Need help with LED Power Supply Design

[EasyGoing1]

Hello,

I am trying to finish a small project...

Here is the schematic:



Here is the video of what the circuit does when I turn it on:

You can see that the LEDs simply flicker off while they continue to draw current ... you can also see that I check components for overheat but nothing in the circuit is too hot to touch.

I'm stumped, no idea whats going on here...

Any clues?

Thank you

[budm]

Re: Need help with LED Power Supply Design

You need to be clear about how many White LEDs you are driving.
No caps' value shown. What is the output Voltage of the regulator? 10.25Vdc?
45V input that means the diff V between input and out out is about 35V (40V TYP.) so when power is first applied, it will have about 45V Vdrops on the IC until te output cap is fully charged.
Why use 32VAC as the supply? You can use lower Voltage and set the 317 as the constant current source instead. LED is current device so you will want to maintain constant current through it.
I.E. http://www.bristolwatch.com/sr/cc_led.htm
http://www.reuk.co.uk/wordpress/ligh...-led-lighting/

[stj]

Re: Need help with LED Power Supply Design

wow.

well 32vac rectified would be as high as 46v
that's too much for an lm317.
you also want ceramic caps between the input & ground and the output & ground close to the regulator

what's your led load??

[redwire]

Re: Need help with LED Power Supply Design

DUDE! You are using a 25V filter capacitor ?!!! It's going to POP
I see a 35V cap on the LM317 output.

The flickering is what I see in LED's that are overdriven or overheating, with broken bonding wires.

[EasyGoing1]

Re: Need help with LED Power Supply Design

You need to be clear about how many White LEDs you are driving.

Ideally, TEN, currently, eight.

- No caps' value shown.
- What is the output Voltage of the regulator?
- 45V input that means the diff V between input and out out is about 35V (40V TYP.) so when power is first applied, it will have about 45V Vdrops on the IC until te output cap is fully charged.
- Why use 32VAC as the supply?

Cap on input is 1000µf, Cap on output is 470µf (the source voltage is VERY noisy!)

Regulator output is 12V

32V AC Input, not 45. I am using 32 VAC as the input because I am retro-fitting these LEDs to light up an LCD display in my soldering iron, and the transformer in the Iron base only gives me 32VAC.

LED is current device so you will want to maintain constant current through it.

I am aware of this.

I.E. http://www.bristolwatch.com/sr/cc_led.htm
http://www.reuk.co.uk/wordpress/ligh...-led-lighting/

Thank you!

[EasyGoing1]

Re: Need help with LED Power Supply Design

wow.

well 32vac rectified would be as high as 46v that's too much for an lm317.

My scope showed (I think) 37 fully rectified, but when I cap it and load it, I'm pretty sure it went back to low 30's.

you also want ceramic caps between the input & ground and the output & ground close to the regulator.

All of my ceramic caps are in pico farads ... nothing meatier than that ... got some you wanna donate?

what's your led load??

Are you asking me for the measured R value of the LEDs when not in the circuit?

[EasyGoing1]

Re: Need help with LED Power Supply Design

DUDE! You are using a 25V filter capacitor ?!!! It's going to POP I see a 35V cap on the LM317 output.

Damnit, you're right! I missed that ... my 50V caps are lower farads but should do the job I'm sure.

The flickering is what I see in LED's that are overdriven or overheating, with broken bonding wires.

You know, now that I think about it ... I might have overdriven these when I was trying to see how bright I could get them ... I'll try a new batch properly choked and see if I get the same behavior.

THANK YOU!


Mike

[budm]

Re: Need help with LED Power Supply Design

My scope showed (I think) 37 fully rectified, but when I cap it and load it, I'm pretty sure it went back to low 30's.
All of my ceramic caps are in pico farads ... nothing meatier than that ... got some you wanna donate?
Are you asking me for the measured R value of the LEDs when not in the circuit?

32VAC RMS after rectification and filtering by the filter cap, the Voltage will be close to 45V due to the filter cap will be charged to peak Voltage. That input filter cap should be rated at least 50V or more since your AC line is not regulated so it can go higher.
The value of the resistors you use will not give you 12V, it will be less but if you get way lot more than calculated value then you need to check your wiring.
http://www.reuk.co.uk/wordpress/elec...ge-calculator/

Are the 10 White LEDs connected in series or parallel? 10 LED in series will give the total Vf around 30V at rated current of the LED.
You cannot check the resistance of the LED, it is not linear device.

[EasyGoing1]

Re: Need help with LED Power Supply Design

Are the 10 White LEDs connected in series or parallel? 10 LED in series will give the total Vf around 30V at rated current of the LED.

The LEDs are in parallel ...

You cannot check the resistance of the LED, it is not linear device.

That's why I asked, because I was unaware of a way to measure resistance of any diode short of looking at the datasheet, which I assume contains those values.

[budm]

Re: Need help with LED Power Supply Design

Can we info about that LED assembly? Are all 10 LEDs directly connected to each other in parallel without any balancing resistor for each LED?
Connecting LEDs directly in parallel is not good idea since Vf of the LEDs will not be the same.

[Curious.George]

Re: Need help with LED Power Supply Design

I am trying to finish a small project...

Why are you adding the bridge, filter caps and regulator?

Just arrange the LEDs in a series string and feed them with a half-wave rectified source behind a suitable ballast (resistor).

E.g., if you have a 32VAC source, your peak voltage will be about 45V at nominal line voltage. Give another 10% for high line and call it 50V; ~40 for low line.

Assume you want to pass ~20mA through the LED string and don't want to dissipate too much heat in your ballast (resistive or electronic).

45 - ~2*N = voltage across ballast (at nominal line -- assumes ~2V for Vf)
0.02 * (45 - 2*N) = power dissipated in ballast

So, with 10 LEDs in series, you're looking at 0.5W in the ballast -- whether that is a semiconductor/active device or a passive. (as Vf increases, Pd will decrease)

Stick a few "extra" LEDs in the chain and you can move some of that power (heat) out of the ballast. E.g., with 15 LEDs, you're at 300mW (the other 200mW has been "relocated" into the extra 5 LEDs).

Omitting the filters means your LEDs will "blink" at 60Hz. <shrug>

Omitting the regulator means intensity will vary with line voltage (do you really think you will notice??)

[budm]

Re: Need help with LED Power Supply Design

OP said the 10 LEDs are connected in parallel. I have a feeling they are not in parallel.

[eccerr0r]

Re: Need help with LED Power Supply Design

My opinion of the circuit in pictures:

[budm]

Re: Need help with LED Power Supply Design

Yep, all of that just to light up LEDs.

[EasyGoing1]

Re: Need help with LED Power Supply Design

Yep, all of that just to light up LEDs.

I changed out the LEDs with 5730's and ended up using this circuit which works like a champ EXCEPT that there is some very slight but noticeable flickering in the light ... I figured maybe another cap would help ....

[R_J]

Re: Need help with LED Power Supply Design

You should have a cap on the lm317 output, around 10-47uf should be ok, these regulators can go into oscillation without it.

[mariushm]

Re: Need help with LED Power Supply Design

(i have a horrible keyboard as my preferred one just died recently and new one didn't arrive yet , until i replace it i may miss some letters or not use uppercase, the keys are truly awful, sometimes they don't trigger even when i press the keys ... my bad typing is not meant as disrespect, i apologize)

The whole circuit is wrong.

First of all, leds are current driven devices, they only have a minimum voltage (forward voltage) from where they are fully on and start producing light. once the supplied voltage is above the forward voltage, you have to limit the current going through the leds otherwise they can blow up due to overheating.

in your circuit you're using the two resistors 250 ohm and 1.8 kohm to set the output voltage of the regulator. the formula used for that is:

VO = VREF (1 + R2/R1) + IADJ R2 where Vref = 1.25v and I adj is practically nothing so you can ignore that.

So basicaly in your case, Vout = 1.25 ( 1 + 1800/250) = ~ 10.25v DC.

your output voltage must be at least number of leds x forward voltage of each led.

For 10 white leds with an average forward voltage of 3v, I would configure the output voltage to around 32v (10 x 3v + ~ 5% tolerance)

you have the formula voltage = current x resistance so you can use this to determine the current going through the leds.

V = I x R => 32v (input voltage) - 10 x 3v (10 leds with forward voltage of 3v) = I x 260 ohm

so current is I = (32-30) / 260 = 0.0076 A or around 7.6 mA

if you want each led to be brighter you have to reduce the resistor value to allow more current to flow.
let's say you'll want 10mA through the leds.... in that case 0.01A = 2 / R => R = ~ 200 ohm , so you'd either go with two 100 ohm in series or a standard 210 ohm

the power dissipated by the resistor would be P = I x I x R where in your case current is 0.0076 A and your R = 260 ohm , so p = 0.0153 watts , so a simple 0.25w resistor would be suitable, but using a 0.5w resistor wouldn't be a bad idea.

Now, another thing that's important t keep in mind is that if you check the regulator datasheet, the LM317 requires a minimum load in order to be stable. Most datasheets say the average minimum load required is 3.5 mA and maximum is 10mA.

The datasheet recommends using a resistor in the 240 ohm range to adjust the output voltage because that causes the regulator to waste a few mA internally. I'm too lazy to do the math now but I suppose it's somewhere in the 5mA-ish range.

Basically, your configured current for the leds at around 7.5 mA is high enough that together with the internal current consumption of the regulator, the total consumption should always be higher than 10mA so the regulator should be stable.

However, if you think you're gonna reduce the current below let's say 5mA or you want to be absolutely sure the regulator will be stable, then you may want to change that 250 ohm resistor to a smaller value, like let's say 100 ohm and also change the other resistor to keep the output voltage where you want it.
With smaller resistor there, you'll allow a bit more current to be wasted inside the regulator.

Now... the regulator needs at least [output voltage + ~ 1.5v..2.5v ] at the input, in order to output a stable output voltage.
The amount of voltage above the output varies with the current. For only 10mA or so, probably just 1.5v over the output voltage is enough. For 1A+, you'll probably need to use the 2v number.
The minimum voltage also varies with heat, as the regulator heats up, the regulator may be happy with lower difference between input and output voltage but it's not good practice to use absolutely minimums ... hence it's safer to just use 2v when calculating things.

So if you want to output 32v to the leds, then you should make sure to have at least 32v +2v = 34v DC at the input and that's the job of that input capacitor.

You have a 32v AC transformer and you use a bridge rectifier to create Dc voltage ... this means the dc voltage will have a peak voltage of

Vdc pea = V ac x 1.414 - 2 x diode voltage drop.

for low currents like 10mA, the ac transformer will probably output more than 32v AC (probably +10% of that) and the diode voltage drop will probably be around 0.6v but let's assume you're making circuit for various currents and go with a more sane 1v per diode... in this case :

Vdc peak = 32 x 1.414 - 2v = ~ 43v

Note though that at low currents like 10mA like I said the transformer may output way more than 32v AC so your peak DC voltage could get very close to 50v, and that's why you should use input capacitors rated for 63v or higher.

now you now the peak dc voltage is 43v, you need at least 34v at the input of the regulator, and you know the current you want the leds to use (~5mA inside the regulator and the current through the leds, in your case 7.5mA , so let's say 15mA in total or 0.015A ... but just to allow for adjustments later, let's calculate for maximum 50mA of current or 0.05A )

So you can estimate the capacitor size with the formula :

ac frequency i assume is 60 hz for us:

C = current / [ 2 x ac frequency x ( Vdc peak - Vdc min) ] = 0.05A / 2 x 60 hz x (43-34) = 0.05 / 1080 = 4.6296296e-5 Farads, which I think works out to either 46.2 uF or 462 uF .... that's the minimum capacitor size you should use to guarantee voltage will be above 34v on the regulator.

keep in mind also that the regulator really won't be designed to handle such high voltage, there are versions of these regulators designed to work with up to 60v but they're more expensive.

There are LED driver chips designed for such purpose, which work much easier and better, here's some easy to use examples:

(up to 150mA configurable current) 0.77$ each AL5812 : https://www.digikey.com/product-deta...DKR-ND/3678111

https://cdn.badcaps-static.com/pdfs/...476041f3ab.pdf

1$ each but easier to solder even by hand , up to 70mA of current NUD4011 : https://www.digikey.com/product-deta...SCT-ND/2122046 (check datasheet, esp. pages 4 and ... https://cdn.badcaps-static.com/pdfs/...b18cf442b0.pdf )

more expensive but switching regulator design so less heat produced and can do up to 550mA current, you need an inductor and some resistors besides the chip but easy to find and datasheet explains well

2$ each ZXLD1356 ET5TA https://www.digikey.com/product-deta...5TR-ND/1986324 datasheet : https://cdn.badcaps-static.com/pdfs/...533b95ad90.pdf

[PeteS in CA]

Re: Need help with LED Power Supply Design

LEDs are current-operated devices. If parallel-connected, nothing forces the devices to share the current equally. An LM317 can be used as a constant current source as well as a voltage regulator.

[budm]

Re: Need help with LED Power Supply Design

I changed out the LEDs with 5730's and ended up using this circuit which works like a champ EXCEPT that there is some very slight but noticeable flickering in the light ... I figured maybe another cap would help ....

That is only 16mA of constant current source.
1.25Vref/80 Ohms = 16mA
And Yes, it needs to have filter cap on the output.

[EasyGoing1]

Re: Need help with LED Power Supply Design

That is only 16mA of constant current source.
1.25Vref/80 Ohms = 16mA
And Yes, it needs to have filter cap on the output.

The resistor I actually chose was not 80 ohms ... I just didn't change the details in the image, only the configuration.