• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Focal Clear Review (headphone)

Joined
Sep 18, 2020
Messages
24
Likes
8
I did AB tests, not ABX (I was alone to dial in the different filters).
Here's from the RME ADI-2 DAC FS manual, regarding the "NOS" filter:
The DAC includes another filter which is called Super Slow in its data sheet. The impulse re-
sponse looks perfect, but checking the output signal with an Oscilloscope reveals steps that are
more typical for so called Non-OverSampling (NOS) devices, so we renamed it NOS within the
DAC filter menu.
 
Joined
May 4, 2020
Messages
239
Likes
269
To be "clear" C or Z weighted...in the picture seems C
General setting is C weighted yes but I am more interested in the LZpeak value (3rd small number) that represents the unweighted peak input.

Naturally the average SPL (large number) will be considerably lower, due to being C weighted and due to the musical piece only reaching the peak values for a fraction of a second at each drum hit.
Do you have any way of measuring the voltages at which the driver reached its maximum excursion?
Unfortunately I do not possess a precise enough voltmeter.
Also, since the Voltage needs to be measured in parallel to the running driver, I would have to either disassemble the Clear or McGuyver some measurement rig to access the electrical signal.

Keep in mind that my soundcard has a 35 Ohm Impedance, so it will elevate the bass response. I don't think you can easily combine Amir's measured frequency response and my measured peak value mathematically.

I measured the FR a while back:
Unbenannt.png
 
Last edited:
Joined
Jun 29, 2020
Messages
55
Likes
45
Location
Paris, France
Besides all this, the Clear had got qualities that the HD650 doesn't have; people have already commented on them, but the greatest for me is that transients are better reproduced. With the Clear, suddenly music is breathing, living, moving in a way that I couldn't hear before, and thanks to this we are given the opportunity to hear with more fidelity the intention of the different engineers and the artist and the band that created that music. This is just awesome.!
I have found that particularly revealing in this track :D
 

DjBonoBobo

Active Member
Joined
Jan 21, 2020
Messages
126
Likes
215
Just FYI i finally found some time to test Amir´s EQ-settings with the Clear Professional and RME ADI-2 DAC and i think i like and keep them. Thanks for that!
I am still not experiencing clipping even with the bass heavy tracks that were suggested. So far, there have been no problems at any volume that is acceptable to me, and no problems at any track. I usually use the IEM-port with low gain and -10db max, but most of the time less.

But i also tried some of the tracks with Amir´s settings for the ADI-2 (phones output, high gain and so on) and i could indeed produce the clipping as described by Amir.
So the conclusion for me now is that the different experiences with the Clear are simply based on different volumes. For me there doesn't seem to be any problems so far, not even with the EQ settings, but I respect why Amir didn't make a recommendation.

PS: One of the tracks i listened to - fits my taste much more than the R&B/Hip hop or whatever that other genres are called...
 

JIW

Active Member
Joined
Nov 11, 2019
Messages
252
Likes
362
Location
Germany
General setting is C weighted yes but I am more interested in the LZpeak value (3rd small number) that represents the unweighted peak input.

Naturally the average SPL (large number) will be considerably lower, due to being C weighted and due to the musical piece only reaching the peak values for a fraction of a second at each drum hit.

Unfortunately I do not possess a precise enough voltmeter.
Also, since the Voltage needs to be measured in parallel to the running driver, I would have to either disassemble the Clear or McGuyver some measurement rig to access the electrical signal.

Keep in mind that my soundcard has a 35 Ohm Impedance, so it will elevate the bass response. I don't think you can easily combine Amir's measured frequency response and my measured peak value mathematically.

I measured the FR a while back:
View attachment 104120
Since Amir has also measured the impedance of the Clear relative to frequency, and us thus knowing both the Clear's impedance at 20 Hz (84.4 Ohm) and 30Hz (122.7 Ohm), respectively, and the output impedance of your soundcard (35 Ohm), the relative changes in level can easily be calculated by treating the soundcard's output impedance as a resistor of equal value in series with the Clear connected to a zero output impedance source such that the source's output voltage is divided between them.

At 20 Hz, the decrease in level is -3.0 dB, while at 30 Hz, the decrease in level is -2.2 dB. Thus, the output level of the soundcard would have to be set to give 0.71 V (~-1 dBu) peaks at 20 Hz and 1V (2.2 dBu) peaks at 30 Hz into a high impedance load, e.g. 10 kOhm.

Regarding measuring the voltage, if I read one of your previous posts correctly, your soundcard is a Creative Labs Sound Blaster X-Fi Titanium HD. If so, based on this review and this review, the maximum level of the headphones output is 1 V RMS, while the maximum level of the analog input is 2 V RMS. Thus, a 1 V RMS sine from the headphone output should be recorded at -6 dBFS RMS according to AES-17 or -9 dBFS RMS relative to the highest sample value. No need to McGyver anything. Just use a 3.5 mm to 2xRCA cable.

Also, to avoid the effects of the soundcards high output impedance while still controlling volume digitally, you could use a headphone amplifier with a low output impedance - around 1 Ohm or less - at 1x (0 dB) Gain and full volume as a voltage buffer.
 

nhatlam96

Active Member
Forum Donor
Joined
Jan 1, 2020
Messages
204
Likes
72
Location
Mannheim, Germany
When comparing these two EQ settings:
AutoEQ:
Preamp: -2.6 dB
Preamp: -5.8 dB
Filter 1: ON PK Fc 16 Hz Gain 4.2 dB Q 0.46
Filter 2: ON PK Fc 195 Hz Gain -2.2 dB Q 0.72
Filter 3: ON PK Fc 1269 Hz Gain -4.6 dB Q 1.34
Filter 4: ON PK Fc 1331 Hz Gain 1.4 dB Q 0.23
Filter 5: ON PK Fc 8034 Hz Gain 5.0 dB Q 1.85
Filter 6: ON PK Fc 3504 Hz Gain -2.9 dB Q 3.22
Filter 7: ON PK Fc 4240 Hz Gain 4.4 dB Q 4.30
Filter 8: ON PK Fc 10288 Hz Gain 1.7 dB Q 3.01
Filter 9: ON PK Fc 12638 Hz Gain 1.6 dB Q 1.68
Filter 10: ON PK Fc 19870 Hz Gain -9.5 dB Q 0.36

Oratory1990:
Preamp: -5.8 dB
Filter: ON LSC Fc 105 Hz Gain 5.5 dB Q 0.71
Filter: ON PK Fc 90 Hz Gain -4.9 dB Q 0.3
Filter: ON PK Fc 1330 Hz Gain -6 dB Q 1
Filter: ON PK Fc 2200 Hz Gain 1.4 dB Q 2.5
Filter: ON PK Fc 3000 Hz Gain -2.6 dB Q 3
Filter: ON PK Fc 3700 Hz Gain -2.1 dB Q 5
Filter: ON PK Fc 4300 Hz Gain 2 dB Q 4
Filter: ON PK Fc 5950 Hz Gain -4.1 dB Q 10
Filter: ON PK Fc 6600 Hz Gain 2 dB Q 7
Filter: ON PK Fc 7500 Hz Gain 1 dB Q 3

I find oratory1990 more V-shaped, while AutoEQ is more mid focused, despite both aiming at harman target 2018.

Note: I find the Oratory1990 preset quieter than AutoEQ, therefore I added preamp -2.6 dB for AutoEQ.
 

LTig

Major Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
2,567
Likes
3,914
Location
Europe
I did AB tests, not ABX (I was alone to dial in the different filters).
Here's from the RME ADI-2 DAC FS manual, regarding the "NOS" filter:
Here are two older measurements of a 20 kHz sinus 44/24 @ -80dBFS of the RME ADI-2 PRO fs, output measured by a DSO. I don't remember why I measured at such a low level but what I want to show can be seen anyway:

First the SD-sharp filter:
RME ADI-2 PRO fs 20 kHz -80 dBFS @ +19 dBu 44-24 SD-Sharp.png

Now the NOS filter:
RME ADI-2 PRO fs 20 kHz -80 dBFS @ +19 dBu 44-24 NOS.png


Does not look like a 20 kHz sinus any more, does it? I think the NOS filter is no filter, so all the aliasing components are not suppresses. What you see is the sum of 20 kHz (valid signal) and 24.1 kHz (mirror signal, 22.05 + (22.05 - 20) kHz).
 

JIW

Active Member
Joined
Nov 11, 2019
Messages
252
Likes
362
Location
Germany
Here are two older measurements of a 20 kHz sinus 44/24 @ -80dBFS of the RME ADI-2 PRO fs, output measured by a DSO. I don't remember why I measured at such a low level but what I want to show can be seen anyway:

First the SD-sharp filter:
View attachment 104143
Now the NOS filter:
View attachment 104144

Does not look like a 20 kHz sinus any more, does it? I think the NOS filter is no filter, so all the aliasing components are not suppresses. What you see is the sum of 20 kHz (valid signal) and 24.1 kHz (mirror signal, 22.05 + (22.05 - 20) kHz).
It is AKM's super slow roll-off filter.

Source: https://velvetsound.akm.com/us/en/technology/

Here is the RME's manual.
Screenshot 2021-01-06 at 18.36.24.png
 
Joined
May 4, 2020
Messages
239
Likes
269
Regarding measuring the voltage, if I read one of your previous posts correctly, your soundcard is a Creative Labs Sound Blaster X-Fi Titanium HD. If so, based on this review and this review, the maximum level of the headphones output is 1 V RMS, while the maximum level of the analog input is 2 V RMS. Thus, a 1 V RMS sine from the headphone output should be recorded at -6 dBFS RMS according to AES-17 or -9 dBFS RMS relative to the highest sample value. No need to McGyver anything. Just use a 3.5 mm to 2xRCA cable.
Hmm. While I am somewhat confused, as to why you want me to measure this here is what I did:
Connect the line-in to the HP out.
Set HP out as output, Line-in as input in REW and display the levels indicator.

Levels.png


I am unsure whether I set the correct level of the 20Hz signal, since my memory is unclear. Sorry, didn't document the Windows volume because I did not expect it to be needed again. I do remember the 30Hz one though.
 
Joined
Sep 18, 2020
Messages
24
Likes
8
Here are two older measurements of a 20 kHz sinus 44/24 @ -80dBFS of the RME ADI-2 PRO fs, output measured by a DSO. I don't remember why I measured at such a low level but what I want to show can be seen anyway:

First the SD-sharp filter:
View attachment 104143
Now the NOS filter:
View attachment 104144

Does not look like a 20 kHz sinus any more, does it? I think the NOS filter is no filter, so all the aliasing components are not suppresses. What you see is the sum of 20 kHz (valid signal) and 24.1 kHz (mirror signal, 22.05 + (22.05 - 20) kHz).
Wow, that’s a ton of aliasing! Thanks for the graphs. [Edit: is this really aliasing? 20kHz is below 22.05kHz!]

What would have been nice is the same 20kHz tone but at 96kHz SR.

What JIW mentioned above is correct: « NOS » is a LPF, but it’s very slow. It looks like a first-order one.
 
Last edited:

ShiZo

Senior Member
Forum Donor
Joined
Sep 7, 2018
Messages
491
Likes
320
I like to blast my headphones too @amirm. With the pair I have, there is no clipping (at least as far as I push it, which is ear-bleeding levels with bass boosted tracks). But I've had like 7 pairs of clears and more than half exhibited the clip, which is why I returned them until I got one that didn't. The older the batch the more likely it seemed. I loved the tonal balance more than any other headphone (I don't eq) so I put a lot of work into getting one that didn't clip.

Like we've been talking about, maybe we can get focal or headphones.com to send you a second pair to review. Not to defend focal because they should have better QC on such an expensive product (even acting like it is a feature). But this test has only one sample. I think between the descriptances another unit should be reviewed, one from a new batch.
 

JIW

Active Member
Joined
Nov 11, 2019
Messages
252
Likes
362
Location
Germany
Hmm. While I am somewhat confused, as to why you want me to measure this here is what I did:
Connect the line-in to the HP out.
Set HP out as output, Line-in as input in REW and display the levels indicator.

View attachment 104154

I am unsure whether I set the correct level of the 20Hz signal, since my memory is unclear. Sorry, didn't document the Windows volume because I did not expect it to be needed again. I do remember the 30Hz one though.
I want to calculate the voltages since I consider the SPL measurements less reliable.

If I read the meters correctly - the red bar being the peak indicator and the colored bar the RMS level - then the level of the headphone output for the 30 Hz sine is -4.7 dBV with -1.7 dBV peaks. Since the voltage across the Clear is reduced by about 2.2 dB due to the soundcard's output impedance, the peak voltage required to make the Clear's driver reach Xmax at 30 Hz is about -3.9 dBV (-1.7 dBu) or about 0.64 V.

Given the measured SPL, this gives a - hypothetical - sensitivity of about 113.6 dB for 1 V at 30 Hz. Since according to Amir's measurement, the Clear is 4 dB less sensitive at 30 Hz compared to 1 kHz, this gives a sensitivity of about 117.6 dB for 1 V at 1 kHz. This is both about 1 dB higher than according to manufacturer specifications and 6 dB higher than third party measurements. However, from owning an UMIK-1 myself I recall the default sensitivity being 6 dB too high.

The level for the 20 Hz sine is 3.9 dB below that of the 30 Hz sine while the peak SPL is 5.8 dB lower. This would require a 1.9 dB lesser sensitivity at 20 Hz. According to my calculations based on Amir's measurements, the difference in sensitivity while connected to your soundcard is 2.8 dB. However, your measurement shows that it is pretty close to 2 dB. Thus, I think the level of the recorded 20 Hz sine wave is set correctly.

For the 20 Hz sine, the output level is -8.6 dBV which is reduced by 3 dB across the Clear due to the output impedance to -11.6 dBV. The peak voltage required to make the Clear reach Xmax at 20 Hz is thus -8.6 dBV (-6.4 dBu) or about 0.37 V.
 
Joined
May 4, 2020
Messages
239
Likes
269
However, from owning an UMIK-1 myself I recall the default sensitivity being 6 dB too high.
Huh? You mean a UMIK measures things 6dB too loudly?
That doesn't make sense, I mean they have calibration files and all that jazz. +6dB is a pretty large error that would render them pointless.

Not to defend focal because they should have better QC on such an expensive product (even acting like it is a feature).
Agreed, the unit to unit variation on this is inexcusable.
 

ShiZo

Senior Member
Forum Donor
Joined
Sep 7, 2018
Messages
491
Likes
320
Huh? You mean a UMIK measures things 6dB too loudly?
That doesn't make sense, I mean they have calibration files and all that jazz. +6dB is a pretty large error that would render them pointless.


Agreed, the unit to unit variation on this is inexcusable.
I had no choice but to return until I got a perfectly working one. Out of all headphones I've bought my focal clear for open and elegia for closed have been my favorite. I do not eq though. Now that I think about it I had to return the elegia once too. If I have a headphone that clips at my listening levels, which are quite high, I return them.

This is compared to my hd 650, lcd-2 closed, tr-x00 ebony, k701, he400, hd 598.
 

JIW

Active Member
Joined
Nov 11, 2019
Messages
252
Likes
362
Location
Germany
Huh? You mean a UMIK measures things 6dB too loudly?
That doesn't make sense, I mean they have calibration files and all that jazz. +6dB is a pretty large error that would render them pointless.
Yes, but that was a setting in the operating system (macOS). I just checked the calibration file and it's both sensitivity and frequency response.
 

ezra_s

Active Member
Forum Donor
Joined
Mar 23, 2020
Messages
162
Likes
130
Location
Spain
For those combining the Clear with the RME DAC (pro or non-pro): which DAC filter do you prefer?
I find the two first SD filters (sharp and slow) not working for me; the phase linear Sharp filter sounds too aggressive to me; NOS ain't working for me, I have a "bizarre" sensation with it; the SD LD, I don't know if I like or not but definitely the one I like is the phase linear Slow: I lose a tiny bit of very top end but overall it's the filter I enjoy listening music on.

What about you fellow sound passionate?
I can't feel a difference no matter which filter I choose. So I tend to set Sharp
 

LTig

Major Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
2,567
Likes
3,914
Location
Europe
Wow, that’s a ton of aliasing! Thanks for the graphs. [Edit: is this really aliasing? 20kHz is below 22.05kHz!]
The correct name is mirroring. The audible spectrum is mirrored at half of the sampling frequency (fs/2, and this pattern repeats into infinity). Therefore all frequencies equal to or higher than fs/2 need to be filtered out, similar to what the antialiasing filter does in front of an ADC. Here though, at the output of the DAC, it's called reconstruction filter.

BTW: here is a plot of two sinus signals added to prove that what the DSO shows is the sum of 20 kHz and its mirror signal of 24.1 kHz.
You can create it yourself in gnuplot using these commands:
set xrange[0:5]
set samples 1000
set title "sinus 20x and sinus 24.1x summed up"
plot sin(20*x) + sin(24.1*x)

mirrored-signal.png
 
Joined
Sep 18, 2020
Messages
24
Likes
8
I can't feel a difference no matter which filter I choose. So I tend to set Sharp
I feel Sharp (the linear phase one) is the least pleasing of them all. It over exaggerates the transients and distorts them; it feels aggressive and harsh. And hype.
These are subtle differences but as a sound passionate those differences are a big deal as to how I feel the music that is coming at me.
Slow is really smooth and natural, I like it best with the Clear.
On the other hand, with the HD650 I had the greatest listening experience with the SD Sharp filter.

It’s no wonder why I don’t like the Sharp with the Clear, because the Clear has been designed to reproduce dynamics with great fidelity. That extra transient emphasis/exaggeration of the Sharp filter is just counter-productive and “fights” against the original design of the headphones.
 
Joined
Sep 18, 2020
Messages
24
Likes
8
The correct name is mirroring. The audible spectrum is mirrored at half of the sampling frequency (fs/2, and this pattern repeats into infinity). Therefore all frequencies equal to or higher than fs/2 need to be filtered out, similar to what the antialiasing filter does in front of an ADC. Here though, at the output of the DAC, it's called reconstruction filter.

BTW: here is a plot of two sinus signals added to prove that what the DSO shows is the sum of 20 kHz and its mirror signal of 24.1 kHz.
You can create it yourself in gnuplot using these commands:
set xrange[0:5]
set samples 1000
set title "sinus 20x and sinus 24.1x summed up"
plot sin(20*x) + sin(24.1*x)

View attachment 104182
I understand that. But a 20kHz pure sine wave isn’t supposed to be mirrored at fs/2 when fs is 44.1kHz. ?
 

LTig

Major Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
2,567
Likes
3,914
Location
Europe
I understand that. But a 20kHz pure sine wave isn’t supposed to be mirrored at fs/2 when fs is 44.1kHz. ?
It is, see here. Any signal between 0 and 22.05 kHz is mirrored at 22.05 kHz. That's why you need a filter to get rid of the mirrored signals.

The ideal filter would be down by 96 dB (for 16 bit samples) at 22.05 kHz which nowadays sadly is not the case (most are down 3 dB at 22.05 kHz and below 90 dB at 24 kHz, for whatever reason). In practice though signals close to 20 kHz are much lower in level so the mirrored signals are also low in level, and most people can't hear them anyway.
 

Similar threads

Top Bottom