Side by side speaker tests - Lowther - Tannoy - TP1 - Worden Panasona

[david pinnegar]

might be of interest. For some time I've been measuring and testing different fullrange drive units and the acoustic treatment of them in different cabinets.

Attached are frequency responses of the Worden Panasona and the DX2 in TP1 London

Best wishes

David P

 

[gyraf]

..ruler-flat to beyond 20Hz? Do I read that right, or is it an err?
It would be quite interesting to extend measuring range below 20 to see where what happens - straight lines don't tell us much

 

[JohnRoberts]

Ruler flat at 20Hz tells us a lot but it can be difficult to believe.

JR

 

[david pinnegar]

I think that we're so used to augmented bass that we don't actually remember what flat bass sounds like. Back in the day Lowther used to specify the TP1 to be reaching the lowest frequencies and the bass horn is some 3 or 4 metres. With the Worden Panasona I've tried a number of different drive units and it's the Philips which really reaches audibly properly down into the bass.

 

[JohnRoberts]

Sub woofer design is a major pursuit in professional sound reinforcement. I have never seen a speaker frequency response curve that was ruler flat, without using a ruler.

Speaking about low bass and perception, customers are so accustomed to speaker distortion in the lower registers than clean flat low bass sounds thin to some. Commercial bass enhancers often just add in some LF distortion.

JR

 

[david pinnegar]

Well of course I have not measured distortion here.

Attached are some plots I've done of some commercial speakers and whether using a Behringer measurement mic or an AKG D224 I can't recall.

As a result I have used the Tannoy 611 as a reference for testing microphones but sometimes I get anomalous results in the bass possibly on account of phase issues between the dedicated bass and the bass-midrange cones.

As a result of these I suspect that there is something odd about the Tannoy DC2000 which should have its tweeter reversed.

I prefer using the D224 as Cardioid so as to eliminate 2/3rds of the room acoustic but of course without anechoic facilities one will always be subject to such influences. Other frequency responses however, made sense so I'm reasonably happy with the validity of the Panasona and TP1 graphs.

Best wishes

David P

 

[JohnRoberts]

Yup I've been inside both Bozak, and Peavey's anechoic chambers. Don't spend too much time in there.

I've seen the lads testing big dog loudspeakers out in the parking lot.

I saw a photo in one of my dad's old engineering notebooks of rig for measiring speakers out in the woods (back in the 30/40s). The microphone rotated on a pivot making a circle and avoiding standing waves or cancellations/reflections (i didn't say it was a good idea).

JR

 

[david pinnegar]

Yes - here's the plot of a DAP 15" and the 15" Tannoy bass in a Tannoy stage wedge. Bearing in mind that this is what I'd expect from that cabinet and the same for both drive units I believe that the flat graphs for the Worden and the TP1 London are not flawed.

Best wishes

David P

 

[abbey road d enfer]

I believe these graphs are produced from an FFT. It looks like the frequency resolution for those in post #6 is about 500Hz, which explains the flat response between 20 and 500 Hz.
No current speaker can go flat down to 20 Hz.
Those on post #8 show a better resolution. You must have changed the software settings in between.

 

[david pinnegar]

THANK YOU so much. Yes - upon checking the software was applying an impulse gating and needed resetting to Frequency Response Smoothing, which I've set on 8th Octave.

 

[moamps]

It would be very good to also state which equipment you used for measurement (what the measurement setup looked like), what distance the microphone was from the speakers, then which microphone was used and the characteristics of the room (or dimensions) in which you measured the speakers.

According to the frequency curves shown, it is obvious that there are certain problems in measurement (and then in listening experience) due to early reflections and standing waves at the microphone position, where a certain pattern occurs for all speakers except Tannoy. That bump at 100Hz got me interested in watching your video, and there I think I found an explanation, the speaker is placed on the floor.

Also the D224 is an extraordinary microphone (if it is still fully functional), but it is not IMO suitable for general measurements that can serve others as a kind of reference.
However, in your case, this measuring system can be very useful for optimizing the acoustic characteristics of the listening room, finding the optimal speaker position and sweet spot position for the listener.

 

[abbey road d enfer]

I've seen the lads testing big dog loudspeakers out in the parking lot.

That's how they did it at JBL and EV. Much easier in sunny Northridge, CA than in windy Buchanan, MI.

 

[david pinnegar]

Thanks

Most helpful.

Distance to speakers was around 6ft.

Of a number of D224s I've collected the one that I use for measurements was measured flat against the Behringer measurement mic . . . but using the erroneous software setting. I've been hunting through the measurements to find that test so as to apply the appropriate frequency response setting but without result and will have to do the comparison again. Thank you for pointing out my error as some measurements which were puzzling now make sense.

Attached are the graphs with the settings revisited. On a couple I've done measurements at speaker grill, 1m, 2m and 3m.

Having excited my own curiosity I'll be doing measurements with the same setup with Michael Wallis version Voigt Horn and Lowther Audiovector and together with the same example recordings to add to the YouTube video.

Best wishes

David P

 

[JohnRoberts]

This is so old I probably won't recall it correctly but the hot new thing for speaker measurements back in the 1980s was TDS (Time delay spectrometry). A time windowing technique that ignores room reflections.

Doing a web search for TDS gear reveals paid ads for water testing equipment. I did find an old AES paper about TDS.

JR

 

[JohnRoberts]

No current speaker can go flat down to 20 Hz.

With active EQ and much tweaking it is possible to develop significant output down to and below 20Hz, but even that will not make a ruler flat response plot.

For 20Hz response look to the sound reinforcement industry and expect large, heavy boxes.

This is a specialty woofer for lowest octave

The Danley DTS 20 Specifies a half power bandwidth -3dB @ 18Hz to 100Hz, with -10dB @ 16Hz. Thats how real speakers behave.

JR

 

[Gold]

Doing a web search for TDS gear reveals paid ads for water testing equipment. I did find an old AES paper about TDS.

JR

There is a compendium book with all the Richard C. Heyser papers. He was the inventor of TDS among many other things.

 

[abbey road d enfer]

This is so old I probably won't recall it correctly but the hot new thing for speaker measurements back in the 1980s was TDS (Time delay spectrometry). A time windowing technique that ignores room reflections.

Doing a web search for TDS gear reveals paid ads for water testing equipment. I did find an old AES paper about TDS.

JR

Most current loudspeaker test softwares or workstations (Klippel, CLIO, REW, NTI, Aurora...) use their own variation of TDS. The grandson of the original TEF machine is available from GoldLine as the TEF25.
However, reflections are ignored as long as the next obstacle is more than a half-wave away. That's why most do not produce adequate results below 200 Hz, in a typical room.
Using the parking lot still requires the use of a 30ft crane for proper 20 Hz measurements.
The alternative solution is either to bury the speaker in a pit for half-space measurement (see how the Danley is spec's in half-space), or use proximity measurements that must by post-processed.

 

[abbey road d enfer]

The Danley DTS 20 Specifies a half power bandwidth -3dB @ 18Hz to 100Hz, with -10dB @ 16Hz. Thats how real speakers behave.

That's how ONE speaker behaves.
Most of the subwoofers I measured that were supposed to go down to 30Hz miserably failed at 40 some.

 

[JohnRoberts]

I trust that Tom Danley's specs are accurate, that said we all wrestle with consumers who try to replicate product tests without using correct procedures, then accuse the manufacturer of lying. Power amps are another poster boy for inaccurate consumer testing.

Sadly after decades of success in the professional loudspeaker business, Tom is now getting into the consumer hifi speaker business because like Willie Sutton famously said when asked why he robbed banks, "because that is where the money is".

JR

PS: I recall when I was managing all product management for Peavey, my Loudspeaker product manager was in constant distress over competitors with less than honest specifications, while simultaneously dealing with Peavey's brand image that didn't get much respect even when we were the one telling the truth.

 

[david pinnegar]

In testing units I've been putting them into an infinite baffle lying on its back on the floor in the approximation to a half-space setup. The software is HolmImpulse which behaves very well but frustratingly assumed that the default Impulse Time-Window gating settings would give meaningful results. Luckily I've saved all measurements so I can revisit the software and extract the relevant frequency response graphs. This is one reason why I've done audio recordings of programme material such as the YouTube video so that one can hear differences rather than merely rely on what can be the arbitrary nature of graphs. Very interestingly I was testing a Visaton BG20 and the result bore strong resemblance to the published curve which is why I had not picked up on the glaring mistake.

Very many thanks to all with experience and advice.

Best wishes

David P