Does using EQ degrade sound quality in Wiim Pro?

[slartibartfast]

Nobody mentioned autogain, at least not that I've seen - the current EQ overload protection implementation looks like soft knee brickwall limiting to me.

I appreciate that some might prefer autogain or manual pregain to this, but note that this kind of implementation has its benefits too.

Either way I hope we can agree it is much better than hard digital clipping (which is what I expected prior to testing).

It sounds very like the "Soft Clipping" of the old NAD 3020.

 

[dominikz]

Why is flat topping below 0dB different from a distortion perspective than hard clipping at 0dB? The waveform must be very similar.

Limiters have a response time (attack/release) at a predefined threshold (probably a bit lower than 0dBFS), and can have a softer knee in the transfer function.
So they can better preserve the waveform shape compared to hard clipping (which is like a limiter with zero attack/response time and a hard knee at 0dBFS so it just cuts the waveform flat).

 

[canard]

drc (in digital stage)
even the gec mode presets "wiim" do not manage the gain

 

[dominikz]

drc (in digital stage)
even the gec mode presets "wiim" do not manage the gain

I don’t really understand if you're agreeing or disagreeing with my analysis.

I agree that it seems WiiM is using some kind of DRC (=dynamic range compression) in the digital domain just after EQ - if that is what you're trying to say? Note that a limiter is a type or DRC.

This is IMHO a lot better than clipping; though I can understand that some people would prefer autogain which would reduce the overall level instead and avoid any DRC. Either approach has merit.

 

[canard]

I don’t really understand if you're agreeing or disagreeing with my analysis.

I agree that it seems WiiM is using some kind of DRC (=dynamic range compression) in the digital domain just after EQ - if that is what you're trying to say? Note that a limiter is a type or DRC.

This is IMHO a lot better than clipping; though I can understand that some people would prefer autogain which would reduce the overall level instead and avoid any DRC. Either approach has merit.

just a form of discrete circumvention of a subject which could be regulated differently, in a classic autogain way, with the famous 32bit volume which Wiim boasts of... be careful there is also a strong variability of engagement of the limiter linked to the correction values, that's annoying....
here i use just +2db.. but can more early if correction is more and if "a zone of fr "with level.. (and ls/hs more than q0.7 is a problem)

(ps we can always observe the inaccuracy of the graphic curves often pointed out curves in Wiim Home "55")


....but as I said above...it's not a problem for me who can calculate, very fast, the acceptable gain quickly and apply it manually... for correction headphone for me
(can also be done in software like rew, the peq part etc as has often been pointed out over the years.... )
so everyone will see the situation like he want...
no more to say for me..
;-)

 

[canard]

(Aside
Just a fun little thing...very, very often forgot...in many cases, test signals can also be listened to... Not just measured...
it's sometimes interesting but most of the time forgotten...
IMD in particular... It was part of their interest....
it was done a lot in the 70s and 80s... ;-)

)

 

[onlyoneme]

Looks like return of the limiter indeed How does it work now for a full scale signal with 0 dB flat PEQ enabled?

 

[canard]

Looks like return of the limiter indeed How does it work now for a full scale signal with 0 dB flat PEQ enabled?

yes...(and seem not) ;-)
"el limiter"
but the opportunity to keep yourself busy "healthily" ;-)))

I don't want to stop you from having fun
gooooo

;-))

(but there is also the monitoring of the resampling modes... that is more "annoying" and time-consuming... the story of the limiter is easy-quick to do... (especially with my quick and easy approach in the way of sweep measurements ;-) and screen impact of the ultra...in details..? ;-))

(I remind you that I told you, at the time, that in the absence of autogain... like domikz, I think that it was not the worst of the "workaround" solutions in gec or peq mode... " mieux que rien")

 

[harkpabst]

and screen impact of the ultra...in details..?

Sure.

 

[onlyoneme]

I'm not a fan of the limiter here, I still remember a TD in the range of -40 to -50 dB for IMD test. And for me it looks like a hard knee limiter rather than a soft knee. I'll try to make some tests on my own also when possible.

 

[canard]

I'm not a fan of the limiter here, I still remember a TD in the range of -40 to -50 dB for IMD test. And for me it looks like a hard knee limiter rather than a soft knee. I'll try to make some tests on my own also when possible.

if in the digital domain, limiter, it etc is rather easy...
but for the screen for example it is possible that the margin of the cosmos-( dac rme , okto ) is needed that I do not have for the ana...
you will have fun ;-)

 

[Nanook]

IMHO it is a very reasonable compromise - average level of most content will not be near 0dBFS anyway, and it is similarly unlikely that most users would keep volume at 100%/fixed level all the time.
For the rare cases where these conditions do happen, WiiM implementation of EQ still avoids clipping by using a graceful form of limiting output, without having to decrease playback level. Even in such cases the change in tonality due to this implementation might not be very audible, since it would likely only affect waveform peaks in the content.

Have to say it is a very interesting and clever approach! I'm impressed, nice work @WiiM Team!

I was not fully happy to see these FullScale frequency responses in @dominikz 's post, although the implementation is very clever and clean. It avoids clipping and it doesn't lower the overall maximum level which might be a limitation in one or the other setup.

My plan is to use the WiiM Pro Plus as streaming source and as "Room-Correction-insert" in the tape-loop (record out -> Line In, Line out -> tape input) of my integrated Arcam amplifier. The analog Line-In of the WiiM allows me to route Tuner and Phono through the room-correction equalization as well as my CD-player (when connected to the Toslink input). So RC is performed on all sources.
Volume control will be done on the Arcam and I initially thought I will run the WiiM at Vol=100% and would thus have the potential level limitation shown in dominikz's post when using boosting EQ settings.

I did some measurements playing with EQ, incoming digital level, output voltage setting, Volume-Limit and Volume.
Signal chain was REW -> SMSL PO100AK -> Toslink -> WiiM Pro Plus Line out -> Cosmos Scaler+ADCiso -> REW



(1) digital 0dBFS, Vout=2Vrms, no EQ, VolLimit=100%, Vol=100%: This is the maximum output level. (I have no idea why the 0dBFS frequency response shows artifacts around 5kHz. The REW then stopped the sweep because it detected excessive distortion. I will have to look into this)
(2) digital -3dBFS, Vout=2Vrms, no EQ, VolLimit=100%, Vol=100%: 3dB down as expected
(3) digital -6dBFS, Vout=2Vrms, no EQ, VolLimit=100%, Vol=100%: 6dB down as expected
(4) digital -6dBFS, Vout=2Vrms, +12dB@1kHz, VolLimit=100%, Vol=100%: Limits level to FS as already shown by @dominikz.
(5) digital -6dBFS, Vout=0.5Vrms, +12dB@1kHz, VolLimit=100%, Vol=100%: Limits level to FS, which is consequent if you want to reliably limit the max. output level and thus max. power.
(6) digital -6dBFS, Vout=2Vrms, +12dB@1kHz, VolLimit=90%, Vol=100%: seems to be exactly -6dB compared to 100% and together with the -6dBFS digital input level this thus allows the extreme EQ to pass without getting limited.
(7) digital -6dBFS, Vout=2Vrms, +12dB@1kHz, VolLimit=80%, Vol=100%: another 6dB down resulting in 6dB headroom as expected
(8) digital -6dBFS, Vout=2Vrms, +12dB@1kHz, VolLimit=100%, Vol=80%: about 0.5dB lower than (6)
(9) digital -6dBFS, Vout=2Vrms, +12dB@1kHz, VolLimit=80%, Vol=80%: about 6.5dB lower than (7)

Observations:
The volume slider still goes from 0% - 100% when Vol-Limit ist set below 100%. I like this approach because it allows a finer volume adjustment and it would be irritating getting stuck at e.g. 80%.
It seems like Vol-Limit and Vol scale differently but it's likely that I have accidentally changed the digital level when adding measurements (8) and (9) later on.

I now plan to limit the volume to e.g. 80 or 90% to account for the boost PEQ filters I have in my room-eq. The APO-EQ that I currently use calculates and recommends a Preamp gain of -7dB. Given the excellent AK4493SEQ based DAC in the WiiM Pro Plus I have no concern at all to attenuate a couple of dBs using "Vol-Limit" to allow for headroom for the EQ and for intersample peaks.

I assume, that in case an external DAC is used, the behavior at the digital outputs will just be the same as for the analog output. I will look into this when I find some time.

 

[Nanook]

The digital output (I used Toslink) behaves identical.
Volume and Volume-Limit indeed scale differently. See attached graphs with 0dBFS Sinewave and +12dB graphic EQ @ 1kHz.

Volume-Limit decreases level by 6dB every 10%


Volume uses different stepsizes:


Volume and Volume-Limit attenuate "on top of each other", so you can just add the attenuation.
I do like this behavior. One of my setups just consists of a WiiM Pro + DAC + active speakers. When I want listen music just in the background the 1% step of the volume still is too loud (assuming 110dB SPL @ FS we still have 60dB SPL at the lowest volume setting). Using a Volume-Limit of e.g. 50% in that situation allows me to get another 5*6dB = 30dB down.
(I'm fully aware, that the signal quality suffers when using this much digital attenuation, but it's a 24bit signal-path and when listening at these low levels artifacts due to the quantization are not audible anyways).

 

[Nanook]

@Burnside: You're right - Volume scaling was already addressed in a different thread: https://forum.wiimhome.com/threads/volume-scaling.6771/#post-116665
I should rather have posted results in that thread, but wanted to make the interaction of boost-EQ and the volume / volume-limiting on the potential (but very rare) case that a near full-scale frequency occurrence coincides with a boost-EQ visible.

Those, who use heavy boost-EQ and do volume adjustment further downstream (and may thus operate the WiiM at fixed output = 100%) might consider to use the Volume-Limit setting as a Pre-gain just like the APO-EQ and other EQs do.

I was actually surprised by the way the Volume-Limit works. The name of this setting suggests, that it limits the range of the volume control, but the implementation rather is something I'd call a "Pre-gain".
For me this implementation is just perfect, I would just give it a different name.

The output level setting (2Vrms down to a couple of 100mVrms) rather does something like limiting. It indeed makes sure the output does not go beyond the setting even with heavy boost-EQ ( see trace 5 in post #72).

 

[dominikz]

@Burnside: You're right - Volume scaling was already addressed in a different thread: https://forum.wiimhome.com/threads/volume-scaling.6771/#post-116665
I should rather have posted results in that thread, but wanted to make the interaction of boost-EQ and the volume / volume-limiting on the potential (but very rare) case that a near full-scale frequency occurrence coincides with a boost-EQ visible.

Those, who use heavy boost-EQ and do volume adjustment further downstream (and may thus operate the WiiM at fixed output = 100%) might consider to use the Volume-Limit setting as a Pre-gain just like the APO-EQ and other EQs do.

I was actually surprised by the way the Volume-Limit works. The name of this setting suggests, that it limits the range of the volume control, but the implementation rather is something I'd call a "Pre-gain".
For me this implementation is just perfect, I would just give it a different name.

The output level setting (2Vrms down to a couple of 100mVrms) rather does something like limiting. It indeed makes sure the output does not go beyond the setting even with heavy boost-EQ ( see trace 5 in post #72).

Thanks a lot for making these measurements! I find it really surprising that the "Volume Limit" seems to use a different curve in the WiiM Mini compared to Ultra and Pro! You can see here that on the Mini "Volume" and "Volume Limit" both use the same steps.

One more thought - there's also the per-input "Pre-Gain" control in the "Audio Input" menu. This one is already in dB and has a range +/-10dB, so it doesn't require conversion. Another benefit is that it is assigned per-input - just like EQ is assigned per-input. So far I didn't test how it interacts with EQ, but I suspect it will be equivalent to "Volume" and "Volume Limit".

 

[dominikz]

(I'm fully aware, that the signal quality suffers when using this much digital attenuation, but it's a 24bit signal-path and when listening at these low levels artifacts due to the quantization are not audible anyways).

Note also that the WiiM internal processing is actually 32bit (source), so there is no additional quantization noise added by the digital volume control.
System noise floor won't be affected by WiiM volume/volume limit so you're good as long as you can't hear the system's noise floor at your listening position.

In practical terms, assuming reasonable gain-staging, there's really no detriment to signal quality at all by decreasing the volume or volume limit in WiiM.

 

[Nanook]

@dominikz: Thanks for pointing me to the "per input Pre-gain".
I do as well expect it to work like the Volume or Volume-Limit does, I will give it a try how it interacts with EQ.
The Volume-limit function is really nice because it allows a broad span of attenuation, it just deserves a different name and entering the setting in dB would be nice.

For me it came as a surprise that the Vout setting has limiting functionality such that it limits an output voltage that would be higher due to a boost-EQ. But I agree, that this is the most sensible implementation.

 

[Nanook]

System noise floor won't be affected by WiiM volume/volume limit so you're good as long as you can't hear the system's noise floor at your listening position.

Thanks, I'm using a Pro and an external DAC in the setup where I need a lower volume from time to time. This is how the 24bit slipped in.

I fully agree - as long as you do not hit noise floor or hum, even large volume adjustments can be made in the digital domain without audibly compromising the signal. Sure, technically S/N suffers, but if the noise SPL is way below human hearing thresholds it's not relevant.

 

[canard]

But I agree, that this is the most sensible implementation.

It has the merit of existing... but to judge it "But I agree, that this is the most sensible implementation."... no, it's just a discreet workaround for the lack of known autogain solutions on peq etc... but the discussion on ISP saturations etc. omnipresent in the most popular music questions its impact on operation in these situations... a real subject of study in this specific case....
who can perhaps be much more present than many suppose.


;-)

 

[slartibartfast]

but the discussion on ISP saturations etc. omnipresent in the most popular music questions its impact on operation in these situations... a real subject of study in this specific case....

I have no idea what this means