Most HPhones are not flat freq response at the ear... just like few speakers are flat FR at the ear. There is little point in getting a super good headphone amp while the HP freq response is all over the map. But with an over the ear headphone, they can be EQ'ed just like speaker/room can. Using a small mic inside the ear cup (Panasonic capsule) or a probe mic (B&K) and an auto EQ like Audessey Pro (and others) you can listen to headphones with a flat FR at the entrance to the ear canal..... or to even near the ear drum if you like.
Here is a sample headphone (Dr Dre) before and after EQ:
View attachment Dr Dre Headphone EQ with Audyssey -1.pdf
This was also described (how to) in my headphone amp article in Linear Audio article.
THx-RNMarsh
Here is a sample headphone (Dr Dre) before and after EQ:
View attachment Dr Dre Headphone EQ with Audyssey -1.pdf
This was also described (how to) in my headphone amp article in Linear Audio article.
THx-RNMarsh
Last edited:
But headphones are never flat at the ear, and shouldn't be.
To the ear entrance you have the head and pinna modifying the response, and to the ear drum you have additional resonances like from the ear canal.
For example at 30° azimuth, in an HRTF this can sum to a 15 db difference between 1 kHz and 3 kHz.
If you EQ that away it would be like putting a -15 dB peak filter on flat speakers. Essentially, your whole upper midrange and treble range would be messed up.
Even if you account for that, you will still need a slight downward tilt of the FR to get a more natural response. The same happens to the sound of flat (in the free field) speakers in a real room.
Last edited:
But this is not a matter of opinion. These are just facts about human hearing. The pinna, even the head, has a significant effect on the FR that will reach your ear entrance. You cannot just ignore that.
I have first tried this years ago and it sounded as described. It may be more or less bad with different types of headphones, but it's just not the right way to do headphone equalization.
I'm not saying you cannot enjoy such kind of equalization, but it has nothing to do with flat FR.
Right, which is still quite different from what your ear entrance receives. It's not flat at all. Again, EQing flat at the ear entrance is like taking calibrated speakers and distorting the FR with peak filters.
Last edited:
I agree it is not what the ear drum recieves. The HAT tests have a microphone where the ear drum is located within a dummy ear/head/torso. That includes all the reflections and resonances of the ear tube/canal leading to the ear drum.
My EQ is like a speaker EQ... measured where the head/ear is during listening. But EQ of Headphones sending its sound to the ear.... loaded with same air volume as when worn on the head. In fact I was told that my measurement 'before' was very close to a test of that headphone using a dummy ear.
We EQ at a place close to the head (typ 1-2 inch from the ear canal) and we include the ear shape and volume enclosed for loading the HP. But not any closer.... not including the ear canal. It is measured outside the ear. With the HD800 I have on my desk now..... the driver is about 1.5 inch away from my skull.
We do not need to EQ for the ear canal -- we always have that effect... speakers or headphone listening. Now, IN-Ear reproducers would be very different and would give you the ear canal resonance(s) and a probe mic located inside the ear and near the ear drum is used.
THx-RNMarsh
My EQ is like a speaker EQ... measured where the head/ear is during listening. But EQ of Headphones sending its sound to the ear.... loaded with same air volume as when worn on the head. In fact I was told that my measurement 'before' was very close to a test of that headphone using a dummy ear.
We EQ at a place close to the head (typ 1-2 inch from the ear canal) and we include the ear shape and volume enclosed for loading the HP. But not any closer.... not including the ear canal. It is measured outside the ear. With the HD800 I have on my desk now..... the driver is about 1.5 inch away from my skull.
We do not need to EQ for the ear canal -- we always have that effect... speakers or headphone listening. Now, IN-Ear reproducers would be very different and would give you the ear canal resonance(s) and a probe mic located inside the ear and near the ear drum is used.
THx-RNMarsh
Last edited:
comment --- as soon as we put a closed, over the ear headphone on, we no longer have the head related que's and timing differences between ears etc etc. This loss of realism is a reason to Not use stereo recordings with headphones. Rather, for headphones, binaural recordings (using a dummy head/torso) does this for you. Today, it is being called 3-D sound. But it is still the same Binaural recording deleoped long ago for headphones. https://en.wikipedia.org/wiki/Binaural_recording
New HD binaural recordings downloaded and with a flat FR headphone is amazing.
But if you do want to use stereo recording with such headphones, some cross- feed and eq is necessary. However, that does not mean you dont need to start with flat FR headphones.
THx-RNMarsh
New HD binaural recordings downloaded and with a flat FR headphone is amazing.
But if you do want to use stereo recording with such headphones, some cross- feed and eq is necessary. However, that does not mean you dont need to start with flat FR headphones.
THx-RNMarsh
Last edited:
If we measure from inside the ear and have a mic where our ear drum would be, we, of course, do not have a flat response any more There. But that is a whole other matter.
A typical ear drum FR from a flat FR HP or speaker source is more like this:
Or to say it another way...... if you have a flat FR speaker/headphone, you get this FR at the ear drum.
THx-RNMarsh
A typical ear drum FR from a flat FR HP or speaker source is more like this:
Or to say it another way...... if you have a flat FR speaker/headphone, you get this FR at the ear drum.
THx-RNMarsh
Last edited:
In my last post I haven't even touched on the ear drum, I just talked about the ear entrance, since you said this:
but you also continue:
I've just explained the objections to these statements.
I'm very aware of dummy heads, binaural, HRTFs etc. I've developed DSPs in this area.
That's of course a different story.
This means you're not plugging the ear canal and just put the mic capsule in front of the ear with headphones on (which have to be big enough for this).
I can see how this would work better than (occluded) ear entrance or even near ear drum equalization to a flat FR.
Last edited:
This forum subject relies on having the Linear Audio article. If you read the LA article, i show the setup with mic in an HP cup. The other comments here are simply that you could do it other ways but it wouldnt be the same thing... and not elaborated upon.
THx-RNMarsh
OK, at last, after a long time, i tried a shunt reg with this amplifier.
The reg is this one: Zerozone HP 2 Super Class A Parallel Power Supply Kit L158 6 | eBay.
I had an hard time with the transistor that came with the kit(fakes?) and i was forced to replace most of them(all in the CCS stage). I've also replaced all resistors with Vishay-Dale.
Set the current at 250 mA and the voltage at +- 18V for each channel using two separate regs.Hot heatsinks.
And, drum roll..... the sound improved and not by a small margin. More natural and effortless sound.
Ok, for someone lm317-337 are deadly boring,but with them the results were pretty good and now are better, now this amp really sings.
So,IMHO, if you do not fear more heat and more cost, shunt reg is the way to go.
Thanks again Mr. Marsh.
I bought this one Class A Parallel Type Regulated Power Supply Finished Board | eBay which looks less sophisticated, but still a parallel Class A, and am hoping it will work with some tweaks (as per the thread, I'll need to reduce the resistance of the CCS set resistor to get more current out). What voltage/va transformer did you use for your +- 18V out? I'm curious how much drop I need to plan for across the shunt regulator to do +-24V?
Now i'm using 4 xfmrs(one for every + V and - V of each channel) rated at 18V(9-0-9)/830 mA that i bought cheap. When i tested them i found about 22-23 V unloaded; after rectification and filtering(2 stages) i had more than 30 VDC(unloaded). The shunt reg circuit itself drops about 7 V. I set the voltage initially with a load resistor of about 70 ohm/ 10W and i needed only minor changes, when i feeded the amp. Xfmrs are barely warm. I tested with my scope the final noise and it's about 250-300uV. The amplifier alone is dead silent. Maybe i shall try to increase the voltage to + - 24 V and if the xfmrs will seat, i shall replace them.
I don't know if these numbers will be useful for your circuit, but from my experience with other uses, the LM 317 as CCS drops at least 5-6 volts lightly loaded. I would prefer 32-33VDC before them.
You should definitely increase the DC voltage rails to +-24V as everyone who has done it says that the amp gets even better sounding. And you won't be dumping as much power out in heat. After looking at several sellers of your power supply, I'm getting an itch to pull the trigger on it. Maybe after I play with this one for a bit. After a bit of research, I'm thinking a 18 - 20 V out transformer should do the job. Just have to source one..
I'm wondering if the heatsinks are big enough for dissipating the heat. I'm assuming 20 - 24V output voltage and 500 mA overall current via CCS.
For example one version of my shuntregulator I use to feed my CFA design looks like this:
You can compare the heatsinks. And believe me, they become hot!
You're right; those flimsy heatsinks that came with the kit(also with my kit
)aren't up to the task. I had to replace them with something beefier to limit the temperature rise to 30-35°C over room temp. Not so simple given the space constraints.I apologize for a mistake: i set the current to 300 mA, not 250.