Like in my other post, I am making a graphic equalizer. I was having trouble getting the op amp to work with the LM3915, but I fixed that. Now, having gotten a normal op amp to work, my bandpass filter won't work with the LM3915. Despite the infinite contempt this project seems to have for me, though, I think I'm actually learning quite a bit with this.

I decided to make a separate test circuit with a bandpass filter straight off the internet from here: http://sound.westhost.com/project63.htm
The only difference is that I didn't bother to put a bypass capacitor on the power supply, which is from 2 9V batteries put together to make a split power supply, and I didn't bother to use Resistor X (it was optional, so I just used a wire instead of the resistor. The op amp used is one from a LM248. I know it's a noisy chip, but I'm just trying to get the stupid thing to work right now. I can worry about sound distortion later. The input signal is directly from my electric guitar.

I used the standard LM3915 circuit from the datasheet (page 2 to be exact, and I'll include the datasheet with the post). The only difference is that I changed the resistor values. I know this part works because I can feed my unfiltered input signal directly from the source and it works fine.

Basically, the output of the filter SEEMS perfect. The sound is fine. I even hooked the output out to my laptop and used a free oscilloscope program. The AC signal looks fine. I know the LM3915 can take an AC input, so I'm sure that's not the problem (although it's only half the signal with AC). I compared it to the unfiltered signal straight from the source, and that had about the same power level (both were around 45mV for V eff according to the oscilloscope program). There did not seem to be any DC voltage (neither my multimeter nor the oscilloscope program showed any DC voltage). However, if I put a diode (1N4148, if I remember right) between the filter and the output, the oscilloscope program picked up a completely normal signal (unrectified) when oriented one way, and no signal whatsoever when oriented the other way. I assume this may have something to due with the low voltage of the signal, but I would have thought AC would merely be flipped more or less by changing the diode's orientation.

I've tried sticking a 10 uF capacitor between the output and the LM3915 input, but that didn't do anything.

Oh, and if anyone cares, the Velleman project kits are on clearance at Radioshack. I can confirm firsthand that it was at 2 different stores, so there's probably a good chance that they're on clearance at all the stores (but it could just be a weird coincidence). I scored a K1823RS power supply unit for $5 (normally $20, there were a few others with the same deal). So now's the time to get them if you want to give them to a kid for a present or something.

 

From what I can see, the LM3915 requires an input of about 45xVref mV for the first output and about 63xVref mV for the second output. Maybe the output from your amp is too small.

 

I'm not sure if that's the case. Right now my oscilloscope program seems to be picking up random noise of 15.734 kHz of about a third of a microvolt for some reason, even when the cable's not connected to anything, so I'll have to check again later. I tried putting an LM386 with a gain of 20 after the filter, but then all I got was 9 LEDs lighting up regardless of what I did with the volume potentiometer (and a lot of static-y noise). Maybe there's some sort of radio wave source at 15.734 kHz or something screwing with my circuit.

 

Do you have a television nearby? 15,734 Hz is the horizontal sweep frequency.

 

I used the standard LM3915 circuit from the datasheet (page 2 to be exact, and I'll include the datasheet with the post). The only difference is that I changed the resistor values. I know this part works because I can feed my unfiltered input signal directly from the source and it works fine.

What resistor values did you change ?
What is your unfiltered signal source ?

 

What resistor values did you change ?
What is your unfiltered signal source ?

I made R1 500 Ohms, and R2 680 Ohms.

My signal source is simply my electric guitar. I have heard something about electric guitar signal sources being high impedance, if that has any effect here.

As to the other guy, I DO have a small television about 1-2 feet from my breadboard, but that's a flattish one (I'm quite sure it does not have a CRT), and I don't think I've turned it on for at least a month or two, if not more, so I'm quite sure my television isn't the source. My sister has an HD flatscreen television in her room, but I would think that's too far away to do anything (besides, it has to go through the walls of my house).

Now that I think about it, I think the oscilloscope program was picking up some 60 Hz noise somehow. My laptop was plugged into the charger, but the probe thingie (basically a 1/8" auxiliary input cable plugged into the mic of my laptop, with the other end going into a thing that converts it a 1/4" end, and then using the wires on a 1/4" female jack (It may be a cheapskate redneck method, but it works quite well for my purposes considering I need the audio jack for the circuit anyway). I was only getting the 60 Hz noise when I checked again later, and that was when the auxiliary cord wasn't connected to the circuit. I figure the cord (about 3-4 feet long) is acting as an antenna and picking up noise from the wires in the walls.

Oh, and when I tried the oscilloscope program with just the filter again, I noticed some obvious clipping. I don't particularly care about it right now (Right now I just want the thing to just work, and I can always fix clipping later). I just thought it was strange considering the peak to peak didn't even reach 2 volts (or it might have just been amplitude instead of peak to peak, but still strange), when my power supply to the op amp (1/4 of an LM248 op amp) in the filter was +- 9V.

 

Like in my other post, I am making a graphic equalizer. I was having trouble getting the op amp to work with the LM3915, but I fixed that. Now, having gotten a normal op amp to work, my bandpass filter won't work with the LM3915. Despite the infinite contempt this project seems to have for me, though, I think I'm actually learning quite a bit with this

I made R1 500 Ohms, and R2 680 Ohms.

My signal source is simply my electric guitar. I have heard something about electric guitar signal sources being high impedance, if that has any effect here.

Yes, the guitar is high impedance which means it won't drive the filter very well, and depending on the instrument, can have an output as low as the 100mv range. The guitar neeeds a proper preamp. What was your op-amp circuit that seemed to work ?

Have you tried guitar --> op-amp circuit that worked --> bandpass filter --> LM3915 ?

Try changing R1 to 1k (500 is driving the Leds' with over 20MA) and try a smaller value for R2 to reduce the signal voltage required for full-scale.

B.T.W I think what you are building would more properly be referred to as a graphic analyzer, or graphic display. Graphic equalizer typically refers to narrow band tone controls controlled by slide pots.

 

Yes, the guitar is high impedance which means it won't drive the filter very well, and depending on the instrument, can have an output as low as the 100mv range. The guitar neeeds a proper preamp. What was your op-amp circuit that seemed to work ?

Have you tried guitar --> op-amp circuit that worked --> bandpass filter --> LM3915 ?

Try changing R1 to 1k (500 is driving the Leds' with over 20MA) and try a smaller value for R2 to reduce the signal voltage required for full-scale.

B.T.W I think what you are building would more properly be referred to as a graphic analyzer, or graphic display. Graphic equalizer typically refers to narrow band tone controls controlled by slide pots.

IIIIIIIITTTTTTTT WOOOOOOOOOOOOOOOOOOOOOOOORRRKKKSSS!!!
THANK YOU VERY MUCH.

Finally, I got the thing to work. I just needed the op amp in front. I could have sworn I had already tried that, but I must have done it wrong or something. I know I had tried an op amp AFTER the bandpass filter, but that didn't work out very well.

As to the LEDs, I don't think it's actually giving all that much current to them. I had calculated that with all 11 bargraphs (10 for each filter + 1 for volume) going at full blast with 20 mA to each LED piece, I would be drawing 2.5 - 3 Amps (the LEDs certainly won't be battery powered), but when I actually hooked my multimeter up to it, I was only getting about 1.4 Amps. The LEDs are certainly not blinding me with light, and I'd really rather not have to modify the resistors in the LM3915 circuit. I'd have to unsolder 110 wires, replace the resistors, and then resolder the 110 wires. I know pins or something else would have been better, but they didn't quite work out when I needed them to be soldered on upside down on the perfboard. I'll include a picture of the bargraph + LM3915 thingie to show you.

As to the name of the project, I'm not really sure what to call it. It's certainly an equalizer, and it's also an audio analyzer. I thought it would be awesome to build an equalizer effect pedal with an audio analyzer built into it. I've just been thinking of it as a graphic equalizer audio analyzer combo thingamabobber.

 

Great you got it working !!

Here's a guitar amp from the ESP site with an example of a guitar preamp using the TLO72 op-amp:
http://sound.westhost.com/project27.htm

It's important to have a preamp with a high impedance input for a guitar. Otherwise your tone and signal strength will suffer greatly.

The op-amp circuit you posted works but does have issues. For experimenting it will work with some tweaks, but you should eventually build a preamp similar to Rod Elliot's circuit

Regarding your op-amp circuit: I don't' know what the pot value is but put a resistor in series between its gnd end and gnd to limit the max gain.
something like 2-5% of the pot value.

Connect your guitar input to the + input instead of the - input, with a series cap, resistor and a resistor to ground as in Rod's preamp.

 

I see. I'm planning on replacing the LM248 with a lower noise chip and a single rail supply, like an MC34074, so that should allow it to work with a single 9V battery/adapter instead of needing +/-15 V like in the diagram.

By the way, will I need to put some sort of decoupling thingie between the audio output and the bandpass filters? I've been thinking lately that if I just send the LM3915 input directly from the filter output and send the filter output directly to the audio output, the outputs would probably go backwards into the other filter outputs and screw with the LM3915 input for each band.

 

I see. I'm planning on replacing the LM248 with a lower noise chip and a single rail supply, like an MC34074, so that should allow it to work with a single 9V battery/adapter instead of needing +/-15 V like in the diagram.

By the way, will I need to put some sort of decoupling thingie between the audio output and the bandpass filters? I've been thinking lately that if I just send the LM3915 input directly from the filter output and send the filter output directly to the audio output, the outputs would probably go backwards into the other filter outputs and screw with the LM3915 input for each band.

A single supply op-amp circuit is great as long as it's biased properly. With single supply op-amp circuits a coupling 'thingie' is required on both the input and the output because both will be sitting at 1/2 the power supply with no AC (audio) signal. The 'thingie' is a capacitor of the proper value.

Would you post a block diagram of your circuit idea ?

 

Here's my block diagram.

The first one is just how I was originally planning to do when I first started. The second one is what I mean by 'decoupler thingie'. I also just tried the thing with 2 filters, and I'm pretty sure the signal from one leaked into the other through the audio output connection, because the displays for the 2 bands were pretty much identical.

I'm thinking the decoupler would have to be something like another op amp with a gain of 1 or a transister, or possibly an optocoupler (unlikely but probably effective).

 

Yes, you have the right idea. But, let's refer to the "decoupler thingie' as a buffer stage.
Here's is ESP's project for a graphic display:
http://sound.westhost.com/project136.htm
The use of a buffer circuit to drive multiple display filters is explained. The NE5532 IC is used because it's able to drive lower impedance's than typical op-amps but also has very good audio performance.

You can drive a number of the filters shown in your first post using this.
If the input resistance is around 40k per filter as shown, then 10 filters (for example) in parallel would equal a load of 4k which is much higher than the output impedance of the buffer circuit. You always want the load impedance to be higher than the source impedance.

In your 2nd image, instead of the 'Decouplers' you could put the buffer after the preamp, drive multiple filters/displays with it and also take your straight signal output from that point through it's own output buffer.

EDIT: I'm basing this suggestion on the idea that this is a graphic display for various frequency bands because of the filter circuits you are posting which don't have boost/cut capability....

 

OK, so 'decoupler thingie' = buffer.

I really should learn about impedance one of these days. About all I learned about it in my Engineering Physics II class was that it was dependent on frequency and could be calculated with capacitors, resistors, and inductors. I've heard about it doing some stuff, but I never thought about it much before. I think I'll spend a good chunk of today looking it up.

Flipping through my electronics book I got when I was in high school, I saw something about an Impedance Transfer Function (Obviously, I did not understand the fancy math (aside from stuff in Calculus I) in the book when I was in high school, but I didn't have any money to make anything at the time anyway). Glancing at the math involved in the next few pages, I'm feeling really glad I got an A in Differential Equations last semester (and really glad that I didn't sell my textbook). The math looks rather tedious but doable. When I was leafing through it for information a month or two before, I assumed it was unnecessary when simpler math was available, but now I'm rethinking that assumption.

I revised the block diagram, if you'd like to take a look at it. This project is really getting to require a buttload of op amp chips (but hey, I'll feel more accomplished when I'm done). Ideally, the volume control will control the boost/cut as in a normal equalizer. The volume control will probably be either a simply potentiometer or another op amp with the potentiometer controlling feedback. I didn't want to stick the potentiometer on the filter op amp because I'm pretty sure that'll mess with the frequency of the filter.

 

Oh, and this is what I'd consider the front panel to be, if you're wondering.

Back when I was planning this, I DID only intend it to be an equalizer and nothing more. But then I got the "Let's add MORE features! MORE MORE MOOOOORREEEE!" overengineering bug and decided to throw in the audio analyzer just for fun. I had actually even initially tried to make the sensitivity level of ALL of the LM3915s controllable by a single potentiometer, but that didn't work out too well. Turns out you can't easily find a 10 gang potentiometer (plus it would take up a crapload of space), and 10 digital potentiometers would just be making it even more complex. A simple volume boost would probably suffice for that purpose, though.

It's surprisingly difficult to tell yourself, "No, I DON'T need to make every single detail adjustable in this thing."

 

Here"s a graohic equalizer circuit with boost/cut capability:
http://sound.westhost.com/project75.htm

I've included this link to show what is considered a true graphic equalizer with boost/cut and flat response setting (slide pots at midpoint. I think that's what you are thinking of in post #15.

There is also a link to a guitar graphic equalizer (project #64). This works similar to your drawing in post 14. But, this type of design does not provide a flat response setting. Instead it runs the signal though multiple band pass filters with overlapping frequencys. This lets you mix together these different frequency bands which works great for guitar "sounds", but will always add tone coloration.

The traditional graphic allows a flat EQ setting to start, with only subtle changes in sound if desired.

I hope this makes some sense.

Your guitar drawing is kind of cool.

 

Ooh, that's perfect! I'll have to mess with the voltages (probably change the op amp chip as well for a convenient single supply) a little to fit a 9V battery/AC adapter, but it should still be fine.

I was initially planning the equalizer for a bit of tone coloration to begin with, so I appreciate the second equalizer reference as well. It turns out to be rather difficult to get the tone just right with only 3 knobs (or I'm just not using the right settings or completely different equipment, causing an inevitable change). And I looked at some other projects there as well. I almost want to try to tackle the 1500W amplifier (#117) just because. Too bad that common sense and lack of money prevents me from doing so. I think a simple head phone amplifier or robotic litter box (seriously, changing cat litter sucks and I always forget to do it so it gets rather smelly) will suffice for my next project.

Thanks for the compliment on my guitar drawing. I tend to suck at drawing things on computers, unless it's something absurdly simple, like a box or a stick figure.

It'll be at least a couple weeks until I have the cash for new components to work with (I'm hopefully getting a tiny drill press soon, so I may as well make a pcb layout as I've learned the hard way that soldering tiny solder trails in small areas seriously sucks balls), so I should be out of questions and problems for now.

Seriously, thanks for the help. I really appreciate it. Before this weekend, I was seriously considering just eliminating the audio analyzer part of the equalizer out of constant frustration.