Hz equations: EQ tuning

[no one ever]

Could anyone be so kind as to tell me what equations to use in figuring out the best component substitutions for this eq? I suppose its just simple R/C tweaking...


help much appreciated. 

[Transmogrifox]

Check this out.  It's called an active 3-band Baxandall Equalizer.  Scroll down until you find the circuit:
http://www.headwize.com/projects/equal_prj.htm

Appears to be a pretty neat site.

[no one ever]

Appears to be a pretty neat site.

agreed 

[no one ever]

 

[no one ever]

it seems that the frequency varies inversely with the frequency (shoulda known that.  ) although i thought small caps were treble boosters.

i'm trying to get this thing down to 800hz.

[Transmogrifox]

it seems that the frequency varies inversely with the frequency (shoulda known that.  ) although i thought small caps were treble boosters.

i'm trying to get this thing down to 800hz.

Uhhh...it appears to me that frequency varies inversely with period.  Not sure what you're trying to say here.

Anyway, what band are you trying to get down to 800Hz?  Center of the mid-band?  High-pass cut-off?

In either case, the filter's cutoff frequencies are interactive with the adjustment on the pots.  If you move any of the 3 knobs, the cut-off frequencies start to move around.

I recommend going to a gyrator-based EQ if this is a problem.  Unfortunately with the gyrator EQ, the bandwidth of each band gets more narrow as you approach boost/cut, but the center frequency stays pretty constant.  I guess there's a plus and minus to any design you choose.

I think PSPICE is your best friend for optimizing the Baxandall type EQ.  Otherwise you have to keep juggling things around, and computers do that so much more quickly.

[no one ever]

it seems that the frequency varies inversely with the frequency (shoulda known that.  ) although i thought small caps were treble boosters.

i'm trying to get this thing down to 800hz.

Uhhh...it appears to me that frequency varies inversely with period.  Not sure what you're trying to say here.

Anyway, what band are you trying to get down to 800Hz?  Center of the mid-band?  High-pass cut-off?

In either case, the filter's cutoff frequencies are interactive with the adjustment on the pots.  If you move any of the 3 knobs, the cut-off frequencies start to move around.

I recommend going to a gyrator-based EQ if this is a problem.  Unfortunately with the gyrator EQ, the bandwidth of each band gets more narrow as you approach boost/cut, but the center frequency stays pretty constant.  I guess there's a plus and minus to any design you choose.

I think PSPICE is your best friend for optimizing the Baxandall type EQ.  Otherwise you have to keep juggling things around, and computers do that so much more quickly.

gah, brain dysfunction. i meant frequency varies inversely with capacitor value... or the other way around... i don't know, i did it hours ago.



i'll look up pspice, but i doubt its free 

[no one ever]

oh, and yes, the mid band i'm aiming for is 800hz.

[Transmogrifox]

The student version is free.

http://vorlon.cwru.edu/~flm/eecs245/pspice.html

Look under the "Local Downloads" heading and download PSPICE version 9.1.  The student version is more than enough to simulate simple circuit modules such as this.  It just has a limit of 64 nodes, and you can't use some of the more complicated models.  You're right about PSPICE not being free, otherwise.  A full version is in the $7,000-$15000 range, depending on the options you include.

For stompbox EQ's and 5-transistor fuzzes, the student version is more than enough.

To bring the center frequency down, scale all the capacitor values by some fixed amount.  That amount should be greater than 100%, for example, multiply them all by a factor of 1.45.  The equations given on the Headwize page (as described) are not for an *exact* answer.  They just give you an acceptable approximation to experiment from.  If you scale all the capacitor values by a fixed amount, then you stay out of the trouble of causing bands to overlap and create "humps" when the EQ is supposedly set flat. 

Set it up on your breadboard and run set of calcs for capacitor values.  Crank the mids and cut the highs and lows to start.  Start changing out cap values according to your calculations (the direct scale value) and let your ears tell you when you got it right.  If you're able to use PSPICE to get the center near 800Hz that's great as long as you know that's a range for the mids that complements your rig real well, and your ears agree.

[George Giblet]

> the mid band i'm aiming for is 800hz.

As transmographics has said the frequency varies with the mid pot positions (and other pots) - it's hard to define the centre frequency with a single number.  Also if you play with the mid caps the bass and treble behaviour will change - it's all tied together, which is my there's no simple equations for the 3-band.

[no one ever]

i got the student version, and copied the schematic at headwize into their schematic editor. a few quibbles: it doesn't seem like there's any real voltage pins for the opamps?... just inverting, non inverting inputs and an output. i also don't know how to set the input voltage (acmag, dcv), or how to run the optimizer or simulator or probe.


don't have a breadboard yet 



i was aiming for an 800hz center frequency band, but i suppose that doesn't change anything...

[Transmogrifox]

PSPICE has an op amp model that is an "ideal" op amp.  This one is fine to use for this type of simulation since a good op amp will not cause the results to vary to a great degree.  It doesn't have a power supply.  PSPICE just assumes it has an infinite power supply.

In further simulations where you want a more realistic model, select the menu Draw--->Get new Part--->LM324.  This part does have power supply pins, and you would then need to select a DC voltage source to simulate it.

They might have a 741 in there (I can't remember).  It doesn't come stock with any great models, but unless you're doing criitcal design where a chance of 5% error would ruin the circuit, it won't be much problem just using these models. 

Here's what you need to do to set up a frequency plot:
ACMAG:  Just set this one to 1V
Find the drop-down menu that lists "Markers", or select from the tool bar a Voltage Probe.  It's like a little arrow with a circle on the end with a "v" in the middle.
Go to Analysis-->Setup
You should be able to get a dialogue box that allows you to select a simulation type.  Select "AC sweep".  Sweep type such as log/linear can be left to your own preference.
Set your start frequency (I usually select 10Hz), and set your end frequency.  For this simulation, 10 kHz is probably ok since you're more interested in what's happening with the mids around the 1kHz range.
The number of points is just how many frequencies it uses to calculate the plot.  The default of 500 points should be fine.  If you see your plots getting relatively crude, then you can select more points at the cost of a little bit longer simulation time.

When done with that, click on the simulation button, or Analysis-->Simulate.

Let me know if things aren't working out for you, and I'll actually open the program and go through it step-by-step to make sure I'm not missing something.

[no one ever]

  there was a question here, but i should think before i post.

Code Select Expand

ERROR -- Missing or invalid expression
RB_R8        $N_0011 $N_0009 {(500k*0..25)+.001}

[no one ever]

alright, i got it to work with the simulator (and got the midband center freq to 800hz! using unheard-of values, however) , but:

the output's trace is nearly identical to the treble trace! (i've set the graph to read from the output of each band and the summed output) and putting 4.7M pulldown resistors on the thing affects the frequency response.


and the treble control appears to be much too powerful, and the site notes that the equations do not apply to the treble circuit. how can i fix that?

[no one ever]

 

[davebungo]

Be aware that in PSpice 4.7M is 4.7 milliohms NOT megaohms.  Use 4.7Meg if you want megaohms.

[Transmogrifox]

alright, i got it to work with the simulator (and got the midband center freq to 800hz! using unheard-of values, however) , but:

the output's trace is nearly identical to the treble trace! (i've set the graph to read from the output of each band and the summed output) and putting 4.7M pulldown resistors on the thing affects the frequency response.


and the treble control appears to be much too powerful, and the site notes that the equations do not apply to the treble circuit. how can i fix that?

Excellent. I'm glad it's working for you.  Doing a voltage trace on the treble "output" may not be very useful.   For the individual frequency response for each band, you'll want to run a current probe on the connection to the (-) terminal of the op amp.  This will tell you the actual current through each band.  The curve you get from that should show you the contribution from that band.

Be careful adding plots in PSPICE, as well.  It doesn't consider phase on the addition (though I think there may be some tools for doing the phasor addition such as atan(x) such that you can use the phase probe with the current probe to do this).  The effect is that you'll get large humps at the crossover bands, that in reality, are not there because of the phase relationship of the signals added.

In the end, it doesn't matter, because the output shows what the addtion of each individual band is.  What you want to do is get each band's -3dB (0.707) cut-offs at the same frequency as the adjacent band.  When all added together, the output response will be flat.

Again, I recommend getting the 800Hz center frequency by scaling only the capacitors by a certain amount each time....and all of the capacitors by the same ratio.  This will help to keep the bands at their respective relationship to eachother.  You may come up with some non-standard values.  Then you will need to decide how close to 800Hz you need it and try to get as near to 800Hz with as many standard values as possible.  From there, you will need to come up with series/parallel configurations to get the non-standard values.   

[no one ever]

Be aware that in PSpice 4.7M is 4.7 milliohms NOT megaohms.  Use 4.7Meg if you want megaohms.

thanks

Excellent. I'm glad it's working for you.

thanks a lot for your help, transmogrifox! 


i have, however, decided to leave this circuit off until i get more experience in electronics, for now i'll just use a 'stupidly wonderful tone control'