Some Test Results (SE vs Balanced vs Portable SE)

[Dreadhead]

Well I just got done doing tests on the HD650 using my ECM8000 measurement mic and dummy head and room EQ Wizard (that some of you have seen). I kept the mic and headphones in the same spot and tried the following:

1. HD 650 running Balanced from the GS-X (of DAC3 Balanced)
   
2. HD 650 running SE from the GS-X (of DAC3 Balanced)
   
3. HD 650 running SE from AE-2 plugged in with wall wart(off DAC3 SE)
   
4. HD 650 running SE from AE-2 running off battery (off DAC3 SE)
   

I attempted to volume match all 4 configurations but ended up having about .3 db (at 1kHz) variation that I could take out using the software so they all have the same value at 1kHz.

Attached are the frequency response curves for the four configurations. The second one has a vertical separation introduced so you can see that there are different curves.

Amazingly Identical. The most difference I can find is on the order of less than 1db.

[luvdunhill]

Well I just got done doing tests on the HD650 using my ECM8000 measurement mic and dummy head and room EQ Wizard (that some of you have seen). I kept the mic and headphones in the same spot and tried the following:

1. HD 650 running Balanced from the GS-X (of DAC3 Balanced)
   
2. HD 650 running SE from the GS-X (of DAC3 Balanced)
   
3. HD 650 running SE from AE-2 plugged in with wall wart(off DAC3 SE)
   
4. HD 650 running SE from AE-2 running off battery (off DAC3 SE)
   

I attempted to volume match all 4 configurations but ended up having about .3 db (at 1kHz) variation that I could take out using the software so they all have the same value at 1kHz.

Attached are the frequency response curves for the four configurations. The second one has a vertical separation introduced so you can see that there are different curves.

Amazingly Identical. The most difference I can find is on the order of less than 1db.

you should post the impedance curves. these mean a lot more than the frequency response does, as the frequency response and phase data can be recovered via a Hilbert transform. This is the beauty of the Thiel-Small work.

[recstar24]

Awesome. Thanks for sharing. I do not have much experience with balanced drive, other than 15 minute auditions at meets. The past chicago meet I was in love with the HP1000 APS balanced from the TTVJ 307a. Probably the best I have ever heard them. I would attribute that more to the 307a and the APC cabling than the balanced part.

[Dreadhead]

Here are the THD results for a 997.6 signal for cases 1,2,3 in that order.

the two SE cases it's 0.347% and for case 1 it's .351%

[Dusty Chalk]

Dummy head? You mean you bought or rented a binaural mic of the dummy head variety?

[Dreadhead]

you should post the impedance curves. these mean a lot more than the frequency response does, as the frequency response and phase data can be recovered via a Hilbert transform. This is the beauty of the Thiel-Small work.

Unfortunately not doing this with an AP system so don't have those curves. I do have the impulse responses and I'm plotting those as we speak:

The first attached is the Impulse response for all 3 versions where the phones are being driven SE.

The last if for all 4 and you can see that the response is slightly different for the Balanced line (the line that's been added).

[Dreadhead]

Dummy head? You mean you bought or rented a binaural mic of the dummy head variety?

Nope. It's a block of modeling foam with an omnidirectional measurement mic shoved through it. I would love to get a true binaural head but I've not got the money and to be honest it would be of limited use to me. In the end whatever problems this would have should show in all the measurements so the comparison is still valid.

Oh and all measurements are the average of 4 sweeps.

[Dreadhead]

Unfortunately not doing this with an AP system so don't have those curves. I do have the impulse responses and I'm plotting those as we speak:

The first attached is the Impulse response for all 3 versions where the phones are being driven SE.

The last if for all 4 and you can see that the response is slightly different for the Balanced line (the line that's been added).

I should add the differences are at such low amplitudes I can't say they are trustworthy though they are repeatable.

[Dusty Chalk]

Well, no the head portion of the binaural mic is modeled to have similar acoustic properties as the human head. Foam does not (well, not most of us, anyway ). Maybe Billy.

[luvdunhill]

you don't need something like AP. Just download Speaker Workshop and use the play back / recording features built into that. You may need a simple cable with some known impedance, but it should work very well.

[grawk]

I've never thought frequency response would be impacted by going balanced. Control, slew rate and power, which are hard to show in a static graph, are what I thought was improved by going balanced.

[Dreadhead]

you don't need something like AP. Just download Speaker Workshop and use the play back / recording features built into that. You may need a simple cable with some known impedance, but it should work very well.

To be honest you're missing the point. I am not showing the results for the headphone I'm showing that the headphones produce the same sound out the other end with all these different combinations of amps.

I will look into the speaker workshop but to be honest I don't see how the various amps would be seeing a different impedance if the response was the same.

[grawk]

It only misses the point if the only thing impacting how headphones sound is frequency response.

[Dreadhead]

Well, no the head portion of the binaural mic is modeled to have similar acoustic properties as the human head. Foam does not (well, not most of us, anyway ). Maybe Billy.

Yeah I know that. It's not the point though since I can't see how it being a head would make any difference to the comparison. The curves would be different but they would still lie on top of each other unless you think the head has an effect on the driver.

I'm not making binaural recordings I'm using it to calibrate to flat at the entrance of the ear not to the ear drum. It's essentially like a coupler that a lot of people use for the same purpose.

[Dreadhead]

It only misses the point if the only thing impacting how headphones sound is frequency response.

Grawk, Actually I plotted the IRF which is both frequency and phase response and that's all sound is. They are identical.

From my understanding control is the ability to deal with the resonances of the phones and that appears to be the same or they would show up in the frequency respons. The IRF is not a static graph by any stretch of the imagination.

I'll be honest and say I was very very suprised by these results and I'm trying to figure out another test but I can't see one right now.

[grawk]

So you'll be selling the GS-X in favor of the AE2? How good a deal will you give me?

[Torpedo]

Very interesting Dreadhead. To support your findings, I paste here the measurements we got using a calibrated Earthworks cardioid placed into a dummy ear, driving the signal into an Apogee Duet and using a Macbook Pro with some software which I don't know. It's a friend's system.

The signal was fed into a Rudistor RPX-100 (the only balanced amp I had at that moment) driving the HD600 in Single-ended (black trace) and in balanced (clear brown trace). The main difference is a bit more than 1dB in the 1200-2500 Hz area, and also in the harmonic distortion levels (second harmonic is the top pair of traces at 40-50Hz, third harmonic is the lower pair) specially the second harmonic in the whole treble, which seems lower using balanced mode.

[Dreadhead]

Very interesting Dreadhead. To support your findings, I paste here the measurements we got using a calibrated Earthworks cardioid placed into a dummy ear...... snip

Torpedo,

Thanks for the measurements. I'll have to look around more. The software I have computes the the harmonic distortions as well but I didn't want to type all of them in. I only checked at the 500 Hz that Stereophile usually uses and at 1000 but I can do it at any frequency.

It's just not as organized as your friends and I can't get much above around 11-12kHz because the harmonics are above my allowed maximum of 24kHz (the software only supports 48kHz sampling).

If lower THD is what you get that sounds good to me I'll take a look tonight.

[grawk]

The trick is you're testing with tones, instead of with music. Music is affected by how the speaker transitions from one frequency to another. You need to pick a challenging piece of music, and plot it over time to get the whole picture.

[Torpedo]

He also measured impulse response and energy storage-liberation, but those measurements were virtually identical in both modes.

The program he uses is not RTA, so it takes some time to process the data, but the whole procedure seems quite "kosher". As a matter of fact I rely more his measurements than Headroom's, not only for the quality of the microphone, which costs about 3000 euros for its extreme frequency accuracy and balance, but also for not using unknown ponderation dBs, but just the ole good SPL ones.

Let's see if your findings on harmonic distortion match ours. Distortion products are frequently overlooked, but they can explain a few things we can hear, just have a look at this graph from the K701:

That 2nd harmonic peak at 400-600Hz...

[Dreadhead]

The trick is you're testing with tones, instead of with music. Music is affected by how the speaker transitions from one frequency to another. You need to pick a challenging piece of music, and plot it over time to get the whole picture.

An IRF is not a test with tones it's a impulse (which is very wide banded in frequency space) and tests the speakers (or rooms or panels or ships) response over a range of frequencies at the same time and the transition between these frequencies. The only thing the IRF doesn't really capture is the THD because that's a nonlinear effect in general.

[Dreadhead]

He also measured impulse response and energy storage-liberation, but those measurements were virtually identical in both modes......

Yeah that's my finding too. The software I'm using is Room EQ Wizard (freeware) and it does everything using a sweep and impulse. Using 4 sweeps and impulses etc it takes over a minute to generate one curve. The THD calculation is done separately but within the same software.

I'm using a Edirol FA-66 and a Behringer ECM8000 measurement mic with and have the calibration curve for the mic as done by a lab and posted on the REQ website.

Are you sure you didn't move your headphones at all in the balanced/unbalanced comparison, I thought I saw those curves before and that you had switched cables at the headphone. I used an adapter specifically so that I could avoid touching the headphones at all, because my experience is that if I move the headphones a little bit it makes a sometimes order 1dB changes to the response I measure out the other end.

[grawk]

There's more to music than how an "impulse" is reproduced, tho.

[Torpedo]

Yep, I posted those graphs at HF too some time ago. We didn't have an adapter, so we needed to swap the wires at the headphone's end. We put much care to place the phones in the very same position, if you look at the curves, they are very well matched being the only significant difference the 1-2KHz dip in SE mode, so we assumed we didn't a bad job. Otherwise there had been other greater differences at other frequencies too.

Your Behringer mic is widely used for this type of measurements, however its quality isn't comparable to the Earthworks' M30 my friend uses for his field measurements. BTW it's an omni, not a cardioid as I stated previously. I think the Behringer would take less accurate distortion measurements for its own distortion products being added to the signal the computer analyzes.

Grawk, measuring using a musical signal would be too complex to be practical and meaningful, that's the main problem with it. If you apply averaging and statistics, you spoil the "real time" part of the signal, and if you try to measure in real time, the signal is too complex to let you know for sure if you're comparing the original signal to the measured one or you're also measuring spurious signals added by the environment, the own cans or measuring devices... However I agree that these measurements don't say everything we can hear, that's sure. It isn't a bad starting point though. In any case I rather trust my ears than a measurements set. I mean that I can hear the HD600 don't sound the same in SE than balanced and were measurements all equal, I'd still trust my ears since they'd be probably finding something that we still haven't learned to measure.

[Dusty Chalk]

Yeah I know that. It's not the point though since I can't see how it being a head would make any difference to the comparison. The curves would be different but they would still lie on top of each other unless you think the head has an effect on the driver.

I'm not making binaural recordings I'm using it to calibrate to flat at the entrance of the ear not to the ear drum. It's essentially like a coupler that a lot of people use for the same purpose.

You're right, I was getting distracted into binaural recording territory by the fact that you were using a head. You were just using it for convenience for measuring headphones, I see that now. No, I don't think it has any impact whatsoever.