Modified 3-band Baxandall circuit

[LyosNorezel]

Greetings all.

For several years now... I've been trying to find a set of headphones/earbuds that would deliver a satisfying sound.
Not being able to afford the "audiophile" versions (and knowing they could never match my hearing preferences)... I turned to the DIY community to see if I could find something that would work.

My search eventually turned up the "jackhammer" headphone mod. This gave me a base design that sounded somewhat decent, though nowhere near perfection.

Having spent all that time searching... I now knew why audio sounded so terrible to me with most headphones/earphones. Earphones could never truly do well with lower mid range and bass signals... mid size headphones did well in the mid range, ok in the treble and terrible on bass. I recently acquired a set of headphones that use 40mm drivers... and I finally had the pleasure of hearing bass via headphones, but the drivers were terrible with treble and upper mid range.

This finally brought me full circle... I could have great sound with three different drivers... but could never enjoy them together.

It was this realization that made me remember why speaker boxes for stereos inevitably had three drivers. It was also this that got me thinking about modifying the jackhammer idea to use three drivers and a baxandall circuit (that I remembered from my earlier searching).

However... the baxandall circuit I had found ([image]http://gilmore2.chem.northwestern.edu/images/eq3.gif[/image] ) was far from perfect... for the 3-driver solution would require some level of frequency separation.

Having recently created the "Mint-Tin Amp" (found on Make, design from: tangentsoft), and having previously had success with other audio projects, I decided to try my hand at modifying a circuit to suit my needs.

This is what I came up with: [image]http://www.oslb.net/Baxandall-modified.png[/image]

As you can see... it's clearly based on the Baxandall circuit of Chu Moy's design, with the only difference being the separation of the channels with amps for each.

Btw, I am aware that the circuit depicts a flawed power design (I'm still learning the Eagle program)... and I will actually be using the power supply circuit from tangentsoft's mint-tin amp. The "GND"s on the + side of the opamps will, therefore, be attached to the virtual ground created by tangent's power supply, and the "GND"s in the image attached to the power supply of the chip will actually be the negative voltage required. I am also aware that this is a "mono" circuit, I will build two of these circuits for the final stereo design.

All that said, however, I am still fairly new to the circuit design side of diyaudio... and would like some advise/friendly criticism before I continue with this circuit.

I would also like some alternatives to the 100k ohm and 500k ohm potentiometers the circuit calls for... as I'm finding it really difficult to find dual linear pots in those values.
If I could use them without changing the other components of the circuit much... I would love to use the tiny 500 ohm dual pots you see as in-line volume controls on many earbuds, as that would allow me to fully integrate the Baxandall circuit into the headset with the drivers (obviously without the power supply, as batteries take up far too much space).

I would appreciate any help/advise ya'll could give on this.

Thanks.
Lyos Gemini Norezel

[LyosNorezel]

I've created a new version of my circuit which fixed the power supply issue, here: http://www.oslb.net/Baxandall-modified2.png

I also changed the capacitors and they now display their values, as well as changed the "GND"s on the + side of the opamps to "vGND"s for more accuracy in the circuit.

Could still use some advise on this circuit... if there's anyone here who can help.

Lyos Gemini Norezel

[gootee]

If you're using three drivers then you are basically just needing to implement a "tri-amping" type of scenario, with line-level crossover (filter) networks.

So just design such a circuit that has the correct relative output levels for the three outputs, so you won't need any potentiometers.

You only need to know where the cutoff frequencies should be and what slopes the filters' rolloff characteristics should have, and what the relative output levels should be for your particular drivers. Then design each of the three filter/amplifier chains.

I would try to not use any high-value resistors, anywhere, since they contribute more noise.

[AndrewT]

The NFB has variable gain. This changes the closed loop and and the loop gains.
This in turn makes it very difficult to ensure stability of the amplifier at all adjustment positions.

Settle for one fixed gain for each amp to take account of power amplifier gain and speaker sensitivity. A small range of attenuation on some of the outputs allows for final "tuning" of the relative SPLs.

[gootee]

Originally Posted by AndrewT

The NFB has variable gain. This changes the closed loop and and the loop gains.
This in turn makes it very difficult to ensure stability of the amplifier at all adjustment positions.

Settle for one fixed gain for each amp to take account of power amplifier gain and speaker sensitivity. A small range of attenuation on some of the outputs allows for final "tuning" of the relative SPLs.

Yup. Or, if these are the final amps for the headphone drivers, then put the attenuators (or pots) in series/shunt with the input signals.

If that causes input-impedance problems, you can always add a unity-gain buffer upstream from each attenuator.

You might want to download the free filterpro software, from ti.com . Also, here is a great source of information:

http://www.analog.com/library/analog..._handbook.html

[LyosNorezel]

Originally Posted by gootee

If you're using three drivers then you are basically just needing to implement a "tri-amping" type of scenario, with line-level crossover (filter) networks.

So just design such a circuit that has the correct relative output levels for the three outputs, so you won't need any potentiometers.

The problem with that is the fact that these drivers will be inside of a jackhammer type headset. Using fixed values for each driver wouldn't be ideal, given the strange relationship of audio to the ears when they're so close.

Originally Posted by gootee

I would try to not use any high-value resistors, anywhere, since they contribute more noise.

Given my source of audio (Sansa Clip+), I thought this particular Baxandall circuit would be near perfect. Are you suggesting I lower the values of some of the resistors?

Originally Posted by AndrewT

The NFB has variable gain. This changes the closed loop and and the loop gains.

What is NFB?

Originally Posted by AndrewT

Settle for one fixed gain for each amp to take account of power amplifier gain and speaker sensitivity. A small range of attenuation on some of the outputs allows for final "tuning" of the relative SPLs.

Given the oddities of my hearing, it's actually better for me to be able to set the gain of each range separately.

Originally Posted by gootee

Yup. Or, if these are the final amps for the headphone drivers, then put the attenuators (or pots) in series/shunt with the input signals.

I may, or may not, use a CMoy amp in front of (between audio source and the eq) this circuit, depending on how well this circuit amplifies the source on it's own.

I'm not entirely sure I understand what you mean, however.

I do want to be able to control each filter/band boost/cut separately, simply because of the peculiarities of my hearing.

Thanks for your input/help so far.

Lyos Gemini Norezel

[AndrewT]

NFB = Negative FeedBack

[AndrewT]

Originally Posted by AndrewT

Settle for one fixed gain for each amp to take account of power amplifier gain and speaker sensitivity. A small range of attenuation on some of the outputs allows for final "tuning" of the relative SPLs

Originally Posted by LyosNorezel

Given the oddities of my hearing, it's actually better for me to be able to set the gain of each range separately.

Read again. I am suggesting you allow for channel level adjustment.

[LyosNorezel]

Originally Posted by AndrewT

NFB = Negative FeedBack

Ah... ok. That make sense.

Originally Posted by AndrewT

Read again. I am suggesting you allow for channel level adjustment.

Obviously I'm not fully understanding what you're saying/writing. Any chance you could explain visually?

Thanks again for your help.

Lyos Gemini Norezel