stellarelephant
100+ Head-Fier
Hi everyone.
After buying my first set of Grado headphones, the SR-80e, I simply couldn't get used to the uneven harshness of the upper midrange and low treble response, despite positive impressions in other areas, such as detail, transient speed, bass, and the sweet "open" sound. Based on published frequency response measurements taken of several Grado models, coupled with plenty of listening and tuning by ear, I have designed two custom EQ curves that drastically even out the frequency response of the SR80e. They will likely work for several other Grado models, because according the the measurements I found, they all have a similar sound signature. For me, these EQ settings sound more natural and reduce listening fatigue a ton, while adding some of the missing air back in on top. I am interested in hearing others' impressions of these settings. Use them as is, or tweak to your liking.
If you have Rockbox on you player, it's super easy! Download my cfg files here. Or enter values manually if you want.
http://www.dropbox.com/sh/0aclc5cza4pvtu5/AABIpsHmUnsOH4lHmFB_0aSLa?dl=0
If not using Rockbox, just paste these values into your favorite EQ plugin.
Anybody have a reference mic and want to test these?
There are two different EQ curve settings. The first one I did all within Rockbox. It was just me tweaking according to this one excellent response graph I found on the InnerFidelity site, which was made for another headphone in the Grado lineup, the RS-1. I was surprised to hear resonances in my SR80e model that were exactly the same as those in a much more expensive model! You can really hear a headphone's resonances if you boost a narrow frequency band by several dB and then sweep through the problem region. It will blare out at you when you nail it and make you cringe. Then you apply a cut in that area and tune by ear. Anyhow, that was my method for the first curve, which I'm pretty happy with. The cfg file is called "G FLOW". Here are the EQ settings:
name boost/cut frequency Q
LS: 0.0 60 0.8
PK1: 0.0 110 3.5
PK2: 2.0 550 6.4
PK3: -4.0 2050 1.0
PK4: 5.5 3430 5.0
PK5: -2.2 4500 0.3
PK6: 4.8 5600 2.0
PK7: -4.0 7000 1.0
PK8: -3.5 9000 0.5
HS: 8.0 10000 0.7
Precut: -8.0 dB
After listening to "G FLOW" for a while, and quite enjoying the improvement, I decided to plug those values into Logic on my Mac, to get a visual representation of the curve and see if I could make further improvements. I noticed immediately that I had some Q values set a little too wide. After fixing that issue, I found the HeadRoom website, which actually has a response graph of the SR80 model. So now I had two graphs to use, as well as a visual tool to really dial in each cut and boost. I tried hard to really correct each peak and dip with precision. This curve sounds even more neutral to my ears, so much so that it pretty much completely does away with the original Grado sound (which many revere--don't stone me). This cfg file is called "G FLAT". The settings:
name boost/cut frequency Q
LS: 0.0 30 0.7
PK1: -5.5 2050 1.5
PK2: 4.0 3450 1.5
PK3: -4.5 4250 2.5
PK4: 3.0 5800 6.2
PK5: -7.0 8100 0.8
PK6: -3.0 9000 2.6
PK7: 8.0 15600 1.5
PK8: -4.5 17400 2.5
PK9: 0.0 10000 0.7
Precut -8.0 dB
NOTES:
Here are the graphs I used, followed by a screenshot of the curve I designed in Logic, which is sort of an inverse of the frequency response. Based on a lot of listening, I actually think the RS-1 graph from InnerFidelity is a closer match to my headphones than the SR-80 graph from HeadRoom, which uses a LOT of smoothing. Big thanks to Tyll Hertsens for taking great measurements and sharing them!
Strategy:
Most of what I'm doing here is trying to cut the resonant frequencies of the Grado drivers, represented by the spikes in the measurement graphs. So my EQ curves are more or less an inverse of the graphs. I also boosted a tad in a few areas and counteracted the natural high end roll-off of the drivers by boosting above 10K. I DID NOT have any success evening out the bass response. The Grados have a midbass hump around 100 Hz, which is common in many speakers and headphones, and sounds just fine, really, though not "accurate". I tried cutting the hump and boosting below it. What a disaster, even with only 0.5 dB boost. These drivers totally lose their composure with any low bass boost whatsoever. Distortion out the wazoo. Leave the bass alone…the drivers already push as much air as they are prepared to handle, and the untweaked bass sounds pretty darn natural, detailed, fast, and awesome from the get go.
Rockbox Volume:
You will notice that the volume drops when you use these settings. This is because there is a 8 dB Precut applied to both…designed to prevent distortion by counteracting the boosted frequencies. This makes it difficult to AB with no EQ. But Rockbox does come with a "flat" curve preset you can use for that purpose. 8dB is pretty significant, but luckily Grados are very efficient. That being said, if your player doesn't reach your desired volume with these settings, you can lower the Precut setting. If you do this, be sure to also lower all boosts in the Graphical EQ to the same or lesser values as the Precut to prevent clipping.
Limitations of EQ:
Taming the peaks does not mean sonic perfection. Remember that the peaks are not simply louder--they are louder because of resonances. The resonances exist because the driver's physical dimensions allow easier vibration at certain frequencies. So not only is the original signal amplified in that bandwidth, but also a ringing noise is added. When we make cuts into resonant bands with EQ, we arrive at a more balanced sound, but remember that within those bands, we have reduced the amplitude of the original signal to BELOW its original level. Now we have a quieter signal in that band, plus the noise (which is still there but less) which together add up to be the "correct" amplitude. My point is, a bit of the signal is missing, and the noise is still there, though quieter.
Also, EQ introduces some phasing issues. I think I can hear them, especially with the "G FLAT" preset, which has some very steep Q values. It sort of sounds like the signal loses some of its punch. Different frequencies of the same sound are arriving at different times instead of all at once…its like the sound is a little smeared. To my ears, the tradeoff is not worth it. In the end, I like the "G FLOW" setting best, even though it is probably less accurate in terms of frequency response. It is a compromise that still sounds punchy and coherent while eliminating like 80% of the Grado listening fatigue for me.
Conclusion:
This was a fun project for me. It really challenged my critical listening skills, and I came up with some results I enjoy. I gained a working understanding of the Grado drivers' strengths and weaknesses, and I've now got a better grasp on the capabilities and limitations of digital EQ as a tool for tuning headphones.
I hope to hear your impressions, everyone. Happy listening.
After buying my first set of Grado headphones, the SR-80e, I simply couldn't get used to the uneven harshness of the upper midrange and low treble response, despite positive impressions in other areas, such as detail, transient speed, bass, and the sweet "open" sound. Based on published frequency response measurements taken of several Grado models, coupled with plenty of listening and tuning by ear, I have designed two custom EQ curves that drastically even out the frequency response of the SR80e. They will likely work for several other Grado models, because according the the measurements I found, they all have a similar sound signature. For me, these EQ settings sound more natural and reduce listening fatigue a ton, while adding some of the missing air back in on top. I am interested in hearing others' impressions of these settings. Use them as is, or tweak to your liking.
If you have Rockbox on you player, it's super easy! Download my cfg files here. Or enter values manually if you want.
http://www.dropbox.com/sh/0aclc5cza4pvtu5/AABIpsHmUnsOH4lHmFB_0aSLa?dl=0
If not using Rockbox, just paste these values into your favorite EQ plugin.
Anybody have a reference mic and want to test these?
There are two different EQ curve settings. The first one I did all within Rockbox. It was just me tweaking according to this one excellent response graph I found on the InnerFidelity site, which was made for another headphone in the Grado lineup, the RS-1. I was surprised to hear resonances in my SR80e model that were exactly the same as those in a much more expensive model! You can really hear a headphone's resonances if you boost a narrow frequency band by several dB and then sweep through the problem region. It will blare out at you when you nail it and make you cringe. Then you apply a cut in that area and tune by ear. Anyhow, that was my method for the first curve, which I'm pretty happy with. The cfg file is called "G FLOW". Here are the EQ settings:
name boost/cut frequency Q
LS: 0.0 60 0.8
PK1: 0.0 110 3.5
PK2: 2.0 550 6.4
PK3: -4.0 2050 1.0
PK4: 5.5 3430 5.0
PK5: -2.2 4500 0.3
PK6: 4.8 5600 2.0
PK7: -4.0 7000 1.0
PK8: -3.5 9000 0.5
HS: 8.0 10000 0.7
Precut: -8.0 dB
After listening to "G FLOW" for a while, and quite enjoying the improvement, I decided to plug those values into Logic on my Mac, to get a visual representation of the curve and see if I could make further improvements. I noticed immediately that I had some Q values set a little too wide. After fixing that issue, I found the HeadRoom website, which actually has a response graph of the SR80 model. So now I had two graphs to use, as well as a visual tool to really dial in each cut and boost. I tried hard to really correct each peak and dip with precision. This curve sounds even more neutral to my ears, so much so that it pretty much completely does away with the original Grado sound (which many revere--don't stone me). This cfg file is called "G FLAT". The settings:
name boost/cut frequency Q
LS: 0.0 30 0.7
PK1: -5.5 2050 1.5
PK2: 4.0 3450 1.5
PK3: -4.5 4250 2.5
PK4: 3.0 5800 6.2
PK5: -7.0 8100 0.8
PK6: -3.0 9000 2.6
PK7: 8.0 15600 1.5
PK8: -4.5 17400 2.5
PK9: 0.0 10000 0.7
Precut -8.0 dB
NOTES:
Here are the graphs I used, followed by a screenshot of the curve I designed in Logic, which is sort of an inverse of the frequency response. Based on a lot of listening, I actually think the RS-1 graph from InnerFidelity is a closer match to my headphones than the SR-80 graph from HeadRoom, which uses a LOT of smoothing. Big thanks to Tyll Hertsens for taking great measurements and sharing them!
Strategy:
Most of what I'm doing here is trying to cut the resonant frequencies of the Grado drivers, represented by the spikes in the measurement graphs. So my EQ curves are more or less an inverse of the graphs. I also boosted a tad in a few areas and counteracted the natural high end roll-off of the drivers by boosting above 10K. I DID NOT have any success evening out the bass response. The Grados have a midbass hump around 100 Hz, which is common in many speakers and headphones, and sounds just fine, really, though not "accurate". I tried cutting the hump and boosting below it. What a disaster, even with only 0.5 dB boost. These drivers totally lose their composure with any low bass boost whatsoever. Distortion out the wazoo. Leave the bass alone…the drivers already push as much air as they are prepared to handle, and the untweaked bass sounds pretty darn natural, detailed, fast, and awesome from the get go.
Rockbox Volume:
You will notice that the volume drops when you use these settings. This is because there is a 8 dB Precut applied to both…designed to prevent distortion by counteracting the boosted frequencies. This makes it difficult to AB with no EQ. But Rockbox does come with a "flat" curve preset you can use for that purpose. 8dB is pretty significant, but luckily Grados are very efficient. That being said, if your player doesn't reach your desired volume with these settings, you can lower the Precut setting. If you do this, be sure to also lower all boosts in the Graphical EQ to the same or lesser values as the Precut to prevent clipping.
Limitations of EQ:
Taming the peaks does not mean sonic perfection. Remember that the peaks are not simply louder--they are louder because of resonances. The resonances exist because the driver's physical dimensions allow easier vibration at certain frequencies. So not only is the original signal amplified in that bandwidth, but also a ringing noise is added. When we make cuts into resonant bands with EQ, we arrive at a more balanced sound, but remember that within those bands, we have reduced the amplitude of the original signal to BELOW its original level. Now we have a quieter signal in that band, plus the noise (which is still there but less) which together add up to be the "correct" amplitude. My point is, a bit of the signal is missing, and the noise is still there, though quieter.
Also, EQ introduces some phasing issues. I think I can hear them, especially with the "G FLAT" preset, which has some very steep Q values. It sort of sounds like the signal loses some of its punch. Different frequencies of the same sound are arriving at different times instead of all at once…its like the sound is a little smeared. To my ears, the tradeoff is not worth it. In the end, I like the "G FLOW" setting best, even though it is probably less accurate in terms of frequency response. It is a compromise that still sounds punchy and coherent while eliminating like 80% of the Grado listening fatigue for me.
Conclusion:
This was a fun project for me. It really challenged my critical listening skills, and I came up with some results I enjoy. I gained a working understanding of the Grado drivers' strengths and weaknesses, and I've now got a better grasp on the capabilities and limitations of digital EQ as a tool for tuning headphones.
I hope to hear your impressions, everyone. Happy listening.
