Hello im making a led vu meter with a lm 3914 IC. Here is the datasheet http://www.ti.com/lit/ds/symlink/lm3914.pdf.
Im doing the 1st schematic in the datasheet. Im using the 2 formulas right below the schematic for the calibration and LED brightness
My R1 is 660 making Iled=18mA (2x330 in series, because i dont have a 660 resistor) and R2 is 82 ohms (330 220 220 in parallel). The range of the leds is about 1.6 volts. Im powering it with a 9v wall thing
Im planning to wire it between my computer and my 2.1 speakers. Today i figured out the resistor values and tested it with the pc cranked to the max and all leds blinked yey. so far so good but as i was watching the leds blink, it must have been working for 2-3 minutes and when i turned it off the chip was really hot. I checked for shorts, bad wiring and whatever. nothing found. Then i read in the datasheet, that Vled should be < V+ and in my case its the same. I checked the internet for other schematics and all of them were powering the leds from the source that is powering the chip without any dividers resistors or whatsoever. Can someone hint me what im doing wrong? I even tried adding the 10uF cap like the dotted line says, it didnt help.
Power dissipation, especially in bar mode should be given
consideration. For example, with a 5V supply and all LEDs
programmed to 20mA the driver will dissipate over 600mW.
In this case a 7.5? resistor in series with the LED supply will
cut device heating in half. The negative end of the resistor
should be bypassed with a 2.2?F solid tantalum capacitor to
pin 2 of the LM3914.
With a 9v wall thingy, the chip has to get rid of another 4 volts at 200 mA ( 10 LEDs on ) another 800 mW,
So the chip is trying to get rid of 1.4 watts in heat ( max )
Rather use a 6 volt wall thingy, and possibly also the 7.5? ( or 6.8 ? more common ) feeeding the LEDs.
And of course the other component missing ( for this group ) is the Arduino 
spoocter:
My R1 is 660 making Iled=18mA (2x330 in series, because i dont have a 660 resistor) and R2 is 82 ohms (330 220 220 in parallel). The range of the leds is about 1.6 volts. Im powering it with a 9v wall thing
So you're asking the chip to convert the other 7.4 V into heat....and you're surprised when it gets hot?
Get a lower voltage wall thing or put some suitable resistors between the LEDs and the chip. Let the resistors get hot instead (nb. use high wattage resistors).
with a lm 3914
You can probably do it a lot more efficiently with your arduino.
As you may know power (mW) is heat.
A couple more bits from the data sheet:
Power Dissipation (Note 6)
Molded DIP (N) 1365 mWNote 6: The maximum junction temperature of the LM3914 is 100°C. Devices must be derated for operation at elevated temperatures. Junction to ambient thermal
resistance is 55°C/W for the molded DIP (N package).
Gluing a [u]DIP heatsink[/u] would help too.
You can probably do it a lot more efficiently with your arduino.
I'm not sure what you mean by efficient... A microcontroller is total overkill when a single-simple cheaper chip will do the job.
Now, for the VU Meter "effect" I'm working on right now... It reverses, inverts (lights go off with loudness), it has bar-graph mode, dot-mode, and dots-mode (with a random variable number of "dots"). And, it has does several other non-meter audio-driven effects. That design "needs" a microcontroller. I actually started that project a couple of years ago with an LM3914 and a bunch of other logic and shift registers, etc... It got really "messy", and I decided to "go programmable" and do all of the hard stuff in software.