I had no idea this would cause this much rumble among everybody reading this blog 😉
I started this after receiving questions from a couple of guy’s regarding crossovers and filters. Whenever i do not know an answer i’ll try to find out (if i have time to do so 😉 ). Since the Dutch government has decided to put the whole country in lock down i had some “spare time”.
Here are the blog posts before this 1 on the subject https://timobeckmangeluid.wordpress.com/2020/12/06/crossover-filters/ https://timobeckmangeluid.wordpress.com/2020/12/11/crossover-filters-part-2/ https://timobeckmangeluid.wordpress.com/2020/12/16/processors-part-3-lake-lm26/
Here’s a link to an explanation on the Lake raised cosine stuff in GERMAN by FELUSCH
btw: i’m not the first 1 to do a “compare” like this. Bennett Prescott did the same thing a decade ago. His findings can be found at the Sound Forums Network
This post is not to bitch or tell everybody the processor brands i played around with are crap or really good. It’s just to make people aware of copy/pasting settings between brands might give different results which can be easily corrected via measurements.
So first let’s take a peek at a Parametric EQ point: +6dB Bandwidth 1 octave set at 1kHz.
There’s sort of a difference between the 3 brands. I did a “Google thing” on Bandwidth/Q-factor and found this:
Electronic filters The numerical difference between the upper and lower -3 dB points of a band of audio frequencies. Used to figure the Q, or quality factor, for a filter.
So the way this translates in to something meaningful for me: peak is set at +6dB and the Bandwidth/Q-factor would be determined by the 3dB points below the peak. If i take a look at the PEQ +6dB BW 1oct measurement from the Lake LM26 it looks like this
If i then offset the measurement 3dB down i can get sort of scientific and read out the frequencies determining the bandwidth (hope this makes sense 😉 )
Nice thing: if i can do it once i can do it 2 more times 😉 . Here’s the same thing from the Apex Intelli-X2
And now the science crap: offset for the 3 dB below peak thing
And the Linea Research ASC48:
And science stuff:
And all in a slide show with the offset -3dB on the 3 processors with the frequency points 😉
Crossover filters:
So first let’s start with what i’ve learned on the first seminar i attended which started my journey in measurement world:
Per filter order 45º of phase shift at the crossover frequency and over the pass-band a doubling of that. This means if i do a 2nd order Linkwitz-Riley high-pass filter i should (and will) end up with 90º of phase shift at the crossover frequency (2 x 45º=90º) and a 180º over the whole passband (2 x 90º).
Butterworth high and low-pass filters end up at -3dB at the crossover frequency for whatever order you use with the phase response following the 45º of phase shift @ the crossover frequency and a doubling of that on the whole pass-band per filter order (so a Butt 3rd wil have 3 x 45º=135º of phase shift @ the crossover frequency and 270º over the pass-band) .
Linkwitz-Riley always ends up at -6dB at the crossover frequency and only comes in even orders so 90º/180º/270º/360º degrees at the crossover frequency and double on the whole pass-band.
There are more exotic filters (Hardman/Linea Research Elliptic/Meyer Galileo Neve-Tiel/BSS) but i have not compared them between brands so those are not in this blog post.
For the Linkwitz-Riley and Butterworth crossovers from all 3 processors i tested the results where pretty much similar. However with the Bessel filters all 3 processors differed in the response. The Linea gave a response the way i learned these filters but there are only 4 (2nd 3rd 4rd and 8th).
With the Apex the magnitude respons seems to be leading meaning the -3dB point is always at the crossover frequency. You can match them with the Linea by moving the crossover frequency up in case of a high-pass filter and down in case of a low-pass filter. When the magnitude respons matches the phase respons also matches between them.
The Lake LM26 gives the same respons on the Bessel high-pass filters compared to the Linea but the 1st 3rd 5th and 7th order low-pass filters are polarity inverted (no idea why but this comes in handy when doing cardioid sub set-ups https://timobeckmangeluid.wordpress.com/2018/08/24/a-prelude-to/)
I’ll only post the screenshots from the Bessel Low-pass 1st 3rd 5th 7th order from all 3 units. The rest will be available for download at the bottom of the post.
All-pass filters
The first order filters from the LM26 and the Apex Intelli-x2 are similar so a 90º of phase shift is introduced at the frequency of choice in this case 1 kHz and a 180º over the pass-band. The difference in the upper range comes from a 0.01ms time offset (if you wonder why attend a Smaart training 😉 )
The 2nd order All-pass filters from the lake Linea and Apex implement a 180º of phase shift at the frequency of choice and 360º over the pass-band with a Bandwidth control. Again there’s a difference on the Bandwidth when set to 0.33 and/or 1 octave.
With the Linea there’s another option: FIR filters. At the cost of some latency (in this case 1.98ms) you’ll get more control over the phase (if needed).
So first the FIR 1st order All-Pass filters minimal and maximal phase standard and inverted via FIR Designer M
And the FIR 2nd order All-Pass filters minimal and maximal phase standard and inverted via FIR Designer M. There are some examples of a Truncation error in some of the FIR files (afaik not enough taps and filter delay for a clean respons).
Via FIR Designer M there’s a 4rd order All-Pass filter so let’s get crazy for a moment….;-)
The TRF data from Smaart V8.4 can be downloaded below. (it’s a lot 😉 )
Pingback: PROCESSING PART 4: POWERSOFT X8 Dante | Timo's World all around the globe
Halo,
As far as SQ, how would u rate them? I am thinking to get Linea as it has LIR and FIR as well.
Thanks
Al do what they are supposed to do