How does what do what?

 

Searched VU meter on youtube. One result was

It looks nice.
So for LM3915, how many LEDs light up is according to to signal amplitude. You still need band filters to differentiate different frequency bands of the audio, but the mixer has EQ.

 

Here's a different way to light hundreds of LEDs easily

 

The schematic for the OctoWS2811 shows 100 ohm resistors in series with CMOS inputs. This is incorrect, and the resistors do nothing except make the circuit *more* susceptible to radiated noise. The schematic calls into question the integrity of the overall system design.

ak

 

That's 10 lm3915 per channel, or 100 leds.

I have build similar meters from 12F675 + shift registers (2xHC595), or 16 leds. Each transmission is less than 0.3ms. So if I were to expand that approach to 96 leds (12x HC595), I would need 2ms, with a refresh rate of 50hz.

Flicker free to a human eye.

It also allows the ability to pwm the brightness through the HC595s.

 

or simply use the lm3915 to adjust the gain in the transistor for only 10 specific levels of LED brightness, or less. Different combination of resistors can also be used to continue modify the logrithmic function between input voltage levels and LED steps or now about LED brightness levels.

 

The schematic for the OctoWS2811 shows 100 ohm resistors in series with CMOS inputs. This is incorrect, and the resistors do nothing except make the circuit *more* susceptible to radiated noise. The schematic calls into question the integrity of the overall system design.

ak

The resistors are explained here.
Seems to be able to run thousands of LEDs. I have only personally used it for 60, with which it works flawlessly

 

The resistors are explained here.

A resistor in series with a very low output impedance driver will indeed have the effect described in the scope shots. In fact, some CMOS bus drivers and transceivers have 25 ohm resistors built in. This is called source termination, and is one half of a complete line termination system. But the 74HCT245 is not a driver. In this circuit it is a receiver, and one with a near-infinite input impedance. A 100 ohm resistor in series with an input will not terminate the transmission line effects. A 100 ohm resistor to ground would, but it also would cut the signal amplitude in half. Given the speed of the data, it is clear that the misplaced resistors will have almost no effect on the signal reliability. This also is clear from the scope shots, where the transients are less than 1% of the pulse width.

ak

 

Can't find anything in the datasheet of the LM3915. At page 9 talks about current programming, but i can't figure out any adaptation.

A power darlington, or power MOSFET, or multiple devices in parallel can handle any current you want. I think you are wrong about not needing a lowpass filter. You basically are building a visual VU meter. There are chips for this that handle the log function for you. Most are designed to run a bar graph, which is constant brightness and variable size rather than what you have, constant size and variable brightness. Again, this is not a difficult adaptation. Search for the LM3915 datasheet.

Lowpass filter is not the correct term. What you actually want is an envelope detector, something than makes an analog voltage that tracks the overall loudness of the audio signal. Again, for the changes to be visible by humans the output has to have a pretty low bandwidth.

ak