Graphic EQ using EQ3 question

I have a question about the EQ3 to make a graphic EQ bank by chaining them together.

I made what I think is a 15 band graphic EQ by chaining 5 x EQ3’s together but some of the bands I made don’t quite sound right so maybe my numbers are wrong and I am using the EQ3 the wrong way.

In the wiki it’s says this :

“This is a 3-band equalizer with controls for the gain of each band (low, mid, high) and controls for the center frequency of the two outer bands (low, high). You can also think of this equalizer as a 3-band exponential VCA since each frequency band’s gain is modulate-able. The equalizer is constructed from two 4-pole filters: a low-pass filter and a high-pass filter both acting on the input in parallel. Subtracting the outputs of the low-pass and high-pass filters from the original signal yields the mid band”

If I set low = 50 hz, high = 100 what would the middle band be? Would it be 75hz (150/2) - so I have 3 bands at 50/75/100 or would it be 50hz (100-50?) - 3 bands at 50/50/100? based on the way I am reading the wiki description. Or does it behave a different way that I’m not understanding? The idea is to get separate bands spread from a low of 20htz to max 13khz. Except for the top and bottom frequency bands, the bands I want spaced at an interval of a major seventh - 29 / 61 / 115 / 218 / 411 / 777 / 1.5k / 2.8k / 5.2k / 11k

based on my understanding, not yet having an ER301 to confirm, but to create a 15 band graphic eq, you would need 1 low shelf, 1 high shelf, and 13 mid bands. so i think you would need 13 EQ3 units because each unit only has 1 mid band. i think you would have to have them all running in parallel and then summed, so, i guess you could contain each mid band in a mixer unit?

Ok thanks - so basically it’s only the middle band I should be using for each band eq using an EQ3. So if I want 61 I should be setting low mid and high to 61 and using mid only when I boost? I want a fairly sharp boost at that 1 point. When you say parallel do you mean 29 - 61 etc in a mixer, then 115 - 218 etc in anotherr mixer. Or do you mean 21 115 411 on one side in one mixer and 61 218 777 in another mixer, then summed? What’s confusing me is the description - it’s says it’s a 3 band EQ - low, mid, high and you set the low and high to get the mid.

Update : Ok I found a schematic and can see it’s parallel, then summed, so this needs to be 29 115 411 1.5k 5.2k in 1 mixer unit then 61 218 777 2.8k 11k in another, then summed like you say

so based on the wiki, i think the source of misunderstanding here is that when the EQ3 is said to be a ‘3 band EQ’, what might not be obvious is that there are different types of EQ bands, or in other words, filters. you could think of a graphic EQ as being made up of lots of overlapping bandpass filters like in the bottom part of this picture:

in this graph though, you can see that at the top and bottom of the frequency range, there is a low pass filter and high pass filter respectively instead of a bandpass filter. The low and high bands of the EQ3 unit are equivalent to these filters. On some EQ units these are called high and low shelves.

What you’ve been doing is trying to use the high and low bands of the EQ3 unit as if they were bandpass filters. but as the wiki entry says, the mid band on the EQ3 unit is just the subtraction of the low and high bands from the original signal. The mid band is effectively a bandpass filter, but the others are not. so when you boost the gain on a highpass filter set at 115hz, you’re affecting the entire frequency spectrum above that point, not simply the band between 115 and 218, which is what you want.

If we think about your ideal graphic eq unit in terms of filters, what you need is 1 low pass filters, 13 band pass filters and 1 high pass filter, each one of those with adjustable gain, all of them running in parallel. Alternatively you could have 15 band pass filters running in parallel if you were happy with cutting all frequencies above 11k and below 29hz.

You wouldn’t necessarily need to use the EQ3 unit to create this effect: you could use the Ladder LPF and HPF units, and then you could have some control over Q as well, which you wouldn’t get using the EQ3.

Either way, I think it would actually be best to create 15 mixer units. Inside each of them, set the input to be the original signal you want to process, then add an EQ3 unit or, for the bandpass filters, an LPF and a HPF unit in series. You can then use the gain control on the mixer units as the gain setting for that band. Then you may have some fine tuning to do in regards to crossover points if your goal is to have a flat response when all bands are set to unity.

EQ units can be pretty complex. I may have made some mistakes in the way I’m talking about it here, but this is correct to the best of my knowledge.

3 Likes

Ok great, that’s awesome, thanks for the detailed explanation, that makes a lot more sense :slight_smile: I didn’t realise the differences in EQ designs. @Joe 's custom accents units has bandpass filters so will try that and also lay it out differently with the mixer units. I’m just not sure the er301 would handle cpu wise 15 of them, so will try 10. I was looking at this schematic as to the layout of the EQ bands

1 Like

I see, great to know that Joe’s custom units includes a bandpass, but also good to keep in mind that a bandpass is fundamentally the same as a low pass and high pass in series!

This schematic helps give me a sense of what you meant too. I guess the thing to consider if that in the Serge Res EQ as well, all bands are running in parallel (you can see that the input line splits out to each band, whereas if they were in series, then each band would be fed input from the band preceding it).

the confusing thing here is that there are two summing stages. first, the odd numbered bands are passively summed, and the even numbered bands are summed seperately (you can see how their outputs are just joined together), then the two halves are summed after going through what i think is a high pass filter? (my electronics knowledge is not fully up to scratch to know for sure). I think the reason they use two summing stages is so there can be seperate outputs for the odd and even bands - you can see that the signals are buffered before the two ‘comb out’ outputs. I hope that helps!

1 Like

That helps a lot, thanks. I don’t necessarily want to replicate the Res Eq (I already have one). It was more the specific frequency numbers and general layout of an eq design I was interested in. I do want to use those Serge eq frequency numbers though. I just want this to be a mono, serial design, so more a basic chained eq, rather than a Res Eq clone with all the separate summing, odds/evens and different outs.

1 Like

And that’s really what the custom unit is. The frequency controls are tied together and offset in either direction by the bandwidth control.

2 Likes