Full range cardioid

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Full range cardioid, need advice on next prototype

Currently i'm working on a new speaker build, aiming for a cardioid radiation pattern from 50Hz on up. I have made a prototype consisting of a top cabinet which has a A&D R1030 10" woofer in a damped Uframe, and a Hypex PSC 2 400 plate amp (2 ch amp + dsp). I experimented with the opening of the Uframe, adding slots and especially the placement and quantity of damping material. See attached files for the optimal polar pattern. For high frequencies, I have a B&C DE250 on a Dayton 10” waveguide. I modified the waveguide slightly to accept the bolt on compression driver and remove any irregularities in the transition to the waveguide.

For the sub cabinet, I have 2 Peerless XLS10 woofers and a Hypex PSC 2 400. In order to achieve cardioid radiation, I have combined a sealed and dipole enclosure. The dipole is high passed at 50Hz using a 4th order filter in order to limit cone excursion and to have the advantages of a pressure source <50Hz. Free field measurements showed a polar resembling cardioid from 63Hz on up.


However, there are some points for improvement for which I would like your advice!
1. In order to achieve maximum suppression of LF at rear, I had to delay the monopole sub by 1,9ms. This resulted in less than optimal summation of the two drivers at the front. The dipole woofer is actually positioned 20cm behind the woofer in the sealed enclosure.
2. The polar has a maximum which is about 20degrees of center, caused by asymmetric construction of the dipole. I think I will make a new version using a regular H frame.
3. I’m considering adding passive radiators to increase output of the monopole <50Hz for LFE use. I’m a bit concerned this will complicate achieving cardioid radiation in the 50-150Hz region.
 

Attachments

  • Cardioid sub.jpg
    Cardioid sub.jpg
    813.9 KB · Views: 5,072
  • B&C_DE250.jpg
    B&C_DE250.jpg
    399.9 KB · Views: 4,898
  • Gesloten_dipool_cardoid_respons.png
    Gesloten_dipool_cardoid_respons.png
    9.3 KB · Views: 4,728
  • Prototype_Cardiod_sub_80Hz_polar.png
    Prototype_Cardiod_sub_80Hz_polar.png
    38.9 KB · Views: 4,640
  • Sonogram_B&C DE250.png
    Sonogram_B&C DE250.png
    152.7 KB · Views: 4,656
  • Topkast.jpg
    Topkast.jpg
    718.4 KB · Views: 1,965
  • Sonogram_woofer.png
    Sonogram_woofer.png
    116.6 KB · Views: 1,890
Last edited:
For the sub cabinet, I have 2 Peerless XLS10 woofers and a Hypex PSC 2 400. In order to achieve cardioid radiation, I have combined a sealed and dipole enclosure. The dipole is high passed at 50Hz using a 4th order filter in order to limit cone excursion and to have the advantages of a pressure source <50Hz. Free field measurements showed a polar resembling cardioid from 63Hz on up.
1. In order to achieve maximum suppression of LF at rear, I had to delay the monopole sub by 1,9ms. This resulted in less than optimal summation of the two drivers at the front. The dipole woofer is actually positioned 20cm behind the woofer in the sealed enclosure.
The cardioid arrangement you describe will always have less than optimal summation of the two drivers at the front, and can only have maximal attenuation at the rear at one frequency determined by placement and delay.

At low bass frequencies, the cardioid arrangement won't be very effective in small rooms, as the reflected waves off the boundaries changes the phase relationship of the two drivers compared to outdoors or large rooms where boundaries are wavelength distances away.
 
At low bass frequencies, the cardioid arrangement won't be very effective in small rooms, as the reflected waves off the boundaries changes the phase relationship of the two drivers compared to outdoors or large rooms where boundaries are wavelength distances away.

I'm not sure I understand this. I don't think there is any connection between room size, boundary proximity and effectiveness.

I was doing some simulations a couple of weeks ago in the context of killing a wall bounce for a typical wall mounted (but with some finite spacing) system.

In free space we would want to make a cardioid with some range of operation. As long as there is a rear firing null then there is no bounce off of the wall behind (at least on the system axis, not necessarily if you moved to oblique angles).

In the case I was working on I had a small element, say, 6" off the wall and a cancelling unit flush on the wall. As long as I compensated for their respective boundary conditions (one in free space the other boundary mounted) I could achieve full cancelation of the wall reflection.

Everything was tied to the distance of the system from the wall, could be adjusted for any wall spacing, and had nothing to do with an overall room size.

Regards,
David
 
The cardioid arrangement you describe will always have less than optimal summation of the two drivers at the front, and can only have maximal attenuation at the rear at one frequency determined by placement and delay.

At low bass frequencies, the cardioid arrangement won't be very effective in small rooms, as the reflected waves off the boundaries changes the phase relationship of the two drivers compared to outdoors or large rooms where boundaries are wavelength distances away.
Thank you for your reply! Im trying to understand why it would only work at one frequency. Measurements indeed show attenuation around a maximum at 100Hz. See attached off axis repons measured outdoors without gating, at 2meter distance, at 0,45,90,135,180 degrees. Attenuation shows a maximum of over 25dB, which was probably more since the respons was "dancing" on the wind at 180degrees.

Im also having a hard time understanding why the monopole needed around 1,9ms of delay to get max attunuation at the rear, while the monopole is actually positioned more to the front.

In free space we would want to make a cardioid with some range of operation. As long as there is a rear firing null then there is no bounce off of the wall behind
Interesting point. Only, im not sure how close the null would be to the speaker, since the respons of a dipole is different in the nearfield. Tailoring the filter for the dipole to the specific distance to the wall, might eliviate this problem wouldn't it?
 

Attachments

  • Prototype Cardiod sub off axis respons.png
    Prototype Cardiod sub off axis respons.png
    32.3 KB · Views: 756
I'm not sure I understand this. I don't think there is any connection between room size, boundary proximity and effectiveness.

In the case I was working on I had a small element, say, 6" off the wall and a cancelling unit flush on the wall. As long as I compensated for their respective boundary conditions (one in free space the other boundary mounted) I could achieve full cancelation of the wall reflection.

Everything was tied to the distance of the system from the wall, could be adjusted for any wall spacing, and had nothing to do with an overall room size.
David,

Room size will often determine boundary distance to some extent. You adjusted your simulation to account for the boundary distance where the cancelling unit was located, requiring a different set up than what would be effective outdoors or with boundaries at a different distance.

Sounds like you are developing some cool new stuff, how are you liking the new job?

Art
 
Im also having a hard time understanding why the monopole needed around 1,9ms of delay to get max attunuation at the rear, while the monopole is actually positioned more to the front.
These sources will do a lot better at helping you visualize what is going on with your tests than my explanations:

Phase Wavelengths: The End Fire Cardioid Array made visible | Bob McCarthy's Blog

New website Merlijn van Veen, home of the Subwoofer Array Designer calculator

Merlijn really did a nice job with the calculator.
 
David,

Room size will often determine boundary distance to some extent. You adjusted your simulation to account for the boundary distance where the cancelling unit was located, requiring a different set up than what would be effective outdoors or with boundaries at a different distance.

I adjusted to account for boundary distance just because I was modeling a speaker with a given depth and a canceling drive on the wall (a bit like the "pitcher/catcher" bass arrays that have been discussed occasionally). In reality if you create a free standing cardioid pattern then there is a rearward aiming null and the wall can be any distance behind. It is in the null so no rear wall bounce will occur.

Of course this only applies for the listener directly in front of the speaker. As you move to the side you are on a line back to the speaker that is not at the angle with greatest depth of null. Still, for that new angle you could create a hyper cardioid polar from the speaker (with 2 nulls) such that you are once again on the null of the system (actually, relative to the reflected virtual speaker) and the wall will once again disappear.

Sounds like you are developing some cool new stuff, how are you liking the new job?

Art

I'm enjoying it a lot. Thanks for asking.

There are great facilities here and a lot of brilliant people. The environment is very academic and they have me taking an intense DSP class.

The products are very sophisticated in their technology (takes a lot of DSP to get acceptable sound out of a tiny plastic box with 1 1/2" drivers!)

David S.
 
"In free space we would want to make a cardioid with some range of operation. As long as there is a rear firing null then there is no bounce off of the wall behind."

Interesting point. Only, im not sure how close the null would be to the speaker, since the response of a dipole is different in the nearfield. Tailoring the filter for the dipole to the specific distance to the wall, might eliviate this problem wouldn't it?

If anything varies with distance then the distance to the wall is not the issue, but the distance from speaker to listener.

In fact it is the distance from the virtual reflection speaker to the listener.

Remember, what you are setting up is a speaker in front of a boundary. If the boundary is a mirror (for the sound) then you will see both the real speaker and a backwards facing virtual speaker just beyond the mirror (wall). You want to configure that virtual speaker such that at your distance and angle it has created a wide band null that kills its output. By doing so the wall bounce and hence the wall cease to exist.

I don't know how much the dipole varies with distance but I would think that with most realistic scenarios you would close enough to far field conditions.

As to your somewhat narrow null, you need to look at the phase response of your two elements in the rearwards direction. Cancelation requires 2 things: equal strength and opposite phase. I'm guessing the phase curves swing at different angles, so they are only 180 degrees apart for a fairly narrow range.

The easiest way to do this over a wider range is with electronic delay for one of the elements.

David
 
David ... I'm under the impression that neither equal strength or opposite phase is a requirement. I'll site an EV pamphlet which states that the "control" driver (rear) can be 6db lower in output. Drivers (front and rear) are wired "in phase" and require delay applied to the control driver along with appropriate spacing. Since the enclosures are separate, adjusting the position of the rear drivers should shape the null.
 
Last edited:
In general terms we are canceling two vectors so equal strength and opposite phase are a requirement.

If you mean that with the configuration shown, 6dB less drive will get you to the required levels, then that may be true. I don't know enough about this particular configuration.

I've just worked through my own variant of a cardioid array and I did all the tuning by turning one element off, then turning the other off. Adjustments were made until the curves and strengths of the 2 units best matched when viewed from the observation point. Then, when their delay was adjusted for best 180 degree phase spread, the greatest cardioid null was achieved.

Regards,
David
 
These sources will do a lot better at helping you visualize what is going on with your tests than my explanations:

Phase Wavelengths: The End Fire Cardioid Array made visible | Bob McCarthy's Blog

New website Merlijn van Veen, home of the Subwoofer Array Designer calculator

Merlijn really did a nice job with the calculator.
Thank you very much for these links, i'll look into them more closely and try the sub array calculator. Sub arrays for PA use are fascinating business.

I've just worked through my own variant of a cardioid array and I did all the tuning by turning one element off, then turning the other off. Adjustments were made until the curves and strengths of the 2 units best matched when viewed from the observation point. Then, when their delay was adjusted for best 180 degree phase spread, the greatest cardioid null was achieved.
That was my strategy as well, only using outdoor measurements for proof of concept. Only i needed adding delay to the wrong element, not realising the polarity was reversed.

Remember, what you are setting up is a speaker in front of a boundary. If the boundary is a mirror (for the sound) then you will see both the real speaker and a backwards facing virtual speaker just beyond the mirror (wall). You want to configure that virtual speaker such that at your distance and angle it has created a wide band null that kills its output. By doing so the wall bounce and hence the wall cease to exist.
This would mean adding the right amount of delay in order to create a zero at the listener?

But in my case i have one dipole and one monopole. The rear output of the monopole and dipole need to be in phase and of same amplitude at listening position. Im trying to understand how the distance to rear wall or to the listerner would influence the required delay :confused:
 
But in my case i have one dipole and one monopole. The rear output of the monopole and dipole need to be in phase and of same amplitude at listening position. Im trying to understand how the distance to rear wall or to the listerner would influence the required delay :confused:

Okay, if a monopole and a dipole are to give cardioid response then we can ideally use no extra delay and just set their 0 degree (and 180 degree) strengths equal. In theory that will cancel rear firing radiation and give us the cardioid pattern.

In practice the responses don't match as the dipole will be rolling of 6dB per Octave. So lets equalize that to give as close to the same response as possible. Again, if anything varies with distance then we need to view from a distance equal to the distance from the listener to the rear wall virtual speaker (the reflected one). In practice I don't think it matters much as long as your measurements aren't too close to the system. Note that we are going to equalize to match the rear firing part of the monopole, not the usual front firing part.

Now our dipole and monopole aren't coincident in space so we need to adjust for that. It looks like your monopole is in front and your dipole behind. That means, when considering the reversed virtual system (the one we want to cancel) that the dipole is closer and will need some delay applied to it.

In the end, if we can look at the rear radiation of monopole and dipole, set them to the same level, the same frequency response and the same effective delay (but of course of opposite polarity), then we will get perfect broad band cancelation.

Distance to the wall will always be irrelevant since we are finding a way to send no energy in that direction.

Regards,
David
 
Distance to the wall will always be irrelevant since we are finding a way to send no energy in that direction.
David,

There are two sources both sending energy, ideally 180 degrees out of phase which results in cancellation in the rearward direction. Both sources reflect off the rear wall (and ceiling, and floor, and front wall) all with different return path lengths (and relative phase) to the listener position(s). Assuming a stereo pair of LF cardioid speakers and a center listening position, how can one avoid the reduction of the LF cardioid effect from the multiple path lengths, and attendant phase change?

Art
 
We are really just talking about the single bounce case. If we are in half space with only one boundary then there is only one virtual reflected system and we aim the null of that system at our seating position and the boundary goes away (the reflection goes away).

In practice people are trying to use speaker directivity to reduce room modes. If they can get a null in one direction then some of the axial modes will go away. Yes, clearly a device with a single node can not reduce reflections from multiple surfaces and multiple angles from the source but that was never the intention.

It still is true that if we create a null aimed at a wall behind our system then the distance from system to wall is immaterial.

David S.
 
Sigh, it is sad that these theory/simulation compound systems don't work in reality...

First - perfect nulling at 180¤ and perfect summing at 0¤ is possible only if acoustic centers are exactly at same vertical plane. And just believe me, the are not! With delay we can optimize either front or backwards radiation. And then, because radiators are not coaxial, we loose this summing/nulling soon with vertical offset!

Second - a speaker radiates in 3D - not in 2D like simulations (the way we think) Even a perfect cardioid has lots of radiation at 150¤ and less, horizontally and vertically (3D) which means lots of soundwaves sent to walls, floor and ceiling.

Been there, done that! That was my intention... AINOgradient has monopole bass crossed to dipole mid somewhere between 100-200Hz depending on settings. This question is furthermore corrupted with difficulties with indoor measurements...

About frontwall reflections - every speaker has them, but dipoles have most severe. Peaks and dips (summing/nulling) also happen at different frequencies for monopoles and dipoles. The first null is strongest, multiplities are weaker.

However, despite being imperfect systems, cardioid radiation pattern is worth to try!

(Ah, Dave just said this while I was writing!)
 
I think there is a very valid goal in all of this.

If you consider the typical speaker in a typical room situation, the biggest issues in the interaction between the speaker and room are the first 2 reflections: the floor bounce and the rear wall reflection from behind. The Soren Bech tests found these were the only room reflections likely to be audible.

If you deal with those then standing waves may remain but the biggest dips will be gone. A cardioid radiation pattern woofer is one way to eradicate the rear wall bounce.

David
 
In real world (my living room) - I have measured a dipole and monopole speakers, and difference of a sealed box mopole and dipole are not as great as one would think. AINOgradient (avatar) is dipole between 150-3,5kHz.

We see floor bounce at 180Hz and front wall reflection at 320Hz. Here are nearfild measurements (5ms speaker in the middle of the room) and 20/300ms with speaker some 80cm from front wall (actually window!) Peak at 1,8kHz is mystery to me, possibly window glass resonance.

I show also a LEAP simulation that I captured from here The Controlled-Pattern Offset Bipole Loudspeaker
 

Attachments

  • floor front wall reflections.png
    floor front wall reflections.png
    349.9 KB · Views: 1,387
  • 3whifidsp near vs l 5 300ms 148.png
    3whifidsp near vs l 5 300ms 148.png
    86.6 KB · Views: 1,363
  • 3whifidsp near vs l 5 20ms 148.png
    3whifidsp near vs l 5 20ms 148.png
    88.1 KB · Views: 1,355
  • ainogneob v71 near vs far 5 300ms 148.png
    ainogneob v71 near vs far 5 300ms 148.png
    99.9 KB · Views: 1,314
  • ainogneob v71 near vs l 5 20ms 148.png
    ainogneob v71 near vs l 5 20ms 148.png
    90.8 KB · Views: 1,314
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.