Why Do TEF Systems read 3-5 dB high ?

[Art Welter]

While trying to compare hundreds of sub response curves, it seems that half space curves using TEF analyzers often show levels about 3-5 dB higher than other test systems and simulations read.

I have attempted to show the difference in the screen capture below.

The JBL SRX728 comes in “par for the course” when comparing decent dual 18” bass reflex cabinets of similar size, other than the usual “see saw” effect, speakers with lighter cones and higher tunings will be more sensitive up high, speakers with more LF output are less sensitive.
Previously there has been speculation that JBL was using full space for measuring their subs, which would account for a 6 dB difference, but that has been proven not to be the case.

The JBL SRX728 published chart shows about 100 dB 1 watt one meter at 120 Hz.
Comparing the chart next to it, a BFM condensed version of the 2006&7 Tulsa sub shootout, the 728 reads about 3 dB higher than it does on the JBL chart.
Looking at the response of the JTR Growlerx2, by reducing 6 dB (+3 double cabs, +3 double power) we see a single should be within a dB at 60 to 70 Hz as the 728, about 97.5 dB.
The chart above from the 2007 NYC sub shootout shows a single JTR Growler at about 105 dB, the DSL TH-215 just under that level at that frequency.

The DSL published chart for the TH-115 shows the level to be around 102.5 dB in that range, by inference that may be 5 dB higher than we would see if the JBL SRX728 was directly compared.

Another comparison would be the EV2180 dual 18 cabinet, EV’s published chart shows it as 100 dB at 50 Hz, while on the 2007 NYC shootout it comes in 5 dB above the DSL TH-215, a 2.5 dB difference.

From what I have read:

“The TDS technique (used by the Techron TEF system) uses a swept sine wave test signal with an analyzer that has matched tracking bandpass filters that can be offset in time.
The process uses a low crest factor linearly swept sine wave as a test signal which is applied to the test system during the whole measurement/test interval . The test signal characteristics maximize the amount of test energy injected into the system under test for a given peak signal level.”

Is it possible that the TEF test signal “reads higher” than a standard sine wave tone?
How are TEF machines calibrated to be SPL accurate?

 

[Ivan Beaver]

Re: Why Do TEF Systems read 3-5 dB high ?

I don't have time to go over all the details of your post-but will hit a of them.

The TEF is calibrated using a known good claibrator and then using an automatic level setting. Then that level is checked on the RTA display with the calibrator still attached to see if it matches the calibrator level.

I brought up years ago that the calibration was off during at least one of the Tulsa shootouts.

I don't knwo where BFM gets his measurements from, but others have brought up the fact that various other loudspeakers measurements look "off". The curves look as if they were hand drawn with a crayon.

I think there have been a couple of different Growler versions that have shown up at the shootouts (I could be wrong on this). I believe the origional was a little louder, but a latter one was flatter.

Of course you can change the calibration and make it whatever you want.

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

I don't have time to go over all the details of your post-but will hit a of them.

The TEF is calibrated using a known good claibrator and then using an automatic level setting. Then that level is checked on the RTA display with the calibrator still attached to see if it matches the calibrator level.

I brought up years ago that the calibration was off during at least one of the Tulsa shootouts.

I don't knwo where BFM gets his measurements from, but others have brought up the fact that various other loudspeakers measurements look "off". The curves look as if they were hand drawn with a crayon.

I think there have been a couple of different Growler versions that have shown up at the shootouts (I could be wrong on this). I believe the origional was a little louder, but a latter one was flatter.

Of course you can change the calibration and make it whatever you want.

Thanks Ivan,

The BFM chart appears to be measurements from 2006 and 2007 Tulsa sub shootouts. Although they do appear to have been “redrawn”, I spent hours comparing dozens of individual charts, they seem to be correct as far as representing those shootouts.

I’m aware of various measurement calibration issues at various sub shootouts, and the changes of cabinet and speaker types makes for a moving target.

I’m not sure which version of the TH- 215 was at the 2007 NYC sub shootout, there are two fairly different versions (size, weight, power ratings, impedance, frequency response) published from DSL, I did the comparison between the “flatter” version and the EV 2180, it was 2.5 dB off. Looking at the less flat version, the difference is minimal.

However, looking at the upper end of the TH-115 compared to the EV-2181 in the 2007 NYC shootout, the EV has about a dB more response at 100 Hz than the TH-115, yet on the outdoor chart you took around that time, compared to the EV published chart it looks to be 5 dB down from the TH-115 response.

That is a six dB difference between published and side by side in room comparisons, even with the known NYC room problems, that is hard to ignore.

I could point out more references, but it would be far easier to see if these differences are real if you have any straight A/B comparisons of any single unit production subs that have published frequency response charts to single DSL products, preferably measured outdoors.

 

[Steve Hurt]

Re: Why Do TEF Systems read 3-5 dB high ?

The Growler at the Tulsa shootout was a prototype and was not working correctly and has been pointed out in the threads the info you post comes from. Is there a grudge here Art?

I wasn't at NYC, but the pics show 2 Growlers coupled. Subtract 6 db from 105 and you get 99. Their stated efficiency is 100. Sounds pretty close to me.

 

[Silas Pradetto]

Re: Why Do TEF Systems read 3-5 dB high ?

The Growler at the Tulsa shootout was a prototype and was not working correctly and has been pointed out in the threads the info you post comes from. Is there a grudge here Art?

I wasn't at NYC, but the pics show 2 Growlers coupled. Subtract 6 db from 105 and you get 99. Their stated efficiency is 100. Sounds pretty close to me.

When a situation arises where there is something that doesn't make sense and it involves a ton of variables, Art and I obstinately process data and ask questions until we find the answer.

I'm sure Art doesn't have a grudge against the Growler (what did they ever do to him, other than provide beer? :lol, it's just one of the subs where the measurement errors are particularly obvious, and therefore often revisited.

PS- I don't want to put any words in Art's mouth. If I said something out of place, I apologize in advance.

 

[Steve Hurt]

Re: Why Do TEF Systems read 3-5 dB high ?

If the Growler specs are off regarding efficiency, I assume you would all agree that the Danley TH-Mini specs are also off since it's rated at 101?

 

[Steve Hurt]

Re: Why Do TEF Systems read 3-5 dB high ?

And the Yorkville UCS1 at 102 efficiency, that must be wrong too.

 

[Langston Holland]

Syn-Aud-Con

SAC Book of Virtues, Chapter 1, Verse 1:

Always test the testers before testing.

 

[Ivan Beaver]

Re: Why Do TEF Systems read 3-5 dB high ?

As stated-at the NY shootout, the drive voltage was really unknown. There were a couple of different VOM's there, and they all read different, so we had no idea what the drive voltage actually was. But what was important is that it remained the same during the testing-so the results were comparative-if not absolute.

There have been 2 versions of the TH215's. The earlier version-as at the nY shootout, had a lower sentivity, but a higher power capacity. The later version had a higher senstivity, but a lower power capacity. The net result was pretty much the same.

The TH115 had also had 2 different drivers in it. Well the same drive-but the standard impedance was different. At the NY shootout, the 8 ohm version was present. A couple of years later the "standard" was changed to the 4 ohm driver-due to overwhelming customer requests.

I have not had any other standard production subs in the shop and had any time to measure them side by side. If somebody wants to bring some up-they are more than welcome and we can test them and see where everything stands.

But I don't think there is any error in the TEF measurements.

At the Baltimore shootout, Smaart was used (although uncalibrated) and the results were pretty much comparitive to published specs-if I remember correctly.

I don't know what your second graph (the blue one) is or what the conditions are. THe numbers are way off.

 

[Art Welter]

Re: Syn-Aud-Con

SAC Book of Virtues, Chapter 1, Verse 1:

Always test the testers before testing.

Absolutely true.
You are one of my favorite testers (but why don't you return my phone calls ;^() you could probably put this all to rest quickly by simply doing a side by side test on one of your EAW cabinets compared to one of your DSL cabinets.

In fact, you probably already have done those tests, if they were done in the same place with the same drive level, all you would need to do would be to post them, I'll look up and compare DSL and EAW to your graphs myself.

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

If the Growler specs are off regarding efficiency, I assume you would all agree that the Danley TH-Mini specs are also off since it's rated at 101?

I have no grudge against JTR, and neither does Phil Lewandowski, he owns (or owned) both JTR Growlers and JBL SRX 718 cabinets.

I have no grudge against any manufacturer, but having studied a metric shitload of graphs and simulations, it seems that various TEF results look high compared to other tests.

If and when I install DSL product (or any other product), I’d like to know for sure that if DSL sub cabinet “X” graph shows 3 dB more sensitivity than a JBL sub of similar frequency response, it will indeed match the sensitivity of two JBL subs.

If there is a difference of 3 dB (or more), as I suspect, I can't make that assumption.

Phil tested the JTR Growler with the Beta 2010 12” 23 mm Xmax driver, posted results on Thursday, 27 May 2010 21:27 on the old LAB.
JBL SRX718 - Basic Distortion Measurements
http://srforums.prosoundweb.com/index.php/t/56314/19717/
Figures are raised by 20 dB to account for the difference between one meter and ten meters, so they should equate to one watt one meter .
The tests were done using sine wave tones at 85 volts (900 watts at nominal 8 ohms)
55hz - 125 dB
60hz - 126 dB
70hz - 130 dB
80hz - 128 dB

Phil did a JBL SRX 718 test the same day posted Thursday, 27 May 2010 21:47.
55hz 125dB
60hz 126dB
70hz 129dB
80hz 127dB

From that test, it is clear that the Growler and the SRX 718 have very similar output in the 55-80 Hz range.
JBL’s graph shows the 718 as slightly under 95 dB, they rate it as 95 dB one watt one meter.
JTR does not have graphs on his website, but rates the Growler as 100 dB sensitivity, +/- 1dB from 45-90 Hz.
IIRC Jeff had Mark Seaton do his TEF testing.
It appears either JBL is understating their product by 5 dB, or Jeff overstates his.
JBL’s specs are not in whole space, I have confirmed that with my own side by side testing of an SRX 728 to one of my ported dual Lab 12 cabinets.

Art Welter

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

As stated-at the NY shootout, the drive voltage was really unknown. There were a couple of different VOM's there, and they all read different, so we had no idea what the drive voltage actually was. But what was important is that it remained the same during the testing-so the results were comparative-if not absolute.

There have been 2 versions of the TH215's. The earlier version-as at the nY shootout, had a lower sentivity, but a higher power capacity. The later version had a higher senstivity, but a lower power capacity. The net result was pretty much the same.

The TH115 had also had 2 different drivers in it. Well the same drive-but the standard impedance was different. At the NY shootout, the 8 ohm version was present. A couple of years later the "standard" was changed to the 4 ohm driver-due to overwhelming customer requests.

I have not had any other standard production subs in the shop and had any time to measure them side by side. If somebody wants to bring some up-they are more than welcome and we can test them and see where everything stands.

But I don't think there is any error in the TEF measurements.

At the Baltimore shootout, Smaart was used (although uncalibrated) and the results were pretty much comparitive to published specs-if I remember correctly.

I don't know what your second graph (the blue one) is or what the conditions are. THe numbers are way off.

The second “Blue” graph shows the upper end of the NYC sub shootout to show the difference between the TH-115 and the EV-2181. The EV appears to be about 1 dB higher at 100 Hz.

Your graph shows the TH-115-8 as 5 dB higher than the EV-2181 at 100 Hz.

I made a mistake in the post, saying if the published specs are correct, the TH-115 should be around 6 dB higher, I forgot to account for the difference in impedance (8 ohm vs 4 ohm) so the TH-115 should show 2 dB more level than the EV-2181, but it shows 1 dB less, a 3 dB difference.

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

Lest anyone thinks I am picking on DSL, JTR, or any specific company using TEF, that simply is not the case.

I have put together another example of the disparity that seems to occur with TEF systems below, with information taken from the B&C spec sheet for the BC18SW125.

Some background information is included for those not familiar with bass reflex (BR) and tapped horn (TH) designs, which are used in the graphs below.

BR uses a port tuned to the cabinet internal volume, the output of the port combines with the speaker output to bring up the low end response of the speaker.
If the speaker is tuned for flat response, the output down low will be no greater than the upper response the speaker has in a sealed box.

By making the port/cabinet tuning frequency (Fb) higher than optimum for a flat response, it is possible to create a bump in the response above the 3 dB down point (F3), the upper response of the speaker will fall back to the normal level it has in a sealed box of similar size.

The Lab 2x12” ported design in the lower left graph does this, a nominal 92 dB speaker (two 89 dB speakers +3 dB) puts out 94 dB at the bump. Details of the 2x12" design are in the DIY section of this site.

As can be seen in the upper left chart, the 95 dB sensitive BC18SW125 speaker rolls off at about 12 dB per octave in a sealed box. In a “classic” ported box, shown in the simulation in the lower left, response falls at about 24 dB per octave below Fb.
The simulation does not accurately show what happens to the upper response of the speaker, the port creates a “pipe” resonance, internal reflections create peaks and dips.

To get a low F3 requires a low Fb, if you want the speaker to have relatively flat response, the box has to be big. The upper right graph of a B&C BR design shows the falling response of a fairly big box tuned quite low.

Something “funny” has happened to the speaker’s output in that upper right TEF graph, the output should not be any higher than 95 dB at 100 Hz, but it shows about 98 dB, THREE DB MORE THAN IT SHOULD BE.

Now, a very little information on TH, Tom Danley has a full TH white paper on the DSL website for those that want to know more than the very basics. The TH more effectively uses the front and back radiation of the speaker than a BR, the rear side of the driver enters the horn at a location (tap) at a distance far enough away from the throat, where the front of the driver is driving the horn.

The TH that I designed and built, called the Keystone, is about 6 dB louder than a BR cabinet using the same driver, with a similar F3.
Two of those BR cabinets would still be almost 3 dB down from the Keystone TH and take up more truck space.

The lower left graphs show several different cabinet designs and speakers tested at the same outdoor location, same drive level. All except the TH with extender used a BW 24 filter at 25 and 125 Hz.

0 dB on the lower left graph should be about 91 dB, if the published charts of B&C, Eminence, and JBL are correct.

The Lab 2x12” (FB 36 Hz) is nominally 3 ohms, the rest are all nominally 4 ohms.
The BC 18BR38 (FB 38 Hz) is of similar size to the 18 SW 115 simulation, it appears quite close to the 94 dB that sim shows.
The Keystone TH (TH18S no plate) is 45 x 26.5 x 22.5, double the size (volume) of the Lab 2x12”, a bit bigger than a DSL TH-118, a bit smaller than a JBL SRX728.

The JBL SRX728 is rated at 98 dB sensitivity, on the published chart it reaches around that above 60 Hz.

The Keystone TH is almost exactly the same sensitivity as the 728 from 45-80 Hz, louder above and below (other than below 32 Hz).

The Keystone, using the same B&C speaker as the TH-118, being larger, and having a slightly higher LF corner, should theoretically be louder than a TH-118, according to Hoffman's Iron Law.

Given Tom Danley’s design genius, a difference of 3 dB between the two tapped horns (mine being the lesser) seems possible.

The side by side manufacturer’s graphs (I dropped the DSL graph by 3 dB to account for voltage difference) of the TH-118 and the SRX728 show the TH-118 to be around 7.5 dB louder over most of the sub range.

It simply does not seem plausible that two tapped horns of similar size , frequency response, using the same speakers would be 7.5 dB different in level.

Art Welter

 

[Steve Hurt]

Re: Why Do TEF Systems read 3-5 dB high ?

It appears either JBL is understating their product by 5 dB, or Jeff overstates his.
JBL’s specs are not in whole space, I have confirmed that with my own side by side testing of an SRX 728 to one of my ported dual Lab 12 cabinets.

Art Welter

Lets see, you used:
- Home built speakers.
- Measurement equipment that gives you incorrect measurements
- Borrowed SRX728s speakers

And from this you state as incontrovertible fact that speakers that weren't within 1000 miles of your test are incorrectly rated?

Your measurements equipment, the operator, the speaker you used as a reference, the SRX speaker you used... any one of these (or a combination of them) could render your test worthless


And of course the contention that the SRX was rated in full space, as JBL's site seems to say, still is there.
Others have measured the SRX728 and come up 6 db hot.

But still, there is no way that you could be wrong.


I hear there is a position open as fact checker at Fox News.

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

Lets see, you used:
- Home built speakers.
- Measurement equipment that gives you incorrect measurements
- Borrowed SRX728s speakers

And from this you state as incontrovertible fact that speakers that weren't within 1000 miles of your test are incorrectly rated?

Your measurements equipment, the operator, the speaker you used as a reference, the SRX speaker you used... any one of these (or a combination of them) could render your test worthless


And of course the contention that the SRX was rated in full space, as JBL's site seems to say, still is there.
Others have measured the SRX728 and come up 6 db hot.

But still, there is no way that you could be wrong.


I hear there is a position open as fact checker at Fox News.

Thanks for the tip , Steve.

Could you let this "home made speaker maker" in on the details of the fact checker position at Fox News ?

Art (been making "home made speakers" and comparing them to every touring system encountered since 1978) Welter

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

I have not had any other standard production subs in the shop and had any time to measure them side by side. If somebody wants to bring some up-they are more than welcome and we can test them and see where everything stands.

I don't know what your second graph (the blue one) is or what the conditions are. THe numbers are way off.

If anyone would care to take up Ivan’s gracious offer, I’d be glad to pay for you and Ivan’s expenses.

It is very hard to compare traces in the “blue background jumble” (the blue one) from the 2007 NYC sub shootout that Mark Seaton provided which I posted previously, and the production changes in other speakers used there before do cloud comparisons.

Ivan, IIRC, you and Tom had the TEF data from the 2007 NYC sub shootout, if you still have it, could you please post the TH-115 and the EV-2181 2.83 V results?

Art Welter

 

[Ivan Beaver]

Re: Why Do TEF Systems read 3-5 dB high ?

If anyone would care to take up Ivan’s gracious offer, I’d be glad to pay for you and Ivan’s expenses.

It is very hard to compare traces in the “blue background jumble” (the blue one) from the 2007 NYC sub shootout that Mark Seaton provided which I posted previously, and the production changes in other speakers used there before do cloud comparisons.

Ivan, IIRC, you and Tom had the TEF data from the 2007 NYC sub shootout, if you still have it, could you please post the TH-115 and the EV-2181 2.83 V results?

Art Welter

Remember that this is the 8 ohm TH115-not the current 4 ohm. And we are also assuming the current 2181 design/components is the same as the one at the shootout-I have no idea.

What gets weird in comparing the measurements is that the TH115 looks kinda like the TH115 measurement we took (but with the room issues added in). But the 2181 doesn't look anything like EV's measurement on their spec sheet.

It could be that the EV measurement is 1/2 space-which would account for some differences-but not the shape (so much anyway).

One of the misconception people have is that a 1/2 space measurement is simple 3dB higher than a full space measurement.

This is not the case. At some freq it will be 3dB higher, but at others it will be the same (and I am talking about the below 200Hz) range.

So you cannot just "add" 3dB and get a new number. It is actually the shape of the CURVE that changes.

Which just adds to the fact of how important it is to see a measured curve and know how it is measured if you REALLY want to compare boxes-besides side by side of course.

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

Remember that this is the 8 ohm TH115-not the current 4 ohm. And we are also assuming the current 2181 design/components is the same as the one at the shootout-I have no idea.

What gets weird in comparing the measurements is that the TH115 looks kinda like the TH115 measurement we took (but with the room issues added in). But the 2181 doesn't look anything like EV's measurement on their spec sheet.

It could be that the EV measurement is 1/2 space-which would account for some differences-but not the shape (so much anyway).

One of the misconception people have is that a 1/2 space measurement is simple 3dB higher than a full space measurement.

This is not the case. At some freq it will be 3dB higher, but at others it will be the same (and I am talking about the below 200Hz) range.

So you cannot just "add" 3dB and get a new number. It is actually the shape of the CURVE that changes.

Which just adds to the fact of how important it is to see a measured curve and know how it is measured if you REALLY want to compare boxes-besides side by side of course.

Ivan,

Thank you for the quick reply. Looking at two traces makes the comparison much easier to see.

After a close look at the 2007 NYC results, deducting the 3 dB advantage the four ohm EV2181 has over the eight ohm TH-115, and looking at relative levels, both the EV and the DSL are quite close in response shape to the published spec sheets, other than the obvious room influence on both cabinets.

The EV spec sheet chart has a very small frequency axis compared to the DSL charts, which makes the EV look “peaky”, and the DSL “smooth”. When you compare the actual SPL levels, the TH falls off below the LF corner with a steeper slope than the BR cabinet, as it does in the NYC chart.

The TH-115 spec sheet chart, compared to the EV2181, both measured in half space are shown below the 2007 NYC chart .

Since the scaling (and cabinet impedance) difference of the charts makes visual comparison difficult, I have “done the math”, using the charts blown up large so that small differences are visible.
I could not find a chart for the eight ohm TH-115, so the TH-115-4 chart has the dB scale lowered by 3 dB since it was tested at 28.3 v rather than 20 v.

The EV2181, being a 4 ohm cabinet, should be 3 dB louder when driven at 2.83V than 2V (as it was at the NYC test) is shown in the “2183+3” column.

The difference between the TH-115 and the EV2181 spec sheets with 3 dB added ot the 2181 is shown in the “diff.SS” column.

The difference measured at the 2007 NYC test is shown in the “Diff.NYC” column using the measured difference and subtracting 3 dB.

The levels from the charts are rounded to the nearest dB.

The difference in the spec sheets totals 16 dB/9 =1.77 dB, if the spec sheets are correct, the 2181 would average 1.77 dB more sensitive than the TH-115 (average 98.888 dB at 2.83V to 100.666 dB at 2V from 20-100 Hz).

The difference in the 2007 NYC results total 43/9=4.77 dB, exactly 3 dB more than the spec sheets would indicate, the DSL spec sheet high by 3 dB from the comparative measured results.

If the TH-115-4 is not a full 3 dB more sensitive than the TH-115-8 (my guess is it is not, IIRC the B&C 15TBX100-8 is not double the impedance of the 15TBX100-4 ), the DSL TEF sensitivity (and max output) specs would be off by more than 3 dB.

 

[Ivan Beaver]

Re: Why Do TEF Systems read 3-5 dB high ?

If anyone would care to take up Ivan’s gracious offer, I’d be glad to pay for you and Ivan’s expenses.

It is very hard to compare traces in the “blue background jumble” (the blue one) from the 2007 NYC sub shootout that Mark Seaton provided which I posted previously, and the production changes in other speakers used there before do cloud comparisons.

Ivan, IIRC, you and Tom had the TEF data from the 2007 NYC sub shootout, if you still have it, could you please post the TH-115 and the EV-2181 2.83 V results?

Art Welter

Art. We do have a EAW AX box in house. Not sure of the model number off the top of my head.

Since EAW publishes the raw responses of the pass bands, I can take that and do a side by side to some Danely products and see where they stand.

I am REALLY swamped right now with work-so this will not be a calibrated test, just the same signal sent to both boxes and measured at the same distance. It will also have to be a ground plane measurement, as "doing it right" is kinda involved and takes time that I don't have right now.

While not a sub measurement-it should be a decent comparisom. The biggest issue is getting the AX to stand up straight-as all 4 sides are cut on angles.

 

[Art Welter]

Re: Why Do TEF Systems read 3-5 dB high ?

Art. We do have a EAW AX box in house. Not sure of the model number off the top of my head.

Since EAW publishes the raw responses of the pass bands, I can take that and do a side by side to some Danely products and see where they stand.

I am REALLY swamped right now with work-so this will not be a calibrated test, just the same signal sent to both boxes and measured at the same distance. It will also have to be a ground plane measurement, as "doing it right" is kinda involved and takes time that I don't have right now.

While not a sub measurement-it should be a decent comparisom. The biggest issue is getting the AX to stand up straight-as all 4 sides are cut on angles.

Cool, just use an A.D.A.M. (angular discrepency adjustment module, AKA audio log, AKA 2x4 with gaffer tape cover...) to prop it up.

Of course, EAW is one of the few companies that publish sub cabinet response charts in whole space rather than half space, which mixes mediums ...