Bossobass Mini GTG Thread

[Shredhead]

Bosso and I did the experiment using a single raptor module with Nick's HST-15 drivers.  We used the Hulk vs. Abomination punch in the face scene and inched the volume up until the excursion was 35mm (driver's xmax).  We measured the peak Voltage and peak current to the cabinet. 

 

Then we used a 10Hz sine wave and inched the power up until we reached the same 35mm and measured the peak Voltage and peak current.

 

Here is the spec graph of the content.  This is looped, not mic'd and the levels are not important.  I only include this for a visual reference of the source material.

 

10Hz SINE WAVE

135.4V  21.06A  =2851.5 peak Watts (1425.75W to each driver)

 

HULK PUNCH SCENE

168.77V  20.57A  =3471.6 peak Watts (1735.8W to each driver)

 

-Difference of 620.1W (or 310W to each driver)

It takes 21.7% more power to drive the HST-15 to xmax with this scene than with a 10Hz sine wave.

 

 

 

[tuxedocivic]

Interesting. My only question of rigor in this test is the way you measured excursion. You say you increased the volume until you had 35mm. Was this done visually? Any chance of some difference there? A few hundred watts per driver at those power levels is only a few mm of excursion difference,,, off the top of my head.

 

How does crest factor play into this like 3ll3dood is saying. I just read a couple articles on pro sound web, but not sure about the technical details.

[Shredhead]

We set up a static rig to measure excursion.  We could hear if the dust cap was hitting the mark (and see it but we would never only rely on visuals alone for an excursion test) and if so how hard.  It is pretty damned accurate I would say. 

 

I think we can ignore crest factor for this test.  Peak power causes peak xmax and the original idea was that it takes more power to push a driver to xmax with a spread of frequencies than only 1.  I don't mean to sound dismissive to tux or d00d, I guess you will just have to explain your ideas further for me to respond with more explanation. 

[tuxedocivic]

Cool.

 

Not dismissive sounding. My thoughts aren't well constructed at all.

 

I think your test shows that it does take more power, but it doesn't seem like an excessive amount of extra power. Quantifying how much extra it takes to power subs with wide bandwidth signals isn't clear to me, but this puts things in the ball park.

 

Thanks.

[Bossobass Dave]

Please note that this was one scene with one configuration of driver-to-Vb. The difference can vary quite a bit between the steady state signal+power and the dynamic soundtrack source+power. Some soundtrack scenes/sub systems can require 2-3dB more than a steady state test will show to compel the drivers to max excursion.

 

Of course, we selected Xmax as the target so as not to wreck perfectly good drivers for the sake of an experiment.

 

How much power it takes is not knowable because dynamic source is a moving target. But, props and much thanks to John and Abraham, we have a pretty good idea of a window vs having no data at all.

 

The bottom line is that balancing driver/box/amplifier and signal shaped curve involves more than looking at how many volts will push the driver to a safe excursion when driven by a steady state single sine sweep.

 

At the GTG, during playback of extremely low frequency content, Luke mentioned that the clip lights were flashing on the dual A-14K amps. I said, yes, they're doing what they're supposed to be doing, limiting the voltage to the drivers.

 

It took lots of testing and experimenting to get the gain structure, playback level maximums, mains power, Voltage Peak Limiters and signal shaping just right to extract maximum safe performance from the systems, especially when maniacs are in the house and extremely dynamic very wide bandwidth content is loaded into the player.

 

The amps weren't clipping, the voltage peak limiters were engaging at the preset points. Lesser amplifiers cause all sorts of problems that can be avoided by using adequate power. That was the point of the argument and the reason for the test.

 

Thanks to Paul for the measurements and math. There was no doubt in my mind, I just don't have the instruments or the acumen to use them to get the actual numbers.

[3ll3d00d]

We set up a static rig to measure excursion.  We could hear if the dust cap was hitting the mark (and see it but we would never only rely on visuals alone for an excursion test) and if so how hard.  It is pretty damned accurate I would say. 

 

I think we can ignore crest factor for this test.  Peak power causes peak xmax and the original idea was that it takes more power to push a driver to xmax with a spread of frequencies than only 1.  I don't mean to sound dismissive to tux or d00d, I guess you will just have to explain your ideas further for me to respond with more explanation. 

what I was thinking of is, I think, illustrated by the speclab capture you've used. I was thinking in terms of a broadband signal being made up of the long, slow low frequency waveform with a bunch of higher frequency waveforms added on top (thus leading to a combined waveform that places a more sustained load (lower crest factor) on the amp. I got back to my audio PC today and that seems like the wrong way to think about, at least with sine waves, and the effect on peak amplitude dominates. For example, I created a bunch of sine waves (20, 30, 50, 75, 95) to reflect that LMS example from earlier and then summed them. This graph shows the resulting wave form compared to the 20Hz sine wave

 

 

The crest factor of the summed form is ~8dB but the peak is ~11dB higher. Intuitively this surprised me but it makes sense once you actually look at the individual waveforms. 

 

Obviously this is an extreme example, +11dB is rather more than the +2dB shown in the Hulk case, but I would think the same principle applies. 

[Shredhead]

I tend to stay away from RMS figures when measuring stuff because I'm uninterested in averages and the average point varies too much from a sine to a real life source to be useful to me.  I also tend to stay away from posting peak to peak values because it confuses the folks who don't understand what that means.  I stick with the peak values either positive or negative.  Here is what the waveform from Hulk looks like (scaled properly):

 

Here is the scene (properly scaled to the above numbers in post #426) next to the 10Hz sine:

 

So yeah, for sure the hulk scene "places a more sustained load (lower crest factor) on the amp."  -I completely agree.  That's one of the reasons that I just stick to peak values for measuring power, voltage, and current. 

[3ll3d00d]

Nice example, thanks for posting this to illustrate.

[3ll3d00d]

One follow on question... Leaving aside whether the driver can take it, what would be reasonable power headroom to leave over an excursion limited figure? 3dB? 6dB?

[Bossobass Dave]

SHOOTOUT!

 

A Raptor single module was loaded with HST-15s, BHT-15s and LMS-R-15s were run through a few tests to compare in-room results. The HST-15s were dual voice coil, 4 ohms nominal each, as are the BHT-15s. The LMS-R-15s are also dual voice coil but only available in 2 ohms nominal per VC.

 

Player: OPPO BDP-105

 

Signal Shaper: Marchand Mono Bassis XLR version

 

Amplifier: Bossobass A-14K on a dedicated 120V-30A mains outlet

 

Microphone: ACO Pacific 7012, PS 9200, CA4012

 

Interface: Edirol UA1000 24 bit/48k Hz

 

PC: Dell running Windows 8 OS, both 32 bit and 64 bit versions

 

The Raptor Modules were corner loaded on a marked platform so that each of the 3 modules would be placed in precisely the same spot. The close mic, THD test and listening position (PLP) test microphone placements were in marked positions to ensure precisely the same microphone placement for the respective tests for each of the 3 modules.

 

The gain structure for the signal was carefully calibrated to avoid the OPPO clipping problem when redirecting bass to the SW output. The Voltage Peak Limiter on the A-14K was set to 116V, which is doubled when in bridged mode to 232V peak. The A-14K was in bridged mode throughout the testing.

 

We loaded the drivers into the Raptor module and:

 

1) Measured the Re of each of the dual driver modules.

 

2) Measured the native response of each of the dual driver modules.

 

3) Configured a custom L/T curve using a Marchand Bassis and measured the new response of each of the dual driver modules.

 

4) Played the HULK PUNCH scene and raised the OPPO MVL until the the peak dBSPL reading was 115.2dB at the PLP @ 4M distance from the edge of the module to the microphone. We then measured and recorded the resistance, voltage, amps and watts for each of the 3 different driver-loaded modules.

 

5) Played a 10 Hz sine wave tone with the microphone at a marked 1M distance and raised the MVL until the dBSPL read 95dB and recorded the THD and capped the graph. NOTE: We are aware that THD is not accurate when measured in a room full of rattles, etc. We only wanted a relative guage at 1M with a 10 Hz sine and at the PLP with a 100dB sine sweep for comparison.

 

6) Played a sine sweep at 100dB to measure sub-only frequency response magnitude at the PLP showing the FRM trace and the THD results. As stated earlier, each of the 3 driver-loaded modules was the only thing connected and measured. There was no FRM post smoothing EQ or any other measures taken to alter the response at the seats.

 

7) Played the scene in Captain America where the heli-carriers launch at a MVL equivalent to approximately +3dBRL (decibels from Reference Level, in this case, above Reference Level) and captured the performance in SpectrumLab to compare it to a looped version captured straight off the OPPO SW output jack.

 

There were no satellite speakers engaged throughout the tests. The only audio playing was through each of the 3 driver-loaded Raptor modules, on at a time. The frequency magnitude response tests were run with the LPF at 250 Hz, the highest available in the OPPO menu for the SW output. The soundtrack scenes were run with the LPF set at 100 Hz, the highest setting we use the the HT.

 

There was no attempt to measure torture level maximums as it is assumed that the Stereo Integrity HST-15s would win that metric due to its larger spiders and maximum throw capability. We want to return the drivers to Nick in perfect condition, not in a trash bag. The intent was to gauge any differences using soundtrack source at full reference level with 3dB of headroom.

[Bossobass Dave]

reserved

 

 

 

[Bossobass Dave]

reserved

 

Native and L/T'd frequency response magnitudes of each driver in Raptor enclosure, 2 drivers each, dual opposed configuration:

 

Frequency response magnitude at the PLP, approximately 4M from sub to mic and separate FRMs of each sub with THD at the PLP:

 

 

10 Hz THD at 95dB at 1M.

 

[Bossobass Dave]

reserved

 

Animation of CA heli-carrier scene with speclab caps corresponding with linear scale FR graphs:

 

 

Same scene, comparing each sub to the looped cap off the OPPO:

 

HST-15:

 

BHT-15:

 

LMS-R-15:

 

The FR non-linearities (deviation from flat) are reflected in the SL caps.

[Bossobass Dave]

reserved

 

This is where we ran the Hulk Punch scene to 115dB at the seats with each of the 3 driver-loaded modules and recorded the pertinent data.

 

Of course, the difference in impedance in using dual VC D4 vs dual VC D2 drivers is reflected in the results.

 

We did not integrate the subwoofers into the 5.1 system and apply post smoothing EQ to do subjective listening. Both Paul and I agree that when operating the subs within reasonable parameters and with the subs carefully calibrated to equal levels, we do not hear a difference subjectively with these drivers.

 

All 3 drivers are long throw 15" subwoofer drivers of very high quality. The HST-15 has the longest stroke of the 3 by approximately 15-20%, so, if we had driven the systems to destruction, the HST-15 would have given the loudest peaks before death by a few dB.

 

The HST-15 also has the lowest Fs, so it requires the least LT boost to get flat in-room. We chose to use the same boost on all 3 drivers so that's reflected in the overall in-room FR and the reproduction of full BW source.

 

All 3 drivers are excellent performers and can take peaks in excess of 5KW all night long with no audible mechanical distress. The LMS-R-15s were a pair consisting of one slightly older and one slightly newer driver because one of them failed when we originally tested them using the "Coming in Hot" scene from Oblivion. I imagine that that sort of poor QC might be a reason why PE has decided to drop the line.

 

For the record, 2-1/2" of excursion equals 63.5 mm of peak to peak throw and is quite violent motion. The volume of sealed air in the Raptor module is designed to allow for that sort of maximum while making it much harder to exceed that level, requiring LOTS of power to drive the subs to mechanical extremes.

 

In an earlier test, shown previous to this shootout in this thread, we drove the HST-15 drivers to their rated Xmax of 35 mm one way or just about 2-3/4" peak to peak. At the GTG we had them up to 44 mm one way, or almost 3-1/2" peak to peak. We don't recommend pushing the HST-15 (or 11, or 18) that hard. the result will include audible distress and unwanted non-linearities. But, up to its rated Xmax, the HST-15 is well behaved and quite accurate.

 

Getting reference level from a pair of 15s in 3500 cubes is remarkable performance when the content is to 5 Hz, based on dozens and dozens of drivers in the same listening space that have attempted it over the years. Of course, these drivers actually require more power than was applied to get a flat response to 5 Hz in-room, whether you boost the low end (alter the LT parameters) or attenuate the top end (use a separate post LT EQ) to achieve that goal. So, at least 4 x 15" of the caliber of the shootout trio are needed, IMO.

[Infrasonic]

Excellent stuff, Dave! Tuning in for more data.

[Electrodynamic]

Thanks for posting the measurements Dave. Nice to see measurements on all three drivers...particularly your drivers.

 

Besides the BHT driver the results are as I anticipated. LMS driver with better HF response and HST with better LFE and output. 

[Shredhead]

One follow on question... Leaving aside whether the driver can take it, what would be reasonable power headroom to leave over an excursion limited figure? 3dB? 6dB?

Good question d00d.  I don't have an accurate guess having measured the numbers on only 1 test.  That's a better question for Bosso due to his practical experience with many amps, drivers and enclosures.  I think he pretty much sums up the experiment here (post #430):

 

 

 "Please note that this was one scene with one configuration of driver-to-Vb. The difference can vary quite a bit between the steady state signal+power and the dynamic soundtrack source+power. Some soundtrack scenes/sub systems can require 2-3dB more than a steady state test will show to compel the drivers to max excursion.

 

Of course, we selected Xmax as the target so as not to wreck perfectly good drivers for the sake of an experiment.

 

How much power it takes is not knowable because dynamic source is a moving target. But, props and much thanks to John and Abraham, we have a pretty good idea of a window vs having no data at all.

 

The bottom line is that balancing driver/box/amplifier and signal shaped curve involves more than looking at how many volts will push the driver to a safe excursion when driven by a steady state single sine sweep."

[Bossobass Dave]

A word about the THD numbers; The BHT posted the highest THD as a % during the 10 Hz single sine test because of it's higher 3HD @10 Hz. The sine sweep THD numbers are much more similar at 10 Hz, so we'll be looking into that as well, although, in the final analysis, I don't fret harmonics as a percentage <20 Hz as I don't believe they hold the relevance most others seem to think they do.

 

The shorting rings employed in the BHT-15 are designed to lower Le. The primary reason was to lower harmonics in the crossover region where they are most audible in the subwoofer band. While not being concerned with harmonics as a % <20 Hz, I believe harmonics in the cross region are more significant and should be as low as possible.

 

Although the total harmonic distortion is higher as a percentage in the BHT-15 than the other drivers with a single sine @ 10 Hz, its HD is lower than the other drivers above 40 Hz. Here's the comparo between the HST-15s and the BHT-15s >40 Hz. Except for the anomalous little spike in THD at 52 Hz, the output and response magnitudes are, if anything, higher from the BHT-15s in this band with lower HD as a %.

 

This is just the HST-15s and the BHT-15s:

 

[Bossobass Dave]

We just got a cal file for the Edirol interface to use in SpecLab from 3II3d00d that takes the interface roll off out of the speclab caps. Awesome and many thanks to the genius of the d00d.

 

We'll be running more comparo scenes, but I was waiting to get this correction file for obvious reasons.

 

We promise not to tilt the graphs, so just look at them with your heads tilted to the left until it feels right.

[Bossobass Dave]

Thanks for posting the measurements Dave. Nice to see measurements on all three drivers...particularly your drivers.

 

Besides the BHT driver the results are as I anticipated. LMS driver with better HF response and HST with better LFE and output. 

 

The HST drivers kick butt. Anyone familiar with my 15 year posting history will co-sign that I've raved about precious few hardware items over those years. They have been a blast to eval. Nice job, young man.

[Electrodynamic]

A word about the THD numbers; The BHT posted the highest THD as a % during the 10 Hz single sine test because of it's higher 3HD @10 Hz. The sine sweep THD numbers are much more similar at 10 Hz, so we'll be looking into that as well, although, in the final analysis, I don't fret harmonics as a percentage <20 Hz as I don't believe they hold the relevance most others seem to think they do.

 

The shorting rings employed in the BHT-15 are designed to lower Le. The primary reason was to lower harmonics in the crossover region where they are most audible in the subwoofer band. While not being concerned with harmonics as a % <20 Hz, I believe harmonics in the cross region are more significant and should be as low as possible.

 

Although the total harmonic distortion is higher as a percentage in the BHT-15 than the other drivers with a single sine @ 10 Hz, its HD is lower than the other drivers above 40 Hz. Here's the comparo between the HST-15s and the BHT-15s >40 Hz. Except for the anomalous little spike in THD at 52 Hz, the output and response magnitudes are, if anything, higher from the BHT-15s in this band with lower HD as a %.

 

This is just the HST-15s and the BHT-15s:

 

 

The one thing that I've stopped posting about is how distortion in any aspect (whether it be 1'st harmonic, 2'nd, 3'rd, 4'th, etc) is not necessarily a bad thing. And forum members please don't go ninja-editing that last sentence in attempts to misconstrue my point. My point is that distortion may have been (usually IS) there in the design for a reason. Case in point is my shallow BM mkIV - I trimmed the gap on the Xmax:Xmech too close on that design and the driver is VERY linear right up until it bottoms out. Most users  keep turning up the volume because they do not hear any signs of stress until WACK and it's too late. And the discussion could go on forever about distortion but solid 2'nd and 4'th harmonics are a welcome item to have in the playback much like how people like class A amplifiers because of their "warmth" which is actually 2'nd/4'th HD. But I digress....

 

Thanks for posting the GIF's Bosso.  

[Shredhead]

Recently I came across a post by DB member mojave in AVS about different wiring schemes distributing power unevenly: http://www.avsforum.com/forum/155-diy-speakers-subs/2075890-series-parallel-vs-parallel-series-driver-wiring.html

Bosso and I were curious so we broke out 4 dual voice coil 15's and got to work.

 

 

Wiring scheme test:

-The quad of 15" drivers were in free air. 

-All three schemes presented 8 ohms nominal to the amplifier, which was in bridged mode for all three schemes.
-All three schemes used equal lengths of series, parallel and amplifier leads, respectively.
-All three schemes resulted in nominally equal excursion for all 4 drivers (8-voice coils), not measured.
-All three schemes are named starting at the voice coils and ending at the amplifier.
-The total voltage measured at the amplifier for all three schemes was 39.66V RMS, using a 10 Hz sine wave from REW signal generator as input source.
-Wiring scheme 'A' showed a measured difference in voltage to the drivers of <1V, but when the drivers were switched in position, that difference followed the driver and had nothing to do with the wiring scheme.
-Wiring scheme 'B' showed a measured difference of 0.1V between the series pairs.
-With wiring scheme 'C', the first driver on each end of the leads to the amplifier, the positive lead and the negative lead, received 1.14Vrms more voltage, measured with a Fluke DMM. This difference amounted to a difference of slightly more than 2W with 175W fed to the quad of drivers.
-The difference in power to the drivers in the other two schemes was negligible.

 

Here is a video of equal excursion for the 4 drivers hooked up with wiring scheme "A".  My guess was if there was going to be a difference in excursion for the 4 drivers it would be revealed in this wiring scheme and as you can see, power was distributed equally as it was in all 3 schemes. 

 

-The back EMF test consists of manually causing the cone of one driver at either end of the wiring scheme in each of the three cases and observing the other three drivers to see if there is any opposing motion caused by back EMF from the driver being manually put in motion. There was no back EMF contamination with wiring schemes 'A' or 'B', but there was back EMF induced motion visually noted in wiring scheme 'C'. Wiring the voice coils in parallel neutralizes back EMF, so it's the preferred method to series wiring for that reason. Some are of the opinion that a modern amplifier in bridge mode will have enough damping factor to render the back EMF problem moot in actual use, so either wiring scheme is OK.

Conclusion: If wiring multiple drivers to a mono block amplifier, whether in 1/2 bridge or full bridge configuration, parallel/series is preferred to series/parallel. This is to err on the side of caution and based on measured differences between the two schemes. Wiring of multiple single coil drivers may result in slightly different measured outcomes, but that is to be confirmed by others and, in our opinion, not significant.

[SME]

The one thing that I've stopped posting about is how distortion in any aspect (whether it be 1'st harmonic, 2'nd, 3'rd, 4'th, etc) is not necessarily a bad thing. And forum members please don't go ninja-editing that last sentence in attempts to misconstrue my point. My point is that distortion may have been (usually IS) there in the design for a reason. Case in point is my shallow BM mkIV - I trimmed the gap on the Xmax:Xmech too close on that design and the driver is VERY linear right up until it bottoms out. Most users  keep turning up the volume because they do not hear any signs of stress until WACK and it's too late.

This is an interesting point.  I gather a lot of people have blown out their TC Sounds LMS drivers in the same way.

 

And the discussion could go on forever about distortion but solid 2'nd and 4'th harmonics are a welcome item to have in the playback much like how people like class A amplifiers because of their "warmth" which is actually 2'nd/4'th HD. But I digress....

 

I think you mean tube amps rather than class A amps.  Although, I don't doubt that some class A amps have even-order harmonic distortion added to cater to certain tastes.

 

It is true that the addition of harmonics can add warmth and transient definition to the sound.  It is also my experience that bass harmonics contribute a lot of the loudness to the fundamental.  This is a big reason I don't like listening with the sub level hot and prefer a smooth ramp toward the low end that raises the level of frequencies above the subwoofer range as well.  Nevertheless, it is my opinion that such distortion should be avoided if possible.  If the added distortion is indeed harmonic and is in-phase with the harmonics that are already present, then it may be of mild consequence.  However, such distortion is only harmonic with sine waves and signals dominated by a single tone.  More complex signals will cause intermodular distortion, which is much more audible and unpleasant.  Another problem relevant even to purely harmonic distortion is that the distortion harmonics may not be in phase with the harmonics already present in the content.  This can actually cause partial cancellation of those harmonics leading to a loss of the warmth and transient fidelity that the distortion was supposedly adding.

[SME]

Conclusion: If wiring multiple drivers to a mono block amplifier, whether in 1/2 bridge or full bridge configuration, parallel/series is preferred to series/parallel. This is to err on the side of caution and based on measured differences between the two schemes. Wiring of multiple single coil drivers may result in slightly different measured outcomes, but that is to be confirmed by others and, in our opinion, not significant.

I don't think your testing is thorough enough.  The hypothesized problem with series wiring is that any impedance differences in the drivers will cause different voltages to appear on each driver's terminals.  At 10 Hz, you are likely far away from the free air resonances of the drivers, and therefore the impedances of the drivers will be closely matched.  However, nearer to the free air resonance, the impedances of each driver vary much more substantially and any differences in T/S parameters leading to different free air resonances will be exaggerated.  Also, the amount of variation in T/S parameters between drivers may also change considerably between different sets of drivers.

 

As for back EMF contamination between drivers, this is yet another reason to avoid wiring drivers in series.  If all the drivers are wired in parallel, then any back EMF current will flow directly to the amp instead of inducing voltage across the other drivers.  As for why you saw contamination with scheme C but not scheme A or B, I don't have a good answer, but I would generally avoid wiring any woofers in series if at all possible.

 

I believe the best wiring scheme in this case would be "scheme D" (not pictured), in which the voice coils for each driver are wired in series and each of the two pairs of drivers are wired in parallel and connected to separate channels of the amp with the amp not used in bridged mode.

[Bossobass Dave]

Wiring dual VCs of a dual VC driver in series is not a problem because the impedance curve will obviously be identical for both.

 

Back EMF is also not a problem when the connected drivers are high quality (well damped) and the amplifier has sufficient damping factor (very low output impedance).

 

The exercise was plenty thorough. 10 Hz was chosen because of the obvious difficulty in seeing any excursion differences at 20-30 Hz without dumping a considerable amount of power into the drivers in free air. You may feel free to set up and document your own exercise looking for a different outcome than we found in ours exercise.

 

Eight VCs wired in the scheme that was presented in the OP and two others, in an attempt to see the result that was pictured in the vid posted in the cited thread. The conclusion is that the wiring scheme is most likely not the culprit, but could be experimented with as part of the path to a possible answer.

 

Your wiring scheme presents dual 4 ohm loads and is not practical to the posit of 8 VCs presenting a single load to the amplifier channel, whether 1/2 bridge or full bridge, in a tower of 8 drivers arrayed in a column enclosure.