Starting an LM3915 LED VU Meter Project and have many questions.

Thread Starter

Chardi

Joined Oct 3, 2015
28
Hey all,

I just started planning for the LED VU Meter project I've been wanting to do for a while now. However, I have many questions as I'm stuck in the design of the circuit while trying to understand how the IC works. I don't want to blindly go in and copy circuits without understanding the theory of the chip.

  1. If I have LED which consists of various forward voltage values (e.g. yellow = 1.8-2.2V & pink = 3.0-3.2V), do I need to throw in resistors and have the LED voltage at 5V? This way it'll drop the voltage to the required value instead of burning the LED.
  2. Can anyone explain me how the voltage reference implemented in the IC work? I read the information in the datasheet in vain.
I've searched the web looking for information about this project, though I've seen many people experiencing issues with the last LEDs not lighting up. Is this problem connected with the voltage reference?

Last question (for now): I've also seen a lot of people throwing in a capacitor in their circuit. What are they for (other than getting rid of oscillations from the power supply)? Can they be used to make the LEDs "dance" smoother?

Thanks in advance for any advice/help!
P.S. I have attached a picture of the LEDs' specifications which I have bought a while ago (which I plan on using for this project).
 

Attachments

#12

Joined Nov 30, 2010
18,224
1)V LED must be kept below 7V or dropping resistor should be used to limit IC power dissipation.
2) No.
3) We can't guess what, "many people" had difficulty with in getting the LEDs to light up.
4) Capacitor C1 is required if leads to the LED supply are 6″ or longer. Its purpose is to keep the supply voltage to the chip stable during switching moments.
 

Yangui

Joined Oct 3, 2015
8
It's been a long time since I played with an LM3914/5
1)
In the LM3915 data sheet , there is a table which lists out the output voltages for each segment based on a 10 Volt VCC reference that would be a starting point I would look into.
2) Again referencing the data sheet there is a note that says the capacitor is required if the leads to the LED's are 6" or longer.
 

GopherT

Joined Nov 23, 2012
8,009
The LM3914 has a linear response, the LM3915 has a log response and the LM3916 is set up as a VU meter. You should use the LM3916.

EDIT: Corrected error - thanks @#12
 
Last edited:

Thread Starter

Chardi

Joined Oct 3, 2015
28
It's been a long time since I played with an LM3914/5
1) In the LM3915 data sheet , there is a table which lists out the output voltages for each segment based on a 10 Volt VCC reference that would be a starting point I would look into.
2) Again referencing the data sheet there is a note that says the capacitor is required if the leads to the LED's are 6" or longer.
Thank you for the reply!

1) I assume you're talking about the table titled as "Threshold Voltage"?

2) Okay I got it! It's just that I looked into other people's schematics and they always threw a capacitor somewhere without a real explanation.
 

Thread Starter

Chardi

Joined Oct 3, 2015
28
The LM3914 has a linear response, the LM3915 has a log response and the LM3916 is set up as a VU meter. You should use the LM3916.

EDIT: Corrected error - thanks @#12
From what I read about the LM3916, it is more sensitive as the scales are closer together right? If pretty much everything else is identical to the LM3915 (from what I see with a quick glance to its datasheet), the LEDs would be more responsive to the signal input comparatively to the LM3015.

I'll do a little studying about the LM3916 as it seems that you guys recommend it over the LM3015!
 

Thread Starter

Chardi

Joined Oct 3, 2015
28
Hi everyone! I've been working on this project and I have an update (and more issues yay!)!
  1. Though the circuit works, when I tested it with a frequency generator on DOT Mode, the top LEDs (5 last approx.) tend to flash a little bit as if there was a tiny current going through as the lower LEDs are gradually climbing up. I've tried different resistor combinations (including the use of trim.pots) for pin 6-7 and 8 with the help of the I(LED) and V(REF) formula given in the datasheet in vain. Any ideas?
  2. In addition, I've built a simple amplifying circuit with the help of the LM386 (gain 200 from the datasheet), but when I connect the output of the 386 to the signal input of the 3915, more than half of the LEDs remain lit and the others are slightly flashing. Me and my colleagues later found out that it is most likely due to noise (as seen in the oscilloscope). Anyone know a good way to filter out the noise? I am doing research on this too, but you may have some ideas I've missed.
Here's the Diptrace schematic of what I have on breadboard (working) and the LM386 circuit I tried out:

Schematic.png

LM386.PNG
 

GopherT

Joined Nov 23, 2012
8,009
@Chardi

The easiest way to get rid of the noise, is to use an op amp of reasonable quality instead of that jelly bean amplifier.

The LM3915 does not need any power of the lm386 (capable of driving 1W), just the milliamp power of an OP amp is enough. A TL072, NE5532 or NE5534 or countless other low noise op amps are available. A rail-to-rail output op amp will be recommended by someone in a few minutes. I'm sure a good recommendation is coming soon.
 

Thread Starter

Chardi

Joined Oct 3, 2015
28
@GopherT

Thanks for the examples of low noise op amps! I didn't know different op amps would make a significant difference in terms of noise. If I understand well (with the 5 minutes of quick googling), rail-to-rail op amps are designed to take low voltages while outputting the whole range without clipping right?
 

GopherT

Joined Nov 23, 2012
8,009
@GopherT

Thanks for the examples of low noise op amps! I didn't know different op amps would make a significant difference in terms of noise. If I understand well (with the 5 minutes of quick googling), rail-to-rail op amps are designed to take low voltages while outputting the whole range without clipping right?
Right, I hope someone recommends a good logic level rail-to-rail op amp for you. I don't have one off the top of my head.
 

GopherT

Joined Nov 23, 2012
8,009
Ok, now that I think about it, the LM386 is so bad, that almost any Op Amp will give you low enough noise to satisfy your needs. The "go to" rail to rail op amp is the LM324. It is a quad op amp. The LM224 is the same with two op amps per chip.

Almost any rail to rail op amp will do it for you.
 

Thread Starter

Chardi

Joined Oct 3, 2015
28
@GopherT

Wow! That's exactly what I was wondering (whether the LM324 would work or not) as I've got plenty laying around! I'll give it a shot tomorrow since I left the breadboarded circuit at school and will update as soon as it's tested :)

Thanks a bunch for your help though!
 

Thread Starter

Chardi

Joined Oct 3, 2015
28
@GopherT

/*Update on my amplifying circuit:

I switched to the LM324 and calculated the Rf and Ri values based on my input signal measurements done on the oscilloscope (approx. 5mV peak to peak).
I am using 10k for Ri and 100k for Rf, making an amplification of 10.*/
LM324.PNG

Right now everything is good (noise is gone thanks to GopherT), however the LEDs are flashing too fast to my liking. Instead of going relatively smoothly up and down, it quickly (almost instantly) flashes.

In addition, the LEDs aren't lighting up as bright as I want them to...Is this due to the amplified signal not having enough output current? It shouldn't be since the current through the LEDs are done by the LM3915 (and my I(LED) is somewhere around 1mA which is more than enough to be bright).
 

GopherT

Joined Nov 23, 2012
8,009
@GopherT

/*Update on my amplifying circuit:

I switched to the LM324 and calculated the Rf and Ri values based on my input signal measurements done on the oscilloscope (approx. 5mV peak to peak).
I am using 10k for Ri and 100k for Rf, making an amplification of 10.*/
View attachment 93285

Right now everything is good (noise is gone thanks to GopherT), however the LEDs are flashing too fast to my liking. Instead of going relatively smoothly up and down, it quickly (almost instantly) flashes.

In addition, the LEDs aren't lighting up as bright as I want them to...Is this due to the amplified signal not having enough output current? It shouldn't be since the current through the LEDs are done by the LM3915 (and my I(LED) is somewhere around 1mA which is more than enough to be bright).
1 mA seems like enough to be bright when it is on constantly. Short pulses tend to appear dim to the human eye. Try 10 mA if your LED can handle that. Persistence of vision will also make it appear less jumpy.

Look in the datasheet for the LM3915, it is long but, I think there is something in there for a suggestion on how to slow the response. I think that is where I read it. The application suggestions is the place to start in the back of the document.
 

Thread Starter

Chardi

Joined Oct 3, 2015
28
@GopherT

From what I understand, Figure 27 in the datasheet of the LM3915 gives an example of smooth transitioning of the LEDs; however, I believe that it is only achieved when connecting the signal input to a signal generator outputting clean waves (whether they are sine or triangle waves).

I did attempt the resistor configuration proposed in Figure 27 with no luck.
---
I switched up my resistors in the 3915 circuit for a 1.8-2mA current through the LEDs (the max they take is 2mA) and boosted my Vcc to 12VDC. This sort of fixes my issue with the LEDs constantly flashing and not being bright enough.

In addition, I made the gain of the LM324's amplifier be more than I need (I control the signal input with my laptop's volume) to ensure that there is enough of an amplified signal going in the 3915.

I do have videos of the circuit with the audio input playing, but this website doesn't allow me to upload .mp4 files *bummer*.

Nevertheless, I am still looking for a way to smoothen the LEDs and completely forgot about testing the DOT Mode, so I'll be working on that hopefully tomorrow if I have time.
 

Thread Starter

Chardi

Joined Oct 3, 2015
28
I have tried adding capacitors to smoothen the audio signal, but I didn't work. I don't have any other ideas and came to the conclusion that it is what it is (a transistor-based circuit would've given me the desired result).

Also, the DOT Mode doesn't work at all most likely because the audio signal isn't clean enough.

Nevertheless, I will finish this project as it is, without DOT Mode (unless I figure out a way to make it work). The PCB will be designed and etched; the device's body will be built.

Meanwhile, here's the circuit with a floating input:
IMG_20151022_195845_hdr.jpg
Blue LEDs show the lowest decibel value and white LEDs are the highest decibel value.

Another thing I'm not a big fan of is the logarithmic scale of the LM. The bottom LEDs turn on so easily consequently to the top ones which rarely go on.

Note: I have a LM7409 circuit, two trim.pots and a LM786 on the breadboard which are not connected to the circuit.
 

#12

Joined Nov 30, 2010
18,224
I didn't read the whole Thread but I can say that the dot mode doesn't work for audio signals because they are so fast that a human eye can not see the dots go off. If you want to do dots with audio you have to waste the high frequencies to ground and do a rectifier thing with a capacitor to slow the response down to where the chip only sees peaks fading away between beats.

Think about it. People see TV at 30 frames per second and think it's continuous. There is nearly nothing in the audio range that slow.
 
Top