[headdict]

  Here's the problem: When you change from peak/dip to shelving, the algorithm changes. I plugged in your high and low shelf numbers and look what happens to the EQ curve. 
 
Here's as it should be, without the high and low shelves:

 
And here's when the lowest and highest bands were changed to low and high shelves:

 
That's a pretty big difference. What you need to do, is to use a parametric EQ that shows you the EQ curve visually, so you can make sure you're not creating weird anomalies. You can try tweaking the low and high shelf more so they actually match the curve of the original visually, instead of just plugging in numbers (since different algorithms change how the numbers actually affect the curve). 

Ouch! This could very well explain what I am hearing. Id did not expect such a huge difference in behavior across filter types. Many thanks for pointing this out. So the next step will be upgrading my tools.

 

[headdict]

My computer was out of reach while I was traveling. So I made no progress setting up my tooling in the meantime. However I made a quick check to see if the Rockbox shelving filters are really implemented as your graph suggests. This seems not to be the case. I dropped the filter at 2 kHz as this appeared to make the least difference and used this freed up peak filter to replace the 40 Hz shelving filter. The bass was as boomy as before. I double checked with 20 kHz and I could not hear the slightest difference. As shelving filter implementations seem to vary across platforms, it's probably better to avoid using them altogether whenever possible.
 
Maybe it's because I'm bass shy, but it just doesn't sound quite right. It sure sounds a hundred times better than with no EQ applied, though.

 

[Lunatique]

  My computer was out of reach while I was traveling. So I made no progress setting up my tooling in the meantime. However I made a quick check to see if the Rockbox shelving filters are really implemented as your graph suggests. This seems not to be the case. I dropped the filter at 2 kHz as this appeared to make the least difference and used this freed up peak filter to replace the 40 Hz shelving filter. The bass was as boomy as before. I double checked with 20 kHz and I could not hear the slightest difference. As shelving filter implementations seem to vary across platforms, it's probably better to avoid using them altogether whenever possible.
 
Maybe it's because I'm bass shy, but it just doesn't sound quite right. It sure sounds a hundred times better than with no EQ applied, though.

Maybe you have to set up the EQ curve using your computer, so you can compare it to what you're getting with the Rockbox. If it sounds fine when done using your computer, then you know the problem is with Rockbox. Perhaps it has another setting that's screwing up the bass, such as some kind of bass-boost setting?

 

[gideon228]

@Lunatique, I have copied updated HD650 settings to Equalizer APO for Windows 10
 
The graph shows gain to go upto 11dB at lowest frequencies. So I have set pregain at -11 to avoid any clipping.
Is it also viable to change Maximum dB in program settings to 6 so I can set my pregain at -6? Or, due to the low shelf algorithm, will that mess with the frequency range past 119hZ?

 

[Lunatique]

  @Lunatique, I have copied updated HD650 settings to Equalizer APO for Windows 10
 
The graph shows gain to go upto 11dB at lowest frequencies. So I have set pregain at -11 to avoid any clipping.
Is it also viable to change Maximum dB in program settings to 6 so I can set my pregain at -6? Or, due to the low shelf algorithm, will that mess with the frequency range past 119hZ?

You dont' always have to lower the pre-gain exactly as much as the highest boost in the EQ curve. It kind of depends on the content you're playing back. On songs with lots of sub-bass, it's more likely to distort, but on songs that don't have very strong sub-bass, you'll be fine. Of course, if setting it so you can just forget about it is more convenient, as long as your overall volume doesn't end up being too quiet. 
 
If you're after the exact same EQ curve I created for the HD650, it's best not to change anything if you can help it. The difference between 11 and 6 dB is quite a lot. But if you don't mind altering the EQ to fit your own taste and needs, then do whatever you need to.

 

[Malfunkt]

@Lunatique have to say I've been using the EQ curve you put up for the HD650 now consistently. I've got it setup at my work environment through Vox player. After a while you forget you have it enabled, and you'll be listening thinking "wow, can't remember the HD650 being this nice with the bass while being detailed without being too aggressive". Then I realize, right I'm using the EQ.
 
I much prefer this equalization. Thanks!

 

[Solrighal]

  @Lunatique have to say I've been using the EQ curve you put up for the HD650 now consistently. I've got it setup at my work environment through Vox player. After a while you forget you have it enabled, and you'll be listening thinking "wow, can't remember the HD650 being this nice with the bass while being detailed without being too aggressive". Then I realize, right I'm using the EQ.
 
I much prefer this equalization. Thanks!

 
Sonarworks will blow your mind.

 

[Lunatique]

   
Sonarworks will blow your mind.

You can get similar results using my method, and you wont' have to spend a dime. And you learn some important lessons about audio and music production too.

 

[Malfunkt]

You can get similar results using my method, and you wont' have to spend a dime. And you learn some important lessons about audio and music production too.

I just tried the Sonarworks demo, and it is good from the perspective of the presets it provides and for those who want their individual headphones analyzed.

I pulled up a 31 band EQ and was able to get similar results as Sonarworks.

 

[Solrighal]

  You can get similar results using my method, and you wont' have to spend a dime. And you learn some important lessons about audio and music production too.

 
I was using your settings initially when your post first went up but although it was good it wasn't ideal for my taste. I then experimented with the parametric EQ in JRMC and got even better results. Then, a couple of months ago, I saw a post about Sonarworks and downloaded the demo. I then bought it because it was that good. There's a *No Limits* curve available for Sonarworks that really helps the sub-bass & I also have a boost on sub-bass in the JRMC PEQ which really makes the HD 650 sing.
 

I just tried the Sonarworks demo, and it is good from the perspective of the presets it provides and for those who want their individual headphones analyzed.

I pulled up a 31 band EQ and was able to get similar results as Sonarworks.

 
Whatever floats your boat.

 

[Malfunkt]

Whatever floats your boat.

Thanks for recommending Sonarworks. I'm still going to evaluate it over the trial period. If I was more serious about music production and engineering I think it would be very useful. Being able to simulate other frequency curves of standard headphones and speakers is a plus. Also for headfi users you can dial it into a flat frequency response and use frequency 'tilts' to adjust to taste.

 

[Solrighal]

Thanks for recommending Sonarworks. I'm still going to evaluate it over the trial period. If I was more serious about music production and engineering I think it would be very useful. Being able to simulate other frequency curves of standard headphones and speakers is a plus. Also for headfi users you can dial it into a flat frequency response and use frequency 'tilts' to adjust to taste.

 
EQ is a strange thing. I actually don't find there to be a huge difference going from stock to Sonarworks but use Sonarworks for a few sessions in a row then try going back to stock. That's another story.

 

[headdict]

  Maybe you have to set up the EQ curve using your computer, so you can compare it to what you're getting with the Rockbox. If it sounds fine when done using your computer, then you know the problem is with Rockbox. Perhaps it has another setting that's screwing up the bass, such as some kind of bass-boost setting?

I finally checked it on my computer. For me the boost at 95 Hz makes absolutely no sense. Got a much smoother response after I removed it.
 
While I were at it, I made the boost at 300 Hz broader (from Q=3.1 down to Q=2.0) and moved it up to 350 Hz.That again yielded a smoother response from around 300 to 400 Hz.
 
In the treble region it sounds splashy at times. Will have to look into that eventually. However, my current priorities are other phones, such as beyer T1 (1st rev.).

 

[Lunatique]

 
I finally checked it on my computer. For me the boost at 95 Hz makes absolutely no sense. Got a much smoother response after I removed it.
 
While I were at it, I made the boost at 300 Hz broader (from Q=3.1 down to Q=2.0) and moved it up to 350 Hz.That again yielded a smoother response from around 300 to 400 Hz.
 
In the treble region it sounds splashy at times. Will have to look into that eventually. However, my current priorities are other phones, such as beyer T1 (1st rev.).

The 95 Hz is based on the actual measurements of the headphone. If you look at the frequency response measurements, there's a sharp dip at 95 Hz:
http://www.innerfidelity.com/images/AudioTechnicaATHM50B2012.pdf

 

[headdict]

  The 95 Hz is based on the actual measurements of the headphone. If you look at the frequency response measurements, there's a sharp dip at 95 Hz:
http://www.innerfidelity.com/images/AudioTechnicaATHM50B2012.pdf

I did and even checked the inner fidelity measurements of the newer model M50X, which show the same dip. And yet, I can't hear it on my sample. Did you verify that it's actually there on yours?