(2) Simple RC HP Filters in Series?

[jsteiger]

I have a filter question, or actually [two] I guess. My console has a 3 frequency, passive, 1st order HP filter comprised of 2 pushbutton switches. 1st switch is 100Hz, 2nd switch is 200Hz and combining both(paralleling caps) are 60Hz, or so the manual says. Each switch position is actually inserting 2 different R-C filters in series, into the signal path.

For example, the 100Hz filter is a .22uF C with a 13k R for a 55.7Hz -3dB knee. Immediately following is a .022uF in series with 1uF, so .0215uF C with a 130k R for a -3dB knee of 57Hz. Since it's a first order, -6dB per octave, does this simply move the -3dB knee to essentially 56*1.5, or about a 1/2 an octave higher?

Is this something that is maybe common or maybe anyone has seen or used before? Is there any benefit to keeping the circuit like this? I'm sure API had their reasons. I'm always interesting in learning what they were. I think it seems pretty smart of them after I stumbled across what was actually happening.

Any insight is appreciated.

Oh, I have no plans on changing it, just maybe expanding it a bit.

Thanks, Jeff

 

[abbey road d enfer]

I've always been a little puzzled about this filter section in the API mixer.
According to the schematics, they are 2nd order passive.
RC second-order passive filters are always overdamped, which means the slope is always less sharp than a maximally-flat (Butterworth) filter.
What's more, in the API, the second pole being an order of magnitude lower than the first (same C, but R 10x higher), the resulting transfer function is essentially the same as first-order (6dB/octave). The difference becomes noticeable 10 times below corner frequency.
When not engaged, the -3dB corner frequency is governed by C18 (1u) and R19 (130k), which gives ca. 1.2Hz.
The first filter (supposed to be 100Hz) has a corner freq of ca. 65 Hz.
The second filter, 140 Hz.
When pressed together, the corner Freq is ca. 42 Hz.
See attached frequency response curves.

 

[abbey road d enfer]

Here's the comparison between the actual API curve and a standard 6dB/oct.

 

[jsteiger]

Ah. Thanks abbey.    So, 2 cascaded 1st order filters form a 2nd order filter network. I have found a few useful links that are interesting and have helped me understand a little more.

http://www.johnhearfield.com/RC/RC4.htm

And of course this very helpful calculator. http://sim.okawa-denshi.jp/en/CRCRkeisan.htm

...What's more, in the API, the second pole being an order of magnitude lower than the first (same C, but R 10x higher)...

Actually the 2nd cap value does change proportionally. .22uF and then .022uF. I did notice that the schemo switch labels (Filter 1 & Filter 2) are reversed from the actual module.

Interesting plots. Thanks for doing those abbey.  

 

[abbey road d enfer]

My schemo is almost illegible in places; I had not seen the values of the second caps were divided by 10.
So, i've joined the revised curves; they look more useful.
Still, this is not as efficient as an active RC 2nd-order maximally flat filter.

 

[abbey road d enfer]

Comparison of API (in Green), 1st-order (in blue) and maximally flat (Butterworth) 2nd-order.

 

[jsteiger]

That Butterworth sure has a nice looking knee.

What sim program is that you are using abbey?

 

[dandeurloo]

good thread guys!

I use my UAD plugins and they have a number of different curve options.  I seem to like the butterworth for the  lower shelf's like 20-50hz and below but the api look like they would be really musical for higher settings.

Interesting stuff.

 

[abbey road d enfer]

That Butterworth sure has a nice looking knee.

What sim program is that you are using abbey?

LTSpice, the free spice sim for Linear Technology

 

[jsteiger]

After some further investigation, I see the bandpass filters of the 550a use the same basic structure as in the console. Same for the HP and LP. Interesting.

And to dig even farther, the Melcor gme20 eq uses the same sorta filters for the HP and LP. They just use the same exact values for each instead of multiplying and dividing by 10. Those are independent of each other unlike the single "Filter" toggle on the 550a.

I guess I better take a closer peak at a 550, for ***** and giggles.

Who ever said filters couldn't be fun? 

 

[abbey road d enfer]

The resulting curve with equal values is even more damped than the API.
Green: Butterworth, Blue: API, Red: equal values.
So it appears less usefut at removing rumble because it reacts with low-midrange.

 

[abbey road d enfer]

I use my UAD plugins and they have a number of different curve options.  I seem to like the butterworth for the  lower shelf's like 20-50hz and below but the api look like they would be really musical for higher settings.

Interesting comment! Maybe the explanation is in group delay (although audibility of group delay has never been proved, neither un-proved).
See attached curves (Green: Butterworth, Blue: API, Red: equal values). Pls note different scale.
See how the API is the one with the smoothest variation.

 

[dandeurloo]

yeah that is interesting.  the other thing to consider, who knows what is really happening in the plugins!