Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Adam Dempsey]

Purely out of curiosity, does anyone know of a link to, or documentation of, the Forsell-mod NSEQ EQ response curves?
Reason for asking is that the freqs on bands 1 and 4 sound higher to me than indicated, in part perhaps due to their width (specified as Q=1.0 in the stock NSEQ-2, which to my knowledge the mods don't alter) but it also sounds wider than this. Yesterday a slight bell boost on band 1 at 34Hz was doing nice things right up into the lower mids of a female vocal part! Anyway, just curious...

[jdg]

i did an xfer function to find out the actual gain steps once out of curiosity (1/4ish dB per "click"). i can do it again very easily to find the freqs.

i'll do it when im back in the studio in the morning

[Jerry Tubb]

Adam Dempsey wrote on Thu, 04 February 2010 18:16

Purely out of curiosity, does anyone know of a link to, or documentation of, the Forsell-mod NSEQ EQ response curves? Reason for asking is that the freqs on bands 1 and 4 sound higher to me than indicated, in part perhaps due to their width (specified as Q=1.0 in the stock NSEQ-2, which to my knowledge the mods don't alter) but it also sounds wider than this. Yesterday a slight bell boost on band 1 at 34Hz was doing nice things right up into the lower mids of a female vocal part! Anyway, just curious...

Good question Adam, perhaps one for Fred Forssell himself.

I've also noticed they're pretty wide, but don't have a clue as to how to accurately measure them.

A few years ago I ran some tones to find the frequency centers, but that's a different subject : - )

Cheers - JT

[mastertone]

Most probably you have seen this, but its the documentation i could find.

http://www.forsselltech.com/products/NSEQ-F_testData.pdf

You have the curves there, last page.

[Andrew Hamilton]

Adam Dempsey wrote on Thu, 04 February 2010 19:16

Yesterday a slight bell boost on band 1 at 34Hz was doing nice things right up into the lower mids of a female vocal part!

Good day, Adam,

Well, we are talking about a non-linear phase eq.  I find interesting maskings and unmaskings can happen with any eq setting, anyway.  Whenever you boost or cut program with a normal eq, you can't help but change a number of things simultaneously, including the phase relationship of the affected area with the rest of the signal. It's all about what sonic balloon animals you can create (or destroy) with phrequency tweaks.  (The "ph" is for "phase.")


As for the Forsell .pdf, I found the information on the last page particularly illuminating.  (l;




Cheers,
    Andrew

[zmix]

I think that you could obviate a lot of speculation and subsequent half-truths by simply measuring the response.

The Q can easily be measured by using a signal generator and a meter.  

Use the following formula to determine Q:

Q= Fc / (Fh - Fl)  

with

Fc = center frequency.
Fh = high frequency -3dB point of the bell.
Fl = low frequency -3dB point of the bell.  


Alternately a program like Smaart

or Room EQ Wizard

coud generate the required graphs.

[jdg]

here you go:
http://panicstudios.com/tmp/nseq-f/

the low band and high band in "bell" mode
6dB range, each setting max gain.

labeled by freq

[Jerry Tubb]

zmix wrote on Fri, 05 February 2010 10:09

Use the following formula to determine Q: Q= Fc / (Fh - Fl)   with Fc = center frequency. Fh = high frequency -3dB point of the bell. Fl = low frequency -3dB point of the bell.

Aha, there's that simple but elusive formula!

Muchas Gracias Chuck.

JT

[Adam Dempsey]

mastertone wrote on Fri, 05 February 2010 19:10

Most probably you have seen this, but its the documentation i could find. http://www.forsselltech.com/products/NSEQ-F_testData.pdf You have the curves there, last page.

Thanks Jonas, though it's not clear there if the boost/cut graphs are only for the 2 parametric bands (I suspect), or the end bands.

[Geoff Emerick de Fake]

zmix wrote on Fri, 05 February 2010 10:09

I think that you could obviate a lot of speculation and subsequent half-truths by simply measuring the response. The Q can easily be measured by using a signal generator and a meter.   Use the following formula to determine Q: Q= Fc / (Fh - Fl)   with Fc = center frequency. Fh = high frequency -3dB point of the bell. Fl = low frequency -3dB point of the bell.

This formula is valid only for a bandpass filter with zero response at the asymptotic slopes (mathematically the response is as/bs

[Adam Dempsey]

Thanks John! Spot on. Interesting.

Andrew Hamilton

Well, we are talking about a non-linear phase eq. I find interesting maskings and unmaskings can happen with any eq setting, anyway. Whenever you boost or cut program with a normal eq, you can't help but change a number of things simultaneously, including the phase relationship of the affected area with the rest of the signal. It's all about what sonic balloon animals you can create (or destroy) with phrequency tweaks. (The "ph" is for "phase.")

Yes, totally aware of this and I can see on the graph where +6dB gets things shifting 125-250Hz. Thing is I was hearing it with just 2 clicks (+1.2dB with the f mod depending on how the internal jumpers are set).

zmix

I think that you could obviate a lot of speculation and subsequent half-truths by simply measuring the response. The Q can easily be measured by using a signal generator and a meter.   Use the following formula to determine Q: Q= Fc / (Fh - Fl)   with Fc = center frequency. Fh = high frequency -3dB point of the bell. Fl = low frequency -3dB point of the bell.

Yes Q = f/bw (bw measured from the -3dB points). Just never seen the plots published or had the downtime to do the measurements since hearing what I heard.

Geoff Emerick de Fake

It is evident that the formula does not work for any amount of boost/cut inferior to 3dB.

Can you explain this further?

Jerry Tubb

A few years ago I ran some tones to find the frequency centers, but that's a different subject : - )

And thanks Jerry, yes I'd already kept a copy of the centre freq's you'd measured. All-in-all this EQ is seriously all about fine tweakage by ear. Like I said, purely curious. All good. And now here for posterity.  

[Geoff Emerick de Fake]

Adam Dempsey wrote on Fri, 05 February 2010 18:17

zmix  The Q can easily be measured by using a signal generator and a meter.   Use the following formula to determine Q: Q= Fc / (Fh - Fl)   with Fc = center frequency. Fh = high frequency -3dB point of the bell. Fl = low frequency -3dB point of the bell. Yes Q = f/bw (bw measured from the -3dB points). Just never seen the plots published or had the downtime to do the measurements since hearing what I heard. Geoff Emerick de Fake It is evident that the formula does not work for any amount of boost/cut inferior to 3dB. Can you explain this further?

Since this formula implies finding two points on the curve that are 3dB down the peak (for a boost situation), it is clear that the boost needs to be at least 3dB. In fact, if the boost was 3.1dB, the -3dB points would be so far that the calculated Q would be very low.
This point regarding the ambiguity of the definition in the case of equalizers is a serious concern for manufacturers of so-called loudspeaker management systems: BSS, dbx, XTA, Lake, ...they are trying to make the AES interested in the subject and define a procedure. They are concerned because they all have differing definitions and that creates a problem of portability of loudspeaker presets from one brand to another.

[Adam Dempsey]

Of course. I was over-thinking it... that there was a more complex mathematical reason, just like I used to in Physics class many yrs ago.

[Jerry Tubb]

jdg wrote on Fri, 05 February 2010 11:19

here you go: http://panicstudios.com/tmp/nseq-f/ the low band and high band in "bell" mode 6dB range, each setting max gain. labeled by freq

So from these LF curves, I deduce and surmise, according to Chuck's formula that the Q ~ 0.65.

Correctamundo?

JT

[Adam Dempsey]

@ 34Hz:
Q = f/bw
≈ 34/(85-13)
≈ 0.47