GR Research Speaker Upgrade Review (Sierra-2EX V2)

[PristineSound]

Indeed he did. Electrically, that filter is inadequate.


It's not the tools. It's the methodology that is suspect at best. Gated measurements sans nearfield or ground plane stitch, overly smoothed, inadequate measurement suite.

You can do competent measurements with a Clio. Just... Not like this.

And, if you haven't watch the entire companion video end to end give it a watch or use AI to transcribe it. There a lot of issues Amir pointed out.

 

[a4eaudio]

....It's not the tools.

^^THIS!!
I knew it had been said a few times, but I even missed a few posts like Kimmo's #74. I wouldn't have posted my comment on it "not being the tools" if I had realized how many times it had already been said.

Is that meant to demonstrate that Danny's measurements are accurate?

It depends. Does the blue line look like something that is "broken" and "destroyed" and can't sound better than the yellow line? If so, does that hold true for the red line? When I look at the On-axis, LW, and PIR of what Danny did, I am pretty sure I know what he was trying to achieve. If he achieved what he was trying to achieve, why would I think his measurements are inaccurate? The CTA-2034 standards suggest +/-1.5dB tolerance across different measurement setups is good. On the other hand, Ageve did point out some data that does suggest maybe his calibration file is off, his mic has drifted, or as Amirm pointed out maybe its his amp.

It's the methodology that is suspect at best. Gated measurements sans nearfield or ground plane stitch, overly smoothed, inadequate measurement suite...

I think the statement above and some of the comments in the video miss the point that a tool can be used for different purposes and people are looking at it from different viewpoints.

(1) For passive crossover work: (a) You do not have to know the frequency response below 200-300Hz to tweak a crossover around 2-3kHz. (b) While I would PERSONALLY never use 1/3 octave smoothing for PASSIVE crossover work, I bet not one single component would change to Danny's mod if he used 1/6th, 1/12th, or 1/24th octave smoothing.

(2) On the other hand, for comparing completed commercial speakers in a standardized way, follow CTA-2034 (Standard Method of Measurement for In-Home Loudspeakers). Now including the full low-frequency extension down to 20 Hz is absolutely necessary. Smoothing at 1/20th octave is also the recommended minimum to compare resonances (even somewhat minor resonances) across speakers. Note, that some of Danny's completed speakers in fact do perform quite badly due to what seems like him ignoring the correct merging of low frequency and high frequency response - his methodology is flawed but that doesn't seem particularly relevant to his xo mod.

None of this is meant to support Danny's mod, just that the focus should stay on the objective evidence and Science. The distortion and chance of destroying a very good, expensive tweeter, is where I would focus.

 

[Ricardojoa]

^^THIS!!
I knew it had been said a few times, but I even missed a few posts like Kimmo's #74. I wouldn't have posted my comment on it "not being the tools" if I had realized how many times it had already been said.




It depends. Does the blue line look like something that is "broken" and "destroyed" and can't sound better than the yellow line? If so, does that hold true for the red line? When I look at the On-axis, LW, and PIR of what Danny did, I am pretty sure I know what he was trying to achieve. If he achieved what he was trying to achieve, why would I think his measurements are inaccurate? The CTA-2034 standards suggest +/-1.5dB tolerance across different measurement setups is good. On the other hand, Ageve did point out some data that does suggest maybe his calibration file is off, his mic has drifted, or as Amirm pointed out maybe its his amp.


I think the statement above and some of the comments in the video miss the point that a tool can be used for different purposes and people are looking at it from different viewpoints.

(1) For passive crossover work: (a) You do not have to know the frequency response below 200-300Hz to tweak a crossover around 2-3kHz. (b) While I would PERSONALLY never use 1/3 octave smoothing for PASSIVE crossover work, I bet not one single component would change to Danny's mod if he used 1/6th, 1/12th, or 1/24th octave smoothing.

(2) On the other hand, for comparing completed commercial speakers in a standardized way, follow CTA-2034 (Standard Method of Measurement for In-Home Loudspeakers). Now including the full low-frequency extension down to 20 Hz is absolutely necessary. Smoothing at 1/20th octave is also the recommended minimum to compare resonances (even somewhat minor resonances) across speakers. Note, that some of Danny's completed speakers in fact do perform quite badly due to what seems like him ignoring the correct merging of low frequency and high frequency response - his methodology is flawed but that doesn't seem particularly relevant to his xo mod.

None of this is meant to support Danny's mod, just that the focus should stay on the objective evidence and Science. The distortion and chance of destroying a very good, expensive tweeter, is where I would focus.

If you really wanted to know if danny mods sound better or not , the best thing to do is to get yourself a pair of ascend sierra 2 and the danny over mod and let us know how they sound. I bet danny didn’t know how his modded sierra 2 sounded either.

 

[dfuller]

And, if you haven't watch the entire companion video end to end give it a watch or use AI to transcribe it. There a lot of issues Amir pointed out.

Trust me, I have.

(1) For passive crossover work: (a) You do not have to know the frequency response below 200-300Hz to tweak a crossover around 2-3kHz. (b) While I would PERSONALLY never use 1/3 octave smoothing for PASSIVE crossover work, I bet not one single component would change to Danny's mod if he used 1/6th, 1/12th, or 1/24th octave smoothing.

The problem is that his measurements always have what are essentially response anomalies because of his gating and smoothing and he is doing "tweaks" for things that may well not be present in that sub-1000hz range. Remember, he uses a 4ms gate - Even with his wildly oversmoothed 1/3 octave magnitude response, he still loses significant resolution below about 1000hz, and especially below 500hz. And, part of the problem with smoothing that hard is that it creates erroneous information in the "noise". I've seen this happen first hand.

 

[a4eaudio]

If you really wanted to know if danny mods sound better or not , the best thing to do is to get yourself a pair of ascend sierra 2 and the danny over mod and let us know how they sound. I bet danny didn’t know how his modded sierra 2 sounded either.

I don't want to know. But I am also not posting curves and guessing how they might sound.

 

[amirm]

It depends. Does the blue line look like something that is "broken" and "destroyed" and can't sound better than the yellow line?

As I said, the changes don't create a borken or destroyed response. It simply is degraded in the very areas the likes of Danny praise.

 

[amirm]

When I look at the On-axis, LW, and PIR of what Danny did, I am pretty sure I know what he was trying to achieve. If he achieved what he was trying to achieve, why would I think his measurements are inaccurate?

You are telling us Danny aimed to achieve this???

No. He thought the highs were sloping down so boosted them. I and Ascend measured with different, mics, project setup, etc. and didn't see that.

As to PIR, it is rougher now in addition to boosted highs:

Finally, on-axis is screwed up in lower treble:

We are dipping to 79 dB at 3 or so KHz, with the reference at 200 Hz being about 83 dB. The stock response was even.

So I don't really know where you are going with all that. Danny is claiming to aim for precision. He took a speaker that was already there, and made it less precise because he doesn't know or care how to measure the speaker properly. And lacks the judgement to know that the stock response was already highly optimized. He also doesn't measure distortion so screwed things up there as well by messing with the crossover point.

 

[amirm]

I think the statement above and some of the comments in the video miss the point that a tool can be used for different purposes and people are looking at it from different viewpoints.

Sorry, this is misguided. Neither my video or the write about were created in a vacuum. We are discussing the measurement protocol as used in this scenario. Both our knowledge of measurements and specific data show clear mistakes in Danny's judgement of the speaker and supposed fix for it.

Saying that his tools could have been used to generate better results is orthogonal here. They were not used that way, nor does Danny acknowledge that there is anything wrong with the way he is using them.

These are the things that are motivating the comments for which there is no proper defense. Danny needs to sharply improve his measurement protocol and refrain from doing these "upgrades" without it.

 

[amirm]

(1) For passive crossover work: (a) You do not have to know the frequency response below 200-300Hz to tweak a crossover around 2-3kHz. (b) While I would PERSONALLY never use 1/3 octave smoothing for PASSIVE crossover work, I bet not one single component would change to Danny's mod if he used 1/6th, 1/12th, or 1/24th octave smoothing.

That's incorrect. I just watched a Klipsch speaker he modified. He completely changed the system design with respect to woofers, turning it into a 2.5 way speaker. He has done this a number of times. Such work better involve lower frequency response.

Further, even crossover work could benefit from longer gating time. Even without smoothing, his graphs would lack precision up to a few hundred hertz which could very well be the crossover frequency for a tower speaker woofer.

Also keep in mind that he publishes speaker measurements of his own the same way! Here is his $6,000 "Bully" speaker measurements:

See? It also starts at 200 Hz and his software indicating lowest usable frequency is even higher at 250 Hz.

This kind of approach resulted in this kind of disaster:

This was his advertised measurement:

You could take out the smoothing and it would still lack the proper resolution as the dip is at 200 Hz, which is below 250 Hz.

 

[amirm]

None of this is meant to support Danny's mod, just that the focus should stay on the objective evidence and Science.

Objective science says if you are going to measure a human hair's thickness, you don't use a school ruler. And certain not do so to criticize another person who has used high precision tools to measure the same. He clearly didn't know how to read Ascend's measurements and trusted his own testing to be superior. This is objectively wrong. He is owed no benefit of doubt here especially given his highly condescending attitude in both videos. I don't see how you can in any way shape or form defend how he goes about his work.

The man produces a new mod every other week for heaven's sake, criticizing some other company's design. I bit the bullet and bought the Klippel because I saw no reasonable way to conduct such testing. And here you go, chastising membership here? What else you want them to do? Send him a happy birthday card on his birthday?

 

[PristineSound]

Trust me, I have.

Ok, then why the back and forth rebuttal on the issues with Danny's measurement methods AND the tools he used to measure?

For others who haven't watched the companion video, here is a Gemini summary:

Based on the review, Amir identifies several critical flaws in Danny Richie's (GR Research) methodology and equipment that he argues undermine the validity of the speaker upgrades. Here is a comprehensive master list of the issues highlighted throughout the video:

Gated Measurement Limitations: Danny uses gated measurements in a standard environment, which forces a trade-off between reflection rejection and frequency resolution. Amir explains this results in a smoothed-out response that masks important acoustic details and limits data validity below 200 Hz (4:05-6:30).

Obsolete Filtering Protocols: Instead of the modern CEA-2034 industry standard of 1/20th octave resolution, Danny uses 1/3 octave filtering. Amir characterizes this as an antiquated practice that hides small resonances and defects (6:40-7:20).

Crude Sampling Method: Unlike Amir's $150,000 Klippel Near-Field Scanner—which takes 1,000 full-frequency sweeps to solve the wave equation—Danny uses a single-point microphone setup with limited supplemental positions. Amir deems this insufficient for creating an accurate 3D model of speaker radiation (9:30-10:50).

Lack of Distortion Testing: A major omission in Danny's process is the absence of any distortion measurements. Amir shows that these modifications cause the ribbon tweeter to exhibit 5-6% distortion, which he deems both audible and unacceptable (23:00-26:15).

Failure to Verify with Listening: Amir asserts that Danny does not perform listening tests on his modified designs. He further points out that Danny likely never physically assembled the crossover in the target cabinet, as the parts were too large to fit correctly (14:40-15:20).

Incorrect Microphone Calibration or Positioning: Amir suggests that the perceived high-frequency "droop" Danny attempted to "fix" may simply be a measurement error caused by an uncalibrated microphone, a missing calibration file, a broken microphone, or poor positioning off-axis (16:15-16:50).

 

[diddley]

This thread is no.6 on Google first page. I do hope this will slowly drives him out of bussiness, because the industrie needs honest and reliable men and women.

 

[dfuller]

Ok, then why the back and forth rebuttal on the issues with Danny's measurement methods AND the tools he used to measure?

Because, again, for what must be the 12th time in this thread:

It isn't the Clio!

It's not as cutting edge as the NFS, not by a long shot. But it is entirely capable of producing usable measurements. The issue is not the unit, it's how it's being used.

All the stuff with overly smoothing, gated measurements being used below their meaningful resolution limit, the inadequate and/or very poorly done off axis measurements, the improperly calibrated mic... that's all on the user.

The distortion measurements, I will give you that - but the reason Danny isn't doing that (beyond some very suspect reasoning!) is because you either need a very large space (Amir uses this; even so, you can see some minor interaction in the sub-200hz area) or an anechoic chamber to not have corrupting information from reflections.


IOW: the measurement equipment Danny does have is being used... Poorly.

 

[PristineSound]

Because, again, for what must be the 12th time in this thread:

It isn't the Clio!

It's not as cutting edge as the NFS, not by a long shot. But it is entirely capable of producing usable measurements. The issue is not the unit, it's how it's being used.

All the stuff with overly smoothing, gated measurements being used below their meaningful resolution limit, the inadequate and/or very poorly done off axis measurements, the improperly calibrated mic... that's all on the user.

The distortion measurements, I will give you that - but the reason Danny isn't doing that (beyond some very suspect reasoning!) is because you either need a very large space (Amir uses this; even so, you can see some minor interaction in the sub-200hz area) or an anechoic chamber to not have corrupting information from reflections.


IOW: the measurement equipment Danny does have is being used... Poorly.

Where did I say CLIO? Can you screenshot it for me?

I said measurement tools. There is something likely wrong with his mic or Danny is getting so senile that he cannot set a straight line right.

But now, I will also add that there IS something wrong with him using CLIO. . .for what he is trying to accomplish. Which is: measuring a speaker to find "flaws" to fix, speaker that was designed using a high resolution NFS.

So yes, I stand by my statement, for both CLIO or other tools, Danny does NOT have the right toolset to do the job he set out to do, especially for all the money that he claims he makes.

 

[a4eaudio]

Well, my original post was simply to state that is is NOT the tools that are Danny's problems, as several people were trying to claim. (In particular people suggesting that Clio was some outdated, inappropriate tool, which I didn't realize Kimmo had addressed in post #74.) I agreed that his methodology is the problem. I am not defending Danny or his results in any other venture...this thread is about his Ascend mod. I am PRO-MEASUREMENT and it does not further the cause to convolute valid arguments with less valid ones.

(And as I typed this, Dfuller has again stated the same point.)

As I said, the changes don't create a broken or destroyed response. It simply is degraded in the very areas the likes of Danny praise.

Note, I put quotation marks around "broken" and "destroyed". The reason I did that is because....You said "fundamentally broken" at 18:57 of the video and literally have the words "completely destroyed" in your screenshot in the video and again in post # 347 (the very next post after the quoted text above.) If I misinterpreted what you meant than I apologize for that, but I did not intentionally mis-state what you said.

You are telling us Danny aimed to achieve this???

With respect to ER or PIR, no, as I am pretty sure Danny does not take those into account.
With respect to On-axis, yes. Putting a 3dB dip at 3kHz and raising the response above 10kHz a little is very common. Note, that just because it is "very common" does not mean you or I have to agree with that. I wouldn't do it myself, but I know enough to know why Danny might think it is a good idea. The question is also not about whether the mod sounds better to you or me, Danny is a grifter and is creating a mod to sell to other people. We may not like that, but that is a completely different issue than he doesn't have the appropriate tools. (And also different than whether he knows how to use those tools accurately or not.)

No one has answered the question, but does the red line in post #337 look broken, destroyed, bad, not good, or whatever negative adjective you want to pick? Because it looks a LOT like Danny's blue line.

That's incorrect. I just watched a Klipsch speaker he modified. He completely changed the system

Post #349 is irrelevant to what I said, as my statement only applied to the Ascend mod. For the Ascend mod, with a crossover point of a two-way near 2-3kHz both of my statements are correct.

I don't see how you can in any way shape or form defend how he goes about his work

As I have tried to state in almost every post, I am not defending Danny, I think he is a grifter and charlatan. But I do believe in promoting objective measurements in audio, and false claims about Clio and other tools and overstating what is required does not advance that cause. This is not directed at you, just as my original post was not...to create a tweak to a crossover of a two-way speaker around 3kHz you do NOT need more than an inexpensive measurement mic with a good calibration file and a decent 2-in, 2-out audio interface. That also does not defend any other poor quality things that Danny has done in some other circumstance outside of the Ascend mod.

 

[amirm]

With respect to On-axis, yes. Putting a 3dB dip at 3kHz and raising the response above 10kHz a little is very common.

Not with him. He has repeatedly said he is aiming for flat on axis and "accurate" response.

 

[Keith_W]

Here's something else not yet mentioned. Whenever I take a measurement of a speaker, or if I am asked to comment on someone's in-room measurement of a speaker, I always look up published measurements and compare the measurement in front of me against others. Above a certain frequency, the measurements should be identical. This is an essential reality check! I do not believe that I am so "pro" at taking measurements that I can find things that the designers missed, and I certainly do not assume that random newbies on the internet posting REW measurements on ASR have taken their measurements correctly.

If I see something that deviates from published measurements ... the question is WHY. Maybe there is something wrong with the speaker. Maybe the published measurements are off (certainly a possibility ... if the published measurement was taken by someone like Danny!). But most likely, there was something wrong with the measurement and we need to identify what went wrong.

I don't think Danny does that. Otherwise he would very quickly see that his measurement is off. As mentioned, he isn't fixing a broken speaker. He is trying to fix his broken measurement by breaking the speaker. This is the same pitfall many of us who DSP might fall into, only that some of us have the humility to know that we may be making mistakes.

 

[MattHooper]

One important thing to note: your preference and your opinion do not extend to other people, because they are not reproducible. Reproducibility is one of the pillars of The Scientific Method.

One quibble with that: if you are a participant in the type of preference scoring of loudspeaker designs, eg conducted by Toole /Harman, then your preferences in aggregate with others can be extended to other people, in a highly predictive and reproducible fashion.

 

[Jim Taylor]

One quibble with that: if you are a participant in the type of preference scoring of loudspeaker designs, eg conducted by Toole /Harman, then your preferences in aggregate with others can be extended to other people, in a highly predictive and reproducible fashion.

Touche', mon ami!

 

[Dave Bullet]

I don't think Danny has ever done a quasi anechoic measurement to address the frequency range below his farfield gated measurements.

Anyone know why?