[castleofargh]

mechanical parts do impact the electrical behavior too. the impedance changes if we add physical damping, which would in turn require to rethink the RLC model. so I have no idea if the electrical model can fully simulate the actual headphone(and it's an interesting question if anybody knows?), but it's fair to say that there is a strong link and that the electrical model is indicative of some mechanical behavior.
then each headphone can be different, some headphones clearly have a mainly resistive behavior, while other designs don't. which makes it even more obvious that the ideal damping solution needs to be adjusted to the headphone. at least for those who care about ideals.

I was under the impression that increased damping would minimize resonance, did I get that wrong? are there situation where that's not how it works?
about the impact of high Z amp output, it obviously will depend on the relative value of the headphone's impedance, and how the amp itself deals with a given load. but I've experienced cases where adding 100ohm in series didn't seem to alter the sound audibly (aside from volume level), so that makes me think that poor damping ratio alone doesn't have to be a big deal(with everything else being clean and stable, no weird amp design, no protective caps, no disto from matching the volume...). despite that, I'm still of the opinion that following impedance bridging is generally the best choice for the amp, for the headphone, and for general stability.

 

[pinnahertz]

You've measured headphones, right? Ever check for damping effects, either in terms of response or in the time domain? Should pop right out.

A transducer resonant circuit does include the mechanical, as an underdamped "motor" generates EMF, diaphragm motion and back EMF are tightly coupled, and the amplifier would act as a load to that EMF with whatever series impedance might be in between being the limiting factor to damping, and the big variable. I would think the biggest impedance would be the motor windings, and if there's an electrical crossover. In any case it should be easily seen in a couple of measurements.

 

[Mython]

Technicalities aside, I can tell this thread is, relatively-speaking, populated by cans users more than by fans of multi-BA CIEMs.

It's screamingly obvious!

 

[pinnahertz]

Technicalities aside, I can tell this thread is, relatively-speaking, populated by cans users more than by fans of multi-BA CIEMs.

It's screamingly obvious!

How so?

 

[castleofargh]

Technicalities aside, I can tell this thread is, relatively-speaking, populated by cans users more than by fans of multi-BA CIEMs.

It's screamingly obvious!

because BA drivers add 2 often significant situations:
one that kind of falls under the crossover cases. but just a single BA is enough to have a mess of an impedance graph, if only from the tendency of BA drivers to get crazy high impedance in the treble. so a damping ratio change will often directly result in FR change.
and the other potential case to think about is how the amp reacts to a load that's sometimes closer to a short circuit than it is to a load expected by the designer of said amp.

 

[71 dB]

I don't think 6 kHz resonances are due to the transducer, but they are acoustic resonances inside ear cup / ear. Am I wrong?

It depends on the mechanical damping how much output impedance affects sound. If mechanical damping is low, you need strong electric damping which means low output impedance. The damping ratio of my Sennheiser HD 598 without electric damping is less than 0.4 (should be at least 1), so electring damping is needed. It turns out 56 Ω output impedance gives damping ratio of 1, but then again such output impedance would mean about 4 dB of bass boost, so for Sennheiser HD 598 output impedance should be kept under 5 Ω to keep bass boost under 0.5 dB. It seems that flat frequency response sets harder output impedance demands than damping.

This is a simplified model of a headphone (and amp) on electrical side. For Sennheiser HD 598 Rc = 60 Ω, Lc = 0.64 mH, R = 215 Ω, L = 240 mH and C = 10.6 µF. I got these values by iterating such values that the impedance curve of this simple model matches as accurately as possible the measured impedance curve. Since the match is very good, it's reasonable to assume that this simplification works quite well for describing the effect of output impedance on headphone sound. This simplified model incorporates the acoustic impedance that exists inside the ear cup. From these circuit values it's relatively easy to calculate the damping ratio

Here's the effect of output impedance on Sennheiser HD 598 in terms of damping. The signal is suddenly stopping 100 Hz sinusoid, as the resonance frequency for the phones is 100 Hz. As I mentioned, 56 Ω output impedance means damping ratio 1, critical damping. 220 Ω is clearly underdamped. 15 Ω is clearly at least critically damped and 1 Ω is even more damped. I hade some noise issues in this not so professional measurement, but it nevertheless shows the effect of output impedance on damping.

 

[Strangelove424]

You can't generalize resonance. I don't think anyone has tested a manufacturer's driver isolated from the ear cups since they are designed as a unit. So the only way to parse between driver or ear cup resonance is by modifying the ear cup and deducing from results. Every headphone is different. Monolith M1060 have a 5khz ring that people say can be fixed with felt (with measured results). Beyerdynamic's gen 1 tesla driver appeared to have a 10khz ring that's consistent between different cup designs, so I would deduce that to be driver related. One can't generalize the resonance frequency or a source of the problem.

Electrical dampening can be a cheap fix for mechanical dampening, but it's a band aid or sorts. The Beyerdynamic DT880 line comes in 32, 250 and 600 ohm versions. The difference? It's not the driver. It's not the cup. It's a resister inside. A 5 cent part. At one point, the 600ohm version sold for $100 more. The 600ohm version is known for it's better damping, more refined treble, but also being notoriously hard to drive. There's a major trade off when the engineers use electrical dampening as a cheap fix, and when you get back to high SPL you still have most of the mechanical dampening problems to deal with.

 

[SimonPac]

I'm rethinking a little bit from my first post here where HF resonances are concerned. I also don't think these resonances or peaks in the few KHz region have much to do with electrical damping, not in the steady state sine testing situation anyway. They are essentially acoustic/mechanical. If as the author of the InnerFidelity 'Musings..' article implies, headphone behaviour at certain frequencies is dominated by a mechanism which manifests electrically as a series LCR resonance, then we would expect to see a high-Q dip in the headphone impedance curve at the relevant frequency. I don't think you get this, certainly not in general, or if you do it is very minor and the electrical modelling won't allow you to damp the resonance in practice. A series LCR circuit goes low impedance at its centre frequency.

The parallel LCR plus series L and R model as posted by 71dB is more representative and explains the usual dynamic driver impedance curve. At extreme LF we get low Z, set mostly by the series resistance which is the driver coil plus amplifier output impedance added together. We then move toward primary resonance caused by the parallel LCR circuit and total impedance rises; this is what a parallel LCR circuit does at resonance. Total impedance then falls back toward the static value as frequency rises after the resonance, before rising gently at HF due to the series inductance of the voice coil adding a little reactance. The parallel LCR circuit itself is a good model of mechanical factors, with the R representing mechanical damping. The C represents inertia and the L represents suspension and acoustic compliance. It is well known that these factors can be fairly accurately modeled as lumped electrical parameters; look at literature on loudspeaker electrical modelling.

I have some HD518s which share the driver with the HD598. It is interesting that it is indeed possible to get critical damping with a fairly practical value for external series R. I have not tried to analyse the impedance curve into components but concur that around 50 Ohms series resistance gives the best sound with these 'phones for me. I wonder if 71dB has resolved the HD650 impedance curve into components and what amplifier series impedance gives critical damping for that?

I have seen a 500Hz square wave input response for the 518/558/598 and it seems to ring at about 5-7KHz with a fair bit of under-damping. I am inclined to think this is pretty much entirely mechanical/acoustic and is not significantly coupled over into the electrical model of the headphones. This seems to concur with what Sennheiser did with the HD800S. I believe they tackled the HF resonance present on the 800 with an acoustic system.

It seems that with dynamic headphones a flat LF response and critical damping are hard to obtain. The optimum output impedance is often a moot point.

 

[71 dB]

Electrical dampening can be a cheap fix for mechanical dampening, but it's a band aid or sorts. The Beyerdynamic DT880 line comes in 32, 250 and 600 ohm versions. The difference? It's not the driver. It's not the cup. It's a resister inside. A 5 cent part. At one point, the 600ohm version sold for $100 more. The 600ohm version is known for it's better damping, more refined treble, but also being notoriously hard to drive. There's a major trade off when the engineers use electrical dampening as a cheap fix, and when you get back to high SPL you still have most of the mechanical dampening problems to deal with.

Mechanical damping reduces sensitivity. Especially in these days of "mobile audio" sensitivity is a selling point. That's why damping is done so much electrically. I wouldn't call it a cheap fix. On the mechanical side electric damping looks the same as mechanical, the diaphragm can't tell them apart.

There are not "resistors" inside headphones to dictate the impedance level. The electric resistance comes from the resistance of voice coil + the inverse of mechanical resitance on electric side. Larger mechanical resistance means smaller electric resistance and vice versa. Beyerdynamic has the 32 Ω version for portable devices which can't produce much voltage or power while the 600 Ω version is pure home audio to be driven form powerful amps. 600 Ω has much better electric damping, because even larger output impedance is small compared to the impedance of the phones and therefor the damping ratio remains large enough.

How about a headphone with changeable impedance? At home the higher impedance is used and then you have lower level for portable devices. It could be done with double voicecoil. In low impedance mode the voice coils are in paraller and in high impedance mode in series. For example: 70 Ω voice coils. In paraller mode the impedance is 35 Ω and in series 140 Ω. What if the phones had two cables, longer for home use and shorter for outdoor use? Special connectors + different wiring would produce the impedance modes. What if the phones would change from open to close for outdoor use: Closed mode: 35 Ω, open mode 140 Ω ? The voice coils could be even different, one 34 Ω and the other 566 Ω to have 32 Ω and 600 Ω impedance modes. The challenge of course is to have great sound on both modes.

 

[71 dB]

I have some HD518s which share the driver with the HD598. It is interesting that it is indeed possible to get critical damping with a fairly practical value for external series R. I have not tried to analyse the impedance curve into components but concur that around 50 Ohms series resistance gives the best sound with these 'phones for me. I wonder if 71dB has resolved the HD650 impedance curve into components and what amplifier series impedance gives critical damping for that?

Unfortunately not for HD 650, but I have values for HD 600 which is quite similar. I have 151 Ω for it for critical damping. Frequency response criteria (FR error 0.5 dB) For HD 650 is 53 Ω (42 Ω for HD 600).

​

I have seen a 500Hz square wave input response for the 518/558/598 and it seems to ring at about 5-7KHz with a fair bit of under-damping. I am inclined to think this is pretty much entirely mechanical/acoustic and is not significantly coupled over into the electrical model of the headphones. This seems to concur with what Sennheiser did with the HD800S. I believe they tackled the HF resonance present on the 800 with an acoustic system.

I think they tried to emulate the response of torso, head and pinna with this "ringing". Controlling frequences this high isn't very difficult. You need absorbent material and asymmetrical shapes to spread and tame the resonances. I have never had any ringing issues with my HD 598 nor does the measured spectrum look alarming in this sense.

 

[SimonPac]

I think they tried to emulate the response of torso, head and pinna with this "ringing". Controlling frequences this high isn't very difficult. You need absorbent material and asymmetrical shapes to spread and tame the resonances. I have never had any ringing issues with my HD 598 nor does the measured spectrum look alarming in this sense.

Maybe the 598's have better hf resonance management due to housing arrangement and materials compared to the 518. I think 518/558/598's are nicely judged headphones especially for the money.

 

[SimonPac]

Unfortunately not for HD 650, but I have values for HD 600 which is quite similar. I have 151 Ω for it for critical damping. Frequency response criteria (FR error 0.5 dB) For HD 650 is 53 Ω (42 Ω for HD 600).

Would makes sense. I go with 50 to 100 ohms for longer term listening with the 650. Less amp z out sounds more 'impressive' on a really good HD recording though. The balance is dark even at low amp z out; and more so with substantial series R. Still, I like these phones for understated refinement and resolution. They don't really sound particularly 'impressive' on a quick listen.

 

[ev13wt]

Awesome thread development, thanks. I am learning stuff!

 

[71 dB]

Maybe the 598's have better hf resonance management due to housing arrangement and materials compared to the 518. I think 518/558/598's are nicely judged headphones especially for the money.

HD 598 is the only model of these I have heard (and I love it!). I know HD 558 is very similar, but HD 518 is more for "bassheads". Not anywhere near Beats Audio level bass, but maybe 3 dB more than the other models. HD 598 definitely is a great headphone for it's price, especially the soundstage is great and pretty spectacular with proper cross-feed and well recorded music.

Would makes sense. I go with 50 to 100 ohms for longer term listening with the 650. Less amp z out sounds more 'impressive' on a really good HD recording though. The balance is dark even at low amp z out; and more so with substantial series R. Still, I like these phones for understated refinement and resolution. They don't really sound particularly 'impressive' on a quick listen.

Yeah, 50-100 Ω output impedance is reasonable with HD 650. The whole product line HD 580, HD 600 and HD 650 is easy on the output impedance demands.

How do you adjust the output impedance? Personally I am for "low" output impedance, because it means the amp has strong control over the diaphragm movements (good damping and low linear distortion) and the frequency response error is small.

 

[71 dB]

SimonPac, I took the time and calculated the electric circuit values for Sennheiser HD 650. Here they are:

• fo = 67 Hz
• Rc = 310 Ω
• Lc = 1.7 mH
• R = 166 Ω
• L = 800 mH
• C = 7.05 µF

Minimum damping ratio = 1.014 (at least critically damped for any output impedance). More accurate frequency response criteria for Rout < 60 Ω and linear distortion criteria is Rout < 495 Ω. Below are the same values for other headphones models I calculated in 2016 when I made this study.