Floobydust

Materials and Methods

Sixteen bit (0 db reference) stereo sinusoidal .WAV files were digitally created via software at frequencies ranging from 20 Hz to 10,000 Hz in a 1,2,5 pattern, except in the bass region where the files were every 20 Hertz. These files were used to create a test CD which included CDTEXT with the track names. These were also transferred to a 4th generation iPod in Apple Lossless (ALE) format. A digital sound pressure level meter was positioned on a tripod on the center console such that the microphone was at approximately ear level and pointed forward; this was not moved for the duration of the testing (Figure 1). The equalization (bass, treble) settings on the Harmon-Kardon (Harmon-Kardon) system were set to flat and the iPod EQ was turned off for all tests. Data was taken with the test CD in the in-dash player (CD), from the iPod interfaced through the MB OEM iPod adapter, and from the iPod interfaced through the Dension iPod adapter.

A complete set of measurements were taken with the sound program settings in the "Normal" mode and in the "Logic 7" mode. The Harmon-Kardon system volume was adjusted for each test so that the SPL (Sound Pressure Level) was 85 db at 1 KHz. All measurements were taken with "C" weighting in the "slow" response mode. I decided to conduct all tests using a constant reference level (85 db at KHz) instead of at constant system gain because I think the constant level better represents how the system would be adjusted in real use (we just turn it up or down to a comfortable listening level). This is important to note because of the effects of Fletcher-Munson loudness compensation as discussed below. 85 db was chosen because it is it will not overload (e.g. cause clipping) the amplifiers and is within the range for use with "C" weighting, the weighting that gives the flattest measurements (at least with my equipment). The test vehicle was my 2008 ML550 with the latest Command and audio gateway firmware.


Figure 1




Results & Discussion

The measurement results with the system in "Logic-7" and "Normal" modes are shown in Figures 2 and 3 respectively. Several things are evident in both plots.



Figure 2



Figure 3


First, the Harmon-Kardon system is set up to provide Fletch-Munson or "Equal-Loudness" compensation equalization. The human ear does not respond as well to low and high frequencies at low sound pressure levels, so most (and certainly all I have measured) OEM audio systems boost the low and high output to make the system appear to sound flat at low listening levels. More sophisticated systems provide variable boost whereby the amount of boost decreases as the volume level is increased and it appears that the Harmon-Kardon system is set up this way. Note that the plot for the Dension shows slightly more "boost" in the low/high regions where compared with the In-dash CD. A look a raw data shows that the Dension supplies a signal at a slightly higher level than the CD, thus the system gain needed to below lowered slightly to achieve the reference 85 db at 1 KHz; the lower "volume" setting (e.g., fewer clicks on the volume knob), resulted in more loudness compensation being added, thus boosting the high and low response in the plots. Unfortunately for the OEM iPod, the opposite effect occurs due to the vastly lower input level through that system.

Audio purists may pooh-pooh the amount of loudness compensation in the Harmon-Kardon system as it is significantly more than what either Fletcher-Munson or ISO-226 would suggest. However, there is an additional effect that needs to be taken into account - psychoacoustic masking and it is a significant issue in mobile audio systems. Psychoacoustic masking is the phenomenon where the ear fails to perceive some sounds when two sounds occur near the same frequency. This phenomenon is exploited in .MP3 file compression to allow them to create smaller files by throwing away the sounds that you (supposedly) can't hear. I would suggest that the additional equalization in the Harmon-Kardon system would be beneficial in compensating for road noise (both high frequency like wind noise, and low frequency like drive train noise), although the crossover points chosen by Mercedes-Benz and the limited low end tend to make the system sound "boomy", especially when coupled with the digital bass filtering described below.

The effects of the Fletcher-Munson compensation may also be exaggerated because the test files were referenced to 0 db. Because it is the maximum digital value, this level represents the peak level of any musical source, with the average level of a typical musical source being much lower. A lower average level would result in a higher system gain in typical usage, thus lower a lower amount of Fletcher-Munson equalization. I would have preferred to use something like a -18 db reference file level, but I thought that this might put the OEM iPod output below that level that is measurable with my equipment. Additionally, the whole point of this exorcise (pun intended) was to compare the various in put sources, so the loudness compensation is really just a side note.

The equalization seems to be more well controlled in the Logic-7 mode, although the sharpness of the adjustments show that DSP (Digital Signal Processing) is being used as would be expected. The Harmon-Kardon system uses the Texas Instruments TMS320DA610 DSP chip, and while this is a capable processor, it is not as powerful as many home based DSP units. The Normal mode is a classic analog response curve, even though it may be digital in origin.

Both Logic-7 and Normal show what looks like a high frequency roll off, however, I don't believe this is occurring to the extent you might think from the plots. I think it is an artifact and a result of the fact that that the SPL meter has a relatively directional microphone. At high frequencies, the wavelengths are only a few centimeters and very directional ("beaming"). This can result in acoustic interference patterns that can deceive a fixed, directional microphone like the one used. It is also the reason I only tested the system to 10 KHz.

Mercedes-Benz also has a microphone that measures the background noise and attempts to adjust the system volume to compensate. Since the psychoacoustic masking phenomenon is subjective and differs for various people, these systems can't be accurate for everyone. If fact, I find them incredibly annoying. The fact that the Mercedes-Benz system is neither adjustable nor defeatable (without disconnecting wires), like every other OEM system which has this feature, is a huge annoyance and clearly shows what a low priority audio is with Mercedes-Benz. I asked a friend with Mercedes-Benz about this and he said that their marketing studies showed that the majority of their customers were old and technologically unsophisticated. They are buying the cars for the three pointed star versus what they can do. Because of this it is more of a priority within Mercedes-Benz to make things "idiot proof" (to avoid complaints) than perform well.

Second, it is also clear that the Harmon-Kardon has a digitally limited low bass response. This is actually my biggest compliant about the Harmon-Kardon system (followed closely by the OEM iPod abomination) and one which I identified immediately "by ear" (and posted about) with my first ML. The Harmon-Kardon system uses a 6.5" 2 ohm dual voice coil driver (a $10 Chinese made unit) in a relatively low-tech sealed box enclosure. A sealed box has a natural roll off of 12 db/ octave below its design operating range. These measurements show a roll of more in the 35 db/octave range, indicating that a high order (and of course, digital) filter is limiting the very low end. This is a technique that is commonly used to extract as much performance as possible from cheap or undersize drivers. In theory, if a driver could handle an infinite amount of power and had an infinite amount of cone excursion, just adding more amplifier power would allow even the smallest speaker to achieve significant low end response. Of course this can't be, so manufactures push power into the driver up to the point where one or both of these limits is reached and then limit response beyond that with either compression limiters or digital filters (or both). This generally results in a extension of the low end response of an octave or more, but at the expense of virtually no output below the cutoff frequency. In the case of the ML there was absolutely no reason to do this other than cost cutting; there is plenty of room for a larger sub.

While bass limiting does make very effective use of the ten dollar Chinese woofer, it causes the low end to sound weak and contrived to me. It also unnaturally kills and low end "power" notes like you would find in a good Steely Dan recording. It is also problematic for those who wish to correct this deficiency by adding a new sub woofer since it will require significant electronic equalization/ amplification. Even a very large sub woofer would still be missing these low end notes. It is also not clear whether the resulting modifications would even sound good since the missing signal information and the attempts to restore it with counter-equalization could result in issues with transient response and phasing.

As a side note, I have an SL500 which is equipped with the Bose system. While the absolute bass performance parameters for this system are probably not much different that those for the Harmon-Kardon system, the roll-off characteristics on the Bose are better engineered than those in the Harmon-Kardon. The Bose bass response just "sounds" better in actual listening. If there is one thing Bose understands, it's listener perception and bass reproduction. My wife's Accura uses a proper 10" sub and thus has real low end response (albeit that system has other issues).

Lastly, and the primary reason for this comparison, you can see that the Dension very closely overlays the CD-changer in terms of response. I think that the slight difference in response at the high and low end is due to the variable Fletcher-Munson equalization and the slightly different feed levels. In actual listening, the difference is hard to discern.

The OEM iPod interface, on the other hand, clearly has a different response profile. The low end response cuts off an additional octave higher the either the CD or the Dension. If this were the very bottom octave, it wouldn't be as much of a problem, but the lack of response in this range cuts directly into the most common bass ranges for most popular music. Even if you could "boost" in the input level and cause more Fletcher-Munson compensation, the bass response would still be missing the extra octave. The net result, is that musical low end just sounds bad through the OEM interface.

The high end response is also problematic for the OEM adapter. It suffers from a larger high frequency roll-off than either the CD or the Dension. The net result of lack of low and high end response is a harsh, tinny sound that is un-listenable for me.

As I have posted before, the analog "AUX" input, which the OEM adapter utilizes, is deliberately limited. I have been told that Mercedes-Benz chose to limit both the input level and the low end response in order limit potential warranty claims associated with high level signal inputs at the "AUX" input. This design goes back to the first introduction of the "AUX input with the first H-K systems. Many manufacturers also limit so called third party inputs, but they seem to do so in a much more intelligent and acoustically friendly manner.

Unfortunately it does not look possible to completely overcome these limitations as the analog circuits present in the Command unit will overload (clip) before the fiscally inspired audio limitations can be eliminated. I have tried this with both the infamous 11 cent Peiker in-line amplifier and a circuit of my own construction. While amplification/equalization can help, it can't restore the full and proper audio dynamic range for anything routed through this input.

One final, non audio note is that while I was testing the OEM adapter I noticed that it presents song names in alphabetical order like a computer would (e.g., based on the ASCII codes for letters) rather the like a dictionary would and like the actual iPod does. The Dension presents song names exactly as the iPod does. I also noted that the steering wheel controls - your only source of control with the OEM adapter - are backwards from the CD. Good work Mercedes-Benz!


Conclusions

While the Harmon-Kardon system appears to be "value engineered" with financial targets being more important that acoustic ones, overall, with the exception of the OEM iPod interface, it does seem to extract the most out of what is there. The small sub woofer and the electronic low bass limitation limits the audio performance of all parts of the system. This also makes aftermarket tuning difficult.

The Dension provides audio performance which is nearly identical to that of the in-dash CD whereas the OEM interface clearly measures as acoustically inferior to both the in-dash CD and the Dension. Furthermore, there appears to be no easy way to "fix" the audio performance of the OEM unit.

The Dension provides a usability experience which can be actually better than that of the in-dash CD or changer. Most CD's don't have CDTEXT (Sony issued CDs are a notable exception), but iTunes provides title and artist information for songs that are transferred to the iPod and, with the Dension, subsequently displayed on the Command screen, thus giving the user a level of usability that exceeds even that of the typically CD.

The user experience of the OEM interface via the CAN-BUS only is dubious at best and reeks of "engineering afterthought". I think Mercedes-Benz would have been better off forgoing any such interface at all rather than foist such an a technically inferior system into the marketplace. In my mind, this is the kind of thing that tarnishes Mercedes-Benz's sterling reputation for engineering.

In short, save your money on the OEM iPod option, and enjoy the music!

Sorry if this post is a bit long, but I just wanted to get my thoughts out there.

- FD

baxterdown

FLOOBY, Thanks SO MUCH for the eloquent explanation!!! Tons of information to digest... -- I would love to take your exorcise a step further and compare the Denison/iPod/CD on the ML vs. the same setup on another MB (an E-Class for example). I had posted somewhere else that IMO the sound in our E is so much better than that in our ML.

ToonArmy

Wonderfully detailed technical article, thank you very much. Once upon a time, consumer/hobbyist magazines used to publish articles like this, but I almost never see anything like this any more.

You dropped a couple tidbits about the Acura system (good subwoofer, but other problems) in there. If you have the time, it would be fantastic if you could share more insights on the quality of the stereo systems of the various manufacturers.

Oh, and one minor correction - I believe it is HarmAn/Kardon rather than HarmOn/Kardon.

Skylaw

Floobydust, thanks - great post, intelligently written and with "hard" documentation.

It also tracks what I hear by ear with my '05 S500 Bose system, with respect to the differing inputs (although no doubt the Bose response curves would be different from the HK system, the quality of the Gateway 500/iPod is at least as good as the built-in CD player or the changer, and superior to MB iPod kits, when playing the same formats) - although I listened to the MB iPod kit in another S500, not my own.

Not to diverge too far, I can definitely hear the difference between mp3 and .AIFF. I burn .AIFF or Apple Lossless on the iPod, avoiding mp3 wherever I can. The iPod recorded in either of those formats is far superior to the mp3 DVDs I burned (with the same music as I have on the iPod), or to mp3 on the iPod.

Of course, a great music track on an original DVD sounds best of all, if the car's system has that capability.

Your points are well taken; if you want the best iPod audio, use higher quality formats when burning, and use a Gateway 500 in preference to the OE MB kit (its fiber optic input avoids the AUX input limitations). I also find its menus easy to use, once you get past a slight learning curve (that could be shortened by better documentation from Dension).

Mine is opinion, but nothing beats your "hard" documentation.

jbondox

what type of meter is that, and when was it calibrated and to what weight? what vehicle was it, and why didn't you add the nav-tv ipod interface to it also

most likely the audio on the GW and the cd are the same because they are both being transmitted optically, where the aux audio is just that an aux audio that has to be deciphered in the command then shipped to the amp via the fiber optics.

I also do not think it has anything to do with fletcher Munson curves either, I believe it is more of a compensation for speaker position. a way to tell that is to take the same reading at different volume levels and see what changes

Did you use pink noise at all for your measurement?

The curves regardless of the source are still crappy.

****, in all actuality you definitely need an RTA vs an SPL meter

UK-C200

Quick question - why would the Denison kit be different from the MB OEM based on your supposition above? (GW vs CD). At the end of the day you have a digital to analogue conversion talking place. In Denison case, you actually have three, right? Ipod converts to analogue, whatever interface converts to digital, this is piped to the HU. In reality, the AUX input is one conversion less, right? Analogue from iPod to HU?

Also - would calibration and weighting of the meter really matter that much - other than giving a better ability for others to reproduce the results - as long as it was not re-calibrated or the weighting changed during the tests?

I would agree that having access to a good rta kit would allow more accurate results. In reality, however, you would probably just end up with more accurate curves - I really doubt you would find a 10db track difference between a SPL meter and an RTA setup.

Anywho, my ears compare with the results above - the DICE kit I have sounds better than the MB OEM kit, and I can't hear the difference between my DICE and my friends Gateway. I do differ from Skylaw in that I can't really hear a difference between VBR mp3's and aiff's, so maybe my ears are shot...

Maybe time for some coathanger tests :P

jbondox

oh god the coat hangers.... the only thing I thought they were good for was the weekend abor... nevermind

jimmydagreek

Wow, great info. thanks! Now I wonder if this will work in my 03 E500 with an 07 Command unit in there and if I can install in myself...

Alexander_B86

So lossless is supported? By both Benz and Gateway adapter? Does song names and artist show up when using lossless with either adapter? How hard is the install for an 05 S? Thanks!

Skylaw

As answered in the Audio & Electronics forum:

Both the MB iPod kit and the Dension Gateway 500 can play back lossless. They can play any music file that your iPod can, because the iPod is what actually does the recognition of the file.

The limitations you have read about apply to DVD and mp3 CD files.

I have no experience with the MB iPod kit. My '05 (and likely yours too) has no AUX input jack at all - something that is required for the MB kit that the Gateway 500 does not. No jack, no MB iPod kit.

With the MB iPod kit, song titles, etc. are displayed on the instrument cluster display (the MFD) and not on the COMAND screen. With the Gateway 500, song titles, as well as sorts by album, artist, playlists or song appear on the COMAND display; you get track number on the MFD.

See https://mbworld.org/forums/sho...d.php?t=194660 and https://mbworld.org/forums/sho...51#post2471051 for what I think is a pretty good rundown.

You will find the MB Bluetooth integrates better than the Gateway 500 optional BT kit. See https://mbworld.org/forums/sho...77#post2788377 especially posts 1, 5, and 6.

praveen

Anyone used a GW500 with audio20(most)? Really need to know if there are any bugs before i buy. I can only find Comand users. Thanks

praveen

Installed the GW500 and dension BT tel kit on my W211 e class audio 20 (no CDC).
The result was perfect. Very high quality ipod audio and basic tel hand free functions for a fraction of original MB price.
Ive posted more info on another thread.
Thanks Skylaw, GW500 and Flooby for your extensive research and encouraging the GW500 install. Am very happy with it.