# Understanding LED Arrays and R(D) Calc

#### brightnight1

Joined Jan 13, 2018
72
I have two questions about LED's and would appreciate some insight. First, I want to build a "high power" LED array with 100 LEDs but I’m not sure what the smartest layout design for the LED's would be. I could do 100 LEDs in series but then if one dies they all will stop working. I could do one LED driver with 10 strings and 10 LEDs each (if I could find one). Or, it seems like having a separate LED driver for each 10 LED string would be the smartest was to go.

Second, I’d like to know if I calculated the LED R(d) correctly. The charts for this LED aren’t the typical ones I’ve seen. The all have pulse times instead of continuous on times (which is how I plan to use them). I calculated R(d) with the values on the datasheet was are :
https://www.digikey.com/products/en?keywords=SFH 4547

V(f) = 1.5V
I(f)= 100mA

1.5/.1=15 Ohms = R(d)

Is this a reasonable R(d) value?

#### Dodgydave

Joined Jun 22, 2012
9,915
Use 10 LEDs in series on a Constant current led driver, that's 15V @ 100 mA, per chain, you can vary the current to alter the brightness..

#### brightnight1

Joined Jan 13, 2018
72
Ok perfect so I'll need 10 LED drives.
Already found a driver the LED6000 and calculations show it will be 91.8% efficient which I'd be pretty happy with.

#### dl324

Joined Mar 30, 2015
12,673
V(f) = 1.5V
I(f)= 100mA

1.5/.1=15 Ohms = R(d)

Is this a reasonable R(d) value?
No. You can't calculate the value for a current limiting resistor unless you know the supply voltage.

If you use an LED driver, you don't need a current limiting resistor because it'll be a current source.
Already found a driver the LED6000 and calculations show it will be 91.8% efficient which I'd be pretty happy with.
Efficiency will depend on the number of LEDs you place in series and the open circuit voltage of the current source. If you put one LED on the current source, it's efficiency isn't going to be 91%.

#### brightnight1

Joined Jan 13, 2018
72
I used Edesign studio from STM to "build" my driver. It requirese R(d) as one of the inputs, I believe it's for the LED. For some reason the max it would let me do is 20 LED's per series but if I double to I(f), V(f), and R(d) values and enter 20 LEDs per string it would be like having 40 LEDs (60V). If I build a bunch of these arrays less drivers/components would save me a lot. Does anyone see anything wrong with this approach? Below are two pics (one zoomed of the circuit) for the driver with a 20 LED string.

#### dl324

Joined Mar 30, 2015
12,673
It requirese R(d) as one of the inputs, I believe it's for the LED.
It seems to be the dynamic resistance of the LED. We normally ignore that because when operating far enough beyond the knee, it's small enough to ignore.

EDIT: it's not dynamic resistance either. Dividing LED voltage by LED current doesn't give you a meaningfull number. To get dynamic resistance, you'd use delta V/delta I.

Last edited:

Joined Jan 15, 2015
5,875
Not sure where you are going with this but you are aware the LED you linked to is a 950NM IR LED correct? Not a visible light LED. Just making sure.

Ron

Joined Jan 15, 2015
5,875
1.5/.1=15 Ohms = R(d)

Is this a reasonable R(d) value?
Not quite because LED resistance is not at all linear and even the Vfwd and Ifwd are approximate numbers for you to work to.

Ron

#### brightnight1

Joined Jan 13, 2018
72
Thanks for the reply Ron, Yes it's for an IR LED array for a camera

I guess I'll still a little confused on the R(d) calculation. Can anyone clarify what it should be for these LED's?

#### ebeowulf17

Joined Aug 12, 2014
3,275
TI application note 1656 explains R(d) and how to derive it from the curves on a datasheet. It was all brand new information to me, but the application note is quite helpful:

This stack exchange description of how to model LED parameters in spice also helped me better understand the meaning of these specs, and why LED curves are as complex as they are:

https://electronics.stackexchange.com/questions/9510/how-do-i-model-an-led-with-spice/9543#9543

#### ebeowulf17

Joined Aug 12, 2014
3,275
I have two questions about LED's and would appreciate some insight. First, I want to build a "high power" LED array with 100 LEDs but I’m not sure what the smartest layout design for the LED's would be. I could do 100 LEDs in series but then if one dies they all will stop working. I could do one LED driver with 10 strings and 10 LEDs each (if I could find one). Or, it seems like having a separate LED driver for each 10 LED string would be the smartest was to go.

Second, I’d like to know if I calculated the LED R(d) correctly. The charts for this LED aren’t the typical ones I’ve seen. The all have pulse times instead of continuous on times (which is how I plan to use them). I calculated R(d) with the values on the datasheet was are :
https://www.digikey.com/products/en?keywords=SFH 4547

V(f) = 1.5V
I(f)= 100mA

1.5/.1=15 Ohms = R(d)

Is this a reasonable R(d) value?
Note that 100mA is listed as its absolute maximum current rating, and it's never recommended to run LEDs at, or near, their max ratings for any length of time. You'd be well advised to cut that number in half and shoot for 50mA, unless you're going to be using this in short bursts with long pauses in between.

As for calculating R(d) based on the application note listed above, it's complicated by the fact that the app note suggest drawing a tangent line and taking its slope, but it does so on a linear graph, whereas the graph in your datasheet is exponential. There may be a clever trick to simplify this, but here's my best attempt at a solution. Seems to me like you have to manually capture a bunch of datapoints from the graph, then plot them on a linear graph and do the tangent slope thing from there. Here's the original datasheet graph:

Here are some approximate data points pulled from that graph;

Here is an exponential plot of them in my spreadsheet just to confirm that I've matched the original curve reasonably well:

And here is the linear plot, along with my rough guess at a tangent line:

The tangent line above moves 500mA per Volt, or 0.5A per Volt. Use Ohms law to solve for R and you get 2 ohms.

So, I'd say that at roughly 50mA, where I'd recommend operating these LEDs, the Vf is roughly 1.39V and the R(d) is roughly 2 ohms. I'm not entirely sure that I've done all of this correctly, even in conceptual terms, and I've definitely been doing loose approximations on the numbers themselves, but maybe it's enough to demonstrate the method and get a ballpark estimate.

Last edited:

#### brightnight1

Joined Jan 13, 2018
72
Wow thank you so much for taking the time to do this, your explanation is incredible helpful!

#### dynamoleddisplays

Joined Mar 28, 2020
3
Hi i am searching for LEDs to occur of ordinary medium based bulbs and floods. any I have found so far have been about $30/ea. I understand its the last buld i'll ever purchase put it would take a lifetime to repay also. They would accompany me when I move #### Reloadron Joined Jan 15, 2015 5,875 Hi i am searching for LEDs to occur of ordinary medium based bulbs and floods. any I have found so far have been about$30/ea. I understand its the last buld i'll ever purchase put it would take a lifetime to repay also.
They would accompany me when I move
I have no idea where you are looking but I can go to my local home improvement store E-26 Medium Screw Base LED Floods (GE Basic 65-Watt EQ LED Br30 Soft White Dimmable Flood Light Bulb (6-Pack)) That's six lamps for $16.98 or about$2.83 per lamp. They are rated 9 years at 3 hours per day. I have a fixture using 3 of them which has been on 24/7 for over two years now and they are doing just fine. I have no idea where you are shopping but here in the US they come in at under \$3.00 USD per lamp.

Ron