Connect with us

Help with 19 LED VU Meter circuit (LM3915/16)

Discussion in 'LEDs and Optoelectronics' started by speedyrb, Feb 27, 2013.

Scroll to continue with content
  1. speedyrb

    speedyrb

    7
    0
    Feb 27, 2013
    Hello!

    I am a newcomer to these forums, and somewhat of an amateur when it comes to electronics. In other words, I'm handy with a soldering iron, and I can build most circuits from schematics, but I don't always understand the theory behind why it works.

    I'm building an 8-channel surround sound meter system, using the LM3915/LM3916 chips. Here is the schematic:
    http://www.hqew.net/circuit-diagram/VU-Meter-using-LM3915_2835.html

    In the breadboarding stage, everything was great. But once I connected all 8 units, this happened (see the top left and top right meters in attachment #1).

    LEDs 8 and 9 go dim during loud program material. It doesn't happen every time the meters peak, but it seems to happen when there are prolonged periods of loud activity, and perhaps when low frequencies are present.

    In the breadboarding stage, I used this to supply power (+/- 12V at 1A):
    http://www.musicfromouterspace.com/analogsynth_new/WALLWARTSUPPLY/WALLWARTSUPPLY.php

    But it was not enough to drive all 8 units - I figure I need apx 2.5-3A. Providing enough power for all 8 units has proved tricky. The solution I came up was this one, using two switching laptop power supplies (18V, 3.4A):
    http://www.awrr.com/2warts2.gif

    I'm so pleased with the way this unit has come out, but this one thing is really bumming me out! Any help would be so greatly appreciated!

    Thanks!
     

    Attached Files:

    Last edited: Feb 27, 2013
  2. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,496
    2,837
    Jan 21, 2010
    Is the same thing happening for the 8th LED of the Surround Left level?

    One guess is that R6/R7 are perhaps not the correct values.

    The other is that there is some difference in time constant between the inputs for IC2 and IC3.

    I would get a signal generator and apply a sine wave signal to the input and slowly turn the level up.

    I would also see if it acts the same over the full audio spectrum.

    I'm not overly confident of my explanation because it should affect the 10th LED too.
     
  3. speedyrb

    speedyrb

    7
    0
    Feb 27, 2013
    Firstly, thanks for the reply!

    I have a video of the meter unit (the image is a screencap) and I just checked it - my bad - it's happening to the 8th and 9th LEDs, but not the 10th (as far as I can tell). (My meters actually leave off the 19th LED to fit in the enclosure better, so it gets confusing when counting from the end.)

    Assuming I've implemented the correct value resistors for R6 & R7 as indicated in the schematic (in other words, assuming the schematic is wrong), what value would you recommend I replace them with? Currently R6=330k and R7=62k.

    I'm not sure how to test this (IC2 and IC3 are tied together with R6): "some difference in time constant between the inputs for IC2 and IC3".

    Unfortunately, I don't have access to a signal generator.

    I've attached exactly how I laid out the circuit (I used a solderable breadboard), if this helps.

    Thanks again!
     

    Attached Files:

  4. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,496
    2,837
    Jan 21, 2010
    The issue I'm suggesting is that the attenuation provided by those resistors is the same as the range of the LM3915. If it's not accurate then the last LED on the first IC may overlap (or have an unusually large step to) the first LED of the second IC.

    This may cause some strange displays.

    Because of the high(er) impedance of the first IC's signal source, any capacitance from there to ground will act as a low pass filter with a different (lower) cutoff frequency than for IC2. If this display happens with higher frequency signals, it could be a cause.

    HOWEVER if that were the reason, the topmost LEDs for IC1 would all be dim, and it appears that they are not.

    Are your +ve rails connected together? That diagram suggests they're not
     
  5. speedyrb

    speedyrb

    7
    0
    Feb 27, 2013
    The two "+" rails on the breadboard layout are not tied together - the top "+" rail is the V+ supply and the bottom "+" is the V- supply (dual supply for the opamp). The "-" (ground) rails are tied together.

    Thanks again for taking the time with this - I appreciate it.
     
  6. speedyrb

    speedyrb

    7
    0
    Feb 27, 2013
    Also, I want to emphasize that when I breadboarded a single unit, and ran it for long periods of time, I never once witnessed it behaving like this (with the 8th and 9th LEDs dimming during loud program material), but now that it is in this 'mostly finished' stage, it happens pretty regularly. I would have noticed if it happened in the breadboarding stage.

    The only things that have changed since then involve the kind of power supply I used, and the fact that I am now running 8 of these off a single power supply. (See the original post for details.)

    Thanks again!
     
  7. speedyrb

    speedyrb

    7
    0
    Feb 27, 2013
    OP here

    After doing some troubleshooting (trying a more conventional power supply, playing with the specified resistor values mentioned in this thread), the problem is still occurring.

    However, I've noticed the following:
    1. When I feed certain program material into the meters for the first time, the problem does not occur. However, if I rewind the source and play the same program material in again after the meters have been on for a short while, the problem does occur.

    2. The LM3915 chip (the first of the two meter driving chips) is getting very hot.

    So... could this be due to the chip overheating?

    Another question: I'm having trouble locating negative linear regulators that can handle 2.5 amps. If I use one that can handle 1.5 amps and put a really large heat sink on it, will that work?

    Thanks for taking the time to help with this!
     
  8. babadono

    babadono

    2
    0
    Dec 12, 2013
    Bypassing?

    Hi Speedy,
    I'm new here. Actually stumbled onto your post about the VU meter before i became a member. Did you ever resolve you problem? Have you tried increasing the bypass capacitor that is right on the supply pin(s) of the ICs?
    babadono
     
  9. speedyrb

    speedyrb

    7
    0
    Feb 27, 2013
    babadono

    It turns out that - surprise, surprise - the answer was in the data sheet! Of course! How many times have I seen in these forums and elsewhere that the answer to most problems is in the data sheets... :eek:

    When I sat down with them and really read them closely, I discovered that the chips can overheat. So as recommended by the data sheet, I placed a resistor to drop the voltage to the LEDs, and that cooled down the chips and solved the problem.

    Thanks for asking!
    speedy
     
  10. babadono

    babadono

    2
    0
    Dec 12, 2013
    Speedy,
    so all is well, cool. So the circuit is exactly like the link in your first post with the addition of limiting resistor in the VLED supply? What value is the resistor? I am trying to build a version of this circuit but when the signal level gets around 0dB I am going to make all the Leds turn from green to yellow. Then when the signal approaches +3dB they will turn to red.
    BTW is that a picture of your 7.1 channel box in the first post? Very cool. We love flashing lights!
    babadono
     
  11. duke37

    duke37

    5,364
    771
    Jan 9, 2011
    A resistor to drop the voltage is not a good idea since the drop will depend on the current through it which will depend on the number of leds which are lit.

    You could use a power supply with a lower voltage or put a zener diode in the line to drop a specific voltage, make sure the zener can dissipate the power.
    An alternative to the zener would be a chain of silicon diodes, each of which would drop about 0.6V
     
  12. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,496
    2,837
    Jan 21, 2010
    Doh!

    You're using the bar graph display, right? That is a well known issue. I'm shaking my head that I didn't think of it.

    However, you found the answer in the right place.

    Duke37, the chip provides a constant current to the LEDs. The issue is that when there are a lot of LEDs operating the package dissipation can get quite high. The resistor acts to reduce the package dissipation at the point that it most needs it by limiting the voltage at higher currents. Sure, a lower voltage supply will work, but in this case so does a resistor.

    Incidentally, this information is hidden on page 2! (so not hidden)

    Just as a bit of working, let's assume you want 20mA for LEDs with a voltage drop of 1.8V from a power supply of 12V. That's 200mA max, and the max resistor in series with the LED power supply would be about 8/.2 = 40 ohms. So let's say 39 ohms max.

    Now, the package dissipation due to the LED current for LED power rail Vl, LED forward voltage Vf, LED current If, a single resistor Rl, and n LEDs lit would be:

    P = (Vl - Vf - (n * If * Rl)) * (n * If).

    For the values listed above, (but no resistor) the package dissipation ruses like this:

    Code:
    LEDs   Power
    0      0W
    1      0.204W
    2      0.408W
    ...
    10     2.04W
    I've left out lots in the middle because it's linear.

    Note also that the peak dissipation is 1.365W, so you can easily exceed it.

    With a resistor of 39 ohms, it changes to

    Code:
    LEDs   Power
    0      0W
    1      0.188W
    2      0.346W
    3      0.472W
    4      0.566W
    5      0.630W
    6      0.662W
    7      0.664W
    8      0.634W
    9      0.572W
    10     0.488W
    The amazing thing is that the dissipation starts to fall as you add the last few LEDs. Note that the resitor is dissipating the difference, so with 10 LEDs, it is dissipating 2.04W - 0.488W = 1.56W (so you had better make it a 2W resistor)

    Sure, you can also have 1 resistor per LED, but with constant current drivers for the LED, we don't need to do that.
     
    Last edited: Dec 13, 2013
  13. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,496
    2,837
    Jan 21, 2010
    And here is a spreadsheet you can use to do the calculations
     

    Attached Files:

  14. speedyrb

    speedyrb

    7
    0
    Feb 27, 2013
    Wow - thanks for all the replies to this thread (which I admit I thought was dead :D.)

    I am using a single 47 ohm 1 watt resistor at the head of the LED power supply line, based on the attached schematic for a similar meter kit sold by Velleman (it's at the bottom right on the image).

    Based on your comments and calculations, (*steve*), I hope it doesn't go up in smoke... :)

    Yes, babadono, that enclosure, including the front and rear panels, is my design. When I am finished with all the units I'm building I plan on taking more pics of the entire build, which I will share here.

    Thanks for your interest and all the advice!
     

    Attached Files:

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-