Tannoy System 600 Speaker Review

[Francis Vaughan]

But I don't think a 10v test or any other voltage-based test is the right method. We don't listen based on voltage level. We listen based on output. Therefore, if Amir is to do further multi-level non-linear distortion testing then stick with the basis for which we listen: dB. Distortion at 90dB and 96dB at 1m equivalent. 102dB would be an optional test if you want to push things a bit more.

Absolutely. Not doing the tests based upon output level simply advantages speakers with lower sensitivity. Which is useless.

No-one would consider a test for an amplifier based only upon input level as reasonable. Distortion versus output has never been questioned, so the same should be true for speakers.

ETA. I see Amir has already answered. But even a rough tweak to the drive levels once the sensitivity is known would be enough. We can assume that the distortion is reasonably insensitive wrt small changes in input level, so the results will be valid, even if the levels are a little bit off.

 

[DeeJay]

I agree but that requires more work and licensing cost to do....

Can you explain why licensing cost?

 

[amirm]

For example, see here:
https://www.erinsaudiocorner.com/driveunits/audiofrog_gb25/

That's not the correct example though. The distortion graph I showed is relative to the SPL, not absolute level as you are showing there. So the SPL differential is factored out.

 

[amirm]

Can you explain why licensing cost?

The Klippel hardware I have has most of its software features separated and sold at pretty high costs. My original intent for this system was frequency response measurements so have those licenses. Plus the underlying basic which allows me to produce the graphs you see. They have optional software modules that can keep SPL constant and produce distortion, and a bunch of other fancy graphs related to it. All of that costs money, running into thousands of dollars for software licenses. Given the fact that distortion plays a distant roll to frequency response when it comes to speaker preference so I can't justify spending a ton of money for those features.

That said, I will keep working on getting some of those capabilities within what I have licensed.

 

[hardisj]

That's not the correct example though. The distortion graph I showed is relative to the SPL, not absolute level as you are showing there. So the SPL differential is factored out.

The notion is the same. The math of delta fundamental vs component will give you percent. For example, -30dB from fundamental is 3%. -40dB = 1%.

But, here’s another example of a multi-level test I did using Klippel’s TRF module where the math of fundamental vs component is provided as %. Klippel gives you the option of how you provide the data. You can provide it both ways if you want.

https://www.erinsaudiocorner.com/driveunits/jl-audio_c5-400cm/

 

[amirm]

Klippel gives you the option of how you provide the data.

I know it does. I have shown it in most of my reviews. My interest at this point in distortion metrics is not to look at them for the sake of distortion but as a diagnostic tool to figure out if there is something fundamentally wrong. And correlate that with frequency response measurements. For this, it has been working well.

 

[DeeJay]

The Klippel hardware I have has most of its software features separated and sold at pretty high costs. My original intent for this system was frequency response measurements so have those licenses. Plus the underlying basic which allows me to produce the graphs you see. They have optional software modules that can keep SPL constant and produce distortion, and a bunch of other fancy graphs related to it. All of that costs money, running into thousands of dollars for software licenses. Given the fact that distortion plays a distant roll to frequency response when it comes to speaker preference so I can't justify spending a ton of money for those features.

That said, I will keep working on getting some of those capabilities within what I have licensed.

Thanks for the clarification. Can you set a custom voltage level for all measurements with current license?

 

[amirm]

Thanks for the clarification. Can you set a custom voltage level for all measurements with current license?

Sure.

 

[DeeJay]

Sure.

So after measuring with 2.83V you could calculate mean SPL value based on FR from 250Hz to 6kHz, calculate voltage for SPL difference (eg. 100dB - mean SPL) and redo measurements with calculated voltage. That way we could see correlation between distortion and FR at standard input voltage, "realistic" SPL and a clue of power linearity. I know it's more work for you. Sorry.

 

[amirm]

Repeating what I already know is not helpful. What is helpful is showing how the extra effort is worth it at the expense of measuring another speaker. And remember, it is not just time to measure but capture the results, format, annotate and publish. And field questions about it like I am doing now.

 

[Prana Ferox]

While the discussion of distortion vs SPL is interesting (and probably should be in its own thread) in this case the change seems to be only from port compression, which is related to Amir's comment about the noticeably high port velocity / turbulence.

 

[QMuse]

Exceptionaly smooth directivity indexes allow for effective EQ. This filter variant has been optimised for flat on-axis response:

On-axis:

LW & PIR:

 

[Kyle / MrHeeHo]

Sound and Vision as well as stereophile have measured different Tannoy speaker models and they had equally atrocious measurements. I think it's safe to write off Tannoy speakers https://www.soundandvision.com/content/tannoy-dimension-surround-speaker-system-measurements

https://www.stereophile.com/content/tannoy-churchill-loudspeaker-measurements

 

[DeeJay]

Repeating what I already know is not helpful. What is helpful is showing how the extra effort is worth it at the expense of measuring another speaker. And remember, it is not just time to measure but capture the results, format, annotate and publish. And field questions about it like I am doing now.

Aren't you trying to show which speakers are well designed? Measurements that you currently do can show which speakers are not but that does not mean others can automatically be qualified as good. For that you need to do more. Eg. power linearity (who listens at 2.83V?), acoustical phase response/distortion (very important especially at crossover points). So there are speakers that perform well on your measurements but will fail on measurements that you don't do. So, IMO extra effort is worth. I know it's a lot of work for you but I would always choose quality over quantity.

 

[MZKM]

@Kyle / MrHeeHo
Conversely, none of the Tannoy measured that I have seen have such an egregious dip in the crossover region, especially when the horizontal/vertical performance is amazingly controlled in that region. That coupled with the fact that the frequency response error is much higher than spec (and they state using an anechoic chamber) and that it exceeds the distortion rating at just ~2W leads me not wanting to add it to the master list of preference ratings. I also can’t find an MSRP for it (going for <$500 used for the passive model).

That said:
Preference Rating
SCORE: 1.6
SCORE w/ subwoofer: 4.5

https://docs.google.com/spreadsheet...oXRw9QbVq3i4wIquIpjHd3eSeS8qaO7Rtk6rW/pubhtml

 

[amirm]

Aren't you trying to show which speakers are well designed?

As it relates to listener preference, yes. And that, based on every research shows that THD is not a factor.

In listening tests, I comment on power handling.

I know it's a lot of work for you but I would always choose quality over quantity.

Quality is superbly there in the spinorama measurements. They are state of the art and comply with the best of what we know to characterize listener preference. Your request is to pile on more quantity with little value relative to work already being done.

It is not like we are sitting on data from 1000 speakers already to spend time drilling down on a few. We need to get good coverage of performance from all the major brands to know who is and who is not building good speakers.

 

[QMuse]

Aren't you trying to show which speakers are well designed? Measurements that you currently do can show which speakers are not but that does not mean others can automatically be qualified as good. For that you need to do more. Eg. power linearity (who listens at 2.83V?), acoustical phase response/distortion (very important especially at crossover points). So there are speakers that perform well on your measurements but will fail on measurements that you don't do. So, IMO extra effort is worth. I know it's a lot of work for you but I would always choose quality over quantity.

I don't agree. Current measurements are done according to CEA-2034 standard which is based on parameters proven to be the most relevant for SQ.

 

[KSTR]

I own the active variant of those speakers (SYS600A) and those are much better with regard to frequency response, I found the published pretty flat FR to be correct, notably zero issues in the XO region.

Tannoy aways seem to have skimped on XO in their passives, though. I own the bigger brother SYS1200 passives, too, and their stock XO is definitely very compromised, though not as bad as shown here. I'm using a DSP XO on those now.

The 600As were lacking in other regards than FR. Port noise, rear panel not air-tight at the connectors/switches, electrical hum/buzz from bad power-supply design... also, while the XO summed flat it wasn't phase-coherent. In the end I ripped out the drivers and built a completely new (sealed) active speaker.

My take at Tannoy is, being a big fan of coaxials: Use the excellent drivers and build something new with it (notably a new XO).

 

[QMuse]

@Kyle / MrHeeHo
Conversely, none of the Tannoy measured that I have seen have such an egregious dip in the crossover region, especially when the horizontal/vertical performance is amazingly controlled in that region. That coupled with the fact that the frequency response error is much higher than spec (and they state using an anechoic chamber) and that it exceeds the distortion rating at just ~2W leads me not wanting to add it to the master list of preference ratings. I also can’t find an MSRP for it (going for <$500 used for the passive model).

That said:
Preference Rating
SCORE: 1.6
SCORE w/ subwoofer: 4.5
View attachment 53312View attachment 53313View attachment 53314View attachment 53315View attachment 53316

https://docs.google.com/spreadsheet...oXRw9QbVq3i4wIquIpjHd3eSeS8qaO7Rtk6rW/pubhtml

Can you calculate the Preference rating of the EQ-ed version if I send you the filter?

 

[Vuki]

Sound and Vision as well as stereophile have measured different Tannoy speaker models and they had equally atrocious measurements. I think it's safe to write off Tannoy speakers https://www.soundandvision.com/content/tannoy-dimension-surround-speaker-system-measurements

https://www.stereophile.com/content/tannoy-churchill-loudspeaker-measurements

Although tested speakers were plain catastrophy, churchills don't look too bad. Pretty flat, wide bw and high sensitivity- all good