3" Audio Nirvana 2-way project

[eventhorizon]

For starters I would like to thank member StevenZ for completing my "mini monitors" using the AN 3" driver. From 80 Hz and up they are perhaps the best sounding and most balanced speaker I've heard to date at low levels. I am reluctant to really drive this speaker to higher levels as it really not meant to be a massive bass producer.

Its imaging characteristics in the midrange and wonderfully smooth highs is what I like most about it. Its midbass is also surprisingly good. Unlike most other full rangers, its on-axis response is very controlled and non-fatiguing.

I would now like to tackle building a 2-way speaker using this driver and an appropriate woofer. My preliminary search has led me to the Seas Prestige ER18NRX 7" woofer priced at $92 from Madisound as used in the Zaph ZR71 speakers. The efficiency and impedance's are very closely matched.

ER18NRX

Speaker Building Supplies from Madisound

Audion Nirvana Super 3

http://www.commonsenseaudio.com/an3cfspecs.jpg

Almost all the competent 2-way designs that I have come across use a woofer and a tweeter crossed in the 1,500 Hz-3,500 Hz region. My outside the box thinking is wondering what a 2-way speaker will sound like crossed much lower, perhaps as low as 400 Hz.

I first thought of the concept "Double Driver Dynamic Loading for Low Q Drivers" from this website. From my memory all they did was run the drivers in parallel to "bump up" the bass output.

http://www.fortunecity.com/rivendell/xentar/1179/theory/dddllqd/dddllqd.html

I cannot seem to find this article any longer, and if anyone has saved this information or know of a different source to locate it, I would greatly appreciate it!

Most of the woofers I have looked at do not roll off sharply enough below 400 Hz to allow the woofer to roll off naturally, so a choke circuit of some kind will need to be used. I'd like the crossover to be as minimalistic as possible with a single inductor and capacitor network. If this proves to be the wrong method I will of course consider other alternatives.

Anyone with any suggestions or comments would be most appreciated.

 

[sreten]

Hi,

Search on FAST's. Ideally you want a bass unit 3 or 4 dB more sensitive
than the FR and cross it over 1st order series at around the baffle step
mid point to give around 3 or 4 dBs worth of baffle step compensation.

rgds, sreten.

 

[planet10]

What sreten said.

Some examples that we did;

http://www.diyaudio.com/forums/full-range/144099-thread-tysen-variations-fast.html
http://www.diyaudio.com/forums/css/176790-el166-mtm-ml-tl.html

There are more than a few others, particularily with OBs which almost force this approach.

Here is a 1st pass of our current FAST project.

http://p10hifi.net/FAL/images/uFKTwT-wWoofs-comp.jpg

dave

 

[eventhorizon]

Where would I reference the FAST principle. I can't seem to find the threads with a simple search

 

[tuxedocivic]

Here's one I recently did. Unusual goal in this one though.

http://www.diyaudio.com/forums/full...ge-speaker-photo-gallery-218.html#post3152228

Simple networks are very easy for a fast, except you do have to consider the Fs hump of your mid tweet. With only one cap, you'll likely still get a big output at Fs (or what ever the impedance peak is in the box). At higher levels this will sound really grunty. Also on the woofer is the natural inductance impedance rise. A single choke will only give 6db/oct at best. Often less if the woofer impedance rises quickly. Consider a zobel, or a second order filter.

Also consider that woofer breakup needs to be wayyy down on that woofer if you don't want to hear it. If you crossed at 300hz 1st order on that woofer, the break up would be down 24db. But consider that it's about 12db above the output at 300hz, it will actually only be down 12db. And it'll sound nasty.

I used second order on the speaker I linked to combat these issues.

 

[tuxedocivic]

Here is an example of the Fostex FE83. Blue is the box plot (just a simulation, not the real thing) and Red is the driver output with a single 30uF cap in series with the driver. At the impedance peak, the cap is very ineffective at the driver is almost outputting its full sound right there. Also has hardly any power handling as a result. May as well be running full range.

It's a similar, but better result with a series cross over.

 

[eventhorizon]

Thanks for those plots. Would a second order Linkwitz-Riley network be the best bet for this situation and proposed drivers? Think this combo could "work" without noticeable differences in response with the correct crossovers?

 

[tuxedocivic]

Either a second order or an impedance compensation circuit. Dave might know a handy way to knock down the impedance with clever box design, but I can't help you there. Cap and coil would definetely solve the issue. For the Fostex FE83 I ended up with 30uF and 2.2mH. That gave a smooth roll off at 12db/oct.

And trust me, that woofer breakup either needs a notch, second order, or both. Again, maybe Dave knows how to treat the thing to make the break up benign. My silver flute woofers were treated and when I listened to them full range the break up was extinguished. There was still a measurable peak in the response, but the nasty sound was gone. I used 3mH and a sorta zobel. 72uF and 2ohms. Gave a little peak in the 150hz area, but it had to be done to achieve my other goals.

 

[Bob Brines]

All of the above pretty well demonstrates why FAST has to be active crossed. Then you do not care about the efficiency of the two drivers. All is handled in the DSP or by adjusting the volume of the separate amps.

Bob

 

[eventhorizon]

I've had to use an active crossover on most of my other speakers for similar reasons, but was hoping a passive network could serve the same purpose.

 

[Bigun]

All of the above pretty well demonstrates why FAST has to be active crossed. Then you do not care about the efficiency of the two drivers. All is handled in the DSP or by adjusting the volume of the separate amps.

Bob

Bob - there are many who scoff at putting so much silicon in the signal chain - what has been your experience of the sound quality that can be achieved with a DSP in the picture - and are you using the mini-DSP that can be found on this forum ?

 

[tuxedocivic]

I agree with Bob that active is the way to go with FAST. For the cost of passive components alone it's worth it.

As for the sound quality. I haven't gotten to really gauge that yet unfortunately. I usually use my minidsp for testing things. Or when I run out of coils I'll try and do an A/B comparison tonight.

 

[Bob Brines]

Bob - there are many who scoff at putting so much silicon in the signal chain....

Do you have any idea how much silicon is in the chain that made the recording?

As you know from reading my posts, I personally do not like the lush sound that tubes add to the music. This is not a condemnation of tubes, just a statement of personal preference.

- what has been your experience of the sound quality that can be achieved with a DSP in the picture

A decent DSP works wonders at fixing problems with speakers and rooms, But, be realistic in your expectiations. Try to fix the problems at their source. Now, is a flattened FR better than any noise created in the DSP? You will have to try it youself to decide.

- and are you using the mini-DSP that can be found on this forum ?

Yes. However, to avoid as much signal degridation as possible, use the miniDigi SP/DIF interface. Otherwise, you are going DAC>ADC>processing>DAC. The miniDigi eliminates the first DAC>ADC conversions.

Bob

 

[planet10]

All of the above pretty well demonstrates why FAST has to be active crossed.

To passively XO a FAST one has to VERY carefully choose drivers and boxes. It is much easier to go active.

We have usually been able to use PLLXOs (no silicon) on our active FASTs, again careful with drivers and boxes.

If one needs to go beyond the capabilities of PLLXO, then cascaded sections buffered with FETS likely does the least harm. There is at least 1 thread on this using Nelson's B1 as the buffers.

dave

 

[sreten]

I used second order on the speaker I linked to combat these issues.

Hi,

Contemplation (or simulation) of a 1st order series (rather than parallel)
x/o will indicate the bass drivers inductance rise helps the series case
rather than hinders it as in the parallel case. Similarly the Fbox peak
of the FR is less of an issue, it doesn't peak up as it does with parallel.

Second order parallel is more flexible, as you can adjust Q and offset
the electrical x/o points, but sometimes 1st order series works well.

rgds, sreten.

 

[tuxedocivic]

Hi,

Contemplation (or simulation) of a 1st order series (rather than parallel)
x/o will indicate the bass drivers inductance rise helps the series case
rather than hinders it as in the parallel case.

Right, thanks for pointing that out. Woofer breakup can still be an issue though, imo.

Similarly the Fbox peak of the FR is less of an issue, it doesn't peak up as it does with parallel.

You still get the peak. Sometimes it's wose, sometimes it's better than the parallel circuit. It's dependant on the woofer. And usually not as bad.


I confess I don't know much about series circuits, so if I'm wrong, I'm open to correction. I've alway found parallel to offer better control over the drivers individually.

I'll try and do an A/B comparison tonight.

Bob mentioned some good points about dsp. Last night I compared my active settings with minidsp to a passive circuit I'm working on, on a two-way. It's not a fair comparison because my passive xo is a work in progress, as they always are, where the dsp can give pretty much theoritically perfect results with some button clicks.

And that's exactly the sound I got. The active sounded theoritically perfect. Almost to perfect. No FR abberations, just perfect tonal balance. But I did feel it was almost robot like. Sometimes I preferred it, other times I did not. For the most part though, it was better.

There was some very very quite buzz in the system, only when I put my ear right up to the tweeter. The passive didn't have this.

 

[eventhorizon]

Thanks so much for all the replies I have a simple Gemini active crossover that I could try, but I don't have anything built just yet, just comtemplating the idea and if its feasible. I like what both of these drivers would offer me.

Has anyone here had any experience with the AN 3" driver and could chime in?

 

[Bigun]

Do you have any idea how much silicon is in the chain that made the recording?

As you know from reading my posts, I personally do not like the lush sound that tubes add to the music. This is not a condemnation of tubes, just a statement of personal preference.

I'm quite encouraged to try the mini-DSP one of these days. Perhaps in a future pre-amp project.

As for tubes - I wasn't really referring to tubes when I commented on too much silicon - I was simply saying that lots of electronics in the signal path is something that many audio enthusiasts worry about.

 

[Bob Brines]

....I wasn't really referring to tubes when I commented on too much silicon - I was simply saying that lots of electronics in the signal path is something that many audio enthusiasts worry about.

There are various levels of audio purism. Vinyl, tubes, no passive components, exotic wire, yada,yada. My motto is "Do what you need to do to get the results.

Bob

 

[Scottmoose]

+1