Erin's review for the March Audio Sointuva

[kimmosto]

If we look at the ‘perfect’ 9.6/10 speaker we can see indeed that constant directivity will be punished.

That's neither 'perfect' nor constant directive. Just some composed response set giving 9.6.
Preference rating calculated with equation 9 detects on-axis slope indirectly via NBD_ON because NBD is sensitive to slope. Also SM is sensitive to slope so equation 9 prefers response slopes based on some weighted average included indirectly (by mistake?) in NBD_ON, NBD_PIR and SM_PIR. As already told quite a few times somewhere else, this is one reason why equation 9 alone is invalid for speaker ranking, and should not be used on this forum. It biases subjective impressions, opinions and manufactures. At least I don't want to see new Volskwagen-gate due to pig's hole in test system encouraging manufacturers to cheat.

It's tempting to tilt on-axis up to get more linear PIR (and SP) to get better rating on ASR, but usability in practice depends on other things such as acoustics, listening setup (dimensions and angles), features/quality of drivers and radiators, feeding electronics, favorite records and of course personal tolerance and preference. I don't do that anymore in my designs because hot HF is not sustainable decision no matter relative directivity index at HF.

On more thing. Sointuva is not necessarily designed to have step in PIR and too much highest octave as in measured individual. Unfortunately manufacturing tolerance of drivers is not always what manufacturers hope/expect. This applies also to Purifi and SB Acoustics. Speaker individuals can have e.g. 2 dB difference in LF/HF balance and on-axis response at midrange so LF and HF drivers of single speaker (and both speakers of the pair) must be matched and tested before sending to customers in order to maintain "designed" sound balance.

 

[restorer-john]

This is already post revision, Erin worked quite a bit with March Audio to get it this good.

Oh, OK. That's unfortunate.

I figured we had a genuine, warts and all, honest review here, not a bounce back and forth, get-it-right before publication thing.

 

[restorer-john]

e.g. 2 dB difference in LF/HF balance and on-axis response at midrange so LF and HF drivers of single speaker (and both speakers of the pair) must be matched and tested before sending to customers in order to maintain "designed" sound balance.

Basically, that means the QC is garbage, if that is the case. 2dB is an awful lot and enough to reject-bin most drivers.

 

[abdo123]

That's neither 'perfect' nor constant directive. Just some composed response set giving 9.6.
Preference rating calculated with equation 9 detects on-axis slope indirectly via NBD_ON because NBD is sensitive to slope. Also SM is sensitive to slope so equation 9 prefers response slopes based on some weighted average included indirectly (by mistake?) in NBD_ON, NBD_PIR and SM_PIR. As already told quite a few times somewhere else, this is one reason why equation 9 alone is invalid for speaker ranking, and should not be used on this forum. It biases subjective impressions, opinions and manufactures. At least I don't want to see new Volskwagen-gate due to pig's hole in test system encouraging manufacturers to cheat.

It's tempting to tilt on-axis up to get more linear PIR (and SP) to get better rating on ASR, but usability in practice depends on other things such as acoustics, listening setup (dimensions and angles), features/quality of drivers and radiators, feeding electronics, favorite records and of course personal tolerance and preference. I don't do that anymore in my designs because hot HF is not sustainable decision no matter relative directivity index at HF.

On more thing. Sointuva is not necessarily designed to have step in PIR and too much highest octave as in measured individual. Unfortunately manufacturing tolerance of drivers is not always what manufacturers hope/expect. This applies also to Purifi and SB Acoustics. Speaker individuals can have e.g. 2 dB difference in LF/HF balance and on-axis response at midrange so LF and HF drivers of single speaker (and both speakers of the pair) must be matched and tested before sending to customers in order to maintain "designed" sound balance.

I was strictly talking about whether the model punishes a particular directivity pattern more than others. I was not talking about whether a higher score would translate into better subjective impressions.

In that sense (correlation to real subjective impressions) the model is severely lacking, I would not conclude any score difference lower than 2 points as a significant indicator of one speaker being better than the other.

Yet again, it's the only thing we got that is some what objective and backed by blind listening impressions.

As long as the market demand remains non-existent and there is no interest or research to come up with better models this is all what we have.

 

[abdo123]

Oh, OK. That's unfortunate.

Is it? At the end of the day the consumer is getting a better product.

 

[Rick Sykora]

This is already post revision, Erin worked quite a bit with March Audio to get it this good.

I think anything different would be a regression imo.

Yes, at this point, anybody buying a Sointuva is going to have to make adjustments as-is. Whether toe-in or eq and/or something else, it is what it is.

IME, looking at my measurements and comparisons from many others, they may be close, but are usually not identical. This may be due to many circumstances. Differences in equuipment, conditions or even (as @kimmosto mentions) driver unit variations and crossover part tolerances may cause discrepancies. Unless you match drivers, tune crossovers to match and test to verify, you may be close, but you may not either. This is why I said am pretty pleased when I get results that are close to matching others.

 

[restorer-john]

Is it? At the end of the day the consumer is getting a better product.

Pig's ass.

When it is done in the public domain where people have already plunked down their money and got completely different cofigurations, sure, that's very unfortunate. Unwilling Beta testers.

Count me out of that chit. I'll take proper engineering, with prior research, time in anechoic chambers and products that are ironed-out before being sold.

Bear in mind, this speaker is expensive. Fuggin' expensive for a little 2 way speaker. In fact, it's ridiculous. It needs to be earth-shakingly awesome and practically perfect.

 

[abdo123]

Pig's ass.

When it is done in the public domain where people have already plunked down their money and got completely different cofigurations, sure, that's very unfortunate. Unwilling Beta testers.

Count me out of that chit. I'll take proper engineering, with prior research, time in anechoic chambers and products that are ironed-out before being sold.

Bear in mind, this speaker is expensive. Fuggin' expensive for a little 2 way speaker. In fact, it's ridiculous. It needs to be earth-shakingly awesome and practically perfect.

I didn't know the speaker was being sold before the review came out. And i'm not sure if that's the case.

Eitherway, I think the people who already bought it before the review would appreciate a free crossover upgrade. So all in all the choice to improve the design is good for everyone involved.

Whether it was incorrect of people to buy products from an unknown manufacturer with limited resources is an interesting point, but it's their choice i guess.

If the speaker was indeed being sold before the review was published then I agree with the concern that @hardisj should only review products with the state that they're currently in or the state they're received. it's a bit malicious to report the performance of something when the product in review is obviously different than what exists in the market. I really think in my opinion if he wants to improve other people's designs then he should get paid for his time instead of publishing it as a 'review'.

 

[buz]

At the very least disclose you worked on it, it's an obvious conflict of interest. I have no problem with the way crinacle does it with those IEM collaborations but to publish it as a regular review is rather questionable.

At least it says he tested a version different from the retail edition, whatever that implies.

 

[hardisj]

If the speaker was indeed being sold before the review was published then I agree with the concern that @hardisj should only review products with the state that they're currently in or the state they're received. it's a bit malicious to report the performance of something when the product in review is obviously different than what exists in the market.

At the very least disclose you worked on it, it's an obvious conflict of interest. I have no problem with the way crinacle does it with those IEM collaborations but to publish it as a regular review is rather questionable.

(bolded so no one misses this and the questioning of my integrity can end right here)

You guys act like I haven't discussed this already. I did. In detail. In two different threads. On December 9, 2020. In this post on the Sointuva discussion thread and cross-posted in this thread's post. TWO different places.

(@abdo123 , looks like you "liked" that post in the second link, btw)

I even posted about it on my YouTube page which @VintageFlanker provided a screenshot of in the same linked threat (here). Oh, and my Facebook group page. So, four places total. Three of them are wide open for everyone to see. It was always disclosed.

Spoiler: See? Told ya'.

Update on the March Audio speaker.

The speakers sent to me were the initial prototype speaker which are the same as the final design just more 'broken in' and less a final crossover mod (discussed below). I was completely fine with that. No sense in Alan building me a brand new pair just for review that he'd then have to sell at a discount as they had been used. Logical.

When I first listened, something sounded off. Remember, I test in stereo and mono for different reasons. This is important. I took some notes and moved on to the measurements.

I tested one and the results looked really out of whack so I contacted Alan. Long story short, since this was the prototype, the inductor hadn't been zipped-tied in place and Alan forgot to do that before he shipped it and it was knocked loose in transit. I verified by disassembling the speaker and sure enough... wiggle, wiggle. Rather than fix it right away, I put the other speaker on the stand. And... wow. Good wow.

I sent Alan the measurements so he could see for sure that the two measured different and we could tie up that loose end. The (good speaker's) measurement was as he expected so no news there. But, let's visit this quote from earlier...

Received some more info about the speaker - tweeter is indeed SB Acoustics TW29BNWG-4 and there are some preliminary free field measurements made with the following note from Alan:
"Please note that the small lift in response above 9kHz has already been corrected by a crossover tweak. Otherwise its essentially +- 1.5dB. Listening window is very close to the on axis. As with the previous version F3 is around 45Hz and F6 around 38 Hz. In room the bass extends below 30Hz."

Alan knew what to expect with my (good) result and had already accounted for a HF lift in the final crossover (that wasn't in the prototype board). Knowing that I was going to have to fix the busted crossover (for my listening tests) I thought, "hey, why not just have Alan send me the updated crossover so I can test what will be the true final version". So, rather than fix it, I asked Alan to send me the updated crossovers and I'll test the speakers with those.

Stay tuned for the final results. I'm not sure when the new crossovers will get here but I'll post and update when I get them.

In the meantime I can tell you that the measurements which have already been shared are really quite indicative of what I measured. I haven't overlaid them but I'd say to my eye they're practically the same. This might be one of the best measuring speakers I've seen so far, even with the current HF tilt. The sound power response also might be the most linear I have ever seen. Alan has done a superb job with these speakers.

As I said above, I posted the update of the testing on my YT page a couple of weeks ago as well. Spoiler below.

Spoiler: YT stuffs

There. I'd personally appreciate people not jumping the gun on stuff and implying I'm being "malicious" or deceiving when I've been documenting this all along in every place that I know to document it. Speaking objectively, I don't know another reviewer who goes through the lengths I do to discuss all that is going on in a review so publicly, posting in so many different places and formats.











Now that that is out of the way...

So, actually, the design never changed from the the final version. Instead, what happened is I was sent a prototype design that didn't have the final/production xover installed; an oversight on Alan 's part. Which caused me to spend a lot of time wiring up the new boards and installing them. I mean, hell, if anything I'd think people would appreciate the effort I went through to not only update the speaker to the final config to make sure the measurements matched what folks would be getting but also appreciate that I'm documenting all of this stuff along the way.


If you want to see the warts of the initial speaker (the one that wasn't damaged), here you are:

So, as you can see, the only difference between the initial run and the updated xover run is, indeed, the HF tilt which again was already accounted for by Alan in his final design.


Next time, how about just asking me instead of assuming there's a massive conspiracy.

 

[hardisj]

and to make sure you folks caught it the first time when I posted earlier in this thread, I likely will be testing an updated version. One that is updated off my measurements based on the findings of differences in Alan's measurement rig (discussed below).



HF Tilt:
Alan and I went back and forth on this a bit. His measurements weren't showing as much of a lift as mine. I think we were about 2dB apart. You can look back in the Sointuva thread where someone posted his data and see his measurements don't show this lift. I verified my measurements with a different mic. Alan did some digging on his end and tracked down the reason(s) for the difference. Based on his findings and mine, he said he is going to update his measurement rig and also tweak the HF. He also noted a couple other things he is considering to perfect the design based on the measurement differences. I've pasted his reply about those potential tweaks below in the spoiler. And I told him I'd be willing to re-test. This is the same as I have done with other manufacturers/designers when unintended differences were discovered between measurement systems (DIYSG HTM-12 and Neumi BS5P, for example).

Spoiler
I'm building the new test rig tomorrow to improve the microphone alignment. I think with the other corrections that will bring my measurements very close to yours at HF.

As mentioned previously I will tweak the tweeter XO to bring the HF down. Also I have made another value tweak to one of the inductors on the woofer XO which will flatten further the slight hump centred around 700Hz. It will also change to an air core as the reduced value makes it smaller and possible to fit it on the board. This may provide a small reduction in distortion at very high signal levels.

One other change I'm considering is to very slightly increase the box volume. This will flatten out the slight bass hump at 150Hz and provide slightly more LF extension.

If you are up for it I will send a production unit with these final mods through in a month or so.

 

[buz]

FWIW, I don't question your integrity - measurements are what they are and I didn't want imply they were fudged in any way. Apologies if that came across like this.

Subjective impressions are what they are but there I feel a disclaimer is warranted because all that effort can create bias.

 

[Newman]

It’s not about conspiracy, it’s about unconscious bias

 

[abdo123]

(bolded so no one misses this and the questioning of my integrity can end right here)

You guys act like I haven't discussed this already. I did. In detail. In two different threads. On December 9, 2020. In this post and cross-posted in this thread's post. TWO different places.

(@abdo123 , looks like you "liked" that post in the second link, btw)

I even posted about it on my YouTube page which @VintageFlanker provided a screenshot of in the same linked threat (here). Oh, and my Facebook group page. So, four places total. Three of them are wide open for everyone to see. It was always disclosed.

Spoiler: See? Told ya'. :p

Update on the March Audio speaker.

The speakers sent to me were the initial prototype speaker which are the same as the final design just more 'broken in' and less a final crossover mod (discussed below). I was completely fine with that. No sense in Alan building me a brand new pair just for review that he'd then have to sell at a discount as they had been used. Logical.

When I first listened, something sounded off. Remember, I test in stereo and mono for different reasons. This is important. I took some notes and moved on to the measurements.

I tested one and the results looked really out of whack so I contacted Alan. Long story short, since this was the prototype, the inductor hadn't been zipped-tied in place and Alan forgot to do that before he shipped it and it was knocked loose in transit. I verified by disassembling the speaker and sure enough... wiggle, wiggle. Rather than fix it right away, I put the other speaker on the stand. And... wow. Good wow.

I sent Alan the measurements so he could see for sure that the two measured different and we could tie up that loose end. The (good speaker's) measurement was as he expected so no news there. But, let's visit this quote from earlier...





Alan knew what to expect with my (good) result and had already accounted for a HF lift in the final crossover (that wasn't in the prototype board). Knowing that I was going to have to fix the busted crossover (for my listening tests) I thought, "hey, why not just have Alan send me the updated crossover so I can test what will be the true final version". So, rather than fix it, I asked Alan to send me the updated crossovers and I'll test the speakers with those.

Stay tuned for the final results. I'm not sure when the new crossovers will get here but I'll post and update when I get them.

In the meantime I can tell you that the measurements which have already been shared are really quite indicative of what I measured. I haven't overlaid them but I'd say to my eye they're practically the same. This might be one of the best measuring speakers I've seen so far, even with the current HF tilt. The sound power response also might be the most linear I have ever seen. Alan has done a superb job with these speakers.

There. I'd personally appreciate people not jumping the gun on stuff and implying I'm being "malicious" or deceiving when I've been documenting this all along in every place that I know to document it.




Now that that is out of the way...

So, actually, the design never changed from the the final version. Instead, what happened is I was sent a prototype design that didn't have the final/production xover installed; an oversight on Allan's part. Which caused me to spend a lot of time wiring up the new boards and installing them.

If you want to see the warts of the initial speaker (the one that wasn't damaged), here you are:
View attachment 177539


So, as you can see, the only difference between the initial run and the updated xover run is, indeed, the HF tilt which again was already accounted for by Alan in his final design.


Next time, how about just asking me instead of assuming there's a massive conspiracy.

Like i said I wasn’t sure if the speaker was being sold before the review. My comments are only relevant when that’s the case. I hope you understand why.

Companies usually charge more than 1000$ per spin, you usually go out of your way deconstructing speakers changing crossovers and doing several spins and the occasional groundplane measurements. We appreciate the effort, really, but some people could potentially take advantage of your good will. Just my 2 cents.

 

[hardisj]

FWIW, I don't question your integrity - measurements are what they are and I didn't want imply they were fudged in any way. Apologies if that came across like this.

Subjective impressions are what they are but there I feel a disclaimer is warranted.

Disclaimer of what? There isn't a need for such a disclaimer in the actual review because what I reviewed was what the final crossover config was. I just stated all of this above.

In the ideal world I would have been sent a fresh BNIB set of speakers that already had the updated crossover. Instead what I was sent was a prototype setup that Alan didn't update to the final config. I'm not sure of the exact reason why... my best guess is he updated the xover while the speakers were in transit (they were in transit for about 3 weeks; getting held up at customs for about a week, IIRC). But at the end of the day what I tested was his final xover. Forthcoming changes (as I posted above and earlier; twice now) are due to the differences found in his measurement rig. I'll test those because I think it is worthwhile for the community.


- Erin

 

[hardisj]

Companies usually charge more than 1000$ per spin, you usually go out of your way deconstructing speakers changing crossovers and doing several spins and the occasional groundplane measurements. We appreciate the effort, really, but some people could potentially take advantage of your good will. Just my 2 cents.

Yes, I am aware. And if I shared all my emails between Alan and I you would see that I was willing to make the updated crossover swap because I wanted the consumer to get what he was producing.

I'm not R&D. I've had numerous offers for payment for R&D and I have said numerous times to companies and publicly that I'm not that. I've talked about it in videos, too. I don't think Alan was treating me as such (and he apologized numerous times about the issues I was having). Shit happens and I just want people to see what the product they buy is. But I did document all of this in numerous places so obviously I'm going to get annoyed when people insinuate or otherwise state "maliciousness" and question my integrity as was done above.

 

[hardisj]

FWIW, I don't question your integrity - measurements are what they are and I didn't want imply they were fudged in any way. Apologies if that came across like this.

My guy, you literally just said I had a conflict of interest and said I needed to disclose (something) when I already had discussed this in 2 threads here, on my YT page and on my FB group page.

At the very least disclose you worked on it, it's an obvious conflict of interest. I have no problem with the way crinacle does it with those IEM collaborations but to publish it as a regular review is rather questionable.

 

[BenB]

2dB is probably all it needs to bring it back to neutral and non-fatiguing on reasonable recordings.

Adding some sort of "compensation" in the on-axis response for "narrowing dispersion" seems like a recipe for disaster. Creating a new problem, which this approach does, to ostensibly "fix" another problem, does not seem to be a viable long-term, recording-agnostic approach to take. It would be better to just have a flat on axis response and do the best possible with the rest.

Can you elaborate some more on how the "error" fixed a "problem"?

If you read the thread on this site regarding my wall-splasher (variable directivity) speaker experiment, you'll see that on-axis compensation for off-axis sound differences made it much more difficult to identify a preference (or even a difference) between the configurations.

Directivity Experiment: Wall Splasher 2-way

I've been curious about the impact of lateral (side-wall) reflections from speakers. Some people treat them, others don't. I was pretty sure I preferred them. I had a hunch that the ipsilateral (same side) reflections were beneficial for adding breadth to the soundstage. After seeing the...

www.audiosciencereview.com

I don't agree with your assessment that an on-axis response that varies from flat is necessarilly a problem. Quite to the contrary, for certain dispersion patterns it is a solution to the issue of perceived unbalanced sound (in room). I have performed A/B testing of speakers with wildly different dispersion patterns that had similarly different on-axis measurements, yet were perceived to have very, very similar spectral balances.

Once the decision has been made to create a speaker with atypical dispersion patterns (and the sointuva falls into this category), the frequency response should be adjusted to approach what will be perceived as a balanced sound in the typical listening environment. Further balancing can be accomplished by the end-user via EQ to address the ways in which their room may be atypical, or simply to reach their preferred balance in their listening position.

 

[abdo123]

Yes, I am aware. And if I shared all my emails between Alan and I you would see that I was willing to make the updated crossover swap because I wanted the consumer to get what he was producing.

I'm not R&D. I've had numerous offers for payment for R&D and I have said numerous times to companies and publicly that I'm not that. I've talked about it in videos, too. I don't think Alan was treating me as such (and he apologized numerous times about the issues I was having). Shit happens and I just want people to see what the product they buy is. But I did document all of this in numerous places so obviously I'm going to get annoyed when people insinuate or otherwise state "maliciousness" and question my integrity as was done above.

I apologise if i offended you in any way. That was not my intention.

 

[hardisj]

1) This seems to imply that you did hear the rising treble. That indicates a non-natural response on axis, which in many long-term listening situations will end up being fatiguing.

2) This seems to correlate well with you not finding the rising treble to be a problem. Do you have a general preference for a loudspeaker that has a rising treble on your choice of test material that you use during the listening tests?

3) Did the imaging change at all? "Didn't suffer" leaves a lot for the reader to assume what was meant.

4) What was the size of the listening room? How far from the loudspeakers was the listening position? These two factors would have a significant bearing on the power levels that you used during your listening evaluation.

The lowish sensitivity aligns well with the enclosure being relatively compact but with a good amount of bass extension.

I numbered them to make it a bit easier.

1) IME, the ones that tend to stand out as "bright" but are flat or rising on-axis are the ones with wide radiation. This is anecdotal, though. As with every sighted test there are surely biases that we don't even intend to have. (just as there are ones we intend to have; like someone's praise or disdain for a particular brand)

2) I think(?) I answered that above.

3) Good point. Poor writing. In my defense, ya'll are supposed to care about the data and the subjective is really just there for myself to remember later. But, yes, good point. I wouldn't say it got worse. I toed them out about 10-20° and the only difference I noticed is that the soundstage seemed to widen up. But this is exactly the case for wanting/needing blind testing. Something I can't do.

4) Big. I don't have the exact dimensions but I'd say about 20x25 feet in my living room with an open floor plan extending into the kitchen, hallway, dining room and entrance area. Typical furnishings. In my home theater room the dimensions are about 16x27 feet. I listen between 80-95dB in the MLP with my reviews. I forgot to include the levels and I didn't measure the "limits" at the time but it was sufficiently loud for my needs.