tlc5940 and led strips

Hello

i want to do the following. put 4 led strips behind my tv and 1 led strip under my tv table for lighting.

the led strips behind the tv would be 4 times 50 cm of this type: http://www.produktinfo.conrad.com/datenblaetter/175000-199999/182002-da-01-ml-LED_STREIFEN_150LEDS_5M_12V_RGB_CA_de_en.pdf
the other led strip would be: http://www.produktinfo.conrad.com/datenblaetter/150000-174999/154988-da-01-en-LED_LIGHT_STARR_RGB_27_LEDS_12V_DC_8_8W.pdf
both are 12 volts
after some calculations i came up for 50 cm: ~100mA per color(2800mA/5(meters)/2(to get 1/2meter)/3(per color)). the other one would be 180mA per color.
so driving them all from one tlc5940 would be impossible because of the differences in current (and 180mA would be to much for the ic)

after playing around with the power dissipation equation i came up that driving 100mA and 12 volt (voltage drop is sadly not in the datasheets) and having all colors on would result in to much heat for the ic.
am i correct here ?

so my idea was using tlc5940 and one transistor with a resistor on each ouptut. using 5v to base of transistor an 12v trough collector-emittor.
after reading around this sounds like its been done before and would work.

but this increases the number of parts. eg. 1 tlc + 15 transistors + 15 resistors(led) + 15 resistors(base)
so my question is, is there another way ? or another ic ? or any other input on this setup ?

i want to double check this before ordering parts i would not need in the end.

thanks

I thought the TLC5940 could drive (sink) up to 120mA per channel with VCC of 5V.

IOLC Constant output current, OUT0 to OUT15, VCC > 3.6V: 120 mA MAX

Might need to put a good heatsink on the part to do that.
There are 16 outputs - you could also break your strips up some more to spread the current load across more pins so no one output has to work as hard.

thanks for the reply
Yes it can sink up to 120mA but it seems it can do this only if not all channels are on at the same time.

After playing a little bit more with the heat equation, i realized that output voltage matters alot with this chip. But since i dont know the output voltage (since its njot in the datasheets) i just played around.
And figured if i only have 15 channels on an use 20mA per channel (which is for 10 cm strips) and assume of the 12 volts only 6 remain, it would work.
Or the heatsink approach.

but i dont know about heatsinks, are there special heatsinks for ics ? like there are for normal cpus ? or is there more like one heatsink matching one ic ?

if all fails, would the transistor approach work ?

A heat sink can only help a bit. There are calculations where you use a hypercritical infinite heat sink and that is still not big enough.
Have a read of this:-
http://www.thebox.myzen.co.uk/Tutorial/Power.html
And the examples page.
Yes they make clip on heatsinks for ICs if you want to spend the money but they can only help so much.

thanks for the link

it was very interesting and gave me the idea to asjust my power supply.
i figured even if the leds are 12v and i dont know the voltage drop (which i could find out by simply measuring after i get the led strip) i could adjust the power supply to kets say 1v to 2v above the cumulative voltage drop which would result in less voltage to sink by the ic and therefore less heat.
after some calculations with 2v and 20mA per channel i would get far under the power dissipation rating of the ic.

then i would just have to worry about much more wires from the strips but thats less a hassle then transistors and resistors

thanks for the input.

If the LED strips are 12V then that is what they need to be fed with because the current limiting circuit is already on board. The current setting of the TLC5940 would not affect this unless you wanted it to run at lower than the natural current for 12V, so it is that you want to measure not voltage drop. Normally these are 20mA per section of three LEDs.

thanks.

with 12 v and 20 mA and 15 channels i'm way over the power dissipation ratings of the ic
after some calculations i determined that with 20 mA and 15 outputs the voltage would need to be under 7 volt to be 100mW under the max power dissipation.

so its eather other led strip, getting the led strip and measuring, testing or the transistor approach i think.

power dissipation of the chips sounds strange to me, but it seems it is designed to not have everything on all the time.

Are you sure of these calculations? The volts drop is not 12V as I said because you are not using the constant current capabilities of the chip. The volts drop they want in the calculations is the difference between the LED's supply minus the forward volts drop. As you have no external volts drop you only have to consider the internal one.

Well i have done the example calculation from here: Google Code Archive - Long-term storage for Google Code Project Hosting.
and it came out correct.

maybe i didn't describe what i meant clearly enough.
Since in the datasheet only 12v is marked, i dont know how much voltage will sink into the ic for 3 leds (10 cm strip).
so i tried some values and came to the conclusion that if 5 volts goes into the leds, the ic has to sink 7 volts which would be just under the power dissipation limit.
but since i'm only guessing here, its theporetical. maybe i should just order the strips and the ics and measure....

i attached a screenshot of my excel sheet with some tested values.

came to the conclusion that if 5 volts goes into the leds, the ic has to sink 7 volts which would be just under the power dissipation limit.

You don't sink voltage you sink current. These LED strips have series resistors in them that will drop the voltage when current is drawn. I can't see it dissipating very much power at all.

thanks

yes sink voltage was the wrong term. i meant the voltage that goes into the ic after the forward voltage drop of the leds.
because i need to put that in the power dissipation equation.

i decided to buy the ics and the strips and try.

will post my results but this could take 2 weeks

thanks for the input on this matter

I think the problem is that data sheet assumes you are using the chip to control the current and so the FETs in them are operating in a linear mode. What you want to do is to operate those FETs in a switching mode. The heat generated then is just the current through the on resistance of the FET. This is tiny compared with the voltage drop restricting the current so it is not in the equation. So set the chips current control to twice what you strip takes. Then turn one on fully and measure the voltage on the output, preferably with a scope. Use this voltage in that power calculation. This will be a very small voltage.

thanks

Hello

so i got the strip and i cut an 10 cm strip of it and did some testing. but the readings somehow confuse my understanding of how this should work :frowning:

if i measure my power source directly without anything hooked up, i get 12.30 volt. which is ok since its a 12 volt power source
if i connect the common anode and the blue cathode directly to the power source (without resistors or anything) i get 12.16 volts on the blue line. which seems high. shouldn't leds have a higher voltage drop than 14 mV ?

i then used a transistor setup. common anode to the power source, 3 transistors (one for each cathode ), no resistors and cycled the colors with the arduino and i only measured the blue line again.
then i get:
if blue is off, 4.20 volts
if blue is on without any other colors 12.12 volt
all colors on 11.9 volts

this confuses me

has anybody any advice on why this is or if i am doing something wrong ?

thanks

EDIT: i always measure between anode and cathode (holding the measuring wires directly on the anode and cathode wires)

Ok i discussed this with a friend and the result was that i didn't understand what i was measuring more or less.

but now that i understand what i'm measuring i am still not sure what to insert in the power dissipation equation for voltage. i think i will need more testing nonetheless.

CrossRoads:
I thought the TLC5940 could drive (sink) up to 120mA per channel with VCC of 5V.

Only if you're very careful and leave less than 0.8V per channel for the chip to regulate/dissipate.

(ie. use an external resistor to drop the incoming voltage down to under 0.8V, let the chip do the rest...)

0.8V * 120mA * 16 channels = 1.5W dissipation - about as much as the chip can safely handle without adding fancy cooling.

tulpix:
Ok i discussed this with a friend and the result was that i didn't understand what i was measuring more or less.

but now that i understand what i'm measuring i am still not sure what to insert in the power dissipation equation for voltage. i think i will need more testing nonetheless.

Ok now you understand what you are measuring what do you get?
That is with the anode to 12V, blue cathode to TLC5940 output set to full on, what voltage is on that output? Make sure the TLCs current limit resistors set for twice the current you are drawing. For a better measure then use a scope because with the output full omit pulses off for one click cycle and that will artificially raise the on voltage. You should see a small voltage.

hello

sorry that this reply took so long, was swamped with work and needed some time to familiarize with the chip.

i did some testing. first with a 22k Ohm resistor (had no closer value to 2k) theoretically letting 17,7 mA through and second with a 470 Ohm resistor one letting potentially 83 mA trough.
then measured Volts and Ampere

for a 10 cm strip (3 leds) i get between led strip anode an red cathode:

  1. 11,68V (directly attached to the anode and the cathode) 18,9mA
  2. 12,05V (same measuring as above) 18,24 mA

directly connecting anode to +12V and cathode to ground gives 18,25 mA and 12,28 V
power source gives me 12,30V

so am i right that only ((12,30 - 12,05) = 0,25V goes into the tlc chip which by my calculations would mean i am on the safe side for power dissipation?
my calculation is based on: 20 mA (into the chip) 0,25 V(into the chip) and 12 outputs used (4 led strips per chip)

to make clearer what i measured i attached a picture

thanks

tulpix:
hello

sorry that this reply took so long, was swamped with work and needed some time to familiarize with the chip.

i did some testing. first with a 22k Ohm resistor (had no closer value to 2k) theoretically letting 17,7 mA through and second with a 470 Ohm resistor one letting potentially 83 mA trough.
then measured Volts and Ampere

for a 10 cm strip (3 leds) i get between led strip anode an red cathode:

  1. 11,68V (directly attached to the anode and the cathode) 18,9mA
  2. 12,05V (same measuring as above) 18,24 mA

Those numbers don't look right. Normally more volts = more amps.

tulpix:
so am i right that only ((12,30 - 12,05) = 0,25V goes into the tlc chip which by my calculations would mean i am on the safe side for power dissipation?
my calculation is based on: 20 mA (into the chip) 0,25 V(into the chip) and 12 outputs used (4 led strips per chip)

That number, OTOH, does sound about right. The strips are designed for 12V, your power supply is putting out 12.3V... 0.25 is pretty close to 0.3.

0.25V for the chip to dissipate is plenty of margin. You should be fine connecting a string of 15 LEDs to each pin (ie. 5 groups of 3 LEDs, 20mA per group = 100mA).

Most of strips are voltage based, not current based, I mean, they have resistor built-in to limit the current. TLC5940 is a constant current IC so I am not sure you really need it, if you want to control color of the strip, just use PWM to drive three MOSFETs [edit] for each strip. Your power supply must be able to supply enough power, of course, otherwise, voltage at power supply will drop.

Just a thought.