I have been lurking and learning much from these forums for a while now, but the time has come to post my own question.

My brother and I are building a stereo system with our own custom active crossover network. I have had no prior experience with AC circuits, but I have have done a bit with digital and DC.

Our network consists of 8 filters each with ~100 gain; 1 Low Pass with cutoff at 500Hz, 1 Band Pass 500Hz-3.4kHz, 2 Band Pass 500Hz - 2.5kHz, 2 Band Pass 2.5kHz-15kHz, and 2 Band Pass 15kHz-30kHz.

I constructed these in Multisim 11 and used a Function Generator with a sine wave at 100mVp to test them. I connected the outputs to the oscilloscope to check for signs of clipping.

Everything seems to work well in simulation, but I just tried to realize the low pass circuit with a 4ohm speaker, and I get no audio just a snap; as though I had connected it to a DC source.

As I said, I have had limited electronics experience and certainly nothing like this. You guys seem to be very good, so I hope I can get some help.

 

The LM4562 is an audio op amp (600 ohm minimum load), not an audio power amp so it cannot drive a 4 ohm speaker. You need to feed the op amp signal into an audio amp.

 

Rather ambitious for a first project.

Try implementing a single low pass active filter, and put a frequency generator on 20hz-20khz sweep on the input (Or an MP3 of same sweep through an MP3 player), then see what you get for the output.

Step by step.

 

From my understanding, I thought a the purpose of having high gain was to attenuate the desired frequencies to a point where they could be heard while the voltage peaks of the other frequencies went unamplified (and thus unheard). If the outputs need to be fed into another amplifier, won't that also amplify the sound we don't want in that channel? When the simulation runs in the current configuration, the output of the low pass peaks at around 7.5V from an input of a 300Hz 100mA input. Is 7.5V not enough to power a speaker?

I am unfamiliar with the role of a power amplifier (as opposed to op-amps). What affects load that either can drive (e.g. Why can a power amp drive a 4 ohm speaker but this op-amp configuration has to drive at least 600?)?

 

From my understanding, I thought a the purpose of having high gain was to attenuate the desired frequencies to a point where they could be heard while the voltage peaks of the other frequencies went unamplified (and thus unheard). If the outputs need to be fed into another amplifier, won't that also amplify the sound we don't want in that channel? When the simulation runs in the current configuration, the output of the low pass peaks at around 7.5V from an input of a 300Hz 100mA input. Is 7.5V not enough to power a speaker?

I am unfamiliar with the role of a power amplifier (as opposed to op-amps). What affects load that either can drive (e.g. Why can a power amp drive a 4 ohm speaker but this op-amp configuration has to drive at least 600?)?

7.5V is more than enough to drive a speaker if the amp has enough current drive. You need both voltage and current (power is volts x amps). An op amp can only provide a small amount of output current, so it's not enough to drive a low impedance speaker. Per Ohm's law, 7.5V peak into a 4Ω speaker requires that the amp deliver a peak current of 7.5/4 = 1.9A. That's why you need an power audio amp, which can deliver the needed high current.

 

The way filters work:

They attenuate frequencies outside their passband, and do NOT attenuate (or amplify) frequencies inside their passband (unity gain or gain of 1).

Your question on output power is related to impedance and current sourcing ability.

Power amplifiers have many transistors in parallel to get the output impedance below 1Ω so they can drive a 4Ω load with a lot of current. This takes wide pathways for the current to follow, and large heatsinks to dissipate heat.

Line level conversions, such as crossovers, should receive a 1Vrms signal, and output a 1Vrms signal. Any amplification will typically result in distortion or rejection at the power amp.

If you are physically building this, I dearly hope you are not using 741 op amps!

audioguru crutschow will be has been along to actually educate you completely on the topic, I'm just filling in the high spots.

 

Thank you for the information regarding power amps; I guess I was too caught up in trying to get voltage gain, that I forgot I need current.

I made a few revisions to my system, the most prominent being the addition of 8 LM1875Ts (One per speaker, each in the configuration shown in the datasheet under typical use). Unfortunately I couldn't just slap them in, the voltage gain was too great and transformed practically all my audio into square waves (this is all in simulation).

To combat the issue, I modified my filters to have a gain of ~4, with the exception of the low pass filter which now has a gain of ~8. I also modified the negative feedback loop of the power amps to have a gain of ~2.5. These gains have changed from 100 and 20 respectively. It may also be important to note that I have change the input peak to 200mV, because according to the data that I have collected, this is about equal to the peak output of the music source. All these changes combined provide me with an output peak slightly greater than the initial circuit's 7.5V.

I am very grateful of you guys thus far; not only are my filters transitioning better, but I'm getting what appears to be more realistic current (e.g. >8V@~2.2A to a 4ohm speaker tied to the low pass filter being fed a 400Hz sine wave with a 200mV peak).

Hopefully you guys can help me out a bit more and review the circuit for any blatant errors before I go forth and try to build it. (The power amps are pdf due to how long the sheet was, the filter circuits have only changed in component values, but I can post the revised ones if need be).

-----edit------
Thatoneguy: I did take your advice and my filters are now outputting a peak of just under 1V and the simulation looks good so far. Also, I am using LM4562 from Texas Instruments, which is designed for audio systems. Thanks for being on the lookout though.

 

I made a few revisions to my system, the most prominent being the addition of 8 LM1875Ts (One per speaker, each in the configuration shown in the datasheet under typical use). Unfortunately I couldn't just slap them in, the voltage gain was too great and transformed practically all my audio into square waves (this is all in simulation).

To combat the issue, I modified my filters to have a gain of ~4, with the exception of the low pass filter which now has a gain of ~8. I also modified the negative feedback loop of the power amps to have a gain of ~2.5.

The datasheet for the LM1875 says that it is stable with a closed-loop gain of 10 or more. So yours will probably oscillate.

Most of us use non-inverting Sallen and Key active filters with a gain of 1 instead of your inverting ones with lots of gain.

 

I originally went with a high gain design for the filters because I was unaware that I needed another stage of amplification. I have now changed all the filters to Sallen-Key type. I'm actually rather glad you suggested that, the layout seems much easier to follow.

Another benefit is that the output readily follows changes in the input. With the previous design there would be a wobble (for lack of a better term) when the input changed due to the feedback getting the device "caught up. (I assume, if you have a better explanation, I would like to know what was actually happening)

Since the filters are no longer providing gain, I have again adjusted the power amp for greater gain. With this new configuration I get just over 8Vrms @ 2A from the low pass filter being fed a sine wave with 200mV peak @ 400Hz.

Hopefully most of the hurdles are out of the way design-wise. You guys have been a major help so far.

----------------

P.S. I keep uploading the circuits as they progress, just in case they may be useful to someone looking to do something similar. If this is an inappropriate way of doing this, please let me know. Thanks!

 

Your new Sallen and Key filters have strange parts values and have peaks in their frequency response.

Older crossover circuits had a Butterworth frequency response which is flat then has a sharp dropoff. They are made with non-inverting opamps with a gain of 1. The lowpass filter has equal-value resistors and the feedback capacitor has a value that is double the value of the capacitor to ground.

Your filters are simple second-order ones. When combined with two speakers they cause a notch at the crossover frequency. Therefore one speaker should have its phase connected reversed but then when combined there is a +3db peak at the crossover frequency.

Modern crossover circuits use a Linkwitz-Riley response. Instead of Butterworth the response is Bessel. The resistors have the same value and the capacitors also have the same value. When combined with one speaker with its phase connected reversed the response is absolutely flat.

Here is a link to a good second-order Linkwitz-Riley active crossover:
http://www.sound.westhost.com/project81.htm

I show the response of your circuit then I show the response and schematic of a Butterworth filter.

 

If nothing is built yet, take a peek at ESP's Electronic Crossover

Actually, before building anything, read EVERYTHING YOU CAN from ESP's website.

A few hours of reading will result in performance gains you wouldn't believe over slapping together something that worked in simulation. Factors such as dynamic range, noise rejection, distortion, power draw, etc are all covered in different areas of that site. It's like the wikipedia for audio builders, but with better information (and even kits).

 

The "ESP Electric Crossover" circuit has fourth-order filters. The one I linked has second-order filters like what you want.