AVS Forum banner

101 - 120 of 446 Posts

·
Registered
Joined
·
1,981 Posts
Discussion Starter #101
Are you going to do sweeps indoors? I want to see if the response flattens out without any EQ or BEQ.


Yeah, I’ll take measurements inside. You’re not going to see anything you like though, my room is not friendly.

Chris
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #102 (Edited)
Well, the snow stopped early this afternoon and it turned into a bright, sunny day. Back at it, can’t let frozen sky-water stand in the way of science!

To start off, I added another terminal cup and converted the box to two 2ohm loads.

Step one: cut a hole in a box.


Rewired, and verified with dmm:

I also used the 9v battery test to confirm all drivers are correct polarity.

Haul everything back outside and get it all set up again:


I unbridged the amp, and put each 2ohm load on one channel. I used the dmm to measure voltage on one channel. I set up the minidsp to send identical signals to both channels.

Performed the same sweep test. Once again, ran into mic clipping and bad noises at the same time. This time, reran the trouble sweep while standing up at the cab. I discovered that what I took to be driver distress was actually the whole cab rattling on the dolly. I was able to stop it by putting my weight on the cab. Great, there goes yesterday’s results.

Continued running sweeps. The peak got cut off due to mic clipping, but the rest of the sweep continued to increase linearly. Eventually, increase stopped and voltage readings bounced around, I ran out of amp. I never heard driver distress.

I didn’t export this graph, I do still have the data. I consider it to be of limited use due to mic clipping.

I moved the mic to 4 meters, and started up again with the sweeps. I started at higher level this time, knowing it was safe from the 2m testing. 1db increments on sweeps.

I calculated the wattage for each channel using the RMS voltage displayed by the DMM, and the Re of 1.7ohm. I multiplied by 2 to account for both channels. I didn’t convert voltage to peak.
P = V^2 / 1.7 * 2

Here’s the resulting graph:

I left off the final sweep where the amp gave out.

Since I had invalidated my stopping point from yesterday, I popped the hatch and set it back up series/parallel. I verified polarity again with 9v battery test.

I set the amp up bridged, and performed the sweeps again. Started in 3db increments, switched to 1db as levels got high. Once again, I was able to increase until hitting the limit of the amplifier. This time, I just hit mic clipping on the final clean sweep.

Again, RMS voltage was used for wattage calculation. Re of 3.4ohm was used.


Once again, I left off the final sweep where the amp gave up.

Here’s a comparison of the final clean sweeps for both wiring methods:




What did I learn from this data?
Nothing definite. Other than the fact that this cabinet can put out some serious noise.

It doesn’t seem to be hampered by the over-excursion predicted in AkAbak for series/parallel operation. Or at least isn’t showing it.

The wattages I’m recording make me doubt my voltage measurement technique. Unless my amp is highly underrated, or just high distortion at these output levels. I wish I had a an oscilloscope to check waveform.

The spl I’m recording doesn’t jive with the model at all. If I’m correct, 4m mic distance should be down 12db.

Apparently my model-output level match from yesterday was just chance, from where the mic clipped at 2m.

Hopefully someone more experienced can help me interpret this data. As of now I’m pretty lost on what I’ve measured, and where to go from here.

Chris
 

·
Registered
Joined
·
4,832 Posts
Well, i am no expert, but doesn’t this look rather good?
And if 4m is 12dB down, isn’t it pretty much as modeled?

And if the problem was amp or mic related, and NOT driver maxing out, even better :)

But yeah, waiting fo the experts to chime in ;)

BTW: REW has distorsion graphs too ;)
 

·
Registered
Joined
·
2,155 Posts
I believe it is 6db down for doubling of distance, I know for 1m to 2m it is. Man that thing climbs from 20Hz to 30Hz. This thing would do awesome in a room that gains well!
 

·
Registered
Joined
·
4,832 Posts
I believe it is 6db down for doubling of distance, I know for 1m to 2m it is. Man that thing climbs from 20Hz to 30Hz. This thing would do awesome in a room that gains well!
Yes, but the model is 1m right? So from 1m to 4m would be 12dB?
 

·
Registered
Joined
·
2,155 Posts
Yes, but the model is 1m right? So from 1m to 4m would be 12dB?
Yes, 1m on the sims. Sim was about 117db at 20Hz, if you hit 105db that is on target.
 

·
Registered
Joined
·
1,241 Posts
I believe it is 6db down for doubling of distance, I know for 1m to 2m it is. Man that thing climbs from 20Hz to 30Hz. This thing would do awesome in a room that gains well!
THAT'S the measurement I want to see!
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #108
Yes, 1m on the sims. Sim was about 117db at 20Hz, if you hit 105db that is on target.


Yeah, its not really on target. I bumped the models to 1500w just in case my voltage measurements are ballpark accurate.



Shape still looks pretty good, especially to the jbl specs.

Shape looked even better at 2m, but my mic gave up early. Hell, I’m clipping it at 4m.

Chris
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #109
THAT'S the measurement I want to see!


Its one I’d love to be able to show you, lol. I don’t have said room. My room is open to the rest of the house, has weird shapes and stub-walls, is untreated, and is generally unfriendly.

Close-mic of my mini dev’s:


Response at MLP. Red is mini dev:


As I said, not pretty. I’m sure the quad will be similar.

If you know your room gain, its pretty easy to extrapolate from the ground plane model though.

Chris
 

·
Registered
Yamaha RXa2060 - HTM12 / Pa460 3way LCR - Volt10s atmos - HT8s rears - (4) Dual JBL Devs - (2) F20s
Joined
·
430 Posts
Not to derail this subject to far but how would a dual 12 sim out? While you were discussing wiring in series / parallel it occurred to me that doing a box design similar to a dts10 by cutting the dev box down the middle it would be able to fix that wiring problem and at the same time I though it would be neat that you could keep the box depth to around 14 inches for a really narrow dual 12 slim box design. I also realize you will loose some spl but if the design outweighs the negatives then building maybe eight of these may seem like a good design for some people, It would just be two $30 subs with $50 of wood after all.

I know that comparing the dts to the dev would be apple to oranges because they are not all that close in actual alignments and both have a different goal so I’m only trying to use that as an example and not a direct comparison, but it did peak my curiosity.
 

·
Registered
Joined
·
274 Posts
the general shape of the response in the passband really is close to the model to my naked eye....but that first resonance just out of the passband range seems a bit odd. have you done an impedance measurement and compared it to the model at all?
what kind of distortion are you pushing at the higher output levels? this could possibly let you know if you have that odd excursion Josh had alluded to.

great work on getting this thing put together and being so diligent on the back end with measurements and everything!

*edit*
a quick read on diagnosing impedance differences from model:
https://www.diysubwoofers.org/projects/other/POC2/
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #112
the general shape of the response in the passband really is close to the model to my naked eye....but that first resonance just out of the passband range seems a bit odd. have you done an impedance measurement and compared it to the model at all?

what kind of distortion are you pushing at the higher output levels? this could possibly let you know if you have that odd excursion Josh had alluded to.



great work on getting this thing put together and being so diligent on the back end with measurements and everything!



*edit*

a quick read on diagnosing impedance differences from model:

https://www.diysubwoofers.org/projects/other/POC2/


Yeah, I’ve been eyeing that resonance as as well, not sure what to make of that one. Its not shown in any of the models, and stays consistent in all measurements.

I don’t have the ability to measure impedance yet. Looks like I need to order some resistors.

I need to read on distortion measurements as well, not something I’ve ever done.

Chris
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #113
Not to derail this subject to far but how would a dual 12 sim out? While you were discussing wiring in series / parallel it occurred to me that doing a box design similar to a dts10 by cutting the dev box down the middle it would be able to fix that wiring problem and at the same time I though it would be neat that you could keep the box depth to around 14 inches for a really narrow dual 12 slim box design. I also realize you will loose some spl but if the design outweighs the negatives then building maybe eight of these may seem like a good design for some people, It would just be two $30 subs with $50 of wood after all.

I know that comparing the dts to the dev would be apple to oranges because they are not all that close in actual alignments and both have a different goal so I’m only trying to use that as an example and not a direct comparison, but it did peak my curiosity.

I’m guessing it will model exactly like half of this design.
HR response of course doesn’t show the excursion issue, AkAbak does.

The goal here is not necessarily just to prove the viability of this particular design, but of multi- driver dev’s in general.
Chris
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #114
...but that first resonance just out of the passband range seems a bit odd.

I believe the mystery resonance has been identified. I talked to John, @Red Five, who speculated it could be a half wave cancellation off the QW resonator.

Looking at my mini close-mic, what I took to be the first port resonance fits pretty well with this theory:


The quad has a longer QW, resulting in a lower tune, so the resonance is pushed down:


Now to find something with a similar QW length to the quad. I stole some close-mic measurements of v1’s from @harrisu and @automata21. The resonance is much less pronounced in these graphs, but occurs at the same frequency.

Harrisu:


Automata21:


If one of you guys saved those sweeps, I’d appreciate if you could send the REW file to me for an overlay.

Chris
 

·
Registered
Joined
·
5,542 Posts
I unbridged the amp, and put each 2ohm load on one channel. I used the dmm to measure voltage on one channel. I set up the minidsp to send identical signals to both channels.

Performed the same sweep test. Once again, ran into mic clipping and bad noises at the same time. This time, reran the trouble sweep while standing up at the cab. I discovered that what I took to be driver distress was actually the whole cab rattling on the dolly. I was able to stop it by putting my weight on the cab. Great, there goes yesterday’s results.

Continued running sweeps. The peak got cut off due to mic clipping, but the rest of the sweep continued to increase linearly. Eventually, increase stopped and voltage readings bounced around, I ran out of amp. I never heard driver distress.

I didn’t export this graph, I do still have the data. I consider it to be of limited use due to mic clipping.

I moved the mic to 4 meters, and started up again with the sweeps. I started at higher level this time, knowing it was safe from the 2m testing. 1db increments on sweeps.

I calculated the wattage for each channel using the RMS voltage displayed by the DMM, and the Re of 1.7ohm. I multiplied by 2 to account for both channels. I didn’t convert voltage to peak.
P = V^2 / 1.7 * 2

Here’s the resulting graph:

I left off the final sweep where the amp gave out.

Since I had invalidated my stopping point from yesterday, I popped the hatch and set it back up series/parallel. I verified polarity again with 9v battery test.

I set the amp up bridged, and performed the sweeps again. Started in 3db increments, switched to 1db as levels got high. Once again, I was able to increase until hitting the limit of the amplifier. This time, I just hit mic clipping on the final clean sweep.

Again, RMS voltage was used for wattage calculation. Re of 3.4ohm was used.


Once again, I left off the final sweep where the amp gave up.

Here’s a comparison of the final clean sweeps for both wiring methods:




What did I learn from this data?
Nothing definite. Other than the fact that this cabinet can put out some serious noise.

It doesn’t seem to be hampered by the over-excursion predicted in AkAbak for series/parallel operation. Or at least isn’t showing it.

The wattages I’m recording make me doubt my voltage measurement technique. Unless my amp is highly underrated, or just high distortion at these output levels. I wish I had a an oscilloscope to check waveform.

The spl I’m recording doesn’t jive with the model at all. If I’m correct, 4m mic distance should be down 12db.

Apparently my model-output level match from yesterday was just chance, from where the mic clipped at 2m.

Hopefully someone more experienced can help me interpret this data. As of now I’m pretty lost on what I’ve measured, and where to go from here.

Chris

Been busy. Trying to catch up.

According to the models the drivers aren't starting to reach 15mm excursion until 65 volts which is about 1200w maximum into the impedance min. I'm assuming drivers wired for 4 ohm mono. Do not multiply the rms voltage for a peak #. Run a 50 or 60hz sine into the sub and get a reading with the dmm. Whatever it reads is the voltage. I see that you went up to 38 volts maximum? According to the models that's not enough to hit 9mm one way excursion. Just starting to push it at that level.

I agree something is screwy with the SPL if that's all the voltage that's being put into it. Do you have a calibrator or is this a Umik or similar which has SPL based on a mic capsule sensitivity rating?

As far as SPL and FR shape goes, it will be different from the model. It will show increased SPL and tilted response towards the upper frequencies due to the size of the enclosure and baffle. At 28x48 this is a big front surface. EDGE and other programs can simulate this. This would explain maybe 2dB of extra SPL over the model at 100Hz. At 20Hz it wouldn't amount to much.

Back to driver excursion...The models assume everything is linear. Drivers are not in the real world. It usually starts to take more power to produce the expected output as the suspension tightens, the coils heat up and the BL drops off due to less coil in the gap. The JBL specs say these drivers have a 32mm coil length and a 12.1mm gap. The math says the one way coil overhang is 10mm. JBL says 12mm xmax. After this point the driver probably does not respond to power increases as shown in modeling. (It will take much larger inputs from here to get more excursion) 16mm would have only half the amount of coil mass still in the gap. It will likely be producing significant THD by this point. The coil leaves the gap at 22mm one way. Functionally that'd be xmech if there isn't direct physical contact by that point. Distortion would be through the roof and it may even sound like hard clipping or mechanical contact if driven that far. If the models say these drivers hit 15mm excursion at 65 volts in this cab I'd expect it to take more like 70 or even 75 volts real world. Assuming the models are on point.

Tha Akabak sim with uneven excursion loses about 3dB in SPL capability in the 30Hz range compared to the HR model. The Akabak models shows one set of drivers reaching 15mm xmax at 50 volts (750w max power roughly)(vs 65v and 1200w for drivers in parallel). I'd not expect to start seeing this until 55 to 60 volts on the real sub due to losses and non linearity in the drivers though.

If you want to do distortion measurements in the future just change the measurement range to 0-2kHz. That'll capture the 2nd-10th harmonic up to 200Hz. Make sure to use a 1m sweep length for noise rejection.

What amp are you using? It seemed like the first set of measurements were with 38 volts max but the amp was bridged mono? Drivers wired series parallel? That would be 38 volts into a 4 ohm nominal load. 425w roughly. Amp was clipping?

Here are frequency response texts of the AK sims. One is with inductance the other without. Both at 50volts, 1m, half-space.
View attachment DEV4X1250VAK.txt

View attachment DEV4X1250VAKSEMI.txt
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #116 (Edited)
Thanks again Josh.

I’ll try to clarify my measurements here.

According to the models the drivers aren't starting to reach 15mm excursion until 65 volts which is about 1200w maximum into the impedance min. I'm assuming drivers wired for 4 ohm mono. Do not multiply the rms voltage for a peak #. Run a 50 or 60hz sine into the sub and get a reading with the dmm. Whatever it reads is the voltage. I see that you went up to 38 volts maximum? According to the models that's not enough to hit 9mm one way excursion. Just starting to push it at that level.
Saturday, I quit the sweeps at 38v into mono 4 ohm. I clipped the mic at 2m, and heard what I believed to be driver distress. I later discovered it was the enclosure bouncing on the dolly.

Sunday, I started by rewiring the cab into two 2ohm loads and un-bridging the amp. I ran sweeps with the mic at 4m, ending at 39v per channel.
I then retried the sweeps with the cab back to 4ohm mono, and ended at 69v. In both cases, I tried another sweep 1db up, but was out of amp.


Here’s the comparison of the two top sweeps:


The voltage measurements were taken by watching the DMM during the sweep. It generally climbed by half a volt or so over the sweep. Is this ballpark-accurate, or do I need to run 60hz sines to determine voltage?

EDIT: Should I use the measured Re or the nominal impedance to determine Pe?

I agree something is screwy with the SPL if that's all the voltage that's being put into it. Do you have a calibrator or is this a Umik or similar which has SPL based on a mic capsule sensitivity rating?
This is taken with a Umik, with the dip switches set to 130db sensitivity. I am using minidsp’s cal file. The only check I have is that the mic is clipping near 130db, as advertised.



Tha Akabak sim with uneven excursion loses about 3dB in SPL capability in the 30Hz range compared to the HR model. The Akabak models shows one set of drivers reaching 15mm xmax at 50 volts (750w max power roughly)(vs 65v and 1200w for drivers in parallel). I'd not expect to start seeing this until 55 to 60 volts on the real sub due to losses and non linearity in the drivers though.
What do you make of the series/parallel and parallel 2ch measurements showing very similar levels, with the series/parallel actually slightly higher on less power?

If you want to do distortion measurements in the future just change the measurement range to 0-2kHz. That'll capture the 2nd-10th harmonic up to 200Hz. Make sure to use a 1m sweep length for noise rejection.
Thanks, I’ll try that

EDIT: should I throw a 200hz LPF on the signal, or actually let it run up to 2k?


What amp are you using? It seemed like the first set of measurements were with 38 volts max but the amp was bridged mono? Drivers wired series parallel? That would be 38 volts into a 4 ohm nominal load. 425w roughly. Amp was clipping?
The amp is a Stewart HPA-1000. The amp was not clipping at that level, but the mic was at 2m.

The aforementioned second day measurements, the amp was likely clipping. I don’t have an oscilloscope, so just guessing by power levels.

Thanks,
Chris
 

·
Registered
Joined
·
5,542 Posts
Saturday, I quit the sweeps at 38v into mono 4 ohm. I clipped the mic at 2m, and heard what I believed to be driver distress. I later discovered it was the enclosure bouncing on the dolly.

Sunday, I started by rewiring the cab into two 2ohm loads and un-bridging the amp. I ran sweeps with the mic at 4m, ending at 39v per channel.
I then retried the sweeps with the cab back to 4ohm mono, and ended at 69v. In both cases, I tried another sweep 1db up, but was out of amp.
Gotcha. So you were in the range of 70 to 75 volts. This makes more sense with the SPL's measured.

Here’s the comparison of the two top sweeps:
The voltage measurements were taken by watching the DMM during the sweep. It generally climbed by half a volt or so over the sweep. Is this ballpark-accurate, or do I need to run 60hz sines to determine voltage?
When you increase the output level what are you controlling it with? Is it in dB increments? I run a 100w nominal 60Hz sine wave into a cab to get the baseline voltage reading. 20 volts into 4 ohm nominal for example. I write down the gain settings and get a baseline measurement. From there it's all calculated because I use a single digital gain for all level adjustments and keep track of the setting. You are probably fine but it's hard to say for sure. I always use 50-60Hz because that's where most meters, even really good ones, are most accurate.


This is taken with a Umik, with the dip switches set to 130db sensitivity. I am using minidsp’s cal file. The only check I have is that the mic is clipping near 130db, as advertised.
Ok. I'm not sure how close the sensitivity ratings on those are.



What do you make of the series/parallel and parallel 2ch measurements showing very similar levels, with the series/parallel actually slightly higher on less power?
I think the response curves should be matched by the response at 40Hz and above. If you look at them the FR's in this area will almost overlay exactly. It's the 30Hz area and the low frequencies that is the most different between the two. 30Hz area is where excursion is highest and where differences would be expected based on the wiring and modeling. It could be due to the driver coils being hotter on one test vs the other or the amp clipping too though. Hard to say anything definitive.


The amp is a Stewart HPA-1000. The amp was not clipping at that level, but the mic was at 2m.

The aforementioned second day measurements, the amp was likely clipping. I don’t have an oscilloscope, so just guessing by power levels.

Thanks,
Chris
Agreed. This amp is rated at 1000w at 4ohm bridged. 400w at 16ohm bridged. Looks like the voltage ceiling is around 40v per channel based on the 8ohm ratings and current is limited to about 16A short term based on the 2ohm stereo / 4 ohm mono rating. Based on the voltages you were trying to use the amp was probably current limiting and clipping heavily on the loudest sweeps you did. Combined with the mic clipping it becomes hard to tell what is responsible for what. Got a Behringer 3K or 6K?


EDIT: I throw a LPF on for the THD measurement because I'm always running the subs really hard with the top 2 measurements and there is no excursion for cooling outside of the bass range. Throwing full power 300Hz-2kHz is a good recipe to heat coils up really quick. If I recall it's a 240Hz 4th order BW I switch on. I'd have to check for sure but I set it to cut maximum high frequency from the sub while not affecting the level of anything below 200Hz. It has zero effect on the THD results in the range of interest.

See how fun and easy this stuff is!? :D
 

·
Registered
Joined
·
4,832 Posts
^Agreed! Get a NU6K on that sucker ;)
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #119 (Edited)
Thanks again, Josh!


When you increase the output level what are you controlling it with? Is it in dB increments? I run a 100w nominal 60Hz sine wave into a cab to get the baseline voltage reading. 20 volts into 4 ohm nominal for example. I write down the gain settings and get a baseline measurement. From there it's all calculated because I use a single digital gain for all level adjustments and keep track of the setting. You are probably fine but it's hard to say for sure. I always use 50-60Hz because that's where most meters, even really good ones, are most accurate.
I am controlling output level with REW’s sweep level. It is in 1dB increments. I was using 3dB to start,and 1dB at high levels.

Will your 100w nominal method work in this case? I can tweak the minidsp to get close to the 100w, it gives me tenth-dB increments.

I have been keeping track of settings and using only the REW level adjustment for my sweeps.


I think the response curves should be matched by the response at 40Hz and above. If you look at them the FR's in this area will almost overlay exactly. It's the 30Hz area and the low frequencies that is the most different between the two. 30Hz area is where excursion is highest and where differences would be expected based on the wiring and modeling. It could be due to the driver coils being hotter on one test vs the other or the amp clipping too though. Hard to say anything definitive.
The curves are not adjusted, they are as-recorded. Judging by the voltage, the 2ch sweep should be higher than the mono. Inconsistently between amp channels?



Agreed. This amp is rated at 1000w at 4ohm bridged. 400w at 16ohm bridged. Looks like the voltage ceiling is around 40v per channel based on the 8ohm ratings and current is limited to about 16A short term based on the 2ohm stereo / 4 ohm mono rating. Based on the voltages you were trying to use the amp was probably current limiting and clipping heavily on the loudest sweeps you did. Combined with the mic clipping it becomes hard to tell what is responsible for what. Got a Behringer 3K or 6K?
For higher voltage and clean power, I do have Crown DCi 4/1250’s in my theater. They are the analog version, so no voltage readout though. I would prefer not to drag one outside, but can if it will generate useful data. With 2500w on tap, I’m not so sure that magic smoke escaping counts as useful data, lol.

Would it be useful to use your 100w amp calibration to make 1000w sweeps with both driver configurations? This should insure clean power from the amp and demonstrate loss from the series/parallel configuration.

I do have a few automotive amps that are capable of 1000w at 1ohm. I can rig one to a couple batteries if we need to do an all-parallel mono run on the cab. Nothing like redneck audio testing.


See how fun and easy this stuff is!? :D
I’m not so sure I ever expected this to be fun or easy. I did expect it to be easier to make determinations from the results, lol.
I am enough of an engineer and geek that this stuff is fun to me. Well, other than the temps and snow currently.


Thanks again for the help. Judging by data-bass, you have a bit of experience with this kind of measurement.

Chris
 

·
Registered
Joined
·
1,981 Posts
Discussion Starter #120
EDIT: I throw a LPF on for the THD measurement because I'm always running the subs really hard with the top 2 measurements and there is no excursion for cooling outside of the bass range. Throwing full power 300Hz-2kHz is a good recipe to heat coils up really quick. If I recall it's a 240Hz 4th order BW I switch on. I'd have to check for sure but I set it to cut maximum high frequency from the sub while not affecting the level of anything below 200Hz. It has zero effect on the THD results in the range of interest.


I can do that. The minidsp gives me plenty of filter control.

Chris
 
101 - 120 of 446 Posts
Top