Unveiled: Meet the HD 620S

[jlemaster1957]

It was compared to a Dragonfly Black, which I had used for a long time, and the Fiio K11 (OG, non-R2R).

With the HD620S, the quality is K11 R2R > K11 > Dragonfly Black.

Haven’t tried the Fiio K11 R2R but the Cayin N6 mk2 with R01 (R2R) module produces an effect that resonates with what you describe.

 

[kman1211]

Holy moly, just received my Fiio K11 R2R, what a prefect pairing with the HD620S to my ears, I still EQ to fine tune but I can reduce the number of filters and the gain on the ones staying. In fact, I could do without EQ too. The soundstage and imaging are also improved with the K11 R2R running on SE.

I'm totally sold on R2R in NOS mode.

I have the K11 R2R as well, it’s a nice little device. It sounds even better when you hook it up to the FiiO PL50 linear power supply. The K11 R2R with the PL50 sounds better than the Modi Multibit 2(mm2) for the same price basically. Stock the K11 R2R was on par with the mm2 while much cheaper but also a bit easier on the ears than the mm2 even when both are on NOS mode.

The HD 620S sounds really nice on R2R dacs, then use a tube amp with the R2R dac and you just get sonic bliss with the HD 620S.

 

[jlemaster1957]

Cayin N6Mkii R01 (R2R) - directly playing from the 620S FR is very smooth - bass is textured of a bit gloomy, mids a bit forward but never shouty, treble extends well and has adequate sparkle on the 620s. Stage and imaging are quite realistic. NOW add the Cayin C9

Use pre-amp mode to retain N6mkii FR, class A + Korg Nu tubes. Dynamics tighten up- much clearer detail at both ends of the spectrum. Imaging now becomes class leading- not 3-blobs, realistic. Smooth musicality of the ids is retained. There is a sense of tube-iness, I’m guessing the C9 in this mode is delivering second order harmonics to provide this.

I only tried it out when the power supply on my main desktop rig failed today (Burson 3XGT 5A supercharger). Nice discovery.

 

[oratory1990]

HD620S measurement results​

A HeadFi-user recently sent me their HD620S, I thought some people in here might be interested in the results.

Positional variation​

Frequency Response​

The results below show an average across multiple positions of the headphone on the measurement setup.
Also shown are 90% quantiles - meaning 90% of positions fall inside the shaded area. I measured a total of 27 positions.
You can see how for a large range of frequencies, the exact position of the headphone does not significantly affect the result, but especially for higher frequencies there is some variation. The amount of variation is within the expected range for a closed-back, over-ear headphone.

Frequency response compared to the Harman Curve. Low deviation from this curve is generally desirable


Frequency response compared to the appropriate diffuse-field curve. Low deviation from this curve is generally not required. Good-sounding headphones typically exhibit somewhat of a downwards tilt in their diffuse-field compensated frequency response.
(This diffuse field curve was measured on our actual KEMAR dummy head, it deviates slightly from the one published by GRAS)

Group Delay​

This may not be fully clear for everyone: If the magnitude frequency response changes with position of the headphone on the dummy head, then so does the phase angle frequency response, and hence by definition so does the group delay:

Nonlinearity / Sensitivity​

To test the headphone's nonlinearity, I measured at multiple different signal levels (3 dB steps across a range of 27 dB)

​

some nonlinearity can be seen here. To make it more easily visible, I divided the obtained sound pressure by the input voltage. This gives us the voltage sensitivity ("decibel per volt") of the headphone.
In a perfectly linear system, the frequency response would not change with input levels, so dividing the obtained sound pressure by the input voltage would result in perfectly overlapping graphs.
Using the same color coding as above, we can see how for high input levels the sound pressure deviates slightly from the expected value:

Roughly speaking, this headphone has about 110 dB/V voltage sensitivity.
If we plot only the deviations from perfectly linear behaviour, we get the graph below (same color coding as above):

And zoomed in on just frequencies below 1 kHz, and zooming in on the Y-scale, looking just at +/- 2 dB:

So we see that for frequencies below ~100 Hz, the sensitivity decreases (the sound pressure increases by a lesser amount than the input voltage), while at around 200 Hz the sensitivity actually increases somewhat for higher input levels. This is typical for a vented closed back headphone of this design, and not too worrisome.

If we look at just the values at 60 Hz (where the nonlinearity is highest), we can plot the SPL as a function of input level (this is called the "characteristic curve"):


This lets us observe the deviation from linear behaviour.
For comparison, the characteristic curve at 1 kHz is almost perfectly linear:

An even more direct observation of the nonlinearity can be visualized by dividing the sound pressure by the input voltage (looking at the voltage sensitivity), where we can very clearly see how for high input levels the sensitivity drops for 60 Hz, but remains unchanged at 1 kHz:

Distortion​

The above observed nonlinearities in SPL must come hand in hand with an increase in distortion (the nonlinear characteristic curve is the mechanism that creates THD), and to no surprise, we do see high THD at high input levels (same color coding as above):

The apparent peaks at 4 and 7 kHz are correlated with narrowband dips in the magnitude frequency response, they don't show an actual increase in absolute distortion values.

And again, looking at the distortion just at 60 Hz tells us "how much" distortion this headphone has.


We can go to about 102 dB before crossing 3% distortion at 60 Hz (which is a very low estimate for the audibility threshold of THD at such a low frequency).
Listening at 85 dB average level would give us a headroom of at least 17 dB. That's not bad.

For comparison, the same graph for 1 kHz. No issues there.

Unit Variation​

measured 6 units so far, variation in between units is not too bad:

EQ recommendation​

HD620S - EQ to Harman 2018

 

[SmartCheetah]

Thanks, @oratory1990, I'd have two questions:

1. In the APO EQ software with the PEACE GUI, there are three low/high shelf filters to choose from: 'slope in dB,' 'Q as slope,' and 'corner frequency, Q as slope.' Which one works best with your presets?
2. What is your stance on "exclusive/bit-perfect" mode in streaming/audio software? It's considered superior as it avoids Windows resampling and lets the DAC handle the entire signal encoding, but it also disables EQ obviously. I assume you don’t see this as a big deal, and I'd like to understand why.

 

[oratory1990]

Thanks, @oratory1990, I'd have two questions:

1. In the APO EQ software with the PEACE GUI, there are three low/high shelf filters to choose from: 'slope in dB,' 'Q as slope,' and 'corner frequency, Q as slope.' Which one works best with your presets?

"Q as slope" is the one to use if you want to use a Q-factor other than 0.71, and are working with the center frequency of the filter.

The "slope in dB" one works using the slope parameter as opposed to the Q-factor parameter. You can calculate one if you know the other.
The "corner frequency" one uses corner frequency as opposed to center frequency.
Here's the frequency response of a low-shelf filter with the different parameters labelled:

2. What is your stance on "exclusive/bit-perfect" mode in streaming/audio software? It's considered superior as it avoids Windows resampling and lets the DAC handle the entire signal encoding, but it also disables EQ obviously. I assume you don’t see this as a big deal, and I'd like to understand why.

There's no holy scripture demanding a signal's bits to remain unchanged.
But yes, that's one of the reasons why I prefer EQ to be handled on a physical DSP (separate from the OS), and why I am advocating the use of DACs / DAC-amps with built-in DSPs.

Though if in doubt, I'd much rather use EQ to adress issues with the headphones than play bit-perfect signals through headphones that don't match my preferred sound signature.
For the same reason that I'd rather listen to an AAC-encoded file over decent headphones than listen to a FLAC over airline-earphones: Always adress the weakest link first.

 

[potterpastor]

bit hard sounding in the upper mids on others it was spot on. treble was fine for me.
Why same problem again...not put instead more smoother sound with top technicalities without push to far up...why is still not understanding this filosofy...most people still prefer natural organic smoother top end same story with meze 109 pro fanatic headphone bass superior and soundstage wow...midle spot on and for finishing strident treble how put off many people to not buy or after hear it immediately sell it only because of treble to elevated metallic one.....Sennheiser must do opposite not same fault i hope i m wrong with talk about 620s tuning

I listen to the 620s on my iPhone and I use the Rock EQ setting on the Music app and then it sounds the way I wish it would have sounded out of the box

 

[oratory1990]

I listen to the 620s on my iPhone and I use the Rock EQ setting on the Music app and then it sounds the way I wish it would have sounded out of the box

The "Rock" EQ setting on Apple Music is basically a slight V-shape:

 

[jlemaster1957]

HD620S measurement results​

A HeadFi-user recently sent me their HD620S, I thought some people in here might be interested in the results.

Positional variation​

Frequency Response​

The results below show an average across multiple positions of the headphone on the measurement setup.
Also shown are 90% quantiles - meaning 90% of positions fall inside the shaded area. I measured a total of 27 positions.
You can see how for a large range of frequencies, the exact position of the headphone does not significantly affect the result, but especially for higher frequencies there is some variation. The amount of variation is within the expected range for a closed-back, over-ear headphone.

Frequency response compared to the Harman Curve. Low deviation from this curve is generally desirable


Frequency response compared to the appropriate diffuse-field curve. Low deviation from this curve is generally not required. Good-sounding headphones typically exhibit somewhat of a downwards tilt in their diffuse-field compensated frequency response.
(This diffuse field curve was measured on our actual KEMAR dummy head, it deviates slightly from the one published by GRAS)

Group Delay​

This may not be fully clear for everyone: If the magnitude frequency response changes with position of the headphone on the dummy head, then so does the phase angle frequency response, and hence by definition so does the group delay:

Nonlinearity / Sensitivity​

To test the headphone's nonlinearity, I measured at multiple different signal levels (3 dB steps across a range of 27 dB)

​

some nonlinearity can be seen here. To make it more easily visible, I divided the obtained sound pressure by the input voltage. This gives us the voltage sensitivity ("decibel per volt") of the headphone.
In a perfectly linear system, the frequency response would not change with input levels, so dividing the obtained sound pressure by the input voltage would result in perfectly overlapping graphs.
Using the same color coding as above, we can see how for high input levels the sound pressure deviates slightly from the expected value:

Roughly speaking, this headphone has about 110 dB/V voltage sensitivity.
If we plot only the deviations from perfectly linear behaviour, we get the graph below (same color coding as above):

And zoomed in on just frequencies below 1 kHz, and zooming in on the Y-scale, looking just at +/- 2 dB:

So we see that for frequencies below ~100 Hz, the sensitivity decreases (the sound pressure increases by a lesser amount than the input voltage), while at around 200 Hz the sensitivity actually increases somewhat for higher input levels. This is typical for a vented closed back headphone of this design, and not too worrisome.

If we look at just the values at 60 Hz (where the nonlinearity is highest), we can plot the SPL as a function of input level (this is called the "characteristic curve"):


This lets us observe the deviation from linear behaviour.
For comparison, the characteristic curve at 1 kHz is almost perfectly linear:

An even more direct observation of the nonlinearity can be visualized by dividing the sound pressure by the input voltage (looking at the voltage sensitivity), where we can very clearly see how for high input levels the sensitivity drops for 60 Hz, but remains unchanged at 1 kHz:

Distortion​

The above observed nonlinearities in SPL must come hand in hand with an increase in distortion (the nonlinear characteristic curve is the mechanism that creates THD), and to no surprise, we do see high THD at high input levels (same color coding as above):

The apparent peaks at 4 and 7 kHz are correlated with narrowband dips in the magnitude frequency response, they don't show an actual increase in absolute distortion values.

And again, looking at the distortion just at 60 Hz tells us "how much" distortion this headphone has.


We can go to about 102 dB before crossing 3% distortion at 60 Hz (which is a very low estimate for the audibility threshold of THD at such a low frequency).
Listening at 85 dB average level would give us a headroom of at least 17 dB. That's not bad.

For comparison, the same graph for 1 kHz. No issues there.

Unit Variation​

measured 6 units so far, variation in between units is not too bad:

EQ recommendation​

HD620S - EQ to Harman 2018

Thanks for this thorough, great work. Reading this carefully, 2 messages come through. Pls correct me if I got it wrong
1) Distortion will be higher at 60 Hz for listening above 102 dB but should be fairly minimal below 85 dB, and not discernible/problematical at 1 KHz at any tolerable SPL
2) The EQ profile provided corrects the native 620S FR to Olive et al.’s 2018 Harmon preference curve for over the ear headphones
Are all your EQ profiles correcting to that preference curve?

 

[oratory1990]

Thanks for this thorough, great work. Reading this carefully, 2 messages come through. Pls correct me if I got it wrong
1) Distortion will be higher at 60 Hz for listening above 102 dB but should be fairly minimal below 85 dB, and not discernible/problematical at 1 KHz at any tolerable SPL

Correct, somewhere at 100-105 dB at 60 Hz is where the distortion might become audible (depending on the music of course). At mid and high frequencies the distortion will not be audible.
Note, "audible distortion" does not necessarily mean that it'll sound worse, just that it might be detectable.
THD does not completely correlate with sound quality except for very high values. 10% THD almost always sounds worse than less THD (all else being equal), but the same can not be said for e.g. 3% THD.

2) The EQ profile provided corrects the native 620S FR to Olive et al.’s 2018 Harmon preference curve for over the ear headphones

The Harman AE/OE curve as applicable to this measurement fixture, correct.

Are all your EQ profiles correcting to that preference curve?

No. The target is specified in each EQ setting.

 

[Soundizer]

Does it have less isolation then DT1770 PRO?

 

[kman1211]

Does it have less isolation then DT1770 PRO?

I believe it has more. Isolation is a strong point of this headphone.

 

[Soundizer]

I believe it has more. Isolation is a strong point of this headphone.

I was hoping someone who tried both could tell us. As the DT1770 has excellent isolation for studio use, but guessing the 620s is aimed at general consumer where isolation is no way as important.

 

[Soundizer]

Basically I don’t like noise cancelling headphones much and so want a good sound isolation headphone. Trying to decide which has better isolation better 620s and DT1770pro.

 

[zilax002]

I also can't provide a comparison, but the 620S has the best passive isolarion of all the closed backs I've tried.