Sennheiser HD800S Review (Headphone)

[Blumlein 88]

US $1,323 is really a funny price for this. Considering this is an open back headphone you will be listening to/with alone in a room, why not just buy a pair of Neumanns KH 80 (about $1,200) instead which are objectively close to perfect (when it comes to accepted science) and only need a deecent sub to nail it down for good. Why go for a headphone with poor out-of-the box FR and with some kind of cute simulation of openness/soundstage/separation/whatever when you can get the real thing for not much more cash?

Mobility, neighbors etc.

 

[amirm]

My question is - how is the bass without eq? Do you find it goes substantially lower or sounds better than the HD650's?

Perceptually the HD800s sounds lighter and brighter than HD650. Its bass may be the same but has more highs which throws off the balance. Like you, I can't get EQ in all situations (e.g. watching youtube) but I am not critical when using these other devices. For general music enjoyment using Roon, I have it there as a profile which I load in. I really can't see headphones being used without EQ. They are not like speakers where most companies are trying to hit a flat on axis response.

 

[wasnotwasnotwas]

Don’t bother. The HD800 will fry your ears off with treble. HD600 and HD650 are the best sounding Senn’s—well, except for the HD1 which at $60,000 I have not heard, nor have many other people, I imagine. HD700 was good but for some reason it has been discontinued. There is a new HD660 which could be good. But the HD800 series? No thank you.

Very, very different beasts and shouldn't be ruled out for audition.

 

[amirm]

US $1,323 is really a funny price for this. Considering this is an open back headphone you will be listening to/with alone in a room, why not just buy a pair of Neumanns KH 80 (about $1,200) instead which are objectively close to perfect (when it comes to accepted science) and only need a deecent sub to nail it down for good.

I listen in the loft which is open to the rest of the house. No way I can use speakers for hours while I am testing and working. My wife is usually watching TV downstairs. Even though this is an open back headphone, once music plays it provide enough isolation to mask the sound of the TV and some amount of fan noise from my instrumentation.

Also, these headphones deliver bass and loudness that the KH80 cannot remotely reproduce.

 

[solderdude]

given all that it seems that Amir's comments about 94dB being below normal listening level are a little crazy, I don't get that currently.

I think @amirm meant the LF (with the higher distortion) in that case 94dB SPL for the lowest frequencies means about 80 Phon in music listening which is comfortably loud. 104dB SPL for lows = about 90 Phon which is quite loud for average levels.

The 425Hz is where he set the 94dB SPL level to and where he measured the 0.256V = 105dB/V which is the same as RAA measured it.
The 102dB/V Sennheiser specifies is probably obtained differently.
100.8 dB/mW so a fairly sensitive headphone.

 

[amirm]

I have, in the past, posted the question why, given the technical capability of companies like Sennheiser, and staggering sums charged for high-end headphones now, we don't have a headphone which measures well out of the box. I still don't get it. Wouldn't this involve some minimal in-line thing?

It is a matter of design philosophy. I read that Sennheiser calibrates their headphones against the "diffused field." They set up 8 speakers in anechoic chamber and measure and match what arrives at the "ears" of the measurement fixture. No real room has such a sound field yet no one questioned this until Harman came around and made the reference a good speaker playing in a room.

That aside, I can't understand why their HD650 and HD800S measure differently. Surely if they are all designed to comply with that diffused field, they would have similar response but they do not. So it is a puzzle. Likely it is the "sound they like" as they designed it.

 

[amirm]

I think @amirm meant the LF (with the higher distortion) in that case 94dB SPL for the lowest frequencies means about 80 Phon in music listening which is comfortably loud. 104dB SPL for lows = about 90 Phon which is quite loud for average levels.

I actually don't quite know what is going on here. But I do know that my listening observation of level is correct. Note this key thing on the graph:

Once I calibrated the headphone, I measured the output of the headphone amp and that is what I am reporting above: 0.27 volts. This is a quarter of the output of my Samsung S8+ phone output! My standard for sufficient volume for a 300 ohm headphone is 2 volts, not 0.27 volts.

I am using Topping A90 to drive headphones. For HD800S it was in low gain and volume control set to about 12 o'clock to get the above 94 dB. So that should also tell you that the playback level was not loud.

The drive unit I am using for the mic preamps has an amplifier as well. The default is supposed to be unity gain but maybe it isn't. I have to open it to find the setting.

For now, let me assure everyone that 94 dB as tested was not loud at all. Not only did I think music was too soft, I could listen to the full sweep with almost no discomfort. With speakers and at much longer distance, I have to wear hearing protection.

My main purpose for calibration was to get the two channels the same. It was a pain to do with the GRAS fixture as you have to remove the pinna and deal with very delicate wires inside. And the software in AP for calibration is made poorly making the process more aggravating than it should be.

So more work to do to refine the measurements but the results are not impacted.

 

[amirm]

Someone asked for playback chain.

For measurements, Audio Precision is the source with balanced output. This goes to Topping A90 whose volume is changed until I achieve the "94 dBSPL" as measured by the fixture. The output of the GRAS 45C microphones goes through their power supply and conditioner and then to Audio Precision.

For listening tests and EQ, I use my second generation RME ADI-2 DAC which is my "daily driver." I drive headphones directly from it although for this testing, I also routed its output through Topping A90 with it then acting as a DAC. The results were the same.

Equalization is performed in my Roon player. I create a new profile and create the filters manually and by "eye." No automatic filter generation. It is a very time consuming process and I reevaluate the results the next day. Goal is to reduce the number of filters needed. More optimal solution may exist with more filters visually but you better confirm that with how it sounds as filter response is not identical to what is seen on UI graphs.

 

[Robbo99999]

I actually don't quite know what is going on here. But I do know that my listening observation of level is correct. Note this key thing on the graph:

Once I calibrated the headphone, I measured the output of the headphone amp and that is what I am reporting above: 0.27 volts. This is a quarter of the output of my Samsung S8+ phone output! My standard for sufficient volume for a 300 ohm headphone is 2 volts, not 0.27 volts.

I am using Topping A90 to drive headphones. For HD800S it was in low gain and volume control set to about 12 o'clock to get the above 94 dB. So that should also tell you that the playback level was not loud.

The drive unit I am using for the mic preamps has an amplifier as well. The default is supposed to be unity gain but maybe it isn't. I have to open it to find the setting.

For now, let me assure everyone that 94 dB as tested was not loud at all. Not only did I think music was too soft, I could listen to the full sweep with almost no discomfort. With speakers and at much longer distance, I have to wear hearing protection.

My main purpose for calibration was to get the two channels the same. It was a pain to do with the GRAS fixture as you have to remove the pinna and deal with very delicate wires inside. And the software in AP for calibration is made poorly making the process more aggravating than it should be.

So more work to do to refine the measurements but the results are not impacted.

Yes, I think the reported dB levels are inaccurate.

 

[Jake71]

Nice review, thanks. Could you do these charts with a log scale instead of a linear scale? it seems like that would be more useful.

 

[amirm]

There was a lot of comment about Harman curve, too much bass, etc. I removed the reference to Harman in my review hoping this would help with this but did not!

I like to propose a new approach. Please don't comment about the above in each review. Instead, if you own the headphone I am testing, please apply the EQ as I have recommended and then report what your experience is. How you applied bass compensation with another headphone does not necessarily apply to my recommendations here. Only through this kind of experimentation do we learn what is going on as otherwise there are too many extraneous variables.

 

[amirm]

Nice review, thanks. Could you do these charts with a log scale instead of a linear scale? it seems like that would be more useful.

The log scales are not revealing at all. This was the revelation I had based on Klippel choosing linear scale for distortion. Log emphasizes small values at the expense of large values. Here we have large values and we want to see them.

But just as speaker measurements, log is also provided in the other graph in the form of dB scale:

Notice how this graph doesn't tell you as much about the differential of distortion at low frequencies.

I also don't know of any connection to distortion audibility being related to log scale when presented as a percentage.

 

[amirm]

Another request was for measurement export. AP software is different so not sure what all to export. Here is the frequency response measurements:

Let me know if you want something different.

 

[solderdude]

I actually don't quite know what is going on here. But I do know that my listening observation of level is correct. Note this key thing on the graph:

Once I calibrated the headphone, I measured the output of the headphone amp and that is what I am reporting above: 0.27 volts. This is a quarter of the output of my Samsung S8+ phone output! My standard for sufficient volume for a 300 ohm headphone is 2 volts, not 0.27 volts.

I am using Topping A90 to drive headphones. For HD800S it was in low gain and volume control set to about 12 o'clock to get the above 94 dB. So that should also tell you that the playback level was not loud.

The drive unit I am using for the mic preamps has an amplifier as well. The default is supposed to be unity gain but maybe it isn't. I have to open it to find the setting.

For now, let me assure everyone that 94 dB as tested was not loud at all. Not only did I think music was too soft, I could listen to the full sweep with almost no discomfort. With speakers and at much longer distance, I have to wear hearing protection.

My main purpose for calibration was to get the two channels the same. It was a pain to do with the GRAS fixture as you have to remove the pinna and deal with very delicate wires inside. And the software in AP for calibration is made poorly making the process more aggravating than it should be.

So more work to do to refine the measurements but the results are not impacted.

I measured SPL (calibrated) listening to music in order to find out how much voltage is needed for certain listening levels (HD650 which is also 105dB/V) .

The plot below shows part of the recording was playing at a very comfortable level that is a fair bit louder than the soft level. You can listen a whole evening to this level.
On lower DR recordings you like to turn it down a slight nudge, higher DR recordings you may feel the urge to turn it up a tad.

This resulted in the following (rounded off) numbers:

RMS levels = 75mV = 0.018mW = 80dB (average)
Peak levels = 830mVPP = 293mVRMS = 0.26mW = 92dB peak (is what one would measure when sweeping)
calculated DR = 11.8dB for that 10 sec of the song.





The part of the recording below was playing at a comfortably loud level which is easy to listen to for 1 or 2 songs before getting the urge to turn it down to more comfortable levels.

This resulted in the following (rounded off) numbers:

RMS levels = 236mV = 0.174 mW = 90dB (average)
Peak levels = 2.63VPP = 930mVRMS = 2.7mW = 102dB peaks (which is what the sweep would be,
calculated DR = 11.9dB for that 10 sec part of the song

 

[Vini darko]

Hi intresting review thanks. Sorry if this has already been asked. How is 104db or 94db a quiet listening/normal level? Thats permanent hearing damage. Normal listening should be around the 80db range.

 

[amirm]

@amirm , you said you like 800s as "It manages to provide a spooky layering and I guess I should say clarity that is not only surprisingly but delightful" - I think that's a very similar effect to the AKG K702 (especially when EQ'd to Harman), (the soundstage effect) so might be worth measuring & trying one of those too, it's significantly cheaper than the 800s so I think that would be a real find for people.

THanks. Went to get one and it is $289. So pretty expensive for a quick test. Does anyone have one that they would want to loan? I agree with those large cups it may create that effect or at least be part of the answer.

 

[solderdude]

Hi intresting review thanks. Sorry if this has already been asked. How is 104db or 94db a quiet listening/normal level? Thats permanent hearing damage. Normal listening should be around the 80db range.

80dB average means around 94dB SPL peak and is what one would get if one sweeps at 94 dB SPL and not touch the volume control and play music.

94dB SPL average (the measurements in the graph) means in reality the sounds are continuous (not music) and may have peaks that are a lot higher

 

[amirm]

Just in case this hasn't been posted. AutoEQ settings for Sennheiser HD 800S

Wow, a lot of filters in there. Is it auto generated or by a human?

 

[amirm]

There was a request for post EQ distortion measurements. Right now I don't have a procedure for that. I thought of a way to do that when I was going to sleep last night. Whether it works or not, I will not know until I try it.

 

[amirm]

Someone also asked about impedance measurements. These are done elsewhere and I have to build yet another fixture to measure it. With low impedance headphone amplifiers being so common now, I don't see the need to do this measurement. We know the nominal impedance as published. But if there are strong feelings about it, I will build the fixture to test.