HE400i - to PEQ or not to PEQ..

[Cahudson42]

I do all my serious listening - classical and vocal - using Amazon Music HD, a LG V20, Liquid Spark, and HE400i. I am very happy with it. But for one issue.. the HE400i resonant peak at 7-8k. When playing, say a wide dynamic range recording like the Kleiber Beethoven 7th, very occasionally, I think the existence of that peak almost makes me wince when it's triggered..

So, I am considering adding a miniDSP IL-DAC to cut it. Unfortunately not yet reviewed by @amirm . By necessity, this will be replacing the LG 110db Quad DAC with an (estimated based on specs) 102db IL-DAC. My guess is - I likely should not notice this. Would you agree?

But my real concern is distortion. I hate, really hate distortion. And after 50-some years, I have become very sensitive to it. I believe it's a major reason I like the planar low distortion HE400i, compared to my casual/tv SHP9500 with a fairly similar fr.

So my big question: How is my listening distortion going to be affected by a DSP? Let's consider the IL-DSP being similar to the miniDSP 2x4 HD in what it does for comparison purposes.

Is there an overall increase in distortion even if I leave everything 'as is' with no PEQs at all? If so, how much? How noticeable? (IYO). Now let's PEQ out the 7k peak. What happens there? Even though the peak is reduced, has distortion been introduced because of it? Again if so, how much? Or will distortion actually he reduced because the resonance is 'gone'?

Now let's PEQ a shelf to add, say, 3db of bass. Now do we have a transducer excursion extension that increases distortion? If so, is this only due to the increased SPL, or is the device/DSP also adding its own distortion?

Any thoughts, references, studies etc. giving any insight to what I can expect to happen to my 'perceived listening distortion' on my HE400i should I add a miniDSP, will be greatly appreciated..

 

[MRC01]

If you apply PEQ in DSP and use reasonable settings, especially for cuts not boosts, for all practical purposes there is no added distortion. If you go crazy with the PEQ with steep slopes / narrow filters having big amplitudes, then sure you can make the headphone distort. But in your case you're talking about a PEQ that is something like -6 dB @ 7500 Hz, Q=2.0. This is quite simple and transparent. Also if you boost, remember to reduce the overall level before the EQ is applied, to avoid clipping.

There are any number of ways to do this. VLC player has built-in PEQ. PulseEffects on Ubuntu has PEQ. You can go crazy and get a separate component like the Behringer DEQ2496 to do PEQ.

 

[solderdude]

see here

 

[MRC01]

Nice link. As for PEQ correction:
If the first graph from SBAF is right, it looks like -6 dB @ 8400 Hz Q=3.0 to snip off the top, or -12 dB @ 8500 Hz Q=1.414 to kill the entire bump.
If the second graph from Sonarworks is right, it looks like -6 dB @ 9 kHz Q=4.0

 

[StevenEleven]

If you apply PEQ in DSP and use reasonable settings, especially for cuts not boosts, for all practical purposes there is no added distortion. If you go crazy with the PEQ with steep slopes / narrow filters having big amplitudes, then sure you can make the headphone distort. But in your case you're talking about a PEQ that is something like -6 dB @ 7500 Hz, Q=2.0. This is quite simple and transparent. Also if you boost, remember to reduce the overall level before the EQ is applied, to avoid clipping.

There are any number of ways to do this. VLC player has built-in PEQ. PulseEffects on Ubuntu has PEQ. You can go crazy and get a separate component like the Behringer DEQ2496 to do PEQ.

I have had a Behringer DEQ 2496 for many years. It’s interesting, after reading through this site now for a year or so I understand what it does much better.

It‘s been quite a while since I bought headphones (except for one $30 pair). Nowadays I just listen to my headphones plugged into a device and I don’t sweat the HiFi tweaking stuff. But I do have presets I made many years ago for each of my headphones on the DEQ2496, and when I switch them in and out, I think, hmmm, that does sound better with the preset. FWIW. I have no recollection of what I was thinking when I made the presets, other than basing them in part on FR graphs (sort of a half-baked calibration to smooth out the FR I guess) and then after that more broadly to get the tonal balance where I want it (tone control to preference).

 

[MRC01]

I have a DEQ2496 too. One thing I like about it is that if you use the digital in, and digital out, then everything it does is in DSP and you bypass its internal DA and AD converters.

It's not uncommon for headphones to have a resonance peak typically between 5k and 10k, and a simple PEQ to cut this often does make them sound better. Most also have other deviations, but this alone can be sufficient. You don't have to chase down every wiggle in the response.

If you're listening on a computer, then PEQ in DSP is readily available often for free.

 

[Cahudson42]

see

Thank you. I hadn't thought of solving the peak in the analog domain

 

[pwjazz]

I also highly recommend doing a frequency sweep with something like https://apps.apple.com/us/app/tone-generator/id1207009001?mt=12. There's a lot of ear interaction in the treble and when I listen with my own ear I often hear peaks in a different spot than what's shown in measurements.

 

[sejarzo]

I love the HE400i except for that peak.

In Foobar2000 I use the Math Audio Headphone EQ component that allows entry of PEQ filters and started using the correction found here for the frequency, gain, and Q:

https://github.com/jaakkopasanen/AutoEq/blob/master/results/oratory1990/harman_over-ear_2018/Hifiman HE400i/Hifiman HE400i ParametricEQ.txt

The Headphone EQ component makes it easy to tweak the variables to adjust the EQ if you find the result too bright/dark/whatever.

My take is that even with leather pads that don't leak as much low end as the stock velour pads, the above correction renders a final tone that's bass shy for many recordings, so I add an additional low shelf filter with frequency 90 Hz, gain 3 dB, Q 0.9.

This is what it looks like: