I make NF impulse response of every driver. It is an excellent tool to confirm that a driver is working correctly. The NF impulse response can be used to make a CSD or burst decay plot, as well as a NF frequency response scan. When combined with an impedance sweep, this information will reveal almost every kind of driver defect.
When I am studying the far field response of a driver, it is very helpful to have the NF data. If I see some frequency perturbation in the FF response, the NF response can confirm that it is (or is not) inherent to the driver.
With woofers, we know it is necessary to merge the NF response to the time-windowed FF response. But I do this for midrange and tweeter drivers as well. With tweeters, my expectation is that the merged response will be nearly the same as the measured FF response, so if I see a difference, it may indicate an error. Sometimes the tweeter crossover is low enough that I need good tweeter response data down to 400 Hz, so the FF+NF merged data may be more appropriate. After looking at both the FF response and the FF+NF merged response for a tweeter, I make the decision on which one to use for simulation.
I have never worked with a dome midrange, so I don't know exactly what to expect. Considering a 500 Hz crossover, I think I will need a valid response down to 100 Hz, which means I will need the FF+NF merged data.
Since I could not get the mic within 2 mm of the dome for the NF scan (best practices), I experimented in varying the distance. When I repeated the scan with 7 mm additional distance, the resulting response was the same except for a small change in the high frequency response. This tells me that my NF scan is valid for low frequencies, even though I was outside of the recommended NF mic distance.
j.
Not only driver defects show up, also internal cabinet resonances are revealed by NF impulse measurements. I think the impulse analysis, with the distortion analysis, are the most important assets of modern measurement systems. In the early days we had to settle for quite rudimentary level and OK impedance, the rest was up to our ears. So use the gear while you have it, I'd say.
I performed another set of horizontal polar scans, making sure there was no tilt. During the test setup, I checked that the +/- 45 degree scans had the same impulse start time and had the same frequency response. I also checked the +/- 90 degree scans for the same thing. At +/- 90 degrees, there was a timing difference of 0.01 ms, which is 3.4 mm. I estimated that this difference was insignificant.
So here we have valid data. This is windowed far field scans taken every 15 degrees from 0 to 180, merged with the near field scan after adjustment to 4-pi.
j.
So here we have valid data. This is windowed far field scans taken every 15 degrees from 0 to 180, merged with the near field scan after adjustment to 4-pi.
j.
Thanks, very informative!
There is potential in the widening around 1-1,5kHz to be more uniform, I would say...
There is potential in the widening around 1-1,5kHz to be more uniform, I would say...
Just based on the looks of it, i think the narrowing around the 700Hz region make it look like there is a widening in the 1 to 1.5kHz region. Around 700Hz wavelengths the only thing that matters might be the cabinet width, depth and width-depth ratio.. Probably those dimensions need to be targeted if we need to do something about it..
But then again, a crossover in that region might balance it out/make ot worser a little bit i guess..?
But then again, a crossover in that region might balance it out/make ot worser a little bit i guess..?
This driver has a natural 2nd order roll-off at about 500 Hz. This means it will not be possible to implement a 2nd order acoustic lowpass response, because there would be little or no electrical filter protection for the driver. If I apply filtering and EQ so the driver has a 400 Hz LR2 response, there is no resulting electrical high pass filter
... and the filter gain transfer function
That filter transfer function would soon destroy the driver.
So applying a high pass 2nd order electrical filter to this driver is going to result in a 4th order net acoustical response.
Here is something that looks more realistic. This is an acoustic 400 Hz LR4 low pass:
And the resulting filter gain transfer function:
It is better, but it provides a mere -10 dB of attenuation at 200 Hz (one octave below Fs). This may be pushing the edge of the envelope of this driver's low frequency capability and durability.
By moving the crossover target from 400 to 500 Hz, I get -17 dB of attenuation at 200 Hz, and this seems a more comfortable environment for the M74A.
Lot's of things for me to think about...
j.
... and the filter gain transfer function
That filter transfer function would soon destroy the driver.
So applying a high pass 2nd order electrical filter to this driver is going to result in a 4th order net acoustical response.
Here is something that looks more realistic. This is an acoustic 400 Hz LR4 low pass:
And the resulting filter gain transfer function:
It is better, but it provides a mere -10 dB of attenuation at 200 Hz (one octave below Fs). This may be pushing the edge of the envelope of this driver's low frequency capability and durability.
By moving the crossover target from 400 to 500 Hz, I get -17 dB of attenuation at 200 Hz, and this seems a more comfortable environment for the M74A.
Lot's of things for me to think about...
j.
Distortion (dB) with different topologies will teĺl more. % will look devilish...Very high spl tolerance will be a challenge anyway
Hadn't Yevgeniy already helped out here?
From How To use: "The confirmed low cutoff frequency for all models is 600 Hz with at least 2nd order filtering. It is possible that 500 Hz will be quite normal if you reduce the requirements for maximum SPL. The upper cutoff frequency is around 6-7 kHz for the aluminum"
https://hificompass.com/en/reviews/bliesma-m74a-6-m74b-6-m74p-6-and-m74s-6
From How To use: "The confirmed low cutoff frequency for all models is 600 Hz with at least 2nd order filtering. It is possible that 500 Hz will be quite normal if you reduce the requirements for maximum SPL. The upper cutoff frequency is around 6-7 kHz for the aluminum"
https://hificompass.com/en/reviews/bliesma-m74a-6-m74b-6-m74p-6-and-m74s-6
Sure, I have read the Hificompass reviews, and I respect his opinions. I also know that the Bliesma spec sheet says the usable bandwidth is "Recommended frequency range Fs - 6 kHz", and the power rating is based on "IEC 268-5, 2nd order high-pass Butterworth filter, 400 Hz".
???
Will it play 100hz at 80dB? ..... 200hz? I doubt that it would, but that might be required if the music is at 105dB at 1m. Can you work backwards from your maximum SPL?
If it measures right and sounds wrong, you're not measuring the right things.
If it measures wrong and sounds "right", you might be happy with what you hear but it's a lie.
The art lies in measuring the right things and having the sound reflect what's actually there. The task is tougher than it looks.
As Gloria Steinem said, "The truth will set you free, but first it will p*** you off."
Is there any 3” dome that does 400Hz highpass comfortably? The ATC and Volt soft domes don’t either iirc, let alone the cheaper Scan Speak, Morel and other brands imho. But since a 4 way is the subject, it really isn’t any issue to me. It’s not unusual to cross these domes at around 1k or higher.
I think that hifijim has intention to cross M74 with some subwoofer drivers under the dome? This will really be a challenge, I think that with some 8 or 10 inch midbass filler driver (100 to 600...700...800 Hz) and natural partner T34A tweeter this could be killer combination with very relaxed constraints for developing the crossover. But overall complexity and price go up of course...
And now I see that he made decision to add midbass....so, crossover point can go higher...
And now I see that he made decision to add midbass....so, crossover point can go higher...
Last edited:
Here are some interesting simulations using the actual measured M74A data. At this point I am mostly concerned with achieving an acceptable directivity performance, so I am focussing on the Power and DI plot. I want to ensure that the woofer and tweeter I select will integrate well with the M74A.
For each of the following simulations, I changed the tweeter but kept the same woofer and midrange. I did some optimization to achieve a good on-axis response. This means the sound power and PIR curves are not fully optimized. What I am mainly looking at is the DI curve.
First is a sim of an idealized dome tweeter with an Sd=6.2 cm^2. This is typical of 25mm domes. The DI curve has a shallow dip at 5k, but otherwise is well behaved.
Next I have a sim using measured data from the SBA SB26CDC tweeter. This data is from another project, but the baffle size/shape is similar to this prototype baffle. This tweeter has an Sd of 6.2 cm^2. I find it interesting how closely the DI curve follows the above sim with the idealized tweeter.
Next I show an idealized tweeter with an Sd=9.6 cm^2. This is typical of the larger dome tweeters with 29 - 34 mm domes. The directivity index curve is smoother from 2k - 6k.
Now I show the sim with a Satori TW29TXN-B. This is a large dome tweeter with an Sd=9.6 cm^2. It is taken from another project. The baffle on that other project was about the same width as this prototype, but the edge radius was smaller, and the tweeter was located closer to the top edge. Although the TW29TXN-B data is not fully representative of how the tweeter would measure in this baffle, it gives me an idea of what I could expect. It also helps validate the sim with the idealized large dome tweeter (above). All in all, I think there is good agreement between the above sim and the below sim using real tweeter data.
Finally, I wanted to get an idea of how a small waveguide would work with this midrange. I recently completed a project using the SB26STWGC, a 26 mm soft dome tweeter with a small waveguide. This waveguide begins controlling directivity around 3k, and this is about where the crossover would be for this project.
One of the tweeters I am giving serious consideration to is the Seas 27TBCD/GB-DXT H1499. This tweeter has a small but very effective waveguide. The simulation above gives me an indication that the Seas DXT might integrate very well with this project.
As always, I am interested in any thoughts that anyone might have.
j.
For each of the following simulations, I changed the tweeter but kept the same woofer and midrange. I did some optimization to achieve a good on-axis response. This means the sound power and PIR curves are not fully optimized. What I am mainly looking at is the DI curve.
First is a sim of an idealized dome tweeter with an Sd=6.2 cm^2. This is typical of 25mm domes. The DI curve has a shallow dip at 5k, but otherwise is well behaved.
Next I have a sim using measured data from the SBA SB26CDC tweeter. This data is from another project, but the baffle size/shape is similar to this prototype baffle. This tweeter has an Sd of 6.2 cm^2. I find it interesting how closely the DI curve follows the above sim with the idealized tweeter.
Next I show an idealized tweeter with an Sd=9.6 cm^2. This is typical of the larger dome tweeters with 29 - 34 mm domes. The directivity index curve is smoother from 2k - 6k.
Now I show the sim with a Satori TW29TXN-B. This is a large dome tweeter with an Sd=9.6 cm^2. It is taken from another project. The baffle on that other project was about the same width as this prototype, but the edge radius was smaller, and the tweeter was located closer to the top edge. Although the TW29TXN-B data is not fully representative of how the tweeter would measure in this baffle, it gives me an idea of what I could expect. It also helps validate the sim with the idealized large dome tweeter (above). All in all, I think there is good agreement between the above sim and the below sim using real tweeter data.
Finally, I wanted to get an idea of how a small waveguide would work with this midrange. I recently completed a project using the SB26STWGC, a 26 mm soft dome tweeter with a small waveguide. This waveguide begins controlling directivity around 3k, and this is about where the crossover would be for this project.
One of the tweeters I am giving serious consideration to is the Seas 27TBCD/GB-DXT H1499. This tweeter has a small but very effective waveguide. The simulation above gives me an indication that the Seas DXT might integrate very well with this project.
As always, I am interested in any thoughts that anyone might have.
j.
I tried to generate interest in experimentation with DIYing a DXT based waveguide over on the Open Source Waveguide thread, but it didn't go anywhere. I may try such an experiment for the T25B, but it would be a while before I could get to it. Having the directivity control of the DXT with the distortion levels and high breakup of the T25B strikes me as a great combo.
@profiguy has frequently recommended larger tweeters like the Seas T35 and TW034 as partners for the D7608; perhaps this is one of the reasons, though I'm not sure about the waveguide. OSMC uses a waveguide with the R2904/7000 and Volt mid dome, though I was also not clear why.
Last edited:
A large tweeter with a tall dome like the T35 and TW034 behaves quite differently to a larger dome that is quite flat like the SB Satori's. Within reason the taller the dome the less directive it is at high frequencies.
The DXT creates directivity by introducing intentional diffraction. It has a property that I would try and avoid when desigining a waveguide, that the listening window responses are consistent but messy. They cross over each other at different frequencies and this is the inevitable consequence of diffraction from a sudden change in curvature.
I have never owned a DXT so I have no idea what they sound like, a couple of people whose audio opinions I place some weight in do not like them.
There are also lots of others who enjoy them in Heissman's designs. They remain one of the only off the shelf options for a small tweeter with decent directivity control.
For anyone who is prepared to fabricate their own waveguide themselves or through a manufacturing service there are better available designs for most popular tweeters.
@hifijim can have any of my waveguide designs if he wants to have one printed for himself or I can knock one up if it is for a tweeter I haven't already considered. The only caveat is that if it is not a hard dome the outcome cannot be reliably predicted with BEM.
The DXT creates directivity by introducing intentional diffraction. It has a property that I would try and avoid when desigining a waveguide, that the listening window responses are consistent but messy. They cross over each other at different frequencies and this is the inevitable consequence of diffraction from a sudden change in curvature.
I have never owned a DXT so I have no idea what they sound like, a couple of people whose audio opinions I place some weight in do not like them.
There are also lots of others who enjoy them in Heissman's designs. They remain one of the only off the shelf options for a small tweeter with decent directivity control.
For anyone who is prepared to fabricate their own waveguide themselves or through a manufacturing service there are better available designs for most popular tweeters.
@hifijim can have any of my waveguide designs if he wants to have one printed for himself or I can knock one up if it is for a tweeter I haven't already considered. The only caveat is that if it is not a hard dome the outcome cannot be reliably predicted with BEM.