Help me choose components for my nude dipole build

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The upper part of the response changes, yes, but not the low end extension.

Am i missing something? Any other better way to model H-Frame?

Just a guess here because I am not a user of Horn Response, but is the program just sticking in a straight line below where it is displacement limited? Can you reduce the power input (substantially, because who is going to be using the system at 120db???) since this might push the displacement limited region down in frequency.

I use a measure+model approach to low frequency H-frame modeling:
  • I build the H-frame (even using thin board) to the final dimensions (internal dimensions of the H-frame the same as planning final build).
  • Next, I measure the nearfield response at the midpoint of the front opening, in the plane of edge of the opening. I save this as a minimum phase FRD file (via ARTA). That's it for the measurement part.
    `
    I then use the measured data to do some modeling of the far-field response:
    ,
  • I set up my ACD modeling progam (Excel based) as a set of two driver responses (front, rear) plus the system response (front + rear).
  • I import the FRD file into both the front and rear worksheets.
  • In the worksheet for the rear output, I set the polarity to "reversed' since it is the opposite phase as the front output but we are using the FRD data set collected at the front, which does not capture the phase inversion.
  • In the worksheet for the rear output, I set the delay to a value, D, having an equivalent distance (distance in meters=D*c, where c=345m/s, D in seconds) equal to the total front-to-back length of the H-frame. This is the relative delay of the rear output compared to the front. I want to model the response in front of the front opening, on axis, and away from the front by some distance (e.g. the listening position is 3m in front of the front opening.
  • I need to take into account the fact that the rear output has to travel farther to the listening location, and will therefore be lower in SPL, compared to the front output at the listening location. You can calculate the SPL level for two distances using the online calculator on this web page:
    Sound level distance damping decibel dB damping calculation calculator change distance versus sound level apps reduction drop dissipation SPL sound transmission loss free field loss sound and distance - decrease drop fall sound over distance versus d
    After determining the attenuation for the rear output relative to the front output at the listening position, I set the gain for the rear opening to this value.
With the system now properly described in the modeling program the system response (front+rear) is an accurate model of the low frequency (e.g. <300Hz or so) response from the H-frame including any internal resonances at the listening position and including the actual response of the driver in the cabinet. Since the driver will experience some increased air loading by the H-frame, especially if it is "deep", it is not easily modeled in and of itself and I prefer the measurement-based modeling. The ACD spreadsheets are just computing the "in-air" summation of the front and rear waves, taking into account the propagation thru air, polarity of the sources, etc. I like to use ACD because I know exactly what is going on inside it, unlike the "black box" presented by some modeling programs...

So, what does the H-frame response at the listening position typically look like? Most of the time you will not see the "classic" dipole response. The 6dB/oct slope region below the "dipole peak" is usually steeper because the drooping response of the driver adds on top of it. The dipole peak is often much broader because the line resonance of the H-frame is also appearing in this general region of the response. The first null above the dipole peak may not be all that well formed, probably because of the rear wave attenuation. This depends on the listening position.

You can model other locations/listening positions by adjusting the relative delay and amplitude of the front and rear output accordingly.

One interesting aspect of this approach is that you can see how a listening position that is close to the front opening of a large H-frame will have "boosted" bass output compared to a far-field location. This is due to the relative SPLs of the front and rear wave. When you are close to one source and farther from another, the closer source is relatively louder. You will get less low frequency cancellation as a result.
 
H-frame is basically just a way to make the (open) baffle smaller. Dimension from the center of the driver to the edge is what counts. In a H or U-frame you can't reliably count or estimate this radius/distance, and the frame is not a tube with constant diameter normally...

DIY-dipole-1

I have no idea how Hornresp works, and should be "informed", but if I remember correctly it supposes that the driver has ruler-flat response.

Nearfield responses don't show dipole effect, you must measure at minimum distance of 2x baffle diameter to get reliable low freq result. Does someone know where Hornresp's virtual mic is located?
 
Charlie,

Thanks for the detailed how-to. Those spreadsheets - are they available for download?

Juhazi,

From what i remember, virtual microphone is placed at 1 meter distance.

Here is a good quote from the Hornresp thread:
Pressures are calculated at a point 1 metre from the centre of each of the two sound power sources, taking into account the solid radiation angle. The resultant combined pressure is then determined, taking into account the differences in level and volume velocity phase of each source, and the acoustic path length between the two sources as specified by the user. The resultant combined pressure is then converted to SPL.

I doubt that Hornresp assumes ruler-flat response for each driver - drivers with different t/s parameters model quite differently.

BTW, i made a mistake in a previous example of how H-Frame is modeled compared to a nude driver. When calculating combined power response, Hornresp requires the distance between two radiating sources to be provided. So, in an H-Frame model the distance was correct, but in a nude driver it was still the same as for the H-Frame which makes it an incorrect model.

Here is what i get with 1 watt power for 15LW1401 in an H-Frame:

attachment.php


and nude:

attachment.php
 

Attachments

  • Frame 1 watt.jpg
    Frame 1 watt.jpg
    48.3 KB · Views: 325
  • nude 1 watt.jpg
    nude 1 watt.jpg
    48.2 KB · Views: 361
Your "nude" has path length of 38,7cm. That means from the center of the driver to nearest edge - radius - and diameter of 30"! Obviously you counted path length twice?

Perhaps the same mistake for the H-frame? In HR there is an option to show the baffle, check that!

I am more familiar with The Edge, but it uses only flat response for drivers.

The 15LW has Fs of 42Hz, how low do you want to push it?
 
Last edited:
Juhazzi,

In my correspondence with David, he provided this illustration on how to calculate the offset:

764654d1561356710-hornresp-attach_1-png


From the dialogue one may conclude, that the distance is taken from center to center of the speaker. In the case of H-Frame (or compounded horn as it is presented in the Hornresp), the distance is measured from the mouth of that horn.

According to this method, provided i get it right, one should calculate the distance between center of the front of the driver and center of the rear of the driver, which happens to be the diameter of the woofer - about 387mm.

Perhaps the same mistake for the H-frame? In HR there is an option to show the baffle, check that!

Will do!

The 15LW has Fs of 42Hz, how low do you want to push it?

I want it to play to about 55-60Hz with authority. Below that i will use a monopole woofer.

I have a nice car audio sub with JL Audio 12W6V2 in it. Vented, tuned to about 20Hz. This will get me covered to 20-20K. What's the deal with hate on vented subs in dipole world, btw?
 
Then you must divide the distance by 2 (look at Kreskowsky link I gave in post #22)

A nude and baffled 15" driver simmed with the Edge here. Perhaps you should measure a nude driver quickly to verify this. I believe in Edge!
 

Attachments

  • horn vs edge 360.jpg
    horn vs edge 360.jpg
    213.1 KB · Views: 51
  • horn vs edge 720.jpg
    horn vs edge 720.jpg
    222.7 KB · Views: 53
By the way, the technique that I described for constructing the far-field low frequency response of an H-frame (it can be used on any OB system, really)...
...my measure+model approach to low frequency H-frame modeling...

... was, I believe, first described by John Kreskovsky. He has a web page about the technique with the example being a midrange driver:
measure a dipole
The web page includes very little info, but if you click the icon (it looks like this "<>") at the beginning of the line just above the first plot just before the text "Download zip file (38 k)" the browser will download a zip file containing a Word document with more detailed info.

I primarily use this approach for subwoofers because trying to accurately measure responses at low frequencies in the far-field directly is very involved and typically requires an outdoor measurement or inside a very large space, if you want to exclude room interactions/resonances, etc.

When nearfield data is collected on a driver in a boxed loudspeaker (a monopole) it does not capture the baffle-step and this must be added back in via modeling. This is, for example, the approach used in the FRD Blender software that I developed with Jeff Bagby:
FRD Blender and Minimum Phase Extractor
But with a dipole/OB type speaker, the interaction between the front and rear output strongly suppresses edge diffraction (assuming the "edge" is in the null at 90 deg off axis, which is typically 20dB down or more WRT the on-axis response) and so you do not need to model or include it when you are generating the far-field response.
 
Very interesting stuff, Charlie. Thanks a lot.

BTW, what do you guys choose as a material for H-Frames? I am thinking 18mm MDF for a planned 50cmx50cmx50cm (19,6x19,6x19,6). I can easily make it thicker or two-layered. I was even thinking about making a sandwhich of two 9mm MDF or birch plywood with 3mm aluminium in between to make it both heavier and tame all the resonances.

What do you guys think?

I am also going to use metal frame with sand inside the metal profile so that it doesn't "sing".
 
Then you must divide the distance by 2

Not if the actual locations of the "acoustic centres" are taken into account when determining the acoustic path length - see link below:

Frontiers

Key comment: "The baffle structure has a physical length L = d1 + d2 and width and height W. The acoustic length is larger by about W/2 on each end, similar to the end correction of the vent of a Helmholtz resonator."

Further references to acoustic centres:

"The Acoustic Center: A New Concept for Loudspeakers at Low Frequencies" - John Vanderkooy (AES Paper 2006)
"Applications of the Acoustic Centre" - John Vanderkooy (AES Paper 2007)
"The Low-Frequency Acoustic Centre: Measurement, Theory and Application" - John Vanderkooy (AES Paper 2010)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.