advertisement


Digital Equalisers, Phase, and other possible Damage to Sound

Hipper

pfm Member
Digital Equalisers, Phase, and other possible Damage to Sound

Please note I used 'Digital' in the title. I'm not interested in analogue EQ.

I use a Behringer DEQ2496 only in digital mode, located between my CD Transport and DAC. I therefore am only interested in the effects of digital EQ. I use it just for completing the EQ effects of positioning and room treatment.

One of the criticisms levelled at the use of digital EQ is that it only works in the frequency domain but not the time domain. I have also been told it ‘smears transients‘. What does this mean?

Transient

A musical note consists of '.... the attack, sustain, and decay of a sound. Attack transients consist of changes occurring before the sound reaches its steady-state intensity. Sustain refers to the steady state of a sound at its maximum intensity, and decay is the rate at which it fades to silence.' So the transient is the first thing you hear and is very noticeable on percussion (e.g. bongos), piano, and plucked guitar for example.

Each transient will consist of a number of frequencies (known as harmonics) and in order to get the proper effect they should be heard at pretty much the same time. In other words, when they reach your ears at your listening position they should ideally be heard together in one sharp ‘thud’. If these frequencies arrive spread out over time the transient is said to be ‘smeared‘. You don’t get that ’thud’ but a longer 'thudddd'.

Smeared Transients

How, in theory at least, can an EQ smear transients? Well, apparently they work by changing the phase of the frequency that is adjusted:

‘In the beginning all equalizers were analogue electronic circuits using capacitors and inductors. These components shift the phase of AC signals passing through them. If you combine a signal with a phase shifted version of itself (after passing through the capacitor or inductor), the frequency response is altered. As one cycle of the wave is rising, the shifted version is falling, or perhaps it hasn't yet risen as high. So when the two are combined they partially cancel at some frequencies only thus creating a non-flat frequency response. Therefore analogue equalizers work by intentionally shifting phase, and then combining the original signal with the shifted version. In fact, without phase shift they would not work at all!

Most digital equalizers mimic the behaviour of analog equalizers, but with a completely different circuit design.
’

http://ethanwiner.com/EQPhase.html

(I think he’s writing about studio EQ but presumably the Behringer and other digital EQs work the same)

So that is why it is said that EQs smear transients. Different frequencies are altered by different dBs and so each one's phase is changed differently. Now all the harmonics of a transient no longer arrive at your ears at the same time. You get ‘thudddd’ instead of ‘thud’.

The problem is that I don’t hear it! I don’t hear the EQ smearing transients. I’ve tried. I used ‘bypass’ on it. I took it completely out of the system. I played test recordings of bongos, piano, a drum kit (from Alan Parsons ‘Sound Check 2‘ CD). I could hear no transient changes. It wasn’t easy because I like and am used to the sound with EQ, but I heard nothing negative.

My question then. Has anyone who has actually used digital EQ in their system heard smeared transients, or other phase issues, or any other artifact that caused the sound to deteriorate?
 
I haven't used a Behringer, but the equipment I have used has never had any detrimental effect whatsoever,in fact the improvement gained by improving the room's acoustics have been huge.
Frequency and time are two sides of the same coin, if you remove a huge resonance , you improve the frequency response and the rt of the room.
Keith.
 
Well all filters, weather that is a crossover, the box the drivers are in, a port, or humble EQ create phase shifts that are proportional to the amplitute shift created by said filter. There is no getting away from that, it's maths and physics in action. It doesn't matter if you do this in analogue or digital. Digital is most definately not trying to mimic how analogue shift phase, either technique produces the same amount of shift. Digital filter have until recently been IIR type filters, recently equipment like the miniSHARC from MiniDSP has allowed us to use FIR filters, they are something of a holy grail in that for the first time is is possible to affect amplitude without a corresponding change in phase, or indeed phase without amplitude. This allows us to create linear phase crossovers or indeed correct the phase of the box tuning of the speaker cabinet. FIR does come with it's own set of problem and drawbacks though.

Stefan
 
I used to own a 2496, and even with it supposedly doing nothing at all, IMO made things sound worse. That's using it digital in from my transport, then onto my DAC.
 
Using DSP you will have at least one of the following issues:

Time delay;
Phase distortion;
Pre/post ringing.

So it will basically be a tradeoff.
 
Whatever the detrimental effects of digital eq may be, they need to be put into context.

If it is not practical or desired to have passive room treatments, then digital is the only option to deal with sometimes (actually often) gross acoustic problems.

Live with maybe +10dB peaks and decay times maybe 1-2 seconds longer at certain frequencies, or live with (IMO) minor issues created by EQ.

Also the other thing to bear in mind is that your speakers and crossovers are buggering around with things like phase big time.

Context.
 
[
One of the criticisms levelled at the use of digital EQ is that it only works in the frequency domain but not the time domain. I have also been told it ‘smears transients‘. What does this mean?

In most contexts and most applications this is all pretty much BS. Big BS.

1)

Analogue filters are minimum phase (MP). This means that their amplitude-frequency and phase-frequency responses are unambiguously linked (Hilbert). A speaker driver is MP. This means that if it does not extend to DC and to infinity it shifts phase. A leg of a crossover is MP. A phono cartridge is (largely) MP. A room is (significantly, in the modal region) MP. (As an aside: a multiway speaker is not MP, it is a mess.) An analogue equaliser, and a digital one which is modelled on analogue transfer functions, as most parametrics are (for good reason), is MP.

All of these distort phase. But if you apply the inverse EQ, then the phase distortion is also compensated. Which is why the correct measurement of the problem and the correct programming of the EQ is rather important.

2)

Yes, non-linear phase distortion is audible. But you need quite an awful lot of it, particularly you need a lot of phase shift over a small frequency band, to hear it. This can be demonstrated easily with a decent DAW. Further, non-linear phase distortion is harder to hear when using speakers in a reflective room and when listening to real music, as opposed to headphones and specific test signals.

There is no reason to worry about phase in decent linear electronics (i.e. amps, DACs, ...) and EQ. Speakers are something else, as I said before. Around crossover they are a mess, but phase is likely not even the worst offender here.

Of course a system's phase response must be identical for the two channels. Inter-channel deviation translates in a skewed and wandering stereo image. The auditory system is sublimely sensitive to inter-aural time differences. One of the major benefits of well-done speaker-room equalisation is that the musicians take their proper places and stay put there!


3)

All filters/EQ work in the frequency domain and the time domain. Quite how the opposite notion got hold eludes me. Well, it does not. Most don't realise that frequency and time are not independent entities, and audio companies cunningly exploit this.

Oh, you can set up an EQ by just looking at the frequency domain, which in the case of room EQ would be wrong because the impact of late reflections should not be included, or you could be smarter and let the time domain inform you. But the EQ itself, it just works.

4)

Digital filters can also be linear phase, meaning they don't distort phase, apart of a constant delay for all signal (BTW, FIRs have been around for decades, and they are not a holy grail). This seems neat. It isn't. Depending on the application this is good or bad. For a speaker crossover it might be good, though likely still not optimal. For compensating a minimum phase aberation elsewhere in the system it is decidedly bad.

And yes, linear phase filters have post-ringing (MP filters do, too), and the 'unnatural' pre-ringing. Pre-ringing can be audible. It is audible when the ringing frequency is in the audible band and when at the same timing the filter slope is sufficiently steep; outside these conditions pre-ringing is not an issue. But this limits the application of FIR filters somewhat.
 
Most of the good RC systems seem to FIR and IIR implementations .. with a healthy dose of "perceptual" corrections thrown in.

I have used sigtech , tact , lyngdorf , krk ergo , z-sys rdp-1 (remeber that tnt thing werner), behringer , Acourate , Dirac , just about all the miniDSP stuff and many other eq/rc systems

The overriding result is better sound ..especially after curing bass issues. Im not that convinced that electronic eq or RC is the answer for anything above 3-500 hz apart from personal taste.

The one thing that most newbies to RC or eq don't get is that it is NOT ok to correct to flat at listening position..you have to have a preference house curve to overlay the "correction" ..lots of ppl have shown me a flat curve they got after manipulation and they wonder why it sounds like doodoo
 
Basically, Werner, there's so much phase mess because of crossovers, room etc., that any possible damage by EQ is negligible.

Is that a fair statement?
 
Sort of. Even in the absence of speakers and rooms normal levels of phase distortion are hard to hear on normal music.

And again, if EQ is used to compensate for something else in the system, then the correct characterisation of that something else and the correct setup of the eq yields net zero phase distortion.

Take LP as an example. During cutting the inverse RIAA eq imparts large phase distortions. During replay the straight RIAA eq corrects these. Given a perfect/ideal cutter and a perfect/ideal cartridge, the result is linear phase. Of course you won't find perfect cutters, and even fewer perfect cartridges ...
 
Basically, Werner, there's so much phase mess because of crossovers, room etc., that any possible damage by EQ is negligible.

Is that a fair statement?
I don;t presume to have Werner's level of understanding but it seems to me that in terms of time domain errors it goes beyond that-

EQ is capable of overcoming some of the errors, particularly at low frequencies: it's not just that the EQ's time domain effects are masked by the room's effects. Broadly speaking correcting for a room mode will mitigate the time domain effects of the mode as well as the frequency domain ones.

Equally it is possible to mess things up audibly.
 
EQ is capable of overcoming some of the errors, particularly at low frequencies: it's not just that the EQ's time domain effects are masked by the room's effects. Broadly speaking correcting for a room mode will mitigate the time domain effects of the mode as well as the frequency domain ones.

Equally it is possible to mess things up audibly.

So, correct the major anomalies in low bass but leave it there?

https://www.youtube.com/watch?v=5dhSdnDb3tk
 
Here I have been comparing passive treatment, tuned membrane traps , DRC ( Dirac) and
EQ ( Illusonic) ,the Illusonic has literally just arrived ,and I will continue to compare, but I a pretty certain I prefer the EQ /passive .I have tried parametric EQ before, but this time the equalisation was created by Christof Faller ( Illusonic) .
SQ now is the best I have ever enjoyed, but all three solutions are vastly more enjoyable than the untreated room.
Keith.
 
So, correct the major anomalies in low bass but leave it there?

In the bass one can use eq to correct the combined speaker/room, based on in-room measurements, and for a limited listening area.

This should not be done in the midrange and treble, for two major reasons:
-the measurement includes the room reflections, which become increasingly non-minimum phase and are incorrigible
-the correction would only be valid for an unfeasibly tiny listening spot

However, given exact knowledge of the speaker's anechoic response on the axis of interest one could use eq for correcting the larger trends in this with eq. What this does to the off-axis response and thus the resulting sound field in the room remains to be seen, though.



Seventies/eighties practice: pump pink noise in the room, set mic at listening position, observe real-time bar-graph spectrum analyser and tweak graphic eq sliders until the analyser is flat. No wonder eq got a bad name.
 
EQ ( Illusonic) ,the Illusonic has literally just arrived ,

'depth/immersion control'

Did they drop a K-stereo processor into their signal flow???


Only last week did I cook up a Lyngdorf hack to extract a proper Hafler-like rear channel. The result was instantly addictive, especially on live material. It would be great to extract K-stereo like ambience and send it to rear speakers. One could be fully immersed in the sound of orthaudoxiophiles howling how sacrilegious this all is.
 
Here I have been comparing passive treatment, tuned membrane traps , DRC ( Dirac) and
EQ ( Illusonic) ,the Illusonic has literally just arrived ,and I will continue to compare, but I a pretty certain I prefer the EQ /passive .I have tried parametric EQ before, but this time the equalisation was created by Christof Faller ( Illusonic) .
SQ now is the best I have ever enjoyed, but all three solutions are vastly more enjoyable than the untreated room.
Keith.
Hi Keith
Have you tried using Dirac with the genelecs and with the Gen's own drc switched off?
 
Not corrected - summarised for my understanding.

Also an excuse to get the best Disney flick in there ;)

@Werner - thanks.
The Disney Flick with the two best bits in it, possibly. Ok- quite a few good bits. The beatles/Vultures, though....
 
Hi Keith
Have you tried using Dirac with the genelecs and with the Gen's own drc switched off?

Adam Hi, no I haven't, Genelecs long gone, I do like them but Thomann sell them at almost the same price we buy them!
Also I much prefer the Grimm LS1's, they excite the room far less than the Gens
and are just far easier to connect ,control integrate etc.
You should come and listen to the new Kii speaker when they arrive, if they live up to their review ,they could be special and not expensive, well not too expensive.
Keith.
 


advertisement


Back
Top