EAC has posted DIY speakers, Samba and Overnight Sensation measurements

[ROOSKIE]

The P.E. Dayton Audio DIY kits did not fair all that well.
Has anyone here heard them? I have not heard either set.

The Samba RS drivers are still obviously extremely good reasonably priced drive units.
The cabinet is solid but that crossover is just really short of a good design.
What gives here, PE has great resources available?
It will be cool to see a reworked crossover and hopefully Erin can retest one day with updated X-overs.

The Overnight Sensations are just really poor in many ways. Not a great a value.

Unfortunately, all in all as finished products both kits demonstrate how good the best of the $200-500 completed commercial designs are now (and factoring sales and clearances this is even more so)and that DIY could benefit from better crossover designers.
The cool things is as DIY products they are easier and ripe to modify.

Thanks to Erin and thanks to PE for taking the risk to test and measure. I love PE but certainly these two speakers need to be re-worked.

Home

Samba MT Bookshelf Speaker Kit with Knock-Down CabinetDesigned by the engineers at Dayton Audio, the Samba speaker kit is a compact 2-way bookshelf speaker that provides a reference quality sound without the usual high price. The Sambas produce an extremely detailed and accurate sound while...

www.parts-express.com

Home

Overnight Sensations MT Speaker Kit PairWhen Paul Carmody first listened to the Overnight Sensations speakers that he designed he exclaimed, "I wish I had a time machine. I would go back to my bedroom when I was in Jr. High or High School and give these to myself. These are the little bookshelf...

www.parts-express.com

I used to have the "Tango", PE speaker with RS drivers (the old, fantastic, RS tweeters and Aluminum woofers) and the crossover was weak there as well with a nearly flat in room response - rising quasi anechoic. I took it out and made a 2 way active with a miniDSP.
In that design the sound was superb and those RS drivers really are good values IMHO. (especially that old RS tweeter, man what a smooth sounding and superbly measuring metal dome tweter. To bad they are gone. I do still have one set to use some day soon in a waveguide.)

 

[aktiondan]

I too was excited to see that Erin was testing some DIY speakers. The Sambas certainly didn't fair as well as I had hoped. I have used both of those drivers in various other speakers over the years (see my avatar to the left for one example) and while I don't have a Klippel measurement system, they certainly have the ability to perform much better than these measurement would indicate. The crossover is definitely the limiting factor here and not the drivers.

Erin did post all of the raw driver measurements to his DropBox and a few folks have already provided some ideas for updated crossover designs over on his Facebook page. I was going to take a stab at it myself but haven't familiarized myself with XSim all that much. I'm still old-school and use Jeff Bagby's Passive Crossover Designer in Excel. And as far as I can see it's not obvious if/how his exported data will import. Either way I am sure that someone will come up with something better. There's potential here for sure.

 

[xarkkon]

Pity! I guess there's always the C-Notes for PE to fall back on to.

Wondering if Erin will test some of their floorstanders such as the Solstice. Those would be interesting!

 

[RickS]

I had higher hopes for the Samba, but hopefully it is just a crossover fix.

The Overnight Sensations are clearly a much older design but was surprised that they are not better on-axis at least. Perhaps it is just a voicing consideration and is how Carmody intended them to be.

Might take a stab at crossover redesign when I have more time, but with post of a VCAD project out there, guessing someone will get to it first. Doing with multiple full spins to validate a redesign is time I would rather spend on Directiva or other newer, more forward projects…

 

[McFly]

Wonder how Paul gonna take this. It’s not bad, but it’s not good.

 

[azzy_mazzy]

I had higher hopes for the Samba, but hopefully it is just a crossover fix.

Wouldn’t you still have the directivity issues? Seems like a Waste to start with a compromised design when you can 3D print excellent waveguides

 

[abdo123]

Wouldn’t you still have the directivity issues? Seems like a Waste to start with a compromised design when you can 3D print excellent waveguides

There are already great attempts at Erin’s facebook group with pretty decent directivity and response overall.

I might get banned for posting them here so if you’re interested you can look then up yourself.

 

[azzy_mazzy]

There are already great attempts at Erin’s facebook group with pretty decent directivity and response overall.

I might get banned for posting them here so if you’re interested you can look then up yourself.

I don’t use Facebook, if you don’t mind can you post a screenshot of the responses that they get in VCAD, most importantly the horizontal polar response?

 

[alex-z]

I played around with the Samba MT data in VituixCAD, and while there is some room for improvement, filling the on-axis dip from 1500-4000Hz results in worsened directivity, so the overall preference score only goes from about 7.2 with a sub to 7.5 with a sub.

If you optimize purely for preference score, it can reach 8.1. The power response becomes smoother, but the upper mid-range/lower treble has a slight trough. Fixing the on-axis response without compromising power response simply requires a waveguide. If someone wanted to shrink the somasonus 8" waveguide down to 6", that would justify spending this much on the crossover.

Dayton RST28 | somasonus

www.somasonus.net

 

[Matias]

I look forward to see them remeasured after the crossover improvements the community is working on!

 

[alex-z]

The Overnight Sensations initially look worse, achieving a 3.9 preference score or 6.3 with sub.

However, the tighter driver spacing allows both on-axis and power response to be improved simultaneously, resulting in an 8.1 score.

I must confess, I slapped together the crossover in 60 seconds and let VituixCAD optimizer do the rest. Works shockingly well if your initial values are decent.

First observation is that this is a cheap mod, the stock 1.1mH and .35mH inductors can be reused. The tweeter will be pushed harder, but this speaker seems strongly woofer limited, and on paper the tweeter can handle it. The port tuning is a touch high, and some polyfill might be worth a shot to soften the 1200Hz resonance.

 

[abdo123]

@alex-z @azzy_mazzy Credit is to Damian Demendecki

 

[alex-z]

@alex-z @azzy_mazzy Credit is to Damian Demendecki

I saw that design, but was highly skeptical. The woofer breakup modes are barely suppressed, nice directivity won't stop the speaker from sounding bad.

 

[abdo123]

I saw that design, but was highly skeptical. The woofer breakup modes are barely suppressed, nice directivity won't stop the speaker from sounding bad.

he also made one with the breakup suppressed. Directivity suffers slightly.

 

[RickS]

he also made one with the breakup suppressed. Directivity suffers slightly.

Parts counts are soaring here though. Anybody try an active crossover?

 

[RickS]

I played around with the Samba MT data in VituixCAD, and while there is some room for improvement, filling the on-axis dip from 1500-4000Hz results in worsened directivity, so the overall preference score only goes from about 7.2 with a sub to 7.5 with a sub.

If you optimize purely for preference score, it can reach 8.1. The power response becomes smoother, but the upper mid-range/lower treble has a slight trough. Fixing the on-axis response without compromising power response simply requires a waveguide. If someone wanted to shrink the somasonus 8" waveguide down to 6", that would justify spending this much on the crossover.

Dayton RST28 | somasonus

www.somasonus.net

View attachment 214076

Actually like yours better. Is the tweeter invert really still needed? Made more sense with the original crossover…

Did you try keeping second order on the woofer?

 

[RickS]

FYI for anyone who may have gotten a head start on a Samba project, check your VCAD data from EAC as he reposted after finding an error.

 

[abdo123]

Parts counts are soaring here though. Anybody try an active crossover?

In that aspect i think i like Erin's crossover the most, hopefully the messed up phase is an error or something.

 

[RickS]

In that aspect i think i like Erin's crossover the most, hopefully the messed up phase is an error or something.

Thanks for sharing!

Hard to entirely judge without knowing target acoustic slopes and seeing phase inversion on tweeter to check driver integration.

Directivity is problematic at the crossover point. Agree phase is better but still messier than I would expect.

 

[alex-z]

Actually like yours better. Is the tweeter invert really still needed? Made more sense with the original crossover…

Did you try keeping second order on the woofer?

Yeah, the tweeter inversion is needed. Erin mentioned in his review that the polarity was inverted, so it just needs to be changed in software. Here is what happens if the tweeter is flipped back on the third order design.

Second order on the woofer is possible, but I felt like suppressing those woofer breakup modes was worth the extra component.