Meridian Explorer: A Case Study of the Effects of Output Impedance on Headphone Listening

Meridian Explorer ($299)
In early February 2013 Meridian released it's new USB powered DAC/headphone amp with much fanfare. By the end of the month, member Purrin had found its performance in listening tests underwhelming and had measured the output impedance of the Explorer at about 48 Ohms. This is not good.

Fortunately, Meridian had already received feedback from early units delivered to dealers and distributors, and had got wind of the problem when driving the low impedance headphones commonly used in portable applications. An engineering change was put into motion. Unfortunately, the wheels were already turning and the product introduction continued on schedule with the original product going into production. From what I could gather, a few thousand Explorers were shipped before the running change could be made. The first run of modified product has just finished production (about March 13th 2013), and is currently in shipment to dealers. I'm told the Explorer is selling very well, and dealers have little stock on-hand from previous production runs. I expect it would be safe to place an order now from any retailer that is currently out of stock, and am fairly certain that the unit received will include the changes.

I received the revised version a few days ago, and measured the output impedance:
With 0.1988Vrms out open circuit, I get 0.1922Vrms with a 150 Ohm Load for a calculated 4.99 Ohm Output Impedance. With 0.1988Vrms out open circuit, I get 0.1723Vrms with a 32 Ohm Load for a calculated 4.92 Ohm Output Impedance.

Bottom line: The revised Meridian Explorer now has a about a 5 Ohm output impedance.

What's the big deal? Let me explain...

Output Impedance, Frequency Response, and Damping Factor
In the illustration above, the circuit to the left models the prior Explorer with a 50 Ohm output impedance, the circuit on the right shows the current version. The resistor shown as Rzout is the output impedance of the amplifier. The word 'impedance' indicates both resistive and reactive (capacitive and inductive) elements and is not actually a simple resistor, but for the moment we'll just think of it like a resistor inside the amp and in series with the output. The Rzload is the voice coil of the headphone. I used the Sennheiser Momentum in my listening tests, which has a 22 Ohm impedance, but I've shown a 25 Ohm load just to make the math easy. With an output impedance of 50 Ohms into a 25 Ohm load, 2/3 of the voltage (and power) in the amp is being lost in driving through the output impedance of the amp, with only 1/3 the power being delivered to the load. Power is lost inside amplifiers all the time, so inneficiency isn't really the problem here. Let's move on to the diagram on the right.

The circuit to the right models the new Explorer with 5 Ohms output impedance. Now the ratio of output impedance to load is much different. In this case only 1/6th the voltage is lost in the amp, and much more voltage is available for driving the load, so this is a more efficient configuration.

The important thing to observe here is that changing the ratio of load impedance to output impedance changes the voltage at the load. Lower the output impedance and more voltage is available at the load. But the converse is true as well: If you raise the resistance of the load, it will get more voltage. That's very important to know because most headphones have impedances that change with frequency. Let's talk about that.

Effects on Frequency Response with Headphone Impedance Changes
Let's look at the impedance curve for some headphones.

Sennheiser HD 600

The purple trace in the graph above is the impedance curve of the Sennheiser HD 600. At 1kHz you can see it's just above 300 Ohms, but the primary driver resonance at 100Hz causes the impedance to rise to 550 Ohms. So at 1kHz the HD 600 looks like a 300 Ohm load, but at 100Hz it looks like a 550 Ohm load. That means that the HD 600 will get a little more drive voltage at 100Hz than at 1kHz, and will be somewhat louder at that frequency dependent on the output impedance of the amp.

In the case of the original Explorer with 50 Ohms output impedance with 1Vrms drive, the headphones will get 0.86Vrms at 1kHz, and 0.92Vrms at 100Hz. That calculates out to a 0.58dB increase at 100Hz. The point here is that the impedance curve of the headphones interacts with the output impedance of the amp causing a frequency response change of the headphones. But because these are fairly high impedance headphones the change doesn't amount to much. This is one of the reasons why it's important to use high impedance headphones with high output impedance amps (like OTL tube amps).

Let's look at another pair of headphones.

Audeze LCD-2

Here's a pair of planar magnetic headphones (Audeze LCD-2). One of the unique characteristics of planar magnetic cans is that they are purely resistive in nature and the impedance doesn't change at all with frequency. Even though these cans have a low impedance and would strongly interact with the high output impedance of the early Explorer, the impedance doesn't change so there will be no frequency response change due to the high output impedance of the amp. The poor damping factor will have its effect on the sound though, we'll get to that in a bit.

Audeo PFE

Here's the impedance plot of the Audeo PFE balanced armature in-ear headphone. Again, the purple line shows the impedance plotted against frequency. It starts out as a 32 Ohm headphone, but as frequency rises, so does its impedance. At 3kHz it has a driver resonance bump to 50 Ohms, and at 20kHz its impedance has risen to 80 Ohms. With the high 50 Ohm impedance of the original Explorer, and the low overall impedance of these cans, we can expect a significant coloration as the high output impedance interacts with the widely swinging impedance curve of these headphones.


In the plot above I measured the frequency response of the Audeo PFE with both the old and new versions of the Meridian Explorer. I then plotted the difference (blue) between the two FR plots in dB on the left hand scale. I also plotted the impedance curve of the Audeo PFE on the right hand scale. Because of the step size of the digital volume control of the Explorer I couldn't exactly match the two levels and ended up with about a -1dB difference at the low frequencies. So the difference reference is at the -1dB line.

We can see that with the 50 Ohm output impedance of the old Explorer, we have a fairly strong effect on frequency response due to the interaction of the headphone impedance curve and the high output impedance of the amp. At the 3kHz impedance bump the Audeo PFE was 1.5dB louder with the old vs. new Explorer, at 20kHz it's nearly 4dB louder. This is a fairly significant change in frequency response and would be easily heard as the PFE being substantially brighter sounding on the old Explorer.

Frequency Response Summary
If you have one of the early Meridian Explorers you will find some headphones more colored than they normally would be due to the interaction between the impedance changes with frequency and the output impedance of the amp. Generally speaking, higher impedance headphones and headphones with very flat impedance curves will suffer less coloration due to the 50 Ohm output impedance of the early Explorer.

There's more to this story though, let's talk about Damping Factor...

110 Greene Street, Suite # 407
New York, NY10012
646 666 0140

lithium's picture

Hi Tyll,

Excellent technical article and really helps in understanding the problem. It seems several manufacturers releasing new products are having this problem. John Grandberg mentioned in the comments of the ZDAC article that they also had a problem in their initial production run and had an output impedance of 40 ohms from the headphone out. They have corrected it to 10 ohms now. This could also be a potential reason for John rating his ZDAC(10 ohms) better than the XDA-2, while Steve Guttenberg felt the ZDAC (40 ohms) to be inferior through headphones.

Can I also suggest more technical articles as I feel there is a lot of false unscientific information floating around. Thanks

John Grandberg's picture

I was thinking the same thing with regards to why I enjoyed the Zdac more than Steve. 

Impulse's picture

Second image of the article doesn't seem to be loading for me, above where it says "In the illustration above", the circuit comparison. Anyone else?

I've also got a question regarding damping factor... Does it apply equally on planar magnetics? The last line on the LCD2 impedance section seems to imply it does, but I've seen a lot of people stating the contrary (supposedly because they lack conventional drivers)... The concept just seems a little harder to understand and study than the way a swinging headphone impedance interacts with the output impedance.

Great article btw! Though Meridian's comments at the end seem a little disingenuous, maybe I'm just reading too much into it but the product IS a portable solution and they're making it seem like low impedance headphones or multi-driver IEM are a rarity in that realm rather than the norm. I imagine even a 5 ohm output impedance would still be an issue with a ton of IEM out there...

John Grandberg's picture

The image is missing for me when I use IE but works fine in Chrome. Weird. 

For LCD-2 and other planars, I notice a slight difference in bass control with a lower damping factor, but it's far less significant than with other types of headphones. 

Impulse's picture

It's actually missing for me on the stock Android browser. Thanks for the note about planars, seems like another good reason to go for a pair of them for use in the living room with my AVR.

Edit: They're showing up on Chrome for me now.

Seth195208's picture

You have a real gift when it comes to explaining technical subjects to laymen. Great article.

Tyll Hertsens's picture

Thanks man. 

You know, I'm not an engineer, at best I'm a decent technician. But I'm an awfully curious guy and have always been interested in the way things work. So in many ways I'm a layman myself...that's just figured a lot of stuff out.

I feel my job here at InnerFidelity is primarily to open up the eyes of hobbyists and enthusiasts to the field before them; to help them get off the ground and learn enough to find the parts of this activity that are of real interest. And to do that in the broadest possible way. Once folks involved in this activity are on a path towards some particular interest, there are a number of folks far more qualified than I to learn from. So your comment strikes deep in my personal sense of mission. 

I want to clearly make the statement that I encourage those who are very qualified----far more than I---on certain topics to reach out to me via PM or email ( if they wish to write articles for this publication.  There are LOTS of things I find myself not feeling qualified to comment on. 

I'll give you an example: I'd love to see a difinitive article on Fostex T50RP modification. I couldn't possibly wite that. I could help with some measurements, but I've very little experience modifying headphones.

The one thing I do want to stress though, is something I try very hard to do in my writing, which is to be authoritative and difinitive to enthusiasts at large. I want to create a publication that when someone asks, "Where do I start?" The knowledgable entusiast says, "First read the stuff at InnerFidelity." The good news here is that if the goal is to provide solid basic information for the beginning and intermediate level enthusiast it's not terribly hard...but you do have to put in the effort to make sure it's spot on information. 

I tell you something truly: I'm not as productive in word count as I'd like to be. When I look at other audio journalists I find myself saying to myself, "How the fuck do they write so many articles?" I'd like to do more, but I end up spending so much time reading the forums and various papers and such that it just takes way more time than I'd like to develop what I would consider 'The Full Story'. 

Uhhh...I've rambled off topic. Thanks for the complement, mate. I'll keep working on it.

ultrabike's picture

Being able to explain concepts in layman's terms is a gift and requires effort, all of which IMO translate into quality articles here at IF.

Seth195208's picture

The beauty of being a "subjectivist" leaning audio reviewer is that you can write all day about what you feel subjective about. Diarrhea of the hand is totally allowed and perfectly OK.

THE "objectivist" leaning audio reviewer is constrained by his inability to allow himself to write about things subjective because he deeply believes that his own personal opinion is just not a good enough reason to potentialy make or break a company's product(Or maybe even a company itself). Measurements at least tell certain truths about a product and he feels that writing about a product within the confines of those truth's is a more responsible aproach. This also means that ultimately, he's going to have much less to say compared to a subjectivist writer because good research and good measurements are  difficult and time consuming.

Readers of objectivist leaning magazines and websites also have less to say in response to these articles for the same reasons. Good research and good measurements speak for themselves. There just isn't much to add other than "thanks for the info!"

Readers of subjectivist leaning mags and websites usually have much more to say on forums because there are no objective truth's to abide by. Ultimately, anything goes..

Impulse's picture

Well said! I think it all points back to the blogger vs journalist argument... The blogger's just chiming in with his opinion, repercussions be damned.

Bloggers aren't held accountable to the same standards (or any standards) and they're not necessarily after the truth of the matter, just a story. The professional journalist or reviewer comes in with a more objective frame of mind because he believes that's precisely the service his readers seek...

Anyone can have an opinion, doesn't take any education or even experience in your field... to the heart of the matter and presenting readers with a properly researched and balanced perspective takes far more work.

Innerfidelity kinda reminds me of Anandtech in that regard (for all you computer geeks out)...

Seth195208's picture

..are kept in line when they know that a guy like John Atkinson is there to compare his measurements with their subjective opinion. 

HiFiGuy528's picture

Thanks for letting us know about the issue.  I appreciate your work. :)



OneMic's picture

I think it is great to see articles written on the interactions between amplifiers and moving coil speakers; so thanks for writing it. 

There is however a very large problem with this article.  It is the assumption that voltage is what drives a moving coil speaker, when infact it is the current from the amplifier which produces the sound.   The equation for the applied force to an electromotive system is F=B*l*i   (Force = Magnetic Flux Density * length of wire in magnetic field * current)    {It should be noted however that electrostatic systems like Stax headphones are voltage controled.}

Nelson Pass wrote a great article about current drive when he released his F1 (80 ohm Z-out) amplifier for single driver speakers which is definately worth the read


xnor's picture

I = V/R ... Ohm's law

Near the resonant frequency efficiency usually rises causing the driver to need less current to output the same sound pressure. Forcing a fixed current across the driver will usually cause peaky, loose or kinda rolled-off bass and other frequency response oddities.

Tyll Hertsens's picture

You are exactly correct...problem is it's much easier to explain with dialog using the voltages. 

But I want bypass commenting on my post to say that I think current drive amplifiers will be the next big thing in headphone amps.  Transconductance amplifiers (voltage in/current out) are tricky beasts, but I think we'll find that the relatively easy load of most headphones will allow makers to design really good current amps with 15 Volt rails and off-the-shelf parts. 

I used to work on scanning electron microscopes, and the only way to get linear rastering of the electron beam is by driving the scan coils with current amplifiers. If you want linear movement of a voice coil, a transconductance amplifier is the way to go.

ultrabike's picture

Thank you for the article linky OneMic! Currently reading...

xnor's picture

A thing is that these full-range drivers are optimized for efficiency, have very low Qts (as low as ~0.2) and big magnets and therefore normally need a horn to produce decent amounts of bass. The low Qts also explains the bass roll-off.

Headphone drivers usually work in an extremely small (relative to driver size) more or less sealed chamber on the inner side. Measurements show most headphones are underdamped suggesting a high Q factor - the opposite of those loudspeakers in that article.

Current drive might work with some headphones where output impedance doesn't matter much in the first place, e.g. those with mostly resistive impedance like planar magnetics ... but for the majority of headphones it will just make frequency response and even distortion worse.

ultrabike's picture

First off all, thanks for the article (Nelson Pass). It made me think through things I have not done so before. Very interesting read.

It makes sense to me that the current and the voltage output of an amplifier will not  track each other well unless the load is resistive. If dealing with a voltage amplifier, the current will be both a function of the source and the reactive load. If dealing with a current amplifier, the voltage will be a function of the source and reactive the load. More than likely the amps will behave differently.

Like the article said, many (if not most) speakers, and perhaps headphones, assume a voltage amplifier and this may be reflected in the balance and smoothness of the Frequency Response (FR) SPL plot when using such amplifiers. If using a current amp, it is quite likely that the FR plot will look very different, and not necessarily for the better.

After a quick read of the Nelson Pass article (thanks OneMic!) I believe that the current amp was exploiting the impedance of the full range speakers to improve on some of their issues in FR. Maybe the sensitivity vs. frequency of some of this drivers (not necessarily all drivers in general) deliver better SPL FR when using a current amp. But that probably was not sufficient, and a parameterized analog equalizer optimized for a particular driver was developed (the RLC circuit in page 5.) Depending on the driver and enclosure, the circuit components are tuned.

It is discussed that the circuit can be reconfigured in series for voltage amplifier applications. Furthermore, some drivers (such as the Jordan J92S) did not require that much RLC circuit tweaking and did not perform much different when driven by a current or voltage amp. It seems its all very driver dependent and basically a full system optimization problem.

I tend to agree with xnor in that headphones (specially crazy impedance ones with decent FR when driven by typical voltage amp) might actually degrade tonality wise if using a current amplifier.

As far as full range drivers for speaker applications, I believe many of this gains from current amplification + analog eq can probably be achieved with the use of a parametric or multi band equalizer. The advantages here are perhaps that an RLC circuit is passive and perhaps cheaper/practical.

Sorry for being so far off topic.

Seth195208's picture easy to hear with most iem's. In fact, in this day and age, I'm amazed that any manufacturer believes that anything above 1 ohm is acceptable for any headphone amp. It ain't like it used to be..

Impulse's picture

Yeah, it'd be one thing if this was a desk amp... Then you could argue (as they tried to) that it just isn't made for certain kinds of headphones. It's a very compact portable product tho!

I would assume the intended market is precisely IEM users and low impedance headphones like the Momentum and other portables. Maybe I'm just way off on that assumption tho...

Tyll Hertsens's picture

I think you're right.  But it's reasonable to cut Meridian a little slack as this is their first headphone product. Sure, we can say they should have known better, but a lot of what people know come from learning from mistakes. So, lesson learned for them.

Impulse's picture

At 'least they did react and implemented a change even as they're selling out on the product.

elmura's picture

As an audio designer, there are always trade offs depending upon the application, target market, and target price.

The lowest typical IEM phone impedances is 16-ohm, with most IEMs measuring much higher than that. Divide 16 by 8 and you get a 2-ohm output impedance as a maximum for those headphones. Therefore, above 1-ohm is acceptable.

The other end of the spectrum, 600-Ohm impedance phones, the output impedance can easily be up to 75 Ohms without having any colouration (assuming the amplifier can drive that sort of load)

The point is, do not dismiss an amplifier design off-hand because of the output impedance - it depends on the application.

lachlanlikesathing's picture

I'm so glad that you posted this. Tremendously great explanation with graphs as well!

At least as far as Head Fi I'm seeing more and more people understand what effect output impedance actually has. With the combination of impedance graphs and output impedance figures on amps we can start to make some educated guesses about the combination of amps. (Compared to when I joined a few years ago and the idea was basically: High impedance? Amp it, more bass! Multi-BA? Amp it, more bass!)

I hope we can see the industry move towards always specifying output impedances on amps.

I'm wondering if you would consider measuring the output impedances of some popular sources? There are bits of information here and there but nothing assembled into one nice place.

And one question: resistance adapters. (This has always confused me). Adding an impedance adapter in series between the headphone and the amp - does this just increase the Z-out of the amplifier and reduce the damping factor? Or does it increase the headphone's impedance as seen by the amplifier? What effect does this have on the sound?

xnor's picture

Both: for the headphone the output impedance increases (reducing the damping factor), for the headphone amp the load impedance increases.

mikeaj's picture

edit: nevermind xnor covered it already

Seth195208's picture

..goes a long way.  If I was designing a product (which takes a Lot of time), I would at least be a little curious as to what people need, especially in a small portable amplifier.  A couple hours on Innerfidelity and head fi is all it would take. What you would get in return for those two hours is priceless.

mikeaj's picture

How much of the difference in the 30 Hz square wave plots is a consequence of the different frequency response?

Tyll Hertsens's picture

Probably not much as the Momentum's impedance plot is fairly flat, and with the 30Hz square wave shape change is probably only important from, say, 150Hz and lower. So I think most of the change to the wave shape change was due to damping factor, but it is just an educated guess.

ultrabike's picture

I could be wrong, but think electrical damping factor might be lumped into the overall SPL FR (DAC+Amp+Headphone).

If I remember correctly, electrical damping factor is defined and derived for 2nd order RLC circuit Transfer Functions (which are related to the Frequency Response by simple variable substitution: s->jw.) Higher order circuits could be decomposed into several 1st and 2nd order TFs, each of which define their own damping factors (maybe sort of like having multiple Q factors in audio.)

In other words, I think there is always a relationship between electrical damping factor( s ) and elecrical impedance FR in electrical circuits.

Furthermore, if the output impedance of the amp is not zero, the coil in the headphone may result in some feedback. But I think that this should be captured by the Amp+headphone electrical impedance Transfer Function (which takes into account C and L storage and time constants)  which together with the Sensitivity vs. Frequency might yield the SPL FR of the whole system including Headphones.

If you notice Tyll, the L and R driver of the Momentums are fairly well matched, but there seems to be slight differences. Me thinks (or hallucinates) maybe enough to affect the 30Hz square wave...