Continue to Site

# [SOLVED]LM350 adjustable regulator system reduce Power LEDs voltage !!

Status
Not open for further replies.

#### jeolex

##### Member level 2
Hello my friends,

I am an academician in Department of Agriculture and I aim to make growth chambers for plants which have light, airconditioning (with computer fan) etc.. One of the important thing is light intensity of chambers. So I use 9 pcs, 3W power led for one chamber and adjust the light intensity with LM350 - potentiometer system (circle below). Leds run with 3-5V and max current is 700 mA.

I use 12 V 16A power supply. As LM350 system, Vout is approximately 11.34 V. It is ok for me and I can also adjust light intensity with potentiometer. Just problem that when I run power leds, I get 8.4 V instead of 11.34V. So this is reduce the max. light intensity. If I run power leds without LM350 system, I get 10 V and it is enough for me. You will see my circle at below. I need your helps.

Thanks..

I suspect your problem is the 'drop-out' voltage of the LM350. At high currents it needs at least 2.75V more at it's input pin than you can get out of it. Starting with 12V it means it could be normal for the maximum to be 12-2.75V = 9.25V and you could be losing more due to wiring resistance.

You should really add a series resistor in each column of LEDs to limit their current and also start with more than 12V at the input.

Brian.

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
Thank you for your clear reply. Best solution is approx. 15V power supply as your suggestion (2.75V more need).

Is that possible to get approx. max 7.6V from LM350 Vout by using 12V power supply. I mean that 2 series and 4 parallel leds are enough for me. So each parallel have to get max. aprrox. 7.6 V for life of leds. So when I adjust light intensity with potentiometer, I need to get MAX. 7.6V from 12V powersupply. Is it possible.

I hope my sentences are understandable.
Thanks..
Ugur

It depends on the voltage regulation of your power source.

Four strings of leds are going to draw more current than three strings of leds. The increased in current load may further drop your incoming supply voltage, or that may not be an issue.

Best to just try it and see.
The increase in current will mean the LM350 will get hotter so just be prepared for that.

And I agree with Brian, each string should have a small series resistor to equalize the current through each string.

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
I suspect you also use the fan to cool the LED heatsinks which may be drawing much more than rate 0.7Amp for each string.

caution

mismatched LEDs can burn out fast in parallel without say 0.1 equalization series resistors or wire resistance on each parallel string. Its called thermal runaway.

Do you have a multimeter? LED p/n's? Wire guage? AWG and length?
What is Vdc at source? At load?
What heatsink?

From the Voltage, I assume these are RED and that you aren't driving these at 16A

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
It depends on the voltage regulation of your power source.

Four strings of leds are going to draw more current than three strings of leds. The increased in current load may further drop your incoming supply voltage, or that may not be an issue.

Best to just try it and see.
The increase in current will mean the LM350 will get hotter so just be prepared for that.

And I agree with Brian, each string should have a small series resistor to equalize the current through each string.

Thank you, I just try to get solution without buy a new power supply I estimate it and my power supply (12V, 16A) run four growth chamber (each of them run with max 2.8A, totally 4 parallel). So I prefer finding a solution without change power supply. I will use resistor but what have to be resistors's ohm ? I also use cooler fan for LM350 in my lm350 system.

So I need a answer of this question : "How can I limit the Vout range of LM350 system maximum at 7.6V from 12V power supply while power leds are run".
The answer will be alternative solution of my problem, because I will try 2 series 4 parallel and see the heat of leds.

Last recipe is get a new powersupply

Thanks again

I suspect you also use the fan to cool the LED heatsinks which may be drawing much more than rate 0.7Amp for each string.

caution

mismatched LEDs can burn out fast in parallel without say 0.1 equalization series resistors or wire resistance on each parallel string. Its called thermal runaway.

Do you have a multimeter? LED p/n's? Wire guage? AWG and length?
What is Vdc at source? At load?
What heatsink?

From the Voltage, I assume these are RED and that you aren't driving these at 16A

I measured voltages of all strings with multimeter and get 8.4V values. Normally, voltage value between two terminal from LM350 is 11.3V. It seems that there is a 2.9V drop-out from LM350 and powerleds. I try to find solution without changing power supply, but I guess I will do it

Thanks.

- - - Updated - - -

Do you have a multimeter? LED p/n's? Wire guage? AWG and length?
What is Vdc at source? At load?
What heatsink?

I didn't know what you mean with AWG. I researched it. I just use jumper cables and bell wire. I guess wiring is important at high current circles ? Maybe my drop-out voltage associated with wire type. Am I right?

All components have a thermal resistance rated by deg C per Watt, and the parts should never get too hot enough to touch by proper heat sink and wire only slightly warm.

Your config if 3x3 or 3S3P for Series , Parallel .

With reg. you get 8.4V or 2.8Vper LED less than full power. This tells me your 3W LEDs are White. (Blue LED + red/yellow phosphor)

Without Reg. you say OK at 10V at LEDs. This means wire drops 2V at unknown current. Depending on LED specs. It may need 2.9 to 3.2 typ per LED, worse 3.5

This is your solution as long as not burning hot>75C. using phone wire as resistors. If high quality LEDs, you only need 2.9V per LED with 0.1V per LED drop on heavy wire. This would mean 4 LEDs in series or 11.6 V at LED and suitable heavy wire for 0.4V per string.

I suggest you measure current on 10A scale then design wire to supply 0.7A per string. With suitable heatsink.

This might mean adding more phone wire in parallel if you stick with 3LEDs in series. To achieve rated current and junction temperature one must use Ohm's law for resistance in both electrical and thermal. (Volts per amp and degrees per Watt) with correct choice of conductors (electrical and thermal)

in any case dont use LM350 with 12V in.
9W...err 27W is too small to consider dimming for plants.

try 100W LED power !! and research better solutions of RED , BLUE and White light affects on plants for different times on different plants with different rhythm cycles of light if you are in Plant Research.

- - - Updated - - -

What LEDs are you using? And heatsink?

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
Every replies give me lot of informations..

Yes you are right my power leds are white (I forgot to tell in all my replies) and max. forward voltage is 4V. Normally I was soldering them 4S3P, but LEDs didn't run full performance and light intensity was not enough. Then I measured it and get 8.4V from one parallel.

I use 125mm PVC (I covered inside of them with aluminium band) and it's height is approx. 33 cm. In full power (without LM350 system and with 3S3P LED design) I measured lux value 15cm far away from LEDs with Luxmeter and it gives 105000 lux and 30 cm range is 75000 lux. I estimate it based on **broken link removed** and consider it s fluorescent. These ppf values are enough to breeding plants (cotton, wheat etc.).

I am sorry to tell that I couldn't understand where I have to use phone wire and why ? I have already have a voltage deficiency and if I use phone wire, isn't that lost some voltage ? Can you give me little more information. You can see led and LM350 design at below, and also you can see drawing of design from my first reply.

Thanks.
Best regards..

There are a number of issues here but they can all be resolved easily.

1. The LEDs are not wired optimaly. A LED is a constant voltage device, beyond it's Vf it draws more current if you let it. This means you are effectively limiting the voltage regardless of what the LM350 is trying to do. Rewire the LEDs so there are only two in series and add a resistor of say 1 Ohm in series with each pair. That leaves one LED left over so if possible add another so you have 10 wired as 5 x (2 LEDs + resistor) in parallel. This will overcome the top voltage limit you are seeing.

2. The LM350 will run extemely hot with that sized heat sink. It dissipates (voltage across it * current through it) Watts. You have to include the 2.75V drop in the calculation. I suspect you should be looking at a substantially bigger finned heat sink with forced cooling to withstand continuous use. There is a high risk of melting that proto board! A much better solution would be to use PWM dimming which would run cold and have almost no voltage loss from the power supply. It is easy to do with a microcontroller or at a push an NE555 and MOSFET. Only a few inexpensive components are needed.

Brian.

There are a number of issues here but they can all be resolved easily.

1. The LEDs are not wired optimaly. A LED is a constant voltage device, beyond it's Vf it draws more current if you let it. This means you are effectively limiting the voltage regardless of what the LM350 is trying to do. Rewire the LEDs so there are only two in series and add a resistor of say 1 Ohm in series with each pair. That leaves one LED left over so if possible add another so you have 10 wired as 5 x (2 LEDs + resistor) in parallel. This will overcome the top voltage limit you are seeing.

I tried to make as you explain. I used 4 parallel and 2 series with 1 ohm resistor. You can see in photo below.

All series get 4.87V after 1 ohm resistor. With resistor 9.85V coming from LM350 system. So 4.8V is not enough for 2 series 3W LEDs. Which value I have to choose resistor at least get 7V for each parallel.

Something is wrong. If the clips are connected to the DVM it tells me 4.98V is dropped across the 1 Ohm resistor so the LEDS must be passing (I=V/R) 4.98/1 Amps. Almost 5 Amps each and the board is drawing 20A total! Also the resistors would be dissipating 5W each and they look like 0.25W or 0.5W types so there would be lots of smoke!

The voltage across the LEDs should be fairly constant at 2 x Vf whatever the current is because they are constant voltage devices. You don't want to put 7V across them as that would instantly destroy them. With a conventional filament lamp the brightness is set by the voltage you supply across them, with LEDs the voltage stays steady and you increase the brightness by letting them pass more current. The 4.87 volts is somewhere close to what I would expect, it's how you get 9.85 without fillling the room with smoke that eludes me.

Brian.

Something is wrong. If the clips are connected to the DVM it tells me 4.98V is dropped across the 1 Ohm resistor so the LEDS must be passing (I=V/R) 4.98/1 Amps. Almost 5 Amps each and the board is drawing 20A total! Also the resistors would be dissipating 5W each and they look like 0.25W or 0.5W types so there would be lots of smoke!

The voltage across the LEDs should be fairly constant at 2 x Vf whatever the current is because they are constant voltage devices. You don't want to put 7V across them as that would instantly destroy them. With a conventional filament lamp the brightness is set by the voltage you supply across them, with LEDs the voltage stays steady and you increase the brightness by letting them pass more current. The 4.87 volts is somewhere close to what I would expect, it's how you get 9.85 without fillling the room with smoke that eludes me.

Brian.

You were right. I check it again and I was used to 100K ohm I solder 0 ohm resistor and 3V from each led. Light intensity (68000 lux) almost enough for me but my aim is 105000 lux in which I have done before with 3S3P design. You already said that LM350 will get warmer and yes It has warmed. Current is so high, I guess it is approx. 4x0.7=2.8A and LM350 max current 3A. So it is not suitable, because I will run this growth system 18 hours in a day. I get 6V from LM350 and it is also not enough for my 12V fan.

I have to make a design at least 3 parallel for reducing current which across from LM350 and need to keep voltage in series of three power led. Probably I need a new power supply. I wonder that is it possible to combine 2 pcs 12V power supplies and get one Vout as 24V ?

If you used 0 Ohms (a short circuit) you are lucky the LEDs didn't burn out. I would not recommend you use a value less than 1 Ohm.

I suggest stepping back and re-evaluating the design. Start by looking at the maximum continuous current rating of the LEDs, taking into account the heat they will produce. Then decide how many LEDs you need to achieve the light intensity you want. The dimming circuit is the least of your problems but consider PWM dimming instead of using a linear regulator for efficiency reasons. In one of my applications I dim 20 LEDs similar to yours using a TO-93 packaged transistor with no heat sink and it runs cold! TO-93 is the small half round plastic package you often see on small transistors.

I would not use two power supplies to get 24V, the LEDs need less than 3V to operate at full brightness so to move from 12V to 24V is not going to help. All you will do, apart from increasing the size and cost, is turn more electrical energy into heat. You must understand how LEDs work, forget any notion that more voltage gives more light. You can't 'force feed' them voltage, you just have to make sure they can draw as much current as you need and no more than their maximum rating.

Brian.

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
After your reply, I wired a multimeter without resistor to one parallel (which have 2 power led on string) as serie and get 0.22A value steadily. Isn't it so low for 3W power led ? I have already measure voltage and each led get 3V.

So is there a any PWM dimming circle which is suitable for my project ? I hope it is easy to make it

If you dont know or say part number of LEDs, do we guess?

From datasheet are important design parameters for better heatsink pads and Vf vs I, but looking at your board, I suggest heatsink resistance is too high.

LM350 is not suitable here for more than 9V out and needs large heatsink with fan for >6W loss.3V*2A. DO NOT USE

I know these are 8mm flat leaded Lumex parts who make UV in same package. https://www.lumex.com/sml-lxl8047uvc but if yours are White https://www.lumex.com/sml-lxl705uwc
UV LED is 700mA but only with perfect heatsink, while white is 500mA max so phosphor doesnt burn up.!!

direct to 4S3P with separate feed ( incl. low R series) to
But heavy gnd wire for 0V drop and better thermal conductance is advised unless using fan close across PCB.

but each string needs series R approx equal to LED Effective Series Resistance (ESR) which will reduce to 1/3 Ohm at 3Watts per LED

This can done with phone (bell) wire to each string of 4S as a low R resistor but use heavy wire so no ground shift at 3A to LED matrix (<<0.1V)

All white LEDs begin to shine at 10% max around 2.8V to 2.85V then rise to Vf@Irated with a wide tolerance unless matched.

The LEDs are poor quality for ESR and thermal resistance so datasheet indicates only 500mA max and Vf= 2.95~3.85 @350mA which is what you should use for your layout with poor thermal resistance. ( can you imagine if your CPU has this for a heatsink? )

i would guess 3.5V average per LED and run all 9 or 12 in a higher voltage that is just above the diode voltage string so a small wire resistance will provide 0.5V drop or less at a suitable hotspot max. of 70'C.

thus consider laptop charger.

if 19.5V and if Vf=3.5V then 5S= 17.5V + If*2V drop for each string of 5 using >1W resistors or long phone wire with measured R value.

if laptop charger is adjustable (universal) then you can dim and adjust current with wire R for Ohms law on drop voltage.

- - - Updated - - -

Your best options are;
Get better , more powerful , cheaper Toshiba or Cree LEDs properly heatsinked

Run when it was Ok at 10V

Improve board heatsink with Alum clad PCB not mickey mouse veroboard.

- - - Updated - - -

This is how I do it with 12.0 V supply with 4Sx 3W LEDs with 5 or more strings on 12V bus, knowing exact Vf at 3W per chip and proper wire feed and heatsink (MCPCB).

https://www.edaboard.com/blog/1946/

but in this application 3W per LED was way too bright, so I chose 1.5W/LED using 18LEDs on 30W 12V supply with NO EXTeRNAL regulator at 50% max rated ledpower. Full power would need heavier supply, AWG and also need a bigger PCB area for 12W array.

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
I'm guessing a single dedicated Al clad PCB is probably too expensive and difficult to solder, based on the present style of construction.

You could try it like this:
Start with a sheet of single or double sided copper clad board, the kind that PCBs can be made from. Using a needle file, carefully make four, 1mm wide cuts through the copper horizontally and three cuts vertically so it has 12 isolated copper squares on it. Place a dot of heat sink compound under each LED and position them flat to the PCB so the cut is *JUST* inside the LED solder pad. The intention is to keep the LEDs as close to the center of the copper squares as possible while still being able to solder one pad to the next square. Across the 4th cut, wire the 1 Ohm resistors. That will keep your 3 x 3 LED matrix and also mount the resistors. You can then link the first and last pads of each column to make the power connections.

PWM is easiest with an MCU, can you program a PIC, for example the PIC12F683?

Brian.

Your power supply is 12v/16 amp with which you can do almost anything in electronics.
From what I have read , though we use constant voltage(regulated) to drive LED , LEDs are current driven.
There are usually the ones - 3.6v, 350ma, 700ma , 1050 ma (I think the 10watt ) etc.
1. if you still want it to be voltage driven, and if all LEDs have the same characteristics , you can use one of this
**broken link removed**
connect the boost converter input to 12v , 16 amp supply adjust the voltage to some around 10 or 20. connect all 9 leds in series and connect both ends to the output of boost converter. Then slowly increase the voltage (you can see the led lighting up), increasing the intensity as the voltage is increased and check the voltage is sufficient by using a voltmeter and stop (if it is a 3v led and 9 in series will be 27 volts required. adjust to 26 or 28 v without damaging the led.) also you can connect the 10amp mode of digital voltmeter is series to check current consumed while voltage is increased and stop when it is around 350ma (as per specs).

2. LM350 along with brothers LM 317 and LM338 can be used as a current source - especially for lighting LEDs.
Two links below may be exactly what you are looking for.

https://www.onsemi.com/pub_link/Collateral/AND8109-D.PDF
https://diyaudioprojects.com/Technical/Current-Regulator/

I haven't tried it , hope it may work for you.

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
From matching your picture and description , these LEDs are now rated only at 1Watt which means 1 Ohm +\-50% for ESR

Pls confirm!,

https://www.lumex.com/content/files/ProductAttachment/SML-LXL705UWC.pdf

If you measured 700mA per string before, they must be either bluish UV or extremely hot or very well heat sunk with fans.

Dont underestimate your thermal resistance from case to sink. Consider an old surplus CPU heatsink.

All power SMT LEDs must be thermally attached very well to substrate. below shows using many thermal feedtrus in centre with very flat surface! otherwise thermal resistance is very poor.

Reconsider doing it right the first time.

A better regulator than the lossy LM350 is called a Buck regulator is common or attach a thermistor to LED sink and regulate case temp with PWM using a MOSFET and hysteretic schmitt inverter clock with thermistor biasing duty cycle or shunting a current sense R that provides the minimum current limit. Adding an inductor smoothens the current.

Sounds hard but easy for some bright volunteer here with more time.

but the simplest method is what I did .. 12.0V to 4S strings with feed wire <0.1V drop.

jeolex

### jeolex

Points: 2
Helpful Answer Positive Rating
I'm guessing a single dedicated Al clad PCB is probably too expensive and difficult to solder, based on the present style of construction.

You could try it like this:
Start with a sheet of single or double sided copper clad board, the kind that PCBs can be made from. Using a needle file, carefully make four, 1mm wide cuts through the copper horizontally and three cuts vertically so it has 12 isolated copper squares on it. Place a dot of heat sink compound under each LED and position them flat to the PCB so the cut is *JUST* inside the LED solder pad. The intention is to keep the LEDs as close to the center of the copper squares as possible while still being able to solder one pad to the next square. Across the 4th cut, wire the 1 Ohm resistors. That will keep your 3 x 3 LED matrix and also mount the resistors. You can then link the first and last pads of each column to make the power connections.

PWM is easiest with an MCU, can you program a PIC, for example the PIC12F683?

Brian.

I didn't try PIC but I made some little project with Arduino. So there are a lot of informations about controlling LEDs with PWM by arduino but first step I need to control it as analog. After I have done it I will combine with Arduino and also measure weight, air temperature and humidity etc.. I can adjust light intensity with potentiometer without heat mosfet or etc., after I hope I can make PCB for power leds and put a heat sink which is enough for cooling, I hope

From matching your picture and description , these LEDs are now rated only at 1Watt which means 1 Ohm +\-50% for ESR

Pls confirm!,

https://www.lumex.com/content/files/ProductAttachment/SML-LXL705UWC.pdf

If you measured 700mA per string before, they must be either bluish UV or extremely hot or very well heat sunk with fans.

Dont underestimate your thermal resistance from case to sink. Consider an old surplus CPU heatsink.

All power SMT LEDs must be thermally attached very well to substrate. below shows using many thermal feedtrus in centre with very flat surface! otherwise thermal resistance is very poor.
View attachment 119151
Reconsider doing it right the first time.

A better regulator than the lossy LM350 is called a Buck regulator is common or attach a thermistor to LED sink and regulate case temp with PWM using a MOSFET and hysteretic schmitt inverter clock with thermistor biasing duty cycle or shunting a current sense R that provides the minimum current limit. Adding an inductor smoothens the current.

Sounds hard but easy for some bright volunteer here with more time.

but the simplest method is what I did .. 12.0V to 4S strings with feed wire <0.1V drop.

My leds are 3W and max current is 700 mA. I research PWM and potentiometer controller system to make my own. I didn't have enough time because of working in University also in field. Firstly I try to make a mosfet-potentiometer controller and will see what happens. I will inform you my friends I hope it will be reach success..

Thanks all of you

Then your LEDs are of unknown type because this 8mm type is unable to dissipate more than 1W safely

Status
Not open for further replies.