BS1000, passive 3 way speaker empirical developments

[dave]

A few years ago I had one of those daft ideas that I would make some NS1000 clones. The principle was if i used similar spec (in the loosest possible sense of the word!) drivers I should get similar results, hollow laugh. This is creation is centered around the Dayton Audio RS52AN-8 2 inch aluminium dome mid, ther nearest I could find reasonable priced to the Yamahas 3" Be dome. I started this on DIY audio (link) a few years ago, but will now continue it here.

Using a combination of:
Xsim
LTspice
Winisd
Holm impulse
ARTA

and some recent personal epiphanies, and after several put it to one side in frustration pauses, I am pushing forwards on this again.

It is now renamed the BS1000, because it has departed too far from the Yamaha, it is own thing now.

I really didn't like the Monacor DT-350NF metal dome tweeter I had previously selected, so I have switched it out with a Dayton Audio AMT2-4, in the hope it will sound less like an aluminium dome (it doesn't). Driver complement is now:

Dayton Audio AMT2-4 (nothing like a yamaha NS1000 tweeter)
Dayton Audio RS52AN-8 (metal dome mid with a built in protective grill)
Volt B250 (looks a bit like the yamaha woofer and I have a few spare)

I have had a fair few plays with the Xsim now, and achieved reasonable measured results, but not that happy with the sound on anything I have cooked up with it as yet. Resent epiphany happened the last time I played using DSP with these speakers, was to stop using 1m measurements and use near field (like 2cm). This technique revealed various lumpy anomalies, which were quite unpleasantly audible. Using the eq capabilities of the DCX2496 to get well behaved nearfield response was definitely progress.

This time around I am revisiting my previous Xsim passive crossover attempt. The near field measurements showed various problems which correlated to unhappiness in the sound. Simulating the Xsim developed crossover in LTspice, showed some of the filters were a bit resonant, where the unhappiness presented. Bouncing back and forth between Xsim and LTspice, some mods to the crossover evolved, which have been very beneficial to the sound.

The other problem is the natural frequency response of the Volt B250 seems to like rolling off a bit more enthusiastically that I would like. I am not the only person to realise this, it seems, Volt recommends bass reflex loading. I am seeing some correlation between my near field measurements and WinISD thiel small simulation results. So as this this now a BS1000, I don't feel quite so conflicted about having a port. Actually a slot type port shelf occupying the bottom of the box seems like a nice solution, and doesn't irreversibly hack up the box. I was at one point wondering if putting a B139 in the back of the box and using it as an ABR. Put that thought to one side now.

 

[dave]

Graph experiments today. Decided had to think about baffle step compensation, so a quick search for free tool, lead to Edge. Playing with that, I noticed some bumpiness in the 1-2kHz area when I put the mid in its appropriate place on the simulated front.



Which is curiously similar to my measured midrange response (below). I had been wondering where those bumps came from, as the near field for the mid dome is pretty well mannered.



Not the best graph, nearfield and before I mostly nulled the 12kHz mid ringing.



Bringing it together, measured at 1 m. Had to invert the tweeter, and still a little dip at 3kHz. At the lower end the slot port is very successful, and with the other tweek from the "Edge", increased the woofers inductor from the Xsim derived 4mH to 4.8mH, now deep rich bass.



The above is without the port fitted.

 

[dave]

Had a few days to listen to the speaker, and generally quite impressed, the mid and the tweeter seem quite happy, the tweeter could be a little "better" but is pretty good. The bass is promising but can sound a bit thick on spoken word sometimes. There are quite a few prototype problems, with the front only fixed to top and bottom edge, and on this at some point decided to try bubble wrap as filler. Maybe I did not put in enough, but listening at the port, can hear it rattling around inside, along with the flapping side panels, but altogether giving a good impression of depth, and bass instruments relayed pretty faithfully.

Going to build a crossover for the second speaker, this has resulted in revisiting the cross over, resulted in tweaking the xover some more, quite pleased at the resemblance between the Xsim impedance and the measurements.


I am going to try and null out the 500Hz impedance peak, as I would like to run these from valve amplifier.