Measurements: Charts, Graphs, Software & Methods


#1

Several of you have written reviews that have nice measurements. I read the EARS section a few times, but it sort of looks to me as if the EARS headphone measuring equipment is meant to be used with the mini DSP.

I also see some charts and measurements of DACs and amps. I’ve got an older digital multimeter, but that’s all I have that’s designed to measure. I’ve got iPhone iPads and a Mac. And a Windows box (my Linux box died a horrible memory death). I have a Blue microphone.

I’m sure there is some software out there that generates some of those nice measurements, but I don’t know what ancillary equipment is needed. I know that there are lots of chips out there that convert analog to digital for experimentation, but I don’t think I’m looking to track temperature or barometric pressure.

So, I’d appreciate it if some of the more experienced reviewers and industry people out there would talk about what they use and what an amateur can use to make meaningful audio measurements for comparisons. Like my xDSD DAC, the TEAC, and the Dragonfly Black. Or my Grado headphones, 1More Triple Drivers, and other headphones.

I can find measurements that others have made online just as well as anybody. But I don’t know if that’s useful, as I often don’t know the background, parameters, etc. And sometimes I know the biases - like if it is measured by a manufacturer.

Any help and discussion would be welcome. I don’t know if I want to spend an extra $50 or $500 or anything on equipment, but I’d like to know what’s involved.


#2

The miniDSP EARS is multiple uses. It can certainly be used to create EQ profiles with a miniDSP box, but it also a more broadly applicable tool for measuring headphones and IEMs.

It has some latent limitations, a few oddities, and a quirk or two, but it is easy to use, comparatively inexpensive, and seems quite repeatable between different copies. It’s worth a separate discussion.

Beyond that … in general … and at a very high level that I (and I’m sure others) will delve into in more detail …

Taking measurements, and production graphs or charts from them, require several things:

First, the item to be measured … generally referred to as the “Device Under Test” (DUT). This might be a headphone, an amplifier, a DAC or some other widget.

Next you need a way to take a measurement. This varies with what you’re measuring. For headphones or speakers you need a microphone (calibrated, ideally), some way of capturing the output from that microphone (which could be a dedicated audio interface, or something as simple as a mic-input on a sound card or phone), and a way to get that information into a measurement device (usually done with an analog-to-digital converter or ADC).

If you’re measuring a DAC or amplifier, then a microphone would not be necessary and instead you’d connect your ADC to the output of the DAC or amp in question.

Then you need a signal generator, which can be a dedicated piece of hardware or it can be software. If it is software, then it’s going to have to feed something that can turn the generated signal into an analog form - so typically this is a DAC (digital to analog converter).

Finally you need software (or hardware) that can drive and synchronize all of this, so that it knows what it is supposed to be measuring relative to the signal is being generated. Capture the data, and present it in a suitable form.

You can do all of this with the built-in audio hardware in almost any PC, laptop, and a good number of tablets and phones. This might take the form of a dedicated tool for audio analysis or it could be a more general-purpose signal-generator/virtual scope arrangement.

Often a better approach is to use an external audio interface, such as the Focusrite Scarlett Solo (about $99 and has DAC, ADC, mic-input/amp and a headphone output … which covers the vast majority of what you want to measure) and suitable software (e.g. Room EQ Wizard or ARTA).

Beyond that are dedicated audio-analyzers that typically offer much better performance (lower noise, better resolution, more automation, higher accuracy, greater consistency/repeatability) but at rapidly increasing costs. Something basic like the QuantAsylum QA401 can be had for about $449, and you can spend thousands upwards from there on tools like the PicoScope III or the APx555.

So that’s a bit on tools … and I’ll post a lot more here as I have time.

Technique and measurement regimen is at least, probably more, important as your tools. It is trivially easy to completely bork measurements, especially when dealing with signals at -120 dB, simply by running an AC cable too close to a signal cable, or taking a measurement when your refrigerator’s compressor is running etc.

After I get my next review published, I’ll do a couple or so posts to illustrate the specific tools and approach to do some simple things like measuring the frequency response and jitter for a DAC, and getting a basic response profile for a pair of headphones or speakers.


#3

Thanks this is a really interesting subject for me. As you say it can be prohibitively expensive for a lot of us and perhaps the best option for me would be the Mini DSP Ears setup. I am sure a lot of us that don’t already own measuring rigs are thinking along the same lines. It will be nice to discuss this subject with everyone and hopefully I will learn plenty too.:grin:


#4

If you’re reasonably handy from a DIY perspective, don’t mind a little (easy) soldering and have some free time, it’s possible to put together a perfectly serviceable headphone-measurement rig for a lot less than the miniDSP EARS costs.

The only real downsides to doing so are that a) it won’t be comparable to any other rig, b) you’ll have to invest some time and energy in getting it dialed in and c) you’ll likely have less overall headroom and resolution.

Something as cheap and simple as a the Behringer UM2 would get you going with a ADC, mic-pre, phantom-power, DAC and headphone output. It won’t be useful for hard to drive headphones, is a bit limited in overall resolution and linearity, but all you really have to add to it is a calibrated microphone, a means of coupling that to the headphone and some free software.

Another approach is to use an existing DAC/amp as the source, and go with a basic calibrated USB microphone, and only worry about how you’re going to mount/couple the transducers to the rig.

Mostly it depends on what you’re trying to achieve. If you just want empirical feedback on personal EQ’ing, or to satisfy some basic curiosity, then the cheap/DIY approach will work well. If you want to do something more definitive or reliable (or more broadly relevant to other enthusiasts), then you have to invest more time and/or money.


#5

Thanks for the input. I think I would probably purchase a ready made setup in line with what a lot of other people buy. I haven’t looked deeply into it yet but will be interested to follow any discussions that crop up.


#6

Thank you very much for taking on this subject. I’m not wholly ignorant, just missing the most recent 35 years or so of technological advancement. Back when I last looked into this, I hung out with some radio folks, and all the equipment was expensive and dedicated to a single task.

I figured that there had to be significant changes, and I wouldn’t need an oscilloscope and separate signal generator.

It might be worth getting the EARS unit for convenience. I don’t mind some soldering, but as you point out, any DIY rig would have to be calibrated, and might not be comparable to other’s measurements. Even though I am in awe of the range of Texas Instruments A/D chips.

My next door neighbor is a piano tuner. He says how most in his profession use an app these days. He might on occasion, but always checks against a set of professional tuning forks. There’s always room in my posts for an analogy or non-sequitor.


#7

A few of us on this forum have the MiniDSP EARS (@Torq, @pwjazz, @Ishcabible, myself) and it’s pretty good for the cost. It definitely has some limitations and isn’t considered industry standard, but it works well enough. I find my measurements are close to supplier data so that makes me pretty confident. I know where the limitations lie.

It’s a good comparison tool at the very least, but I’ve used it to EQ some headphones before. @pwjazz has done a lot of work on the EQ side of things using the free Room EQ Wizard (REW) software with his EARS unit and shown his work on this forum.

Before I got the EARS, I made my own flat plate coupler using a box filled with memory foam with a hole cut out that fit an Audessey calibrated mic (the ones that came with my Marantz AV Receiver) and then leveled it flush through an opening of a CD-R disc. This became the flat coupler plate for headphones. I covered the CD-R surface with a heavy felt to reduce resonance and cabled it through a cheapo $10 USB soundcard stick that had mic input.


#8

At least with the EARS unit it will be at least have fairly comparable results with other EARS users. This unit seems to be what most hobbyists use. Although I am generalising a little. I know @pwjazz does some great EQ work. When I finally get my lazy arse into gear that’s the next thing I want to get into a little more. What with you having experience with my Dap and hie EQ know how I know who to call.:grin:


#9

Some other popular mics used for DIY stuff are the Dayton Audio iMM-6 and MiniDSP UMik1, Torq linked above.

The Dayton is ultra cheap but its not very durable. Many users have posted it falling apart. It’s hand in that you can plug it directly into your phone if it still has a 3.5mm jack.

Its popular with the IEM crowd. You can get some thick tubing, wrap it around the mic head, and then put your IEM through the tubing at the right distance and seal it up with silicone putty and get a good measurement. I believe if you use the MiniDSP mic, and proper thickness of the tubing (8mils IIRC), it actually becomes close to or is standard. Look up crinacle’s measurement database for more info on this. It’s a cheap DIY solution.


#10

The E.A.R.S. has been useful for EQ, but it’s more compass than GPS–it’s best used in conjuction with listening and comparing measurements with headphones you already like.


#11

It’s curious that there aren’t available calibration settings for some of the more common studio and USB microphones like the Blue Yeti, Blue Snowball, or Samson mics.


#12

IIUC the mics would have to be calibrated individually, which seems like a big commitment for the manufacturers.


#13

In addition to @pwjazz’s accurate statement (each individual unit would have to be calibrated), the microphones you’re talking about aren’t really designed for measurements. They’d work in a pinch, but their pick-up patterns (even the switchable ones) would be a long way from optimal for taking headphone measurements.

And unless you had one for voice-recording purposes already, they’re not any cheaper than a calibrated USB microphone anyway.


#14

Yes, Siri lives inside my Mini, and I talk to her using a Blue Snowball. It’s also very good for Skype and the occasional voice recording.


#15

I’ve actually been compiling data comparing the EARS to a proper HATS now that I have measurements.

Here are two examples of the differences between a Brüel & Kjær 4128C and the EARS using the same sample of headphone:

I’ve labeled the measurements accordingly. The HATS comp uses a Brüel & Kjær 4128C diffuse field (DF) compensation as reference. It’s closer than the HPN comp but it’s still pretty problematic in that it doesn’t follow the HATS much after 4k (and before 1kHz it’s pretty far off, but between the HPN and 4128C is likely the difference between the 4128C’s DF comp and whatever the HPN uses so I don’t consider that to be too much of an EARS issue). The end goal of taking EARS and 4128C measurements concurrently was to hopefully be able to make a better compensation, but it seems like the differences aren’t consistent and come down to how the headphone reflects on the ear, which isn’t really fixable.

I’ve had an EARS since the first run and I was really, really hoping it would at least be something people would be able to use to compare between different EARS units, but I’ve heard a few people complain about large inconsistencies between units so not even that seems very viable.

It does get general curves though, like you can tell that the LCD-2C measurement is an Audeze, but it doesn’t seem to really align much with what I hear when you analyze it further because I hear a rise higher than the EARS’ 5kHz resonant rise. What I hear is somewhere between 5-6k because it’s a kind of metallic harshness that stands out versus the otherwise dipped upper midrange. But as you can see in the Utopia measurement, both EARS compensations are really funky past 3kHz in ways that seem very, very different than in the 4128C. They look like totally different headphones.

I can post more as I make more comparison graphs like the above (I wish I thought to do this before averaging the reps because it takes about a half hour of straight clicking to make these) but from comparing other measurements, I’m finding myself less compelled to continue posting EARS measurements because they’re off enough that I feel like I’m misleading people a bit by posting them, and incorrect data seems worse to me than no data at all. I’m hoping miniDSP refines the EARS in the future, but right now it seems to be a bit “too good to be true.” I’ll still be using my EARS to take comparative measurements though in case we do reach some sort of breakthrough. It makes a cool mic too.

But now that I can post a full suite of measurements, I’ll be writing a lot more when I have some spare time. I have like, at least 7 reviews that are mostly done, but I haven’t had the time to compile all the data and analyze it properly. After finals season early next month I’ll be chugging away though.


#16

This jibes with what I’ve found when using EARS measurements for EQ. If I try to make the response “flat” in the treble region, it never sounds quite right. For bass and mid-range, it’s been helpful though.

I would imagine that any simulated ear is going to be different from my own ear. The expensive ones might come closer, buy they’re really mostly useful for comparing and at the end of the day there’s no substitute for listening with your own ears.


#17

Yep. There is definitely something weird with EARS after 4KHz that I’ve noticed. It’s much more prevalent in headphones than IEMs for me though. For IEMs, my biggest issue is getting them to fit properly in the odd hole size, so bass response can vary.


#18

I have the same issue. I’ve noticed that you Medienrente reflect a lot more high treble than my IEM measurements, which to my ear are vastly underreporting the actual treble level.


#19

Sometimes I think the EARS would have been a better product if they just made it a flat plate coupler instead of trying to simulate an ear but not doing a great job of it.


#20

At some point, I would like to re-make my calibration files from extrapolated “good” data and making the proper correlation/adjustments. I think @Ishcabible is already working towards that. I saw on SBAF that purr1n was also on the right track too with compensation curves: