ASR Directiva Open Source Speaker Review

[RickS]

As I thought about this more, I realized I was answering the original question being alleged about the Purifi woofer...

Could it cause these speaker's cabinets to resonate at one or more frequencies? In this case, maybe. The scales are all over the place and need to go back and reconsider mine. In this light, I can see how the speaker (not the driver) data above might be correlated.

The most reliable measure shown is EAC's Sointuva one and does not seem very significant (even at 96 dB).

 

[DualTriode]

As these woofers are not inexpensive, I can certainly understand concerns over their performance. That said, it is important to understand the conditions under which any measurements have been done. Given this, here are my findings:

1. I cannot reproduce Amir's results under any conditions.
2. I tried measurements with both SPK5 and Directiva r1 and did not get consistent results. I will stress conditions matter greatly. The SPK5 is vented with a crossover mounted internally and the Directiva has a PR in a different. Would expect a Sointuva to have yet another set of results.
3. In the end, also looked at EAC's and hificompass's test of the driver and found no evidence of a major issue.

Here is the hificompass measure at 2.83 V and 20 mm...

If you go to the site, you can try higher voltage levels and all are pretty good.

My takeaway is that the Sointuva measures are unique and could be an isolated driver/speaker issue and has some contributing factor for a distortion issue. If Lars was evasive, he may have been avoiding a long, drawn-out discussion. I do not see any repeatable reason for concern at this point.

Hope this helps!

Rick

Yesterday I picked up a pair of Purifi 6-½ inch woofers at my mail locker.

I am interested in making some small or mini speakers for my test bench.

I have been reading the threads about a couple of speakers that use the Purifi 6-½ inch woofers. The Sointuva WG thread identifies disturbance(s) that the software identifies as primarily as H3 distortion at 380hZ. The thread that @amirm tests the speaker that you built shows “H3 distortion” also at 380hZ. The threads both speculate that there may be some sort of mechanicial resonance in the driver in addition to enclosure resonance. In the thread amirm tests the speaker that you built amirm speculates that there may have been equipment sitting on the speaker while it was tested. There is a fair amount of coincidence going on here. Plus the Purifi 6-½ inch woofers manufactures data sheet shows a H3 distortion peak at only 100hZ higher for a similar driver under test. Makes me wonder out loud.

Digging a little deeper into how to use these magic very low distortion Purifi 6-½ inch woofers I am looking at sealed enclosures, reflex enclosures, and resistive port enclosures.

Modeling sealed enclosures the enclosures are tiny at about 0.20 cubic feet. With this tiny volume nonlinear air compression will dominate the speaker distortion. Modeling reflex enclosures the tuned port is 30 or 40 inches long, too much. I do not want all the noise and resonance introduced into the mid-range voices. You have looked at passive radiators, pretty cool.

Resistive ports look to have promise. There would not be any high impedance resonance. A resistive port would largely limit the “high” pressure compression of a sealed enclosure and limit the sealed enclosure nonlinearity.

To limit or capture any possible driver mechanical resonance the driver should be mounted with physical support and damping of the magnet / motor structure. Linkwitz style.

Thanks DT

 

[fluid]

Always should be concerned about mounting. My SPK5s are mounted with bolts on a 1.5 inch thick baffle. Not sure you read in the earlier posts, but Amir may have had a bad woofer. I do not have a Sointuva to measure, but EAC did and his THD measures do not align with @Nuyes measure you cite. As I mentioned, I do have a Directiva with a new woofer and cannot replicate Amir's measures.

Given all the noise in the Sointuva measurement thread some of the good information could easily be overlooked. The issue of mechanical resonances and mounting affects all drivers. In the X version of the Purifi Drivers there is a 380Hz basket resonance that is obvious in all measurements I have seen. What may not be so obvious is that this will not present like motor or suspension distortion and increase with level, it may only be present at a specific SPL. As each complete speaker being measured will be constructed and mounted differently the resonance may or may not show up and it may appear at differing SPL levels. Standard distortion testing at specific SPL levels could very easily not show an issue as the resonance wasn't excited.

Above you showed the high level plot from Hificompass. it is in the 2.83V measurement where the resonance can be seen. This could easily be considered no problem as the level is at -65dB, when SPL rises the rest of the distortion goes up and appears to swamp it out, but if you look carefully 380Hz is actually lower than the surrounding frequencies and most likely the resonance is not being excited rather than being covered by more overall distortion.

The realistic concern should be that as mdsimon2 has said there are now a couple of completed speaker systems that have shown some odd distortion results at this same frequency where something about the construction, mounting or measuring environment was different and appears to have excited this resonance to a much higher level.
The difficulty now being that none of the original measurements can be tweaked or adjusted to get more concrete information as they are out of the hands of the reviewers.

 

[RickS]

After rescaling my Directiva measurements, can say there does appear to be some resonance around 380 Hz. I want to be very specific about any root cause unless I have solid evidence. Notably, on ASR and when a vendor's reputation is called into question. While I concur on seeming high number of occurrences, we have driver data and other data that does not support. For example, here is Amir's SPK5 distortion measurements:

Note this a bass reflex design (and a very robustly built one at that).

I posted the driver reference measures earlier and none of them exhibit an indication that the driver is a source of the resonance. However, every speaker with this resonance is a passive radiator design. Perhaps another plausible explanation may be related to using a PR?

I alluded earlier that I had retuned the SPK5 to use a shorter port. While not as low an f3, the ugly port resonance is now gone. Am still working on the damping material(s) for them, but plan to share the mods soon.

 

[RickS]

Given all the noise in the Sointuva measurement thread some of the good information could easily be overlooked. The issue of mechanical resonances and mounting affects all drivers. In the X version of the Purifi Drivers there is a 380Hz basket resonance that is obvious in all measurements I have seen. What may not be so obvious is that this will not present like motor or suspension distortion and increase with level, it may only be present at a specific SPL. As each complete speaker being measured will be constructed and mounted differently the resonance may or may not show up and it may appear at differing SPL levels. Standard distortion testing at specific SPL levels could very easily not show an issue as the resonance wasn't excited.

Above you showed the high level plot from Hificompass. it is in the 2.83V measurement where the resonance can be seen. This could easily be considered no problem as the level is at -65dB, when SPL rises the rest of the distortion goes up and appears to swamp it out, but if you look carefully 380Hz is actually lower than the surrounding frequencies and most likely the resonance is not being excited rather than being covered by more overall distortion.

The realistic concern should be that as mdsimon2 has said there are now a couple of completed speaker systems that have shown some odd distortion results at this same frequency where something about the construction, mounting or measuring environment was different and appears to have excited this resonance to a much higher level.
The difficulty now being that none of the original measurements can be tweaked or adjusted to get more concrete information as they are out of the hands of the reviewers.

View attachment 215747

For the purpose of full disclosure, I realized the hificompass HD measurement I chose earlier were for the 8 ohm driver (was being consistent with the EAC measurements) and these are for the 4 ohm one. I agree this question needs to by considered in the context of a system. Given this, as you show, the 380 Hz resonance changes with output level.

Even if we assume there is driver correlation, this seems very myopic IMO. Have always felt strongly about root cause, but my work experience has always had to consider the impact to the product/system performance. Whereas, I can see sufficient evidence of correlation, am struggling with how significant the issue is in the context of overall system performance. If this is what you meant by "noise", I wholeheartedly agree!

 

[SDC]

The REAL Problem of March Audio's Sointuva WG (Review, Measurements and Reinforcements with Klippel device)

I’m wondering if this particular pair of speakers would have measured as designed without the flaws would we know who Nuyes is? I sure hope so.

www.audiosciencereview.com

Hello, everybody on this thread. This is measurements of 6.5X NAA woofer.

Yes, this is not NFA, but X series motor, plus NAA has higher Mms than NFA.

This is the best we can get as amateur hobbyist.

Me and @Nuyes also suspected 500hz as new candidate on NAA but level was still low so decided to seize the day. Hope it helps.

 

[fluid]

Notably, on ASR and when a vendor's reputation is called into question. While I concur on seeming high number of occurrences, we have driver data and other data that does not support. For example, here is Amir's SPK5 distortion measurements:

The only damage to Purifi's reputation would come from misunderstanding the measurements, jumping to conclusions and pointing fingers. I have cautioned against that very thing in the other thread more than once. Hopefully the rest of this post would only go to reinforce that message.

To me those measurements show exactly the same issue that at some point there is a distortion peak centred on 380Hz that stands out in comparison to the generally low distortion.

I posted the driver reference measures earlier and none of them exhibit an indication that the driver is a source of the resonance. However, every speaker with this resonance is a passive radiator design. Perhaps another plausible explanation may be related to using a PR?

As far as I can see this is not an inherent flaw in the driver but it is a characteristic it has, and needs to be considered in a system design. If the resonance is excited to the point that is seen in some of the measurements then the design or construction of the system could benefit from a rethink. It is not unusual that high performance parts need special care to extract all the performance they can deliver.

It is certainly plausible that the passive radiators are in some way contributing but in the measurements of the Sointuva one sample was good and one was not, both using the same parts so it seems unlikely to be due solely to the use of a passive radiator.

Adding a brace to the magnet to allow the force to be transferred to the cabinet in a similar way to 'grounding' the magnet works when measuring the speaker with the magnet down, should be effective. This would damp the mass spring formed by having the basket rigidly attached and the magnet floating and should reduce or remove the tendency for oscillation.

 

[RickS]

@fluid and all, I may not being seeing everything you are, but after reading most of the other thread, pretty sure we are mostly on the same page. As this seems to have already to have been litigated ad nauseam on that thread, rather not continue to do so on this review thread.

As Directiva is an open project, if someone wants to try enhancing the r1 driver bracing, they are welcome to try. Currently have little time to explore this any further right now. I think I was actually blissfully ignorant of the recent Sointuva thread and found it somewhat informative but less pleasant than I like.

 

[RickS]

When I have some more time, will revisit this resonance question. As I mentioned, the 8 ohm version does not appear to have the same resonance. I happen to have a couple of the aluminum cone version (also 4 ohm) and they do not exhibit the same resonance either. So before devoting any time to different mounting, would like to understand how voicecoil and cone type play a role in this situation.

Clearly changing key aspects of the driver are enough to result in different distortion profiles despite a common motor design. Hopefully, perhaps Purifi will further elaborate.

 

[RickS]

I see this question of the Purifi woofer mechanical resonance keeps getting brought up. Since Amir's distortion measurement makes it look rather ugly, thought I would measure the same speaker on my rig. This is the look at 86 dB...

To keep it simple, the yellow line is third harmonic distortion and the darker one is the noise floor in my room. As you can see, there is some worsening around 400 Hz, but is generally lower than .05%. If I test at 75 dB, this all this drops below the noise floor and not going to hear it here. While not a Klippel, this measurement does not have equipment sitting on top of the speaker and so is likely much more representative of the woofer distortion performance as used within the r1 design.

 

[fluid]

Rick if you want to know if that speaker will excite the resonance you need to measure across a range of dB levels. The nature of the issue means that it is not a linearly increasing or decreasing problem but will most likely occur at a specific spl level or range and that will change based on box construction and other factors. At the specific resonance frequency in your measurement there is a slight dip in 3rd order distortion which is consistent with measurements of other speakers and suggests there will be an spl level where the resonance has the opposite effect.

 

[DualTriode]

I see this question of the Purifi woofer mechanical resonance keeps getting brought up. Since Amir's distortion measurement makes it look rather ugly, thought I would measure the same speaker on my rig. This is the look at 86 dB...

View attachment 217858
To keep it simple, the yellow line is third harmonic distortion and the darker one is the noise floor in my room. As you can see, there is some worsening around 400 Hz, but is generally lower than .05%. If I test at 75 dB, this all this drops below the noise floor and not going to hear it here. While not a Klippel, this measurement does not have equipment sitting on top of the speaker and so is likely much more representative of the woofer distortion performance as used within the r1 design.

Hello,
0.05% distortion is seriously low.

My brain knows the conversion between %, some times I swap the labels on the plots between % and dBr's

@SIY buried his test microphone in a crate of unused kitty litter and tested the system; microphone, mic-amplifier and analyzer for noise floor.

In my world the analyzer cooling fan and test amplifier fan add to room noise. If I use long cords for the speaker and microphone I can move the DUT across the hall behind two closed doors and lowers the fan noise by many dB's.

I am still gluing together a sealed test enclosure for the 8 ohm 6 1/2 inch Purifi.

Thanks DT

 

[RickS]

Rick if you want to know if that speaker will excite the resonance you need to measure across a range of dB levels. The nature of the issue means that it is not a linearly increasing or decreasing problem but will most likely occur at a specific spl level or range and that will change based on box construction and other factors. At the specific resonance frequency in your measurement there is a slight dip in 3rd order distortion which is consistent with measurements of other speakers and suggests there will be an spl level where the resonance has the opposite effect.

Thanks, did that and as mentioned, it only gets better at lower levels. I chose the worst case at a level that matched Amir’s. I kept the graph simple by publishing one level. Do not plan to do much louder as I value my hearing (and would not make for a happy wife or my dog!)

My point was to illustrate that the distortion magnitude was not on the order of Amir’s measurement. Am not denying the resonance, just its magnitude and whether it is worthy of major remediation. I can show comparable results on my vented version.

EDIT: Just realized Amir did not have much trouble until he hit 96 dB. If I have a chance, can move outside and try.

 

[DualTriode]

Rick if you want to know if that speaker will excite the resonance you need to measure across a range of dB levels. The nature of the issue means that it is not a linearly increasing or decreasing problem but will most likely occur at a specific spl level or range and that will change based on box construction and other factors. At the specific resonance frequency in your measurement there is a slight dip in 3rd order distortion which is consistent with measurements of other speakers and suggests there will be an spl level where the resonance has the opposite effect.

Hello,
I agree strange things can happen, resonance may reach a critical unstable value.

Sort of a related story: I was called to a building, the interior was pulsating. The ceiling was shaking so hard that 2 X 4 tiles were falling to the floors the occupants had evacuated. It was a real Tesla kind of moment.

Turns out the Contractor had turned the building over to the Owner. The Owner's Plant Operation Department refused to do maintenance. The Contractor refused to do Warranty work because the required maintenance was not done. The building was very near the Med-Helicopter pad and the Building HVAC system had Carbon filters to remove the jet fuel exhaust from the air being drawn into the building. As the filters loaded the HVAC fan point of operation moved to the unstable back side of the fan curve. This is fan surge.

https://www.achrnews.com/articles/96839-surge-stall-and-instabilities-in-fans-contractors-mysteries

Thanks DT

 

[Chromatischism]

Moral of the story, mind your filters.

 

[RickS]

Thanks, did that and as mentioned, it only gets better at lower levels. I chose the worst case at a level that matched Amir’s. I kept the graph simple by publishing one level. Do not plan to do much louder as I value my hearing (and would not make for a happy wife or my dog!)

My point was to illustrate that the distortion magnitude was not on the order of Amir’s measurement. Am not denying the resonance, just its magnitude and whether it is worthy of major remediation. I can show comparable results on my vented version.

EDIT: Just realized Amir did not have much trouble until he hit 96 dB. If I have a chance, can move outside and try.

At 96 dB, the center frequency moved closer to 380 Hz and the 3rd HD hits under 0.5%. Even adding equipment does not make it much worse. To do this, I wore ear protection and had to move the mic to 1m distance (the previously posted measure was done at 2 cm). When I redid the 86 dB measure at 1m, the 3rd harmonic is below the noise floor.

I do note that high distortion in Amir's THD measurement at 96 dB is 2nd harmonic. So, even if the driver resonance caused the distortion to peak, in the Directiva cabinet it was a more pleasing distortion. Anyway, the closer I look at this, am less inclined to take any further action. I have bigger dragons to slay.

P.S.

After doing the higher SPL level, also realized that the woofer that Amir measured was not in the speaker. I had taken it out and tried to repair it. It has a slight voice coil rub issue. At the 96 dB level, it exhibits about 2% 2HD. If I had not tried to repair, perhaps it could have gotten to 4% (as Amir's THD test reported).

 

[RickS]

Ok, had an opportunity to try to play with this resonance question more. As am working on my Purifi SPKs, I checked the 3rd harmonic distortion for them. The SPKs have an angle brace the ends close to the top of the internal woofer frame. Here are my initial findings:

1. At 96 dB, the SPK does not resonate at the same frequencies as Directiva r1.
2. My attempt to change the resonance was by adding a raquetball between the brace and the woofer/front baffle. Note that I mention the contact area included the front baffle. This is because it was difficult to find a stable location for the ball between the brace edge and the woofer frame alone.
3. My results were inconclusive from a measurement perspective. However, I did hear some improvement that would motivate me to investigate further. I could try a different location for the additional bracing. but is going to require some design effort. The brace is slanted and so would need to get something stable between it and the woofer magnet.

Please note the repeatable results are tricky here. Every time I remove a driver to change the added bracing, cannot be sure I am not changing some other aspect of the test conditions (like screw torque, gasket sealing and speaker location). I used the SPK I have completed, I am documenting the rebuild of the second one in another thread. Rather than further add to an already bulky review thread, will post my next round there.

 

[Head_Unit]

Amazing work!

Sure, Zelle me at EddieThe'[email protected]...