.1 . Note 2.8V is fixed threshold and each LED will have a tolerance on 16 Ohms....perhaps 20 ...25... so add series resistance of this difference to get 20mA in each parallel string.
Then measure output fiber intensity using a PD with TIA not darlington.. too wide tolerance in hFE and poor tolerance on A/W. PD's are very accurate and cheap. TIA config Op Amp is very cheap and accurate for gain.

Thus from3.3V LDO ... (3.3V-2.8V)/20mA=25 Ohms including 16 Ohms ESR of LED so add up to 10 Ohms for each LED in parallel to 3.3V.

get this LDO and a range of R's around 10 Ohms with 1 Ohm increments.

Your LED efficacy and Darlington efficacy might give a range of 3:1 each max:min in mW/mA and mA/mW respectively.
A PD is very tight tolerance. and PT is very loose tolerance.

If you do as I told, you might be able to calibrate each optical pair to be matched within 5% if you are good.
Recalibrate often.

use this only https://www.digikey.com/product-detail/en/MIC5318-3.3YD5-TR/576-2860-6-ND/1858612

finally i bought two buck **broken link removed** to drive LED(ife93) and Photodarlington (op560)

https://obrazki.elektroda.pl/3320015500_1449803586.jpg my schematics

i am going to use 8 photo darlington in parallel 5v and 400ma and the v1 to v8 output given to 8 channel to amplify the signal using National instruments and to plot output response of the PD

[as sir SunnySkyguy mentioned the response of the PD low in my particular wavelength i planned to amplify the signal using NI or OPamp since the only PD available in lab to couple fiberoptic].

Click to expand...

for 8 LED i use 15v and 40ma to drive it.

View attachment ife93 (2).pdfView attachment op560a.pdf