Velodyne Digital Drive Plus 18 (DD18+) Subwoofer Measurements and Analysis
The Velodyne DD18+ subwoofer was measured outdoors sitting on the ground with the microphone placed 2 meters from the front lip of the cabinet with the grill removed. The driver was facing directly at the microphone. The Jazz preset was used for the tests and configured with the low pass filter disabled, the rumble filter set to its minimum setting of 15Hz and 6dB / octave roll off to allow for maximum response extension and flatness, all EQ was defeated and the subwoofer volume was set to 35. All tests were conducted in this configuration, except for those tests purposely conducted to examine the effects of the built in functions or different operational modes.
Also note that two full test sets were conducted on the DD18+. One with the servo gain setting at maximum of 8 and the other with the servo gain setting at a minimum of 1. All other settings remained the same between the two.
The overall approach to this testing along with the equipment and software used is outlined in the article here.
Velodyne DD18+: Effect of Low and High Pass Filter Settings
Above is the effect of various setting of the low pass and rumble filters on the DD18+’s response. There is quite a large amount of adjustment available with just these two settings.
Velodyne DD18+: Effect of Movie EQ Preset
Above is the basic response of the preset EQ curves available from the factory. Most of them appear to add an overall signal boost. The theater setting adds a slight peak near 35Hz which seems to have a higher Q than one would expect. The Games preset adds a little boost to the 60Hz region.
Velodyne DD18+: Basic Frequency Response as Tested
In the chart above is the basic response of the DD18+ at the two different servo gain settings. The red trace is with the servo gain at its lowest setting of 1. The black trace is with the servo at its highest setting of 8. As can be seen the maximum setting of 8 produces a flatter more extended response while the lowest setting of 1 produces a more rolled off and humped up response. The servo seems to be forcing the unit into a flatter response shape at the highest setting whereas the lowest setting is probably closer to the actual driver and enclosure response combination. Velodyne specifies the DD18+ as 14.4-120Hz +/-3dB and 8.8-300Hz overall with no tolerance given. With the servo gain setting at 1 the DD18+ only fits within a 6dB window from 14.6-70Hz. If the window is expanded to 10dB total the response fits in from 13-109Hz. With the servo gain at its maximum setting the DD18+ response fits within a 6db total window from 13.5-126Hz which is slightly better than Velodyne’s specification. Within a 10dB window the response is 12-150Hz. I suspect that this configuration is what Velodyne specs their response from. By 300Hz the response is about 20dB down and probably unusable regardless of the setting.
Velodyne DD18+: Internal EQ Results
In the above graph we have the results of equalizing the DD18+ in room. The black trace is the raw response in room as measured by my reference Earthworks M30 which as you can see is not very good to begin with and requires some taming being +/-10dB from 14-150Hz. The DD18+ was run through each of its equalization schemes to check the effectiveness and results with the Velodyne microphone in the same exact position as the M30 with the elements about ¾” apart. The result for the Self EQ scheme, which is the blue trace in the graph above, was very strange resulting a dramatic boosting of the upper range and an actual worsening of response to +/-12dB 14-150Hz. Next the Auto EQ routine was tried. This had better results than the Self EQ option but still wasn’t very effective. The worst response issue the 12db peak at 45Hz seemed to be ignored and some smaller cuts were made at 20Hz and 40hz. The resulting response was barely improved and fit within a +/-9.9dB window from 14-150Hz. Next the manual EQ option was attempted. This method allows you to manually adjust the EQ settings while the Velodyne on screen display shows the result on the response. This method had the best result and is represented in the pink trace on the graph. The response was improved to +/-7.5dB 14-150Hz using the manual EQ method. Finally I decided to see what equalizing with my regular unit a Behringer DCX2496 would do. This is the red trace in the graph above and is +/-3.7dB from 14-150Hz.
During the process of using the on board equalizer and display of the DD18+ in conjunction with my regular measurement system I noted that the frequency response plot shown inside of the Velodyne software looked much better than what my manual measurements did and also more curiously that some of the features appear to be shifted. I have provided an example below.
Velodyne DD18+: Software Response Shown Versus Actual Measurement
The above 2 charts show the measured response of the DD18+ with some manual equalization applied. The top graph in the bottom picture is frequency response as reported inside of the Velodyne software. It is showing a response variation of +/-2.5dB, or a 5dB total window from 15-200Hz. Any audio fanatic would be proud to have that response. The picture above the Velodyne chart is what my reference measurement system reported the DD18+ response as. I have scaled the 2 graphs as closely as possible for comparison. The black line is the raw measurement from the M30 with no smoothing and the pink trace is the same measurement with 1/3rd octave smoothing. Looks rather different eh? The black trace is +/-13.4dB from 15-150Hz. Even with 1/3rd octave smoothing the pink trace is still +/-6.4dB over that range. I had thought that adding the smoothing would better reconcile the two but it still doesn’t match up that well. Even more curious is the fact that many of the main features from the two measurements do not seem to match up. The Velodyne chart indicates a dip at 45Hz instead of the obvious peak. Again at 30Hz there is a bulge in output instead of a dip. Down below 20Hz the Velodyne software indicates a rising response, but my personal rig indicates a large drop off. Very odd indeed. I’m not really sure what to make of this. It almost seems as if the response as reported on the graph inside of the Velodyne software is offset ½ of an octave. Again both microphones were in the same position within ½” of each other. Certainly there can be differences attributable to the different test signals used and the way the data is presented but these seen here are rather large. If you do use the internal EQ of the DD+ series the manual method is by far the best option in my estimation.
Velodyne DD18+: Waterfall Decay
Velodyne DD18+: Group Delay
The response of the DD18+ in the time domain is rather uneventful and about as expected. There is clean decay across the full range until about 16Hz where there is a moderate increase due to the boost EQ being employed to extend the response and the corner of the high pass filter. Still it never breaks 1 cycle of delay even below 20Hz. There is negligible difference in this performance metric whether the servo gain is set to minimum or maximum.
Velodyne DD18+ Servo Gain 8: Long Term Output Compression
Velodyne DD18+ Servo Gain 1: Long Term Output Compression
The maximum long term output compression test for the DD18+ with the servo gain set to minimum and maximum both are presented above. Overall despite the response shape differences there is not much difference. Compression and output limiting kicks in first at about 16Hz where the response is being boosted the most and expands progressively from there as the output is increased and the subwoofer reaches the maximum output level that it can produce as configured. The output does not start compressing notably until the 105db sweep level. By the 115dB sweep level the deep bass is being limited heavily everywhere below 30Hz.
This is by far the most demanding test conducted on the subwoofers during testing and will reveal any issues with overload, port compression, port noise, driver distress, creaks, rattles, buzzes, etc. Additionally this is outdoors with just the subwoofer operating so there are no nearby walls or objects to vibrate and no upper frequency masking content from other speakers that would cover up or mask any noises from the sub. Any sort of audible issues with the subwoofer will be readily apparent. The most offensive noise that the DD18+ produced was a very slight warming to the tone in the deepest bass indicating increased distortion from high excursion of the driver. In room with other speakers operating and actual content this would probably never be noticed. That is it. Most subwoofers do start making obvious bad noise at some point. This subwoofer is as stable and composed as I have seen. It simply quits getting louder before it ever makes objectionable noise.
Long time, no speak.
By way of comparison, my own test rig is capable of 1.943V RMS in -10dBV trim and 7.62V RMS in +4dBU trim via the Lynx 2B sound card. Havng tested more Velodyne subs than most, and using them in my own home system, I am all too aware of the potential for overloading them inadvertently.
The gain structure on the Velodyne DD subs is rather high, particularly when driven via the XLR inputs, and the overload threshold is not very generous. By my own measurements, a form of clipping sets in around 1.5V RMS. The odd thing is this distortion starts to creep in below 20Hz to start with. As you increase the drive level by a few dB, the distortion creeps up the frequency scale maintaining a very definite descending slope with increasing frequency. Which suggests to me this is not simple clipping but possibly some quirk in the DSP code at high amplitudes. I have a question about this open with Chris Hagen at Velodyne currently.
The upshot of this is that I am acutely aware of the potential for overload when testing these subs. The sloping nature of the distortion as it starts to appear makes it less easy to spot in swept sine tests as it simply becomes a contributing factor in the background. Spotting it with tone-burst testing would be even harder as it will typically just appear as rather high levels of odd-order distortion. Something for the reviewer/tester to beware of is all I am saying.
However, this issue can also impact on the general user. By their nature these subs will tend to end up in high-end systems, often partnered by equipment that generates signal levels more typical in the professional arena. For example, my own Theta Casablanca DACs are specified to deliver up to 20V RMS via XLR. If you connect a DD sub direct to this and match levels, you typically end up with the DD volume set to something like 3 to compensate for the very high signal level from the processor. In this scenario, it then becomes rather easy to clip the subwoofer on loud material, which is a pity after spending all that money! A higher overload margin and lower gain on these subs would be a good idea, particularly via the XLR inputs.
My advice to anyone using these subs in such a high end system would be to ensure you don't set the DD volume lower than 15 and match the speaker levels by dropping the level of the subwoofer output in the processor setup menus only. If you currently have the DD volume set in single figures, then this overload issue may be a problem for you that you may not even be aware of.
I've been meaning to update this thread for some time and just never got around to it.
To recap, for all Velodyne Digital Drive Plus sub owners, regardless of sub size, there are three sets of inputs on each sub. One set for speaker level inputs, one set for single-ended line level, and one set for XLR line level. The speaker-level inputs and the single-ended inputs have dedicated input level controls, so input stage overload is not a factor, but the XLR inputs do not have level controls.
For those of us driving the DD Plus subs with full-range balanced outputs from our pre-amps, input overload is a very real factor. As cjwhitehouse noted, many owners running XLR cables are using sub volume levels of "5" to "10", when Velodyne ships the sub with a default of "30". I called Velodyne and asked if there was an XLR input sensitivity fix, and they never responded.
Some time ago it occurred to me that the fix was simple - use XLR line level attenuators ahead of the XLR inputs on the sub. Fixed levels (-10db, -20db, etc.) are available from Parts Express for about $10 each, and Audio Technica makes some more expensive adjustable ones. I tried the -10db versions from Parts Express.
What a difference! I remember that first listening session with the attenuators in-line clearly, and it was several months ago. Prior to the attenuators I was running the DD18 Plus's volume control at "8", and now I use "25". So what? Well, two things changed. First, the granularity of volume adjustment is so much finer. Before, "7" was too little and "9" sounded bass-heavy. Now I can fine tune the bass level depending on the recording from "20" to "30" and the subtlety of effect is very nice. Second, the sub's EQ is more effective and appears to be more accurate when I do in-room measurements. I don't know for sure that input overload was causing the problem, but the attenuators made a significant, easily audible and measurable improvement in the sub's usability and EQ performance.
As I said, most users won't care because they'll probably use the single-ended inputs. But if you are using the XLR inputs on Digital Drive Plus subs with full-range pre-amp outputs I highly recommend trying XLR attenuators.
Please look at the distortion chart.
Isn't that too close to the amp when bolted/screwed on? If the amp is hot wouldn't the foam catch on fire? Or is it made out of fireproof material?
Ahhh gain structure...There is a surprising amount of final system performance tied to this and many are unaware of how much.
Maybe AH should write up a dummies guide to setting gain structure? I recall starting a thread about it over at AVS and never coming to a conclusion (at one point an expensive o-scope was required and then building some sort of attenuation circuit instead of the o-scope to use on my Mobile-Pre/REW setup to test for signal distortion as a VMM couldn't do this). Mind you, that was probably more involved as it included an outboard amp and DCX..
Just got too complicated. :o