I decided to measure Average and Peak dB for the following devices against two songs played all the way through.
The test subjects iPhone 7 plus, iPad 6, or MacBook Pro 15" 2016 v.s Sennheiser HD-6xx Headphone
Another test I could run is Apple Music vs. Qubux v.s Tidal, v.s Spotify. but enough looking this morning.
Source: Qubuz.
Here are the measured numbers
Foy Vance “Make it Rain”. - Track encoded: 24 Bit 44 kHz Max Volume all devices except XD05 Plus
iPhone 7 Plus with Lighting to Headphone dongle - 76 dB with Peak of 96 dB
iPad 6 Headphone Adapter - 85 dB average with a peak dB of 107
iPad 6 with Lighting to Headphone dongle - 83 dB average with a peak dB of 104
MacBook Pro Headphone jack 85 dB average with a peak dB of 110
MacBook Pro xDuoo XD05 Plus 3/4 volume Gain Level 1 - 86 dB average with a peak dB of 107
Rodrigo yGabriela Live in France Haunman: Track encoded 16 bit 44 kHz Max Volume all devices except XD05 Plus
iPhone 8 Plus with Headphone Adapter - 82 dB average with 96 dB Peak
iPad with Headphone Adapter Average - 95 dB average 107 dB peak
iPad with Lighting to Headphone dongle - 94 dB average 107 dB peak
MacBook Pro Headphone jack 93 dB average with a peak dB of 110 dB
MacBook Pro xDuoo XD05 Plus 3/4 volume Gain Level 1 - 97 dB average with a peak dB of 110
MacBook Pro xDuoo XD05 Plus 3/4 volume Gain Level 3 - 99 dB average with a peak dB of 118.
One thing I have to say please do not play at these average dB levels we see on iPad and Macbook Pro, it will not give you the best sound nor will it be good for your hearing.
Interesting. I observed a lot of quality variation with my HD-600s between sources – it’d be interesting to see the frequency response profiles beyond the volume.
Bear in mind that ability to reach various SPLs at maximum volume into a specific headphone doesn’t necessarily say anything about what qualitative impacts you encounter by doing so.
Amplifiers tend to exhibit increasing distortion as output level goes up, with a frequent J-Curve or “Hockey Stick” effect as you get close the amplifier’s rail voltages.
So how loud they can play vs. how much distortion you’re getting to reach that volume is something worth bearing in mind. Particularly where the HD6XX exhibit almost double their average impedance around their resonant peak, which further strains power requirements.
For reference, here’s the HD6XX’s power requirements for various SPLs assuming 300 ohms impedance, but note this can as much as double with heavy bass content due to the non-linear impedance/frequency response:
This is something I don’t believe most people are aware of, let alone account for. As important as it is, it completely surprises me that it isn’t included in the measurements people take. I know it takes more than a mini dsp ears rig though, which is probably why.
Most people look at spec sheets and just take for granted that a listed impedance is accurate. It’s usually an average at best. It’s why I love and still reference Tyll’s works to this day.
Thank @Torq I was going to add warning label on this about this post about dB levels. No way would I listen on my ears at these volumes, above. This was more of the curiosity question what do they really measure at since I have seen all kinds of funny posts on this around the HD-6xx. So I decided to measure it. I have been looking at Spectrum as well.
I am aware of the variable Impedance on this headset and dynamic speaker in general. I have been looking into Speaker modeling for a bit. It started as a curiosity after getting UA OX amp Amp Top Box for attenuating and recording Guitar Amps, aka save my hearing when I push Marshall JMP 100 watt head into its sweet spot.
Again, I really have to emphasize 99 dB average in a four-minute-long song means you have longer section above 100 dB, as they say, please do not listen at this level.
My favorite part of the SPL Phonitor X amp is the noise floor is low which also means I can listen to music at much more comfortable dB levels. Nowhere near here and still get great tone.
Johnny Cash Hurt, to me, is a great song to test for distortion, it pokes it head out about 2:50 sec to 3:26 sec.
There is an opportunity to write a very good article on the physical behavior of Dynamic Driver vs Planner driver and what it really to take to drive them. Included things like Impedance. Voltage and Current swings/needs and lot more ( i am always surprised now one measure current in there very technical reviews) Note Ty always measures Impedance vs frequency graphs, he has them for a number of headsets.
It is also interesting to look at the input spectrogram post the DAC vs what you see coming out of the headset.
Keep in mind that for current required for the headphone that can be extrapolated from impedance and dB/(m)W (efficiency + impedance) or just through dB/V (sensitivity).
For amplifiers the current is shown through the power measurements generally, since most manufacturers give their specs as W into a given impedance. Based on where the amplifier goes up/down in power you can extrapolate at which impedance it becomes current limited versus voltage limited. For example an amplifier that delivers the most power at 32 Ohms impedance will be current limited at lower impedances but voltage limited at higher impedances.
For amplifiers the manufacturer data is often somewhat lacking, however, and I agree having more knowledge about which point the amplifier is voltage limited/current limited would be interesting.
Another thing: the inductive resonance frequency impedance spike of the HD6X0-series (and most dynamics) actually make them easier to drive since their sensitivity is still the same, the higher impedance means they demand less current, and since P=U*I is the formula for power it quickly becomes obvious that they actually demand less power here.
The challenge comes when you find the resonant peak, and maximum impedance, is so far off the quoted figure (like it is with the HD6XX) that it makes a material difference in how much voltage swing you need to get the SPL you want at the frequency you care about.
That’s the difference between just getting by with a DragonFly black and a DragonFly Red being a less-than-ideal minimum.
I don’t think so, because the output voltage of the amplifier is constant over the entire sine sweep, the current is what varies depending on impedance since the majority of amplifiers are voltage amplifiers and not current amplifiers.
Regurgitating my point; since the sensitivity (dB/V not dB/mW) is already accounted for in FR (also at the resonance frequency) we simply need to calculate the current, which is naturally lower the higher the impedance goes. Whether or not you are measuring the peak impedance or the average impedance or whichever figure you want to use doesn’t change this, the voltage necessary is constant but the current is not. Measuring the voltage output of the amp into the headphone when doing a sweep should easily confirm this.
Yes, the voltage of the amplifier IS constant over the sweep. What is not is the SPL output from the headphone at any given frequency for that voltage.
As long as the amplifier is not hitting its current or voltage limits, this doesn’t matter much. As it gets closer to one or the other, it will increase distortion, but that’s inevitable.
The changes in impedance don’t affect the amount of power required, but they do change the relationship of V to I. And if they’re extreme enough that they exceed the amplifier’s capacity in either case you’re not reaching the same SPL at that frequency.
Having an excess of current capacity doesn’t help you if you hit the voltage rails on the amplifier. Which is a real scenario with something like the HD6XX and DFB vs. DFR (which has about twice the voltage swing). You’ll hit the voltage limits with bass-heavy music at lower SPL here simply because the impedance swings so far.
This is something I know how to do very well, derive Current from Voltage and Watts. My background is Material Science; I went down the Solid State Physics and Polymer route for a specialty. I spent the last 25 years in CPU & GPU Architecture Processor Development and the supporting software and system. So I play with semiconductors solution part of my day job.
But when it comes to looking at current and voltage, sometimes I would rather measure it and see any behavior beyond what I calculate. For example, peak current behaviors are interesting to understand.
Back to topic, Here is a set of visuals that show the Impedence behaviors @Torq is talking about.
Dynamic Headphone in question
The SPL output of the headphone with a given voltage is just the FR though. Basically the FR tells us how much voltage it needs to do 100dB SPL@80hz or wherever the resonance impedance spike of the headphone is. Because of this you can not possibly clip/distort the amp with this impedance spike without the amp reacting the exact same way to the rest of the impedance curve (since, as we already established, the FR is just SPL at a given frequency with a given voltage).
Basically, unless you have an amp that is somehow current limited at the nominal impedance of the HD6X0 but but voltage limited at the resonance frequency impedance spike this isn’t an issue at all, since most amplifiers put out the exact same amount of voltage for Z300 and Z600 and the potential power difference of these 2 impedances is already accounted for by the aforementioned SPL at a given frequency with a given voltage.
Now, you can of course clip it at the resonance frequency by having a digital signal that hits, say, -5dBFS at the resonance frequency but the rest of the spectrum is -20dBFS, where -5dBFS is enough to drive the analog amplification into clipping with the given load but -20dBFS is not. The problem with setting this up as the problem is of course that you can simply do it the other way around and now the amp clips at the lower impedance instead. (I might be straying a bit from the discussion here, please excuse me for that).
I don’t wish to keep this as a long winded argument, as I believe both our points have been made pretty clear at this point.
TL:DR for people who might not wish to read all this. Torq argues that headphones need more voltage at their impedance spike while I argue this is not the case due to the frequency response measurement already telling us what the given SPL will be at a certain voltage, and the fact that the vast majority of amplifiers put out the same amount of voltage at Z300 and Z600 (or whatever the impedance spike is).
Why my fascination with current and headphones is not for the HD-6xx. For most OpAMP I have looked at the HD-6xx would not be current( mA) limited. It all really started with the Dan Clark Audio Aeon Flow Open which does not need much voltage with reported 13 Ohms and 112.9 dB/V but it little more hungry with current(mA). 94 dB/W
Yay for materials science folks
I have BS/MS degrees in MSE, but work in a totally different industry (aerospace, high perfomance composites/adhesives)
@antdroid very nice to meet fellow MSE. You’re in a very fun space. My dad was in Aerospace at McDonnell Douglas. Composites sure have come a long way. My Professor worked on Spectra, was researching Kevlar, Spectra, Carbon Fiber replacement something with high Tensile and Compression strength.
Here is a good article on modeling/emulating speakers
Designing A Reactive Speaker Load Emulator
General
A speaker presents a varying impedance load to the amp. The stated impedance is usually measured at 400Hz, and can vary widely over the frequency range. Speakers generally have a large resonant peak that can be as much as 5 times the rated impedance, or even higher. The impedance also starts to rise upward following the trough impedance at 400Hz. This rise can occur at varying frequencies and have a varying rate, depending upon the speaker and its enclosure.
The design of a speaker emulator
Following is a design for a circuit that will simulate the impedance variations you get with a 16 ohm speaker cabinet. The real speaker will probably vary depending upon how hard it is driven, due to motional restrictions of the cone, which this circuit will not do.
This circuit will make the amp react to the load, unlike pure resistive dummy loads, but if you want to tap off the input and send the signal to a board, or other equipment, you will need to attenuate it with a voltage divider, and filter it to simulate the frequency response characteristics of the amplifier. Designing a Reactive Speaker Load Emulator
Just out of curiosity, how did you measure SPL in the 6XX?
A few weeks ago I took my BM235 multimeter (wanted a Fluke but this one is quite OK actually) and started measuring voltage and current for fun. I was skeptical against this exercise (not my background) but now cross checking against your table, it actually makes sense. I missed the SPL part on first column though. I just remember it was too loud for me. Played Uprising from Muse. Very compressed song.
Now in these quarantining days I may go back and measure this again. Where did you get this table, btw?
