Dear Dave,
I just checked the offset horn in the new version of the HORNRESP.
There is something I would like to ask.
If I have a Main Horn (which is the horn usualy from the HF driver to the horn exit) and an area S2 as the offset input of the Mid driver into the main horn, at a distance L12 from the HF.
Now if there is a MF driver, it surely has a front volume and an area (hole on the sidewall of the main horn), which is not necessarely the same area, as the area of the main horn at this distance from the HF driver.
So the question is, if I can modell that configuration where the area of MF "compression driver" output is other (usualy smaller) than the area of the main horn at that special location?
Thanks,
Tamas Tako
I just checked the offset horn in the new version of the HORNRESP.
There is something I would like to ask.
If I have a Main Horn (which is the horn usualy from the HF driver to the horn exit) and an area S2 as the offset input of the Mid driver into the main horn, at a distance L12 from the HF.
Now if there is a MF driver, it surely has a front volume and an area (hole on the sidewall of the main horn), which is not necessarely the same area, as the area of the main horn at this distance from the HF driver.
So the question is, if I can modell that configuration where the area of MF "compression driver" output is other (usualy smaller) than the area of the main horn at that special location?
Thanks,
Tamas Tako
Assuming by 'compression chamber' you mean the filter chamber between the driver and horn, then the same way you input a compression horn, input your values in Vtc, Atc and if you're designing an offset driver BLH, then set the compression chamber's Vrc, Lrc = 0, so you can use the COMBINED RESPONSE Tool.
GM
GM
Re: The plot thins 
I measured the parameters of the driver I used, to make sure I was simulating what I was building.
That measured and predicted displacement could differ much puzzled me, since the SPL response was so close to the predictions. And diaphragm displacement is closely related to SPL, as the power delivered by the driver is proportional to diaphragm velocity squared, and diaphragm velocity is again equal to displacement times (2*Pi*f). So if one was off, both should be. I'm happy it turns out that physical relations still holds up.
The losses probably plays a role here, but I think that the fact that the driver is "distributed" can have a greater effect. What I mean, is that the Hornresp model assumes the driver to enter the horn in a point, while in practice it enters the horn over an area. This means that the length of each segment is only defined to the precision of the driver diameter, and the resonances will be somewhat distributed and not as sharp as the model predicts.
There is also a difference between measurements and simulations in the magnitude of the peaks in electrical impedance, this may have the same explanation.
I did make some tests other places on the diaphragm, and there is definitively a difference. One problem with this approach is that the accelerometer is a single axis device, and when mounted on the sloping cone, it will not get full acceleration in the direction of maximum sensitivity. I could try to make a wedge for it, though.
Bjørn
kstrain said:
I measured the parameters of the driver I used, to make sure I was simulating what I was building
.That measured and predicted displacement could differ much puzzled me, since the SPL response was so close to the predictions. And diaphragm displacement is closely related to SPL, as the power delivered by the driver is proportional to diaphragm velocity squared, and diaphragm velocity is again equal to displacement times (2*Pi*f). So if one was off, both should be. I'm happy it turns out that physical relations still holds up
.iand said:
The losses probably plays a role here, but I think that the fact that the driver is "distributed" can have a greater effect. What I mean, is that the Hornresp model assumes the driver to enter the horn in a point, while in practice it enters the horn over an area. This means that the length of each segment is only defined to the precision of the driver diameter, and the resonances will be somewhat distributed and not as sharp as the model predicts.
There is also a difference between measurements and simulations in the magnitude of the peaks in electrical impedance, this may have the same explanation.
iand said:
I did make some tests other places on the diaphragm, and there is definitively a difference. One problem with this approach is that the accelerometer is a single axis device, and when mounted on the sloping cone, it will not get full acceleration in the direction of maximum sensitivity. I could try to make a wedge for it, though.
Bjørn
ttako said:
Hi Tamas,
For an offset driver configuration in Hornresp, the area of the "driver output" into the side of the horn is assumed to be equal to Sd if no throat chamber is present, or equal to Atc if a throat chamber is specified. Note that the area is independent of S2. The schematic diagram attempts to show this.
Hope this helps.
Kind regards,
David
kstrain said:
Thanks, Ken.
This is a follow-up to my theory about the "distributed driver". I ran some simulations, taking the average of the responses for different driver entry points. In effect, I moved the entry point along over a distance equal to the diameter of the driver, both for the front and rear entry points.
The high frequency ripple got smoothed out, but nothing happened to the two first pipe resonances, that sets the dips in the displacement curve. These are set by the quarter wave and three-quarter wave resonance frequencies of the full length of the horn. The dips are still just as deep as the non-averaged response.
It turns out that the idea of damping may be much more credible. I will try to simulate this later.
Bjørn
Greets!
As Tom has implied though, if you use the right driver for the BW, you can end load it (max usable BW) and still have a highly damped response, etc. same as a FLH or 'ideal' BLH in its pass-band.
Here's a 2pi space/1 W 6.5" frame size ~98 dB eff. (calculated) driver in a ~366 L bulk expo sim I posted much earlier in this thread IIRC, so a ~8 dB gain in its ~flat pass-band, though the driver can be scaled up to a larger frame size more suited to high power/SPL if required, but I'll leave it to Ian, PB, et al to define the driver specs required to duplicate (beat?) this theoretical performance in a BR or BP sim of the same bulk if more curious than me.
FWIW, this driver in an optimized FLH of the same Fc outperforms it overall in every Hornresp simmed way plus is smaller at ~276 L, implying it's the 'no-brainer' choice once you move up into the > ~100 Hz BW, though done no research to confirm it, so as always YMMV.
GM
As Tom has implied though, if you use the right driver for the BW, you can end load it (max usable BW) and still have a highly damped response, etc. same as a FLH or 'ideal' BLH in its pass-band.
Here's a 2pi space/1 W 6.5" frame size ~98 dB eff. (calculated) driver in a ~366 L bulk expo sim I posted much earlier in this thread IIRC, so a ~8 dB gain in its ~flat pass-band, though the driver can be scaled up to a larger frame size more suited to high power/SPL if required, but I'll leave it to Ian, PB, et al to define the driver specs required to duplicate (beat?) this theoretical performance in a BR or BP sim of the same bulk if more curious than me.
FWIW, this driver in an optimized FLH of the same Fc outperforms it overall in every Hornresp simmed way plus is smaller at ~276 L, implying it's the 'no-brainer' choice once you move up into the > ~100 Hz BW, though done no research to confirm it, so as always YMMV.
GM
Attachments
Oops! Another 'senior moment' , I forgot which thread I was responding to. IIRC I originally posted it and some other examples in the main TH thread: http://www.diyaudio.com/forums/showthread.php?s=&threadid=97674
GM
, I forgot which thread I was responding to. IIRC I originally posted it and some other examples in the main TH thread: http://www.diyaudio.com/forums/showthread.php?s=&threadid=97674GM
Take a look at the attached pic. Almost all of us are building horns as truncated prisms, but we're telling Hornresp that they’re truncated cones (con).
If you use the "Tapped Horn Wizard" to generate s2 and/or s3, or use the "Schematic Diagram" to compute the horn volume, you're not building what you're simulating. With the gentle flare rates of Tapped Horns, the difference is often small, but several people have had problems.
If you use the "Tapped Horn Wizard" to generate s2 and/or s3, or use the "Schematic Diagram" to compute the horn volume, you're not building what you're simulating. With the gentle flare rates of Tapped Horns, the difference is often small, but several people have had problems.
Attachments
You make an important point, Don. Two straight and two conical sides gives the wavefront expansion of a parabolic horn. Bends etc will also cause the wavefront expansion to differ what is assumed in the simulation, and it will also make it a bit harder to estimate the acoustic length of the horn.
The most important effect of this would be to change the resonance frequencies of the horn. The resonances coincide with the dips in the diaphragm displacement. If one can measure displacement, it is easy to check how far off one is. Unfortunately it is harder to check with electical impedance measurements. At the resonanc frequencies, where the acoustical impedance is at its maximum, the motional impedance (the part of the speaker impedance that is not due to resistance and inductance in the voice coil) is at it's minimum, and the exact spot is masked by the DC resistance.
As can be seen from my measurements, the resonance frequencies differ somwhat from the Hornresp simulation. This is probably due both to length and area expansion differences between the model and the sim.
Regards,
Bjørn
The most important effect of this would be to change the resonance frequencies of the horn. The resonances coincide with the dips in the diaphragm displacement. If one can measure displacement, it is easy to check how far off one is. Unfortunately it is harder to check with electical impedance measurements. At the resonanc frequencies, where the acoustical impedance is at its maximum, the motional impedance (the part of the speaker impedance that is not due to resistance and inductance in the voice coil) is at it's minimum, and the exact spot is masked by the DC resistance.
As can be seen from my measurements, the resonance frequencies differ somwhat from the Hornresp simulation. This is probably due both to length and area expansion differences between the model and the sim.
Regards,
Bjørn
As an addition to my previous post: the sensitivity of the design to changes in expansion law can also be checked by comparing a design using all conical segments, to the same design using all exponential segments, keeping lengths and areas the same. The change of resonance frequencies (as seen in the acoustical impedance plot) will be similar, but in the oposite direction of what would happen with a parabolic expansion. An exponential horn with undersized mouth (like in a tapped horn) will have slightly higher resonance frequencies than a conical horn, while the parabolic horn will have slightly lower resonance frequencies.
If the change in resonance frequencies is large, the horn should be built with this in mind.
Checking my test design this way, the change is so small that the differences between simulated and measured results is more likely due to differences in the acoustic length.
Regards,
Bjørn
If the change in resonance frequencies is large, the horn should be built with this in mind.
Checking my test design this way, the change is so small that the differences between simulated and measured results is more likely due to differences in the acoustic length.
Regards,
Bjørn
@david
I found one minor bug, I guess (or am I just too dull?)
After simulating a "normal" horn with one exp segment, I check "compare previos" and go to schematics. From there on, only the "input parameter" tab is available in "window". Only after calulating once more, all features are available again.
Using V 19.40
I found one minor bug, I guess (or am I just too dull?)
After simulating a "normal" horn with one exp segment, I check "compare previos" and go to schematics. From there on, only the "input parameter" tab is available in "window". Only after calulating once more, all features are available again.
Using V 19.40
Sabbelbacke said:
Hi Sabbelbacke,
As I have said before, no bug in Hornresp is minor, as far as I am concerned
.I am trying to replicate the problem you are experiencing, but have not had any luck so far.
Just to clarify, does the following sequence of events generate the problem for you?
1. Open Hornresp.
2. Click the 'Calculate' button to calculate results for the default record.
3. Select 'Input Parameters' from the Window menu.
4. Re-calculate the results.
5. Select 'Compare Previous' from the Tools menu.
6. Select 'Schematic Diagram' from the Window menu.
If the above sequence of events does not generate the problem, could you please provide more details on exactly what does - thanks.
Q1. Are you using Product Number 1940-080918?
Q2. What is your default result window?
Just a thought - you are not editing the record after the results have been calculated, by any chance? This would reset the Window menu to 'Input Parameters' only.
Kind regards,
David