The XSD Speaker

yes that distortion graph is impressive. Where does the unEQ’ed woofer response roll off from? are there a resonance from the cavity on the back (hollow back)? maybe what the pyramid shaped damping is for?
Thanks, lrisbo.

Here is the raw woofer in SLOB response. This was before I applied Noico mass damped butyl onto the stamped steel basket. There was a 200Hz ringing peak from the basket that is clearly visible here. The roll off starts at 470Hz as predicted by the depth of the slot. This was before I added the eggcrate(pyramid) foam. Although I am not sure that the foam reduced distortion. More to reduce resonance and midrange leakage above the 470Hz rolloff frequency. There is no EQ per se, as this is a passive crossover. I did apply a shelving function with a high value resistor in parallel with the first low pass inductor. I did not experience a slot resonance peak. Might have something to do with a magic combo of poly cone driver, higher Qts, and rather large slot cross sectional area.
1652219601261.png


I couldn’t help but wonder what the distortion plot of an XSD with 8x Purifi PTT6.5 would look like? -110dB at 50Hz :p
 
Last edited:
  • Like
Reactions: 1 user
thanks xrk971

I suppose the slot acts as a quarter wave resonator - looks like the 470Hz matches the depth of 6.5” frame. The drivers must add some damping as you say. The slot thus gives a 2nd order low pass with some Q value (somewhat larger than 1). Funny to see that the harmonics also get killed by this low pass.

The acoustic front /back shorting of the bass is not very pronounced. Maybe a corner around 150Hz? there are of course these back wings to increase the path length form the back. Wonder how the 360deg radiation pattern looks like. the nulls are probably pointing backwards relative to the baffle front plane?

Will show to Lyngdorf who loves OB and PTT drivers. Maybe the 8x PTT6.5 dream becomes reality. In this case it must be active or use air cored inductors to avoid the hysteresis distortion
 
  • Like
Reactions: 1 user
the hysteresis distortion is a different beast than simpler types of distortion. it shows up as odd harmonic disortion for sine waves but when the signal is more comples, it reveals that it has strongly nonlinear memory of the past signal. When the current crosses certain thresholds set by past signal reversals the inductance jumps and produces a transient 'crack'. Our ears seem to be very sensitive to this kind of distortion and we prioritise to bring it way down. The PTT6.5X is at -80dB (1Amp peak, 1kHz), the new PTT8 is at -90dB and we have prototyped a 6.5 with -100dB. It is first something one notices when it gets reduced and then after a short while one does not want to ever go back. We have not had resources to do formal blind tests. You can read more here:
https://purifi-audio.com/2020/04/28/dist/
 
I have experienced this distortion in a choke loaded SE Class A amp - it was dependent on the music playing and showed up as almost a selective “clipping” or a sharp crack sound only on certain crescendos and usually when playing deep bass lines it had memory of the music that came before. When I set up the same amp in a balanced mode using a balanced winding bifilar choke, that went away.

I have not experienced that with this speaker crossover but good to know. I might have to try an air core inductor to test. Only 3mH so not egregiously large.

There is a PTT6.5 with -100dB distortion?! Wow.
 
  • Like
Reactions: 1 user
some high hysteresis drivers even give a crackling sound when sweeping them. the -100dB 6.5is an experimental prototype. hope to release it. its rather addictive.

balanced bifilar should reduce the net H field on the core so that should help.

we have done an experiment wheree we first degauss the speaker motor then play a programming pulse and a read back pulse of skip the write pulse. the write pulse easily makes adifference in the readback signal in the order of 0.1% and you can wait minutes after the write pulse to read and the effect is maintained. pretty scary. Anywa, old days computer core memory was done with magnetic cores...
 
This is a thread dedicated to the eXtremely Sexy Dipole (XSD) speaker that originated as a cardboard prototype 8 years ago in this thread.

View attachment 1049106


The woofers I used are GRS 6.5in poly cone rubber surround 8ohm budget drivers. The floor footprint is 12in wide x 14in deep. It stands about 45in tall

View attachment 1049113

View attachment 1049114

I should mention that the woofers are wired 4x parallel, then two pairs of those in series, then a pair in parallel at the end. Make sure the opposing ones are wired correctly so that they both squeeze air out when the signal is positive. This results in a 4ohm nominal impedance and a sensitivity of about +12dB on top of whatever the nominal 8ohm rating is after baffle step losses. So for an 85dB driver we get about 91.5dB at 2.83v (that’s 2W into 4ohms). What this means is that the drivers barely move for significant SPL and that leaves a lot of headroom so that the distortion is very low.

For anyone interested in building this speaker as a personal DIY project, PM me with your email and I will get you the dimensioned version of this sketch.
View attachment 1049118
What a wonderful project! I understand this is eventually going to be a commercial design with a passive crossover so I apologise if any of my questions are too intrusive.

1. For those of us for whom even a 12" wide speaker is too big (not WAF acceptable) what parameters do we need to consider for woofers that are compatible with this type of open baffle?

We have an 8" GRS woofer available in India with a published Qts of 1.1
https://diyaudiocart.com/GRS-8PR-8-8-inch-Poly-Cone-Rubber-Surround-Woofer?search=GRS
6 of these would result in a nominal impedance of 5 ohms and a sensitivity of about 90db. Do we consider any baffle step loss for this woofer configuration?

Do the "woofer-pairs" have to face in the same direction? Why cant they be oriented so that their baskets are on the "outside" and they both face the "ri-pole slot/cavity"?

Would a woofer with a Qts of 0.4 work? I ask because one option is this HiVi woofer.
https://diyaudiocart.com/Raw-Driver...e-Midbass-Woofer-8Ohm?sort=p.price&order=DESC
8 of these would result in a nominal impedance of 4 ohms and a sensitivity of 92db before any baffle step compensation.

It has a slightly shallower basket which can help reduce the total width of the loudspeaker especially if we can orient the "woofer-pairs" so that the baskets are on the outside.

In keeping with aiming for a narrower baffle the full range I had in mind was either of the MarkAudio drivers below (linked in my order of preference):
https://diyaudiocart.com/Raw-Driver...-(4-inch driver)-Pair?sort=p.price&order=DESC
https://diyaudiocart.com/Raw-Driver...d-Cone-3.5-Full-Range?sort=p.price&order=DESC
https://diyaudiocart.com/Raw-Driver...ne-4"-Full-Range-Pair?sort=p.price&order=DESC

I suspect the mono-suspension 7MS might not work in an open baffle as there would be no "air suspension" to protect the driver and the 85db Alpair 6.2 might not be efficient enough.

The MarkAudio Pluvia 7HD might not need any helper tweeter but if it does the one I have in mind is this one:
https://diyaudiocart.com/Raw-Driver...ter-with-Rear-Chamber?sort=p.price&order=DESC

Thanks
Navin
 
Hi Navin,
The 8in GRS woofer would work well. It would lower the upper frequency of the crossover down to 350Hz which is not a bad place to be. The sensitivity of the speaker will be about -6dB below the peak predicted theoretical due to it being a finite baffle width and having some front to back cancellation. How are you wiring it for 5ohms? I see 1.33 ohms or 12ohms nominal. But 6x 4ohms can get 6ohms. Lower Qts drivers have a more powerful motor so will not be able to reach frequencies below their Fs without active EQ. But if the Fs is low enough already then it’s fine. You can have drivers arranged with baskets all out. If drivers do not have symmetric in and out non linearities then the distortion will not cancel as well. It will still work quite well. If you deviate from my drivers then you are on your own with the XO.

I have discussed with my commercial partners, the possibility of providing a XO kit for DIY’ers interested in making the XSD, and they have agreed. So if you stick with the same drivers and cabinet dimensions, you can have an easy route to a well-voiced speaker.

I will also try to offer the CNC waveguide for the GRS PTT2522-4 tweeter for those who are interested.
 
Last edited:
  • Like
Reactions: 2 users
I have developed a BOM for the crossover using quality components. All Mills 12W non inductive resistors for the mid and tweeter, and 20w non-inductive wirewound 1% Dayton resistors on the woofer shelving filter. Iron core 18ga inductor for the woofer, 18ga air core for the midrange, and 20ga aircore for the tweeter. All capacitors are MKP film (mix of Audyn Q4 and Dayton DPMC) plus one 100uF 400v bipolar electrolytic. You are welcome to upgrade the electrolytic to a film cap on your own if you choose. Th XO kit will be just the components plus the schematic. You will have to assemble yourself as a P2P soldering project. It is not too hard as I will show you a layour of how I do it.

The kit will consist of:

6 Mills MRA12 12W resistors
2 Dayton DPR 20W 1% non-inductive resistors
8 Audyn Q4 MKP caps for midrange and tweeter
8 Dayton DPMC MKP caps for midrange and woofer
2 100uF 400v bipolar electrolytic caps for woofer
2 20ga aircore inductors for the tweeter
4 18ga aircore inductors for the midrange
2 18ga ironcore inductors for the woofer

I am offering the XSD XO kit as group buy and will offer the XO kits (stereo pair) for $375 - majority of this is for the cost of the parts and small amount for my effort to assemble the kits.

Please add your name to a list below if you are interested so that I can see if this is worth pursuing.

XSD Crossover Kit GB Interest List

Name No of kits (stereo pair) Country
 
  • Like
Reactions: 1 users

Bryguy

Member
2022-04-26 5:42 pm
Ohio
I have developed a BOM for the crossover using quality components. All Mills 12W non inductive resistors for the mid and tweeter, and 20w non-inductive wirewound 1% Dayton resistors on the woofer shelving filter. Iron core 18ga inductor for the woofer, 18ga air core for the midrange, and 20ga aircore for the tweeter. All capacitors are MKP film (mix of Audyn Q4 and Dayton DPMC) plus one 100uF 400v bipolar electrolytic. You are welcome to upgrade the electrolytic to a film cap on your own if you choose. Th XO kit will be just the components plus the schematic. You will have to assemble yourself as a P2P soldering project. It is not too hard as I will show you a layour of how I do it.

The kit will consist of:

6 Mills MRA12 12W resistors
2 Dayton DPR 20W 1% non-inductive resistors
8 Audyn Q4 MKP caps for midrange and tweeter
8 Dayton DPMC MKP caps for midrange and woofer
2 100uF 400v bipolar electrolytic caps for woofer
2 20ga aircore inductors for the tweeter
4 18ga aircore inductors for the midrange
2 18ga ironcore inductors for the woofer

I am offering the XSD XO kit as group buy and will offer the XO kits (stereo pair) for $375 - majority of this is for the cost of the parts and small amount for my effort to assemble the kits.

Please add your name to a list below if you are interested so that I can see if this is worth pursuing.

XSD Crossover Kit GB Interest List

Name No of kits (stereo pair) Country
I'll take one! I would also want the waveguides. Based in USA. Name's Bryan. Do you want first and last?


Gonna try to spread the word, because I really want to see this happen!
 
Last edited:

JHClark

Member
2016-09-10 11:13 pm
I have developed a BOM for the crossover using quality components. All Mills 12W non inductive resistors for the mid and tweeter, and 20w non-inductive wirewound 1% Dayton resistors on the woofer shelving filter. Iron core 18ga inductor for the woofer, 18ga air core for the midrange, and 20ga aircore for the tweeter. All capacitors are MKP film (mix of Audyn Q4 and Dayton DPMC) plus one 100uF 400v bipolar electrolytic. You are welcome to upgrade the electrolytic to a film cap on your own if you choose. Th XO kit will be just the components plus the schematic. You will have to assemble yourself as a P2P soldering project. It is not too hard as I will show you a layour of how I do it.

The kit will consist of:

6 Mills MRA12 12W resistors
2 Dayton DPR 20W 1% non-inductive resistors
8 Audyn Q4 MKP caps for midrange and tweeter
8 Dayton DPMC MKP caps for midrange and woofer
2 100uF 400v bipolar electrolytic caps for woofer
2 20ga aircore inductors for the tweeter
4 18ga aircore inductors for the midrange
2 18ga ironcore inductors for the woofer

I am offering the XSD XO kit as group buy and will offer the XO kits (stereo pair) for $375 - majority of this is for the cost of the parts and small amount for my effort to assemble the kits.

Please add your name to a list below if you are interested so that I can see if this is worth pursuing.

XSD Crossover Kit GB Interest List

Name No of kits (stereo pair) Country
JHClark one kit USA
 
  • Like
Reactions: 1 user
Yes, you can get the waveguides separately. The crossover filters will be discussed in more detail in this thread. Basically apply 2nd order low pass on woofer at 470Hz and second order high and low pass on midrange at 470Hz and 4700Hz, both in phase. For tweeter, high pass second order reverse phase at 4700Hz. The key is to use the natural fall off on woofer at 470Hz to get a 4th order electro-acoustic filter. Nice thing with DSP is you can delay and time align and EQ and squeeze more bass extension out of it.

I might have to take out my Dayton 408 DSP and give it a try.
 
Hi Navin,
The 8in GRS woofer would work well. It would lower the upper frequency of the crossover down to 350Hz which is not a bad place to be. The sensitivity of the speaker will be about -6dB below the peak predicted theoretical due to it being a finite baffle width and having some front to back cancellation. How are you wiring it for 5ohms? I see 1.33 ohms or 12ohms nominal. But 6x 4ohms can get 6ohms.

The 6 x 8" woofers are each 8 ohms and would be connected into 3 pairs of 16ohms each. Then wire the 3 pairs in parallel and that should give 5.3ohms.

How does one compute the peak predicted theoretical sensitivity?

I have discussed with my commercial partners, the possibility of providing a XO kit for DIY’ers interested in making the XSD, and they have agreed. So if you stick with the same drivers and cabinet dimensions, you can have an easy route to a well-voiced speaker.
That's nice but none of the drivers in your original design are available in India.

I believe you are using the following:
Tweeter: GRS PTT2522-4 with waveguide
Midrange: PRV 5MR450NDY (8ohms)
Woofer: 8 x GRS 6PR-8 6.5in woofers, 8ohms wired in series/parallel to give a combined nominal impedance of 4 ohms.

The tweeter used to be available but has been out of stock for months.
 
  • Like
Reactions: 1 user