As I said, I will not be evaluating the performance as I have no intentions of buying the hardware. You can use any of the following miniDSP products: dual 2x4 units, in or out of the box, balanced or unbalanced, the 2x8, the 4x10 Hd, or if you want to go all digital the nanoDIGI with your own DACs. There are lots of options for DACs at all prices.
Hi Davey,
Yes, I was able to do the same thing. I can generate the transfer functions that way but I would rather just be able to load the biquad coefficients so I can go directly from my simulations to the dsp.
I see you sent SL a nanoDiGi setup for the LX521. Nice work!. How do you like the PE DACs you hung off the SPDIF outputs? That's the kind of solution I like, cheap!
Otto88,
I went with the Vifa because of better dispersion. And I just like it better than the Neo3 overall. Distortion isn't an issue with it. It's not driven very hard and it really doesn't come in fully until 7k. Plus, it makes construction easier.
Yes, I was able to do the same thing. I can generate the transfer functions that way but I would rather just be able to load the biquad coefficients so I can go directly from my simulations to the dsp.
I see you sent SL a nanoDiGi setup for the LX521. Nice work!. How do you like the PE DACs you hung off the SPDIF outputs? That's the kind of solution I like, cheap!
Otto88,
I went with the Vifa because of better dispersion. And I just like it better than the Neo3 overall. Distortion isn't an issue with it. It's not driven very hard and it really doesn't come in fully until 7k. Plus, it makes construction easier.
Sure.
Here is the original Note with Neo 3 tweeter form 8 to 16k.
Compare to the Note II RS over the same frequency range.
Here is the original Note with Neo 3 tweeter form 8 to 16k.
An externally hosted image should be here but it was not working when we last tested it.
Compare to the Note II RS over the same frequency range.
An externally hosted image should be here but it was not working when we last tested it.
Clearly the dome seem to be the better choice technically but since I've never heard a dome come close to the Neo3's sweet reproduction I'm happy I have the original Notes. A personal preference I guess.
Up to 55° off axis the original Note looks better to me in the diagram. What makes you think that 55-90° would be more important?
Rudolf
Hi Rudolf,
It can certainly be argued that the Neo 3 is better (less variation with frequency) over a limited angle, but the design objective was to improve the over all radiation pattern in the front hemisphere. The Vifa allowed me to do that. Additionally, reduction in cost and simplification of the build were secondary goals. Overall I believe the Note II sounds better than the Note I; the cost of the 4 Vifa tweeters is less that that of one Neo3; and it is easier to make the circular tweeter cutout than the rectangular one. But, as I stated above, if someone wants to go with the Neo3, I'm happy to oblige.
It can certainly be argued that the Neo 3 is better (less variation with frequency) over a limited angle, but the design objective was to improve the over all radiation pattern in the front hemisphere. The Vifa allowed me to do that. Additionally, reduction in cost and simplification of the build were secondary goals. Overall I believe the Note II sounds better than the Note I; the cost of the 4 Vifa tweeters is less that that of one Neo3; and it is easier to make the circular tweeter cutout than the rectangular one. But, as I stated above, if someone wants to go with the Neo3, I'm happy to oblige.
I had my reasons to write "it looks better" and not "it is better"
. As you have stated, there are more good arguments for your change of driver. And I don't think that the less "orderly" looking +/- 55° response of the Vifa is as different to the ear as it is to the eye.
Keeping the focus on the "tweeter bloom" problem, my personal experience goes like this: I hear it in dipoles, if the 30°, 45° and 60° response is not sufficiently attenuated. -1, -3 and -6 dB accordingly would be desirable values - which are (over)achieved in both Naos. To have the 90° response 20 db down instead of a mere 10 dB is more an expression of good engineering praxis than a real acoustic necessity.
Rudolf
I had my reasons to write "it looks better" and not "it is better"
As you have stated, there are more good arguments for your change of driver. And I don't think that the less "orderly" looking +/- 55° response of the Vifa is as different to the ear as it is to the eye.
Keeping the focus on the "tweeter bloom" problem, my personal experience goes like this: I hear it in dipoles, if the 30°, 45° and 60° response is not sufficiently attenuated. -1, -3 and -6 dB accordingly would be desirable values - which are (over)achieved in both Naos. To have the 90° response 20 db down instead of a mere 10 dB is more an expression of good engineering praxis than a real acoustic necessity.
Rudolf
You are obviously correct about what is happening at 90 degrees. Once you get above 1k Hz or so, it's very difficult to get a true dipole response. But what happens at 90 degrees is more a function of front to back symmetry rather that achieving a figure 8 radiation pattern.
With regard to the behavior at 60 degrees and tweeter bloom, I have been experimenting with my NaO II which suffers from that problem. What I have done is EQ the system so that the 45 degree response looks more or less flat through the range where bloom is a problem, rather than having the bump in the response. This does effect the on axis response but I am finding that the trade off of a dip in the on axis response is more desirable and sounds more natural than flat axial response with bloom. It's telling me that the reflected sound may be more important than the direct over this range. This will be incorporated in the update to the NaO II RSa. It', playing back what I have learned from the Note and Note II into the NaO II.
This would seem to reflect, at least in part, the experience of ORION owners, and the seeming interminable "tweaks" to the ORION ASP. It also clearly makes "voicing" rather room-dependent. Both you and SL have resolved the problem (correctly, IMO) in more recent designs. Like "baffle step", tweeter bloom is just not really "fixable" with equalization . . .
John, how much gain is necessary please? The Najda developer Nick is aiming to make the adjustments you detailed, some already in the pipeline it seems.
That is a difficult question to answer. It basically affects dipole woofer EQ. And that depends of the design of the woofer system. For example, depending on which woofer is used and what the cut off frequency is, in one of my dipole woofer systems I need A Q PEQ boost of anywhere between 11 and 22dB.
The difference between using DSP and analog is that with off the shelf dsp, like the miniDSP or Najda, the speaker designer is trying to design the crossover within the constraints of the hardware (or programing software) where as with analog the circuits are designed around the needs of the speaker design. But 12dB is probably too restrictive, although, as was pointed out, multiple stages can be cascaded approximate the required response.
This is a common issue with OB/dipole speakers and LT'd closed box subwoofers, where the driver gets quite a lot of "boost". Why not increase the gain of the power amp for that driver and reduce the boost in the crossover stage? Essentially the filter will retain the same "shape" but becomes cut only, or boost/cut, depending on how much of the total boost you move over to the power amp.
In my Active Crossover Designer, the user can assign a gain setting to each filter stage, e.g. to each biquad. When the IIR biquads are calculated for the filter, the gain is incorporated into the biquad. This allows the user to design a boosting filter in the normal way and then as a final step (before implementing the crossover in DSP) assign a filter gain so that the maximum amount of boost in the biquad is reduced as much as is needed to avoid levels higher than 0dB internally. In this way, the gain structure can be improved, in that you don't need to boost 24dB in one stage and then cut in the next. It's all done in one step. You still need to come up with the missing gain somewhere else - if you cannot modify the power amp gain, the other option would be to implement the boost after the DSP unit using an analog gain stage, which would be trivial to implement.
-Charlie
Hi Charlie,
It's not the over all gain that is the problem. It's how the filters are set up. How the transfer function is specified. Like miniDSp, it's a series of biquads which are shelves, PEQs, HP and LP filters. So there needs to be a way to set them up within reasonable limits. Say a Q boost at X Hz with Q = Y and G = Z is required, where Z is greater than 12dB. It can not be specified in the Najda software. That's the problem in setting up the filters. The over all gain of the channel can be reduced before or after the filter blocks, if required.
That's why I like the ideal of loading the biquad coefficients directly, like the miniDSP and your spreadsheet makes it easy to base them on the usual analog definitions which I design with.
I've been playing with minidsp and dipoles for a week now. I have used the minidsp plugin only, so far. I quite soon realised that I must use negative value for the dipole correction (or whatever is the greatest value needed) - the same princple with passive xos (you can only attenuate from initial level!), It is also wise to start by setting main level for each driver to -6dB or like
I like to see the visual effect of changing the values, it helps me to understand what is happening, because I am poor at mathematics!
I am amazed how small changes sometimes have a huge effect - and sometimes vice versa. It is easy in minidsp to drive a test plot of each driver, this helps to determine which side of xo to change.
With dipoles, it is a very challenging task to get controlled and uniform directivity, which is my main goal with AINO gradient project. It helps alot to be able to make plots of each drivers in angles of 0, 30, and 60¤ (for example)
I like to see the visual effect of changing the values, it helps me to understand what is happening, because I am poor at mathematics!
I am amazed how small changes sometimes have a huge effect - and sometimes vice versa. It is easy in minidsp to drive a test plot of each driver, this helps to determine which side of xo to change.
With dipoles, it is a very challenging task to get controlled and uniform directivity, which is my main goal with AINO gradient project. It helps alot to be able to make plots of each drivers in angles of 0, 30, and 60¤ (for example)
John, with tweeters wired in series pairs at 8 ohms, do you think the MiniDSP miniAMP has enough power for the tweeters? It's rated at 20w x2 into 8 ohms. Above 6 kHz there's not that much power, but I'm not sure if this is enough. It would be convenient to use the miniAMP because it stacks on the 2x8 board and lets you power the remaining 6 channels with any number of 7-channel amps, or else two fewer channels of class-D amp modules.
The key is recognizing that while the on-axis response is determined by equalization/crossover the "directivity" is determined by driver choice (size) and the baffle. The challenge is diminished if you treat response and directivity as independent variables . . . (which they are).