Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Lee Buckalew]

Im saying that a single peak in the response of the cabinet shouldn't be used to determine the rated specs.

It would be misleading to use the 116db peak at 68hz to determine frequency response just as it would be misleading to use that same peak to determine the sensitivity.


Sent from my iPhone using Tapatalk

So what you would like to see, and would consider to be "accurate" would be a smoothed response measurement (since that/those peaks should get eq'd out anyway) but you also think it's not misleading to fail to disclose that smoothing?

If the best representation is to show intended usage then a processed curve should be shown.  A number of manufacturers show both, a raw trace and a processed trace.  This would seem to me to be the best of both worlds.  Here is the raw info, here is processor setting 'X' so that you can see what is happening to get from point A to point B.

Lee

[Scott Carneval]

Did you guys check your logic and critical thinking at the door before you entered this thread?

I'm not sure how many different ways I can say this: A 6db peak should not be used when determining the rated sensitivity of a device. ALSO, that same 6db peak should not be referenced when determining the -3 or -10 response curves.

The -3db points have to be tied to something. To say that the box is 3db down at a certain frequency is meaningless unless it's -3 from a certain reference point. If either of you have a belief to the contrary I would love to hear it



Sent from my iPhone using Tapatalk

[Ivan Beaver]

What on earth does the rated sensitivity have to do with frequency response? When specifying the frequency response of a device, whether it's a speaker or electronics, you have to provide a spec which describes the level range that your specified frequency response fits within. If that level description is "-3dB" I read that to say the lowest level is within 3dB of the highest level. In this particular case I see the maximum level being 116dB at about 68Hz, and the minimum being 105dB at about 35Hz, with other minima and maxima at 115Hz and 26Hz. That looks like ±5dB, or -10dB to me.

I'm happy to be shown how I am misinterpreting this, but to me this is a misleading spec from a company that argues that every other manufacturer uses misleading specs but theirs are not.

Mac

-3dB has to be 3dB  down from SOME point.

If you want to use the -3 as 3dB down from some peak point, then MANY MANY (ie most) products are going to look pretty bad-ESPECIALLY if you look at the non processed response.

For example- one VERY POPULAR monitor (that claims very high output SPL) would have -3dB point in the range of around 1Khz to 2.2KHz)

Why would you EVER buy that product?

The reason for the -3dB points is that the claim of the really high SPL comes from a LARGE peak (10dB above the rest of the freq response of the cabinet) around 1400Hz.

Of course once processed that peak is gone-but it ALSO will not get as loud now-the max SPL is 10dB lower-but that does not look good on the spec sheet.

In fact-using that same "criteria" most subs would have a -3dB  ABOVE 100Hz-because they are rising in sensitivity above 100Hz.

Of course that is EXACTLY WHERE some manufacturers get their high sensitivity numbers from -ABOVE 100HZ=but they don't show a graph showing it.

You HAVE to use the same standard or else it can be meaningless.

You can ALWAYS pull a peak down.  But boosting the low freq at max levels will result in things that are not "pretty"-sonically.

And if you want to go down the 3dB from peak game- you HAVE to specify what smoothing is applied to the graph.

If you remove smoothing from the response of a loudspeaker-it is ALL OVER THE PLACE-ALL LOUDSPEAKERS

So now you have a very narrow freq response.

I would make this argument, if you are going to hold Danley to -3dB specs from the peak in the response- THEN ALSO hold EVERYBODY ELSE to the same standard-and see what you end up with.

Simply LOOK at the UNPROCESSED freq response of other products and THEN start to make conclusions based on that.

[Ivan Beaver]

but Ivan constantly insinuates that every other manufacturer deliberately creates misleading specifications, and I find these specifications to be very misleading. If I was going to interpret that particular graph I would say it is ±5dB 25Hz-170Hz (hard to call the upper freq in the graph), and to me, "-10dB" is the same as ±5dB.

Mac

I HAVE NEVER said that every other manufacturer creates misleading specs.

Some are quite accurate.

But others-hover are EXTREMELY misleading.

ESPECIALLY when they provide NO graphs or data to support the "simple spec"  numbers.

I have ACTUALLY MEASURED some of the products and they don't come close-probably the reason they don't show any graphs.

Your conclusion IS EXACTLY why we include the graph (as I stated earlier), so that YOU-the user can come up with your own ideas of the performance-based on that data.

But YOU HAVE to compare under the same standards the other products.

[Peter Morris]

The question I asked really was, what would be different about the quoted specs if the crest factor was the same for each measurement, at the 3 different load impedances. What it looks like to the uninitiated to world of shifting specs is that the testing was fudged to make the amp appear better. If the amp is intended for a given music style, the crest factor will not change depending on the load impedance on the amp, the end user might want to know how the amp performs with a given crest factor at all impedances. I would.

As a topic swerve, in Danley's speaker specs, what does a frequency response at "-3dB" imply? In the spec sheet for the BC218 for instance, which should be a recent, up to date measurement, the "operating frequency range" is stated as "26Hz - 150Hz -3dB". Looking at the frequency response plot for a single speaker in half moon configuration (the flatter of the 2 configs) it seems like there is a peak at 68Hz of about 116dB, and at 35Hz a minimum of about 105dB. That looks like a 10dB difference to me, so where does "-3dB" come from. There are similar seeming anomalies on many of your spec sheets.

Mac

Looks like 10 dB to me as well …. and the 116dB is at a nominal impedance of 4 ohms at 2.83 volts.

It’s not the same as the 1 w 1m spec that other manufactures use.  You have to be carefully when comparing Danley sensitivity spec’s to other manufactures.

With respect to the amplifier's rating - if you are going to quote the output with unusual crest factors I think you need to publish a plot like this:-

[Ivan Beaver]

It’s not the same as the 1 w 1m spec that other manufactures use.  You have to be carefully when comparing Danley sensitivity spec’s to other manufactures.

Danley uses the 2.83V because it is the only ACCURATE way of stating sensitivity.

While many are ignorant of this- NO LOUDSPEAKER manufacturer applies 1 watt to the loudspeaker.  You CANNOT apply 1 watt-due to the varying impedance.

You can ONLY apply a VOLTAGE that is EQUAL to 1 watt into SOME impedance.  8 ohms is 2.83V.

And since the impedance IS NOT equal at all freq, that voltage would HAVE to be changing with freq to get an actual 1 watt applied to the loudspeaker.

So follow along here for this REAL LIFE situation.

Let's say you have a sub that is rated for 4 ohms-that would be 2V applied to get 1 watt.

But what if the ACTUAL impedance was 5 ohms? (as in the case of our TH115).

Do you apply 2 V OR the voltage that would produce 1 watt into 5 ohms?

If we applied that voltage-then the ACTUAL sensitivity would go up about 1dB-along with the max SPL numbers.

BUT THEN-I would be having a discussion with somebody why the "math" for a 4 ohm subs did not match the numbers on the spec sheet.

Why do we call a 5 ohm speaker 4 ohms?  The answer is VERY simple.  It reduces the time we have to spend on the phone with people asking what amp to use because none of the amp manufacturers provide 5 ohms specs.

WE ALREADY WENT DOWN THIS ROAD with the SH50.  Is is closer to a 6 ohm load.  I got tired of the phone calls-so we changed it to the lower rating (taking the conservative approach again) and the phone calls stopped.

We only rate the impedance in standard numbers (2-4-8-etc), BUT WE ALSO show the ACTUAL impedance graph. so people can make up their own minds.

The same thing applies for the SM80.  THe actual load it presents to the amps is closer to 10 or 12 ohms.  But it is rated for 8 ohms.

If we used "the math" for power capacity and voltage etc, we would end up with a higher power capacity than what it is actually rated at-if the rating was staying at 8 ohms.

So the max SPL numbers would go up.  But we (I) decided to keep it conservative-and went with "math" that would assume it was 8 ohms.

[Ivan Beaver]

So what you would like to see, and would consider to be "accurate" would be a smoothed response measurement (since that/those peaks should get eq'd out anyway) but you also think it's not misleading to fail to disclose that smoothing?

Just to help inform people, there are several different ways to "smooth" a freq response.

The first is to apply a certain amount of smoothing to a graph. 

This is what people normally expect.

HOWEVER- you can have a very smoothed graph WITH NO smoothing applied after the measurement.

This is simply done by reducing the data points that were used to get the graph.  With fewer data points, the peaks and dips get "absorbed" into the average.

So some people like to say they applied little or no smoothing-but yet there was nothing to be smoothed in the first place.

If you look closely at the Danley freq response graphs, at the very top will will see the data resolution used to get the graph.

In almost every case (some early models it may be different), you will see that I tried to go a full decade below the expected "-3dB" point. 

So for a 40Hz box the resolution is 4 Hz.  For a 30Hz box it is 3 Hz.

The finest resolution I can get with TEF (Danley measurement system of choice) is 2Hz.  So for a 10Hz box like the DTS10-it only 2Hz resolution.  Which is not enough to accurately describe what is happening down at 10Hz.  But is it all we have.

So ONCE AGAIN- you HAVE to look closer at the data you are given to see what you are really dealing with.

If all you see is a graph that was generated on a computer and looks nice and smooth-with no details, who knows where it came from.

Danley does not try to hide anything-and how we get the data-the resolution etc is out in the open.

People should ask the same of other manufacturers.

Since there are no standards that manufacturers MUST abide by, each manufacturer is completely free to state whatever they want to-true or not.

But the REAL problem I see is not the manufacturers lying-but as them not telling the TRUTH-as YOU (the consumer) would like to believe.

For example.  With subs.  VERY OFTEN there is a peak in the freq response much higher than the "sub" freq range.  I call that below 100hz.

So they will make the "max SPL" claim at this high freq.

Will the speaker actually produce this SPL-YES.  So they are NOT lying.

But what YOU want to know is how loud will it go in the INTENDED freq range of usage.

That max SPL number given IS NOT for that range-so you do not get the performance you "thought" the spec was giving you.

The same thing goes for full range cabinets.  Unprocessed may cabinets with a large diaphragm HF driver will have a LARGE peak around the 1k-2K range,  This can result in some VERY IMPRESSIVE SPL numbers.

But you would NEVER use the cabinet without that peak being brought down 10dB or so-to equal the rest of the freq response.

So while the cabinet will actually PRODUCE that SPL, it is unusable in terms of the rest of the freq response.

BUT YET people will buy the cabinet with the high SPL simply "thinking" that the WHOLE response will equal it.

Of course the people who "play those games" do no actually show any data to support the claims- you are just supposed to "believe" them because they are so "famous and accepted"-how could they lie?

Well all  you have to do is to buy one of the cabinets and ACTUALLY TEST IT yourself and see.

But most people just want to "believe", rather than actually confirm.

Lying is one thing-deception is quite another.  Most people are deceived, but then later think they were lied to.

The truth is not always as we would like to believe.

[John Roberts {JR}]

Looks like 10 dB to me as well …. and the 116dB is at a nominal impedance of 4 ohms at 2.83 volts.

It’s not the same as the 1 w 1m spec that other manufactures use.  You have to be carefully when comparing Danley sensitivity spec’s to other manufactures.

I leave the speaker plot wrangling to the last honest man in the speaker business. 

With respect to the amplifier's rating - if you are going to quote the output with unusual crest factors I think you need to publish a plot like this:-

The "MAX" power vs. time power plot while better than many amp specs, it still reveals more about power supply design than actual amp performance with complex waveforms.  Yes, the amplifier can only put out huge burst power at low impedance very briefly but the real questions is how much burst power X can the amp make for Y time long while simultaneously putting out Z continuous average power (like real signals).  This is far from simple and X, Y, and Z are arbitrary, but as I mentioned at least one old IHF test (dynamic headroom) provides a framework that could be a starting point to compare platforms with and present numbers we can somewhat understand.

We are still operating in a "trust me" mode, and I do not accuse manufacturers of deviousness, as much as showing their products, that they designed to meet their customers known needs as well as they could, in the best possible light.

JR

[Scott Carneval]

It seems like there's some confusion as to what the -3 and -10 ratings actually represent.  These ratings do NOT claim that the speaker is FLAT within +/- 3db and/or +/- 10 db for the entire range.  Look at ANY manufacturers spec sheet and you'll see that this is not the case. In fact, the rating is usually given as MINUS 3 or MINUS 10, not +/-3db/10db. It's intended purpose is to give the operating range of the speaker. The -3 rating designates the lowest or highest point at which the response of the speaker is still within 3db of the overall response of the speaker.  How to determine the 'overall response' of the speaker is a somewhat gray area, but you HAVE to choose SOMETHING, and it only makes sense that that SOMETHING is also used to give the sensitivity rating of the speaker.  That way you can look at the rating and know that the speaker can produce 110db, for a range of 26-150hz, and be within 3db of that 110db rating for most, if not all, of the range.  But the bottom line here is that you need to look at the graph and determine for YOURSELF if the speaker is appropriate for YOUR application.

The litmus test, in my opinion, is how does the BC218 perform at 26hz, as that is the fundamental purpose of the -3db ratings.

[Mac Kerr]

I would make this argument, if you are going to hold Danley to -3dB specs from the peak in the response- THEN ALSO hold EVERYBODY ELSE to the same standard-and see what you end up with.

I am not suggesting holding Danley to a -3dB range from peak, I am saying it is disingenuous to describe the frequency responses with -3dB from the sensitivity when there are peaks in that range 7dB above that. While I make no claims that any manufacturer uses any specific method to arrive at their numbers, the common format for frequency response is a frequency band limited with a level range with a maximum and minimum. On the speaker I used as an example to start this thread that would be ±5dB since it all fits in a 10dB level window. People will have to make their own judgements as to the veracity of any particular manufacturer, but I am will to give them the benefit of the doubt until demonstrated otherwise.

Mac

[ProSoundWeb Community]