SDA dimentional driver bandwidth

Systems

I'm planning to do some experimentation and have a couple of questions for the knowledgable SDA's folks.


1)After looking at a few of the schematics it appears that the top end of the dimensional driver(s) is rolled off with a 12db filter(originally thought it used just a series coil didn't see the cap) and from the values of the coil/cap combo , can I safely assume that the upper limit is in the 2-3k range?

2) By how many db is the DA attenuated in relation to the SA?

3)Is there any available info detailing research done by Polk regarding the optimum bandwidth for the DA driver(s).

Thanks any input is appreciated.

heiney9

GV#27 wrote: »

I'm planning to do some experimentation and have a couple of questions for the knowledgable SDA's folks.


1)After looking at a few of the schematics it appears that the top end of the dimensional driver(s) is rolled off with a 12db filter(originally thought it used just a series coil didn't see the cap) and from the values of the coil/cap combo , can I safely assume that the upper limit is in the 2-3k range?

2) By how many db is the DA attenuated in relation to the SA?

3)Is there any available info detailing research done by Polk regarding the optimum bandwidth for the DA driver(s).

Thanks any input is appreciated.

It really depends on which generation SDA your speaking about. As far as research, there is always the white paper published in 1984 and then a few of the audio rags touch on some points, but I'm sure anything in-depth is proprietary and will be hard to pinpoint.

Later gens had a very narrow band in which the DA operated.

H9

ben62670

The dimensional drivers are run at the same frequency as the stereo drivers. If you want to build some use a single tweeter.
Ben

Systems

heiney9 wrote: »

It really depends on which generation SDA your speaking about.

Yeah I figured it would have been a evolutionary thing with the newer generations incorporating their lastest tweaks.In some models atleast it appears the DA operates over the exact same range as the SA.

I'm curious as to what they determined was optimum as Im going to experiment with the addition of a dimensional driver (1 per side) identical to the 6.5" midranges in my current speakers.These drivers will be in their own purpose built enclosure and placed to the sides of the main speaks.

The dimensional drivers will operate over the same 300hz-2k range as the main mids ,the R-L signal will be derived electronically using an op amp (fed from the mid out of the active crossover).For now the dimensional drivers will be driven by a pair of LM3785 gainclone.Hopefully this setup will give me a little taste of what the true the SDAs will do.

As far as research, there is always the white paper published in 1984

Thats what Im looking for,is there a link by chance?

Systems

ben62670 wrote: »

The dimensional drivers are run at the same frequency as the stereo drivers.

Yes in some the schematics that was apparent [/QUOTE] If you want to build some use a single tweeter.
Yeah I will be restricting the dimensional part to only the midrange.

daifanshi

GV#27 wrote: »

I'm planning to do some experimentation and have a couple of questions for the knowledgable SDA's folks.

1)After looking at a few of the schematics it appears that the top end of the dimensional driver(s) is rolled off with a 12db filter(originally thought it used just a series coil didn't see the cap) and from the values of the coil/cap combo , can I safely assume that the upper limit is in the 2-3k range?

2) By how many db is the DA attenuated in relation to the SA?

3)Is there any available info detailing research done by Polk regarding the optimum bandwidth for the DA driver(s).

Thanks any input is appreciated.

Hey GV,

It's great that you're interested in the details and not looking for "hand waving" answers to the how Polk implemented the SDA. A lot of answers can be found in the crossover. No need to attribute what's going on to a magical "secret sauce" .

Now to your questions...

1. I've done some simulations of the crossovers and the way it works isn't so complicated once you've studied it for a while. You are right that there is what looks like a 12dB/octave roll-off of the the dimensional drivers. The crossover corner frequency is effectively the same as the local stereo drivers. And the corner frequency is the same for the incoming SDA signal from the opposite speaker. The frequency corner is in the range of 2kHz like you rightly assumed.

There is a super-important complication in the crossover design that most people miss. If we take the 1.2TL schematic as an example, you will find a 18AWG 16mH choke from the "-" terminal of the dimensional array connecting to the "-" terminal of the main terminals.

For the local stereo signal coming into the speaker, this looks like a SERIES choke. But since it is so large, it produces a low-pass shelving response starting around 150Hz and will attenuate the local stereo signal produced in the dimensional array around 6dB at the corner of 2-3kHz. (partial answer to #2) But FULL bass response is included in the dimensional array. The dimensional array is acting like an array of subwoofers for the local stereo signal! The DA is getting the real local bass signal and not sympathetic resonance like I read around here before.

For the incoming SDA signal, this 16mH choke looks like a SHUNT inductor. Confirm this by looking at the schematic. So what we have now is an added highpass response to the original 12dB/octave network. The opposite speaker's BASS signal is now attenuated (more than 10dB), while the SDA midrange is allowed to proceed to the DA not attenuated by much and out of phase by 180deg.

The SDA signal's bass is attenuated to prevent it from cancelling bass from the stereo drivers. Stereo localization doesn't need low frequencies to work and you're just throwing away amplifier power by cancelling bass so this makes a whole lot of sense.

2. So to answer your question, the signal level for the local DA are very near the same at the crossover frequencie when compared to the opposite speaker's stereo array. Which makes sense from a cancellation perspective. But don't forget the bass attenuation for the SDA signal.

3. Can't answer the question regarding what R&D was done at Polk.

So after all this the bandwidth of the DA array is actually a band-passed signal, operating between ~150Hz and ~2.5kHz.

Cheers.

Systems

Hi daifanshi,thanks for the detailed and informative reply.

daifanshi wrote: »

I've done some simulations of the crossovers and the way it works isn't so complicated once you've studied it for a while.

Yes Im realizing that as well now that Im taking a closer look at some of schems.

There is a super-important complication in the crossover design that most people miss. If we take the 1.2TL schematic as an example, you will find a 18AWG 16mH choke from the "-" terminal of the dimensional array connecting to the "-" terminal of the main terminals.

yep seen it,16mh 18awg = hi DCR ,so I guess they needed the extra resistance.:D

For the local stereo signal coming into the speaker, this looks like a SERIES choke.

humm,looks to me to be out of the signal path of the SA section completely? but as you mention it feeds the low pass filtered bass into the DA drivers.

Edit.your correct, you say in series with the stereo signal not the SA drivers,my bad.

But FULL bass response is included in the dimensional array. The dimensional array is acting like an array of subwoofers for the local stereo signal!

It does make sense to put all that extra radiating area to good use.

So to answer your question, the signal level for the local DA are very near the same at the crossover frequencie when compared to the opposite speaker's stereo array. Which makes sense from a cancellation perspective.

Ah ok,it wasn't obvious (to me at least) because some models have a single DA driver combined with 4 SA drivers yet others have 4 DA drivers.Intuitively one would expect the 4 DA driver section to more closely match the level of the 4 SA than the single DA driver.Ofcourse they could be compensating by using drivers with lower sensitivty in the 4 DA models and or series connecting them?

So after all this the bandwidth of the DA array is actually a band-passed signal, operating between ~150Hz and ~2.5kHz.

Great,so I guess I'm on the right track with my intended 300hz-2k range.

Cheers

daifanshi

GV#27 wrote: »

Ah ok,it wasn't obvious (to me at least) because some models have a single DA driver combined with 4 SA drivers yet others have 4 DA drivers.Intuitively one would expect the 4 DA driver section to more closely match the level of the 4 SA than the single DA driver.Ofcourse they could be compensating by using drivers with lower sensitivty in the 4 DA models and or series connecting them?

Great,so I guess I'm on the right track with my intended 300hz-2k range.

Cheers

GV,

All my simulations were based on voltage level and I used complex lumped element equivalent models for the drivers. For the local DA, the simulations needed to include the loading from the opposite speaker crossover and DA.

You're right that we need to know sensitivities and impedances to estimate what the acoustic output will be. But I think the 2.3TL is one of the easier to analyze because it uses all the same woofer drivers for the DA and SA. Any relative calculations should be close to correct even if our absolute estimates for impedance and sensitivity are wrong.

At some point I'll go back to my sims and estimate the acoustic output and confirm how the SA and DA series/parallel combinations look like for that speaker.

Good luck.

Systems

daifanshi wrote: »

I used complex lumped element equivalent models ...

:eek::D:cool:

For the local DA, the simulations needed to include the loading from the opposite speaker crossover and DA.

Sure as they are really a composite due to their interconnected "-" terminals.

At some point I'll go back to my sims and estimate the acoustic output and confirm how the SA and DA series/parallel combinations look like for that speaker.

Cool, if you do be sure to post your findings,this is interesting stuff.;)

For my project the output level of the dimensional drivers will be fully adjustable and completely independant of the stereo drivers so experimentation is possible.

Good luck.

Thanks,it will likely be a few weeks before I can give it a try.

ben62670

He said lump:D

Systems

ben62670 wrote: »

He said lump:D

The significance being?

Systems

I came across a couple of reviews (the SRS and 1c)and both indicated that starting with the SRS model, Polk limited the bandwitdth of the dimensional array to the 200hz-1k range.Previously they ran the DA section fullrange but found it not necessary and in fact by restricting the bandwidth they made gains by increasing the apparent size of the sweet spot.