Headphone usage (Grado GS 1000i and JBL SYNCHROS E50BT)

(Randolf Werner 02/2024)

Using headphone is a great way to listen to music. Typically, many good or even highend headphones offer a far better price / performance ratio than loudspeakers and have little additional requirements. A decent system may just consist of a mobile phone, DAC+headphone amplifier and a headphone. To me a good pair of headphones often can resolve fine details better than most speakers do. Anyway, regardless of the headphone model used the sound experience will always be very different from the one offered by loudspeakers with respect to soundstage and feeling of bass notes.

 

A/B comparing of headphones is somewhat tricky. Even with an optimal setup feeding the same source simultaneously to both models with same volume there is no way to seamless switch between the 2 models, like you can do with electronic equipment or loudspeakers. You always have to take off one model and put the other on.

 

Also, measurements of headphones are tricky. Depending on the headphone fixture used and how the headphone exactly fits on the fixture results can differ significantly. Especially beyond 10khz results become very chaotic. High quality headphone fixtures like the GRAS 45CA are rather expensive and therefore the DIY community has limited access to it. MiniDSP EARS is a very affordable Headphone fixture, but measurement result seam to differ significantly from highend fixtures. Also, the desired target frequency response is more uncertain than for electronic equipment and loudspeakers. When measuring on a headphone fixture this is not just a flat response since the human (or headphone fixture) ear canal and earpiece perform some significant equalization especially between 2-8khz. Harman did a lot of research to figure out the average preference frequency response curve. Over the years (2013-2018) this harman headphone curves have been refined a bit and an extra in ear headphone curve was added. Anyway, these are averaged results. They represent the average listener preference. The individual listener preference may differ to some degree from that:

 

  1. Around 64% of listeners, even those working in the audio field, enjoyed the Harman curve the most. Almost all Harman curve lovers were younger than 50.
  2. 15% of listeners preferred more bass. In practice, up to 6dB elevation below 300Hz, and 1dB boost above 1kHz. This group consisted of younger males, which is also a target audience for JBL headphones.
  3. The remaining 21% of users, enjoyed less bass. The low-end was reduced up to 3dB, while the 1dB boost above 1kHz stayed the same. In this group, you find primarily females and older listeners. 

My personal preference is probably best described by 3). Understanding Headphone Measurements & Hifiman HE-6 Review gives a very detailed up to date introduction to headphone measurements. InnerFidelity Headphone Measurement Procedures also gives a very detailed introduction into headphone measurements.

 

The Bruel & Kjær 5128 HATS seems to be an even more sophisticated and expensive headphone fixture and starts to be used by reviewers. Unfortunately results differ significantly from the GRAS fixture for which the Harman target curve was initially developed.

 

In addition to sound quality comfort is an important factor. I personally prefer light, open back over ear headphones. Even after listening for a long period hardly noticing wearing a headphone is important to me.

 

Grado Labs is a traditional phone cartridge and headphone manufacturer in Brooklyn New York USA with a history back to 1953. In 2014 I decided to upgrade from my Jecklin Float Modell 2. I ended up comparing between a Sennheiser HD 800 and the Grado GS1000i and finally purchased the GS1000i for 950 Euro. So, the GS1000i was not a bargain, but 10 years later I am still very happy with it.

 

One thing one has to consider is the headphone impedance. Loudspeaker nominal impedance is usually within 4-8 Ohms. Headphone nominal impedance varies a lot more between the models ~32-600 Ohm. As with loudspeakers, impedance curves are not always flat, the GS1000i impedance varies between 32 and 60 Ohm. When driven by a low impedance source amplifier there is no effect of the frequency response.

 

My Yamaha R-N803D is rated with a damping factor of 240 into 8 Ohm, so its inner resistance is 0,03 Ohm. In practice of course one needs to add the cable resistance to that to calculate the damping factor effective to the drivers. Unfortunately like most integrated amplifier it doesn’t have a dedicated headphone amp, instead the main amp directly drives the headphone jack via a single resistor, in my case 470 Ohm. This is common practice in many integrated amplifiers, e.g.:

  • Yamaha A-960: 270 Ohm
  • Yamaha AX-590: 560 Ohm
  • Yamaha RX-397: 330 Ohm
  • Yamaha R-N803D: 470 Ohm

These high output impedance of headphone jack in amplifiers seems to go back to “IEC 61938 (Multimedia systems –Guide to the recommended characteristics of analogue interfaces to achieve interoperability), which even in its 2018 version recommends 120 Ohm output impedance for “Interoperability of headphones with stationary amplifiers”. However it now at least recommends <= 2 Ohm for portable audio equipment.

 

So we end up with a damping factor of 0,07 into the 32 Ohm GS1000i. This has two effects:

  • The frequency response is significantly equalized if the headphone impedance varies.
  • The amplifier has very little control on the headphone drivers.

The perfect solution for this issue is to use a dedicated low impedance headphone amplifier. However there also exists a simpler solution by building a voltage devider. I build one with 4,7 and 33 Ohm 5 watt MOX resistors:

ATTENTION: Although building a voltage divider is simple and cheap, please be careful. In its simple form it only works with unbalanced amplifiers with a shared ground. For unbalanced amplifiers a different divider needs to be build. Doing it incorrectly or connecting it incorrectly (wrong polarity) can damage your amplifier or headphone.

 

This way we end up with a damping factor of 6.8 into the 32 Ohm GS1000i, still not very impressive but fair.

 

The frequency response of the Grado GS1000 measured by innerfidelity (https://github.com/jaakkopasanen/AutoEq/blob/master/measurements/Innerfidelity/data/over-ear/Grado%20GS1000.csv  /  https://web.archive.org/web/20170110152448/http://www.innerfidelity.com/images/GradoGS1000.pdf ), the harman target curve (blue) and the impedance curve measured by myself (orange) look like this:

The predicted effect on the frequency response of the Yamaha 470 Ohm resistor (green) and my voltage devider (blue) based on the impedance curve of the GS1000i is:

Comparing measured frequency response (orange), measured frequency response + 470 Ohm effect (green) and harman curve looks like this:

So, using the voltage devider makes a clearly audible effect and lets me enjoy the GS1000i a lot more.

 

Using a simple parametric equalizer (Freq=86Hz, Gain=-3db, Q=1.1) can further improve the situation(violet):

AutoEq is an interesting application to generate equalizations for many headphone models and Equalizers for different target curves. It visualizes frequency responses of many headphone models and target curves. The corresponding github repository contains the raw data for your own use cases (https://github.com/jaakkopasanen/AutoEq/tree/master/measurements / https://github.com/jaakkopasanen/AutoEq/tree/master/targets)

Equalizing headphones has become very popular and I agree that it can improve the sound quality significantly. But it has its price especially when it comes to heavily boosting certain frequency areas. You need to avoid clipping. Therefore, headroom in digital and analog domain is required. Often one ends up with setting an attenuation / gain of -10db or even more, which compromises the dynamic range quite a bit. Finally, you should not overdrive the headphone drivers especially in the bass region to avoid huge distortions. The frequency response is just one aspect of the sound, equalizing will not transform a budget below average 100 Euro headphone into a 1000 Euro device, even if its frequency response may look ok or perhaps even better. Let me give you one example. In 2017 I purchased a JBL SYNCHROS E50BT Bluetooth headphone for just 55 Euro (I think the recommended price was significantly higher), so a real bargain for just 6% of the Grado price. I purchased it just to have some kind of acceptable Bluetooth headphone for traveling. Basically, I don’t like it at all. It is very uncomfortable, sounds very booming with some midrange shout and at the same time some midrange leakage to my ears. I cannot listen for longer periods to it, mostly due to its bass performance and comfort. Anyway, you will find some rather positive reviews of it. Actually, it has one nice feature, although it is a Bluetooth headphone it can be used as an ordinary headphone as well in case the battery is empty or you want to use a better amplifier. Measurement wise it doesn’t look too bad, except for some serious leak in the lower midrange around 350Hz (https://github.com/jaakkopasanen/AutoEq/blob/master/measurements/Rtings/data/over-ear/JBL%20Synchros%20E50BT.csv). This is the measurement (red), the harman target curve (green), the predicted response with my eq settings (blue) and my impedance measurement (orange):

Except for the 350Hz deep dip it is pretty close to the harman curve, especially in the bass region. Anyway, to my ears it sounds as already mentioned. With my eq settings

  • Gain: -8db
  • Low shelf 130Hz -4dB
  • Peak 350Hz +8dB Q=2.0
  • Peak 2000Hz -3dB Q=1.0
  • Peak 5000Hz +4dB Q=1.7

it sounds far better to me and in fact midrange and treble is much closer to the harman curve. However, the bass section is now significantly below it. Can it now compete or perhaps even exceed the performance of the expensive Grado? No way, it is still a huge difference, but it is somehow acceptable to me now and I may use it more frequently.

My conclusion is that the harman curve is a valid target for my ears, except for the bass. This may have various reasons:

 

  1. My ears may not be average.
  2. The used measurement system differs significantly from the one use to determine the harman curves.
  3. Bass performance of headphones highly depends on the head / ear fitting.

The old (57 models till 08.07.2021) Audio Science Review Headphone Review and Measurement Index offers a nice frequency response comparison tool of reviewed headphones (“Goto Freq Resp” button). You can select models in various ways and compare the raw frequency response. The result of all measurements looks like this:

 

And even within the models recommended without using equalization there are significant derivations, especially in the bass section:

The recent headphones index currently with 132 models till 20.12.2023 is located here: https://www.audiosciencereview.com/asrdata/HeadphoneallList

 

I have created FreqRespGraph a simple easy to use tool to generate and compare frequency response graphs from CSV data files. It can be used to generate similar plots. README.md shows some plots from AutoEq data for some famous headphone manufacturers.

References