Geddes on Waveguides

There is no requirement for symmetric filter responses at a crossover. The only requirement is that the sound field be smooth and have as little effect of the crossover as possible. The CD on a waveguide does drop like a stone below the crossover frequency, usually far faster than the woofer does above this point. But, so what? If the sum in the far field at all polar locations is smooth and flat then this asymmetery is irrelavent.

There is usually an undesirable resonance from the Compression Driver at the lower edge of its passband (this comes from an ellectrical impedance peak) and so some form of HP is virtually always required. I sometimes use second order and sometimes third. It all depends on what works best in each particular situation. There is no "golden rule" for crossovers. They are such a necessary evil that anything that makes them work better needs to be used. And any design that minimizes their use is to be applauded.
 
soongsc said:
Playing around with wave guide/horn the impedance peak seems to show two peaks blended, impedance phase crosover actually become lower than F0. This seems quite interesting.
It seems very interesting how a horn/guide can effect the driver impedance. Attached is some results. First curve is the raw driver, second curve is one third of the guide/horn from the throat, third curve is the whole horn/guide assembled together and with an RC zobel to flatten the high end of the impedance curve.

Any comments are welcome.

It is amazing how sensitive the match between the driver and horn/guide can be. Driver break-in seems audibly critical. I hope to do some measurements with broken-in and not broken-in drivers some time next week.

Note: direct radiating driver is used.
 

Attachments

  • horned impedance cimparison.gif
    horned impedance cimparison.gif
    16.3 KB · Views: 940
Of course the "load" on a driver affects the electrical impedance this is just fundamental lumped parameter effects. But I don't see how this can tell us much about either the waveguide or the driver as impedance is a lumped parameter quantity and does not show anything regarding wave motion etc. in the waveguide. One CAN see seveer horn resonances in the impedance curve if they exist, but once these have been elliminated, as they are in all of my designs, then the electrical impedance ceases to show much of relavence.

I have never noted serious "break-in" effects either through listening or measurements (except for a LF device whose compliance does shift slightly with break-in). But in a compression driver or HF device I have not seen this effect to be significant.
 
Dear Dr. Geddes,

Last week I was experimenting with home made planar speakers made of thin polypropylene (springy) and polyamide film (stiff). Voice coil can be deposited circularly or linearly (serpentine pattern). Membrane moves like a bow and edge termination is essential.

Now would this work on OS waveguide and what are the requirements?

Issues:


  1. 1) Membrane area larger than throat area, makes back chamber
    2) Both areas equal
    3) Phase plug for a flat diaphragm, is it necessary?
    4) Springy membrane or rigid (small or large excursion)?
    5) Push-pull (0.5 Tesla) or single ended (0.2T)? (former makes cavities)
    [/list=1]
 
I am trying to figure out how the impedance change will effect driver and system performance. I understand that the lower impedance will allow more current to go through the driver, thus improve the power response at the low frequency.

I think the break-in might relate with the surround material. In the beginning, the driver really sounded harsh, when I added the zobel, it did improve the harshness somewhat. But just sending full spectrum audio signals through the driver for a few days, the sound became much smoother. This is the first time I've experienced such break-in difference in any driver, and I've played around with quite a few wide range drivers.
 
jzagaja said:
Dear Dr. Geddes,

Last week I was experimenting with home made planar speakers made of thin polypropylene (springy) and polyamide film (stiff). Voice coil can be deposited circularly or linearly (serpentine pattern). Membrane moves like a bow and edge termination is essential.

Now would this work on OS waveguide and what are the requirements?

Issues:


  1. 1) Membrane area larger than throat area, makes back chamber
    2) Both areas equal
    3) Phase plug for a flat diaphragm, is it necessary?
    4) Springy membrane or rigid (small or large excursion)?
    5) Push-pull (0.5 Tesla) or single ended (0.2T)? (former makes cavities)
    [/list=1]



  1. I really can't comment on the design issues without some extensive study and thats not really viable at this point. I can say that I would tend to make the throat area the same as the diaphragm area. The rest of the questions would take some seriuos calculations to sort out.
 
soongsc said:

I think the break-in might relate with the surround material. In the beginning, the driver really sounded harsh, when I added the zobel, it did improve the harshness somewhat. But just sending full spectrum audio signals through the driver for a few days, the sound became much smoother.

And maybe your ears just got broken-in to this sound character.
 
Since this is the first waveguide I had made, and only one channel, I thought I'd post some impressions:
1. I start to hear the mid frequency tremble and the player breathing when playing some instruments. This is very nice.

2. Noise in the system is more annoying. The effeciency is really high with a 94db direct radiating driver. One player had some increased noise in the 15KHz ~ 30KHz region. This made the system seem very noisy. Another player that was quite in this region had only a slight hiss.

3. Some harshness in the driver is more aparent. I think it's related with the breakup at around 20+KHz, I look forward to treating the driver after I've done some measurements.

4. I hear some honk in the waveguide, but I'me not sure whether it's the design in general or whether it's generated by the disconinuity between the first third from the thoat and the outer 2/3 to the lip. I will also took into it a bit more when I get some measurements. I made a split there to keep costs down while maintaining better accuracy at the throat.
 
I think that its important to understand that Constant Directivity does not mean that the sound stays at the same level as one moves off axis - and then somehow falls to zero at the coverage angle. Waveguides have a continuous drop in level - independent of frequency however - as one moves off axis up until the coverage angle and then the drop is steeper.

This slow drop with angle is exactly what one needs off axis in the toe-in configuration.

The wider the angle of the device (as above) the faster the initial falloff with angle and the slower beyond the coverage angle and this tedns to not be frequency independent - in other words the wider the coverage angle the more the polar response looks like a piston - not surprising.

At about 90 degree coverage (45 degree wall angle) one gets just about the ideal angular falloff. Narrower than this and within the coverage its not falling fast enough, but then it drops like a stone. Wider than this and the falloff with angle is too great.