LED Resistor Calculations ...

[EasyGoing1]

So I ordered a few PCBs from JLPCB ... among them, is an LED panel I've wanted for a while ... this is what it ended up looking like:



And I mean ... $25 for 10 of them with all the parts installed by them ... how stupid cheap is that? Total no brainer ... but to chose the resistor for each line of LEDs I used an online calculator. I speced the LEDs at 3.0 volts, 150ma with a 19v power supply and this is the design it kicked out with 6.8Ohm resistors. I believe it called for ¼ watt, but I got ½ watt.

Anyways, as I was looking at the actual specs for the LEDs, they are rated at like 2.8 to 3.4 volts ... so I decided to go back to that web site and re-run the calls with the LEDs speced at 3.4 with everything else staying the same. And what it kicked back was kind of interesting to me ... it speced 15 ohm resistors for each line instead of the 6.8 ... and with the 15 ohm resistors it said the circuit would draw a total of 900ma ... but only 750ma with the LEDs speced at 3.0 volts.

Now see this is interesting to me, because since the LEDs have a voltage rating that is a range of voltage, and since they are capable of running at 3.4 volts, does that mean I'll be over-driving them with the 6.8 ohm resistors?

How does that work?

[stj]

Re: LED Resistor Calculations ...

got a link to the leds?
i get 6.6ohm for 3v and 150ma

[EasyGoing1]

Re: LED Resistor Calculations ...

got a link to the leds?
i get 6.6ohm for 3v and 150ma

https://jlcpcb.com/parts/componentSe...rchTxt=C210349

[EasyGoing1]

Re: LED Resistor Calculations ...

got a link to the leds?
i get 6.6ohm for 3v and 150ma

I have the same spec LEDs here cause you get 100 of em for $7 on amazon ... and I soldered up 6 in series with a 6.8 ohm at the end that's reading 7.2 on my meter ... cheap ass resistors ... but the current draw was 170ma.

[EasyGoing1]

Re: LED Resistor Calculations ...

I suppose I could just put a buck converter on there and drop the voltage or find some surface mount resistors and replace them.

[redwire]

Re: LED Resistor Calculations ...

Cree JB5630 0.2W class around 2.8V 65mA, 3.0V hot at 150mA. Should be similar. Surprised to see the Vf is lower than rule of thumb, I thought 3.0V or more at low currents under 50mA but nope.

OP because your string voltage will be close to the PSU voltage, you get a very steep I vs V curve.
For 19V PSU, 3.0V LED's and 6.67ohm resistor say it's 150mA per string you get, or 750mA total.
If the LED's heat up to 3.05V then you get 105mA per string with the same 6.67ohm resistor.
If the LED's are cool at 2.95V then you get 195mA per string with the same 6.67ohm resistor.
If the PSU is a bit high at 19.3V then you get 195mA per string with the same 6.67ohm resistor and 3.0V Vf.

I bought high CRI Cree JE2835 LED's from Mouser, to light up my old Topward TPS-4000 PSU meters (totally worth it!) Same Vf curves, 2.7V@25mA.

[budm]

Re: LED Resistor Calculations ...

I would run those LED's at 80% of rated current (spec of the LED based on 25c temperature).
You could have made the traces larger to help dissipate the heat from LED, you have lots of room for big traces.

[redwire]

Re: LED Resistor Calculations ...

I found you can only dissipate a few watts at most with a board that size, unless you add a heatsink or cooling fan.

The LED's have a large (cathode) thermal pad which is supposed to have lots of copper and thermal vias to the backside to get the full rated power. Even then it is chinese watts so running an LED at their rated 150mA or 0.42W each for 30 LEDs is almost 13W for the entire board! It will burn up your fingers. This is likely why somebody changed the resistor's value.

Cree rates their 5630 LED's half that current 65mA 0.18W for 5.5W for the entire board. That's still too much in my experience, I would add an off board power resistor or pull one LED in every string and replace it with a resistor, to lower the power and the I vs V slope.

[EasyGoing1]

Re: LED Resistor Calculations ...

I would run those LED's at 80% of rated current (spec of the LED based on 25c temperature).
You could have made the traces larger to help dissipate the heat from LED, you have lots of room for big traces.

The LEDs have heat dissipation in the center. I put vias at the center of each LED and made the back all copper.

[EasyGoing1]

Re: LED Resistor Calculations ...

I found you can only dissipate a few watts at most with a board that size, unless you add a heatsink or cooling fan.

The LED's have a large (cathode) thermal pad which is supposed to have lots of copper and thermal vias to the backside to get the full rated power. Even then it is chinese watts so running an LED at their rated 150mA or 0.42W each for 30 LEDs is almost 13W for the entire board! It will burn up your fingers. This is likely why somebody changed the resistor's value.

Cree rates their 5630 LED's half that current 65mA 0.18W for 5.5W for the entire board. That's still too much in my experience, I would add an off board power resistor or pull one LED in every string and replace it with a resistor, to lower the power and the I vs V slope.

After experimenting with the actual boards now and thinking about the problem, what you just said makes the most sense to me.

I'm still a little puzzled about one thing... the LEDs that I had them install on the boards are rated for 2.8 to 3.4 volts ... how is it even possible to rate an LED on a voltage range? Don't the damn things turn on at a specific voltage? .. not "around" voltage X ... they will forward bias and tun on... NO, they should forward bias at a fairly consistent and precise voltage, no?

And when I used the LEDs that i bought from Amazon, which are straight up rated at 3V - PERIOD! Not 2.9 to 3.1 or some shit like that ... they are just THREE VOLTS, and I put 6 of them in series with a 15 ohm resistor, they pulled 200ma with a 19 volt source ... I just dont get it. Then again, I'm relying on my power supply to tell me what they're pulling, I didnt actually measure the current with my meter...

I bought some 1 watt 15Ω resistors from Mouser and I took one of these boards and replaced the 6.8Ω resistors with the 15Ω that I got from Mouser and then I set my power supply to 19 volts and didn't bother limiting the current because that's what the damn resistor is supposed to do ... but when I turned on the power supply they pulled WAY MORE than 750ma ... and 90% of them died almost instantly on that panel...

I'm thinking the LEDs that were installed on the panels are just shit and that I need to put the ones I have on the panels, then use the 15Ω resistor so that they drive at half the rated current.

So since the LEDs I have are 3.0 volts rated for 150ma ... driving a line of 6 with a 19 volt source at a 15Ω resistor at the end of the line should put a comfy 75ma in that line ... no?

And here's a curiosity question (purely theoretical in scope): would it not be better ... (technically speaking and using this design as an example) ... to put 7.5Ω resistors on each end of the line of LEDs instead of a single 15Ω resistor ... I'm thinking in terms of electron flow when the switch is first thrown to power the line ... could there be an instant in time where current flows through the LEDs at a faster rate until the resistor has a chance to do its job ... or actually now that I'm thinking about it ... this might be why we want the resistor at the low of the circuit since electrons flow from negative to positive, the current will already be choked before reaching the LEDs?

OH - and to address one of your concerns about heat ... I think I would rather implement a fan than sacrifice light intensity ... though I am willing to reduce the current in each line as a means of preserving the LEDs themselves ... dead LEDs do me no good.

[diif]

Re: LED Resistor Calculations ...

LEDs work over a range of voltages but you need to do real world testing for the brightness you want and current draw and heat dissipation.
Your panel should really be Aluminium not FR4.

[EasyGoing1]

Re: LED Resistor Calculations ...

Your panel should really be Aluminium not FR4.

Even if I run them at half the rated current?

[diif]

Re: LED Resistor Calculations ...

Even if I run them at half the rated current?

I've been experimenting with some LEDs recently. When working out if I needed an extra heatsink, I set the current limit on my power supply then ramped up the voltage, can you not do the same ?

[EasyGoing1]

Re: LED Resistor Calculations ...

I've been experimenting with some LEDs recently. When working out if I needed an extra heatsink, I set the current limit on my power supply then ramped up the voltage, can you not do the same ?

I can ... but what good is ramping up the voltage if you've got your current limit set ... it's not like it's actually going to apply more voltage ...???

[diif]

Re: LED Resistor Calculations ...

You start with the voltage under it's suggested value.
I trust the values on my power supply far more than the numbers written next to some no name/brand LEDs from China/Amazon so I do real world tests, I was suggesting you do the same as you have LEDs of unknown spec you want to know if they can run at the power you want on that PCB before doing the resistor math.

[budm]

Re: LED Resistor Calculations ...

https://www.badcaps.net/forum/attach...7&d=1618853318
Look at that graph,
@2.65V applied to the LED, the current is Zero, as you increase the Voltage on the LED, it will start conducting current. You will see similar graph for Diode. it will a good learning tool if you have Curve Tracer, just search for how to make Curve Tracer, for example: https://circuitcellar.com/research-d...-curve-tracer/
Semiconductor will not conduct current until the Vf is high enough to start conducting current.
Learn about dynamic resistance.
https://cdn.badcaps-static.com/pdfs/...5dcb4a7018.pdf

[EasyGoing1]

Re: LED Resistor Calculations ...

https://www.badcaps.net/forum/attach...7&d=1618853318
Look at that graph,
@2.65V applied to the LED, the current is Zero, as you increase the Voltage on the LED, it will start conducting current. You will see similar graph for Diode. it will a good learning tool if you have Curve Tracer, just search for how to make Curve Tracer, for example: https://circuitcellar.com/research-d...-curve-tracer/
Semiconductor will not conduct current until the Vf is high enough to start conducting current.
Learn about dynamic resistance.
https://cdn.badcaps-static.com/pdfs/...5dcb4a7018.pdf

I just discovered something note worthy about these LEDs



They have three pads underneath them, Annode, Cathode and in the center is what I use to believe was strictly for heat dissipation. In fact, when you're working with these LEDs with EasyEDA, and you chose one of their parts, you can see that it has three distinct contact points...



So when I designed these LED panels, I took the center contact point from each LED and gave them their own net, then I used via's to connect them to the other side of the board and made that entire side a copper layer where all of those center points of contact were connected ... like a big heat sync ...

I bought the boards with all the parts on them ... they passed engineering inspection before being approved ... and I was even charged to have an engineer look at my design - which they require... and like I said ... it passed...

But what I have now discovered, is that the annode is electrically equivalent to that center contact point. There are not three distinct points of contact on these LEDs ... there are three separate pads - no doubt - but the annode and the center pad are exactly the same.

Needless to say, none of these boards work at all.

I took one of them, shoved it in the oven ... scraped off all the parts then drilled out the vias and then solderd new LEDs on one row with a 15 ohm resistor at the end and they work like a champ.

But that leaves the question of ... how to get rid of the heat in a situation where you have many of these on a single board ... you can't very well connect all the annodes together unless you're running them in parallel.

What a shitty design if you ask me ... I mean ... thinking about it from an engineering perspective, heat is going to dissipate from one of the two nodes, obviously ... but from a practical use point of view, it would be convenient to have the heat dissipate without that node being connected to either side of the diode. How that would actually work? I dunno...

[budm]

Re: LED Resistor Calculations ...

The spec sheet of the LED should indicate what the center pad is. The heat is dissipated through the silicon substrate.
Do you have specsheet of the LED?

[eccerr0r]

Re: LED Resistor Calculations ...

Yes you will have to isolate each one because it would be expensive for the LED manufacturers to isolate and provide optimal heat transfer. They want to get the heat out of the package as fast as possible and give the designer the choice how to get rid of the heat... Now we know that TO-220FP's can't handle as much heat as TO-220s, so make your own judgment call.

I thought most substrate pads should be cathode, but yes, depends on the manufacturer. Datasheet should clearly indicate this, and if not, that's a datasheet error.

[Agent24]

Re: LED Resistor Calculations ...

I'm still a little puzzled about one thing... the LEDs that I had them install on the boards are rated for 2.8 to 3.4 volts ... how is it even possible to rate an LED on a voltage range? Don't the damn things turn on at a specific voltage? .. not "around" voltage X ... they will forward bias and tun on... NO, they should forward bias at a fairly consistent and precise voltage, no?

While the precision of components has improved over the years, no two components are manufactured exactly the same, so no, there is no exact voltage. That's why they give a range, or just specify the average/nominal value, as I suspect your Amazon LEDs do.

If you pay more, you can buy components that have been tested and sorted and matched, but straight off the production line (especially if it's a cheap factory in the back end of China) you will get variances.

Same reason computer CPUs are sold at different speed grades (models) and different ones will overclock better than others. They're not all exactly the same.

Part of good engineering is designing with these component variances in mind, or if you reach a situation where you're really pushing the limits of the design or find the averages aren't precise enough for you, you have to pay more to get components that have been matched, or buy a whole lot and test and match them yourself.