Hornresp

Hello David

That's OK, so in the case of the combined response, the artifacts are probably due to the phase wraping.

Unwraping the phase should solve this.

Could you send me the SPL and phase used for the pulse response for the following example :

http://www.diyaudio.com/forums/atta...tamp=1239376483

or for other examples as the ones illustrated by Sabbelbacke...

Best regards from Paris, France

Jean-Michel Le Cléac'h



David McBean said:
Just to clarify, the phase shown on the phase response chart is not the same as the phase used in the impulse response calculations. The phase for the combined response impulse response is recalculated, and hopefully is the phase of the combined outputs.

The phase shown on the chart does not change because as Eva has pointed out, it simply shows the phase shift through the driver only. It is not updated at any time.

Kind regards,

David
 
Re: Hornresp Version 21.40

AndrewT said:

use dos to swap folder names back and forth.
Have two or three icons, each uses a different dos file to use the folder you need.

Thanks, I was afraid this would be the solution. Fortunately though.........

David McBean said:

Also, the limit on the maximum number of data records has been increased to 9999 - hopefully this will be enough even for GM :).

Thanks, appreciate it! Actually, now that MJK has withdrawn his MathCad software I've been promoting HR as a basic substitute for MLTLs, etc., so if the 2500+ MC sims are any indication plus all the compound horn sims I plan it might not be if I don't die young ;).............

Speaking of which, any hope of getting an adjustable vent location option to go along with the offset driver option?

BTW, in playing with the new features, I did a simple sealed cab sim (attached), looked at the various graphs, clicked on FIND expecting to be asked if I wanted to save it and instead got a pop-up "runtime error '9' subscript out of range", clicked OK and it closed HR without saving the sim. So far though, in looking up a few impulse plots of existing sims it hasn't happened again.

GM
 

Attachments

  • b&c 8ndl51 (dual) 0.5 qtc sealed.gif
    b&c 8ndl51 (dual) 0.5 qtc sealed.gif
    14.7 KB · Views: 641
Eva said:
There is currently no way to know combined phase and group delay. A real pity for tapped horns.

Hi Eva,

The attached screenprint shows the system phase response for the William Cowan 60 hertz tapped horn with Eighteen Sound 10W400 driver.

The system group delay for the same loudspeaker is given in my next post.

How would you use this information to improve the design?

Kind regards,

David
 

Attachments

  • phase.jpg
    phase.jpg
    49.3 KB · Views: 639
Re: Re: Hornresp Version 21.40

Hi GM,

GM said:
I've been promoting HR as a basic substitute for MLTLs, etc., so if the 2500+ MC sims are any indication plus all the compound horn sims I plan it might not be if I don't die young ;).............

Sorry, 9999 records are all you are going to get :).


GM said:
Speaking of which, any hope of getting an adjustable vent location option to go along with the offset driver option?

Very little at this stage, I'm afraid :).


GM said:
BTW, in playing with the new features, I did a simple sealed cab sim (attached), looked at the various graphs, clicked on FIND expecting to be asked if I wanted to save it and instead got a pop-up "runtime error '9' subscript out of range", clicked OK and it closed HR without saving the sim. So far though, in looking up a few impulse plots of existing sims it hasn't happened again.

Many thanks for the feedback. I haven't been able to generate the error myself - hopefully it was a "one-off". Could you please let me know if it happens again. I have checked the relevant code, and as far as I can see everything should be okay.

I have assumed that you are now using Version 21.50.

Kind regards,

David
 
Greets!

Correct, will do. Not knowing what it meant, I wondered if it had something to do with me having previously overwritten sim 999, then deleting another one, dropping it to 998 using the older version, then making another 999 in the latest version causing a one time glitch in the database conversion.

GM
 
GM said:
Not knowing what it meant, I wondered if it had something to do with me having previously overwritten sim 999, then deleting another one, dropping it to 998 using the older version, then making another 999 in the latest version causing a one time glitch in the database conversion.

Hi GM,

The 'subscript out of range' runtime error is generated when an array element is given an index number greater than the allowable array size.

I suspect that in your case, the error was generated exactly because of the sequence of events you described.

With any luck, it will not happen again.

Thanks for providing the additional information.

Kind regards,

David
 
David McBean said:


Hi Eva,

The attached screenprint shows the system phase response for the William Cowan 60 hertz tapped horn with Eighteen Sound 10W400 driver.

The system group delay for the same loudspeaker is given in my next post.

How would you use this information to improve the design?

Kind regards,

David

The phase plot includes propagation delay, thus making it much harder to interpret due to wrappings, and the scale of the group delay plot is too wide (the negative spikes are a calculation artifact) making quite hard to see the small bumps. I can't see clearly on these plots the information that I would need to try to make this design more crossover-friendly, although it seems to be good already.

When designing a horn it's useful to know in advance where are going to be the bumps in group delay because they require compensation in the crossover for proper summing. Knowing in advance the kind of compensation required is useful. The size of front and rear chambers has a great impact on these bumps too, they may be "shaped".

For example, I'm trying to determine the rear chamber size for my midrange horns that results in the most uniform group delay (or a good compromise between size and the resulting peak) and Hornresp currently does not help me at all because the response after the sound travels through the horn is completely different from shown.

For tapped horns (and scoops) a group delay plot allows to know in advance the irregularities that require compensation on the top octave, so that they can be minimized, moved or the way to compensate them can be known in advance. Measuring this in practice is quite difficult.

For front loaded bass horns, having actual phase and group delay plots allows to evaluate the amount of phase and group delay ripple depending on how much size reduction is applied. This ripple causes bad summing in the crossover region and may need external correction. I usually compensate bottom octave ripple by adjusting the size of the rear chamber, and top octave ripple by adding a last extra section to the horn with an higher cutoff. But again, the current phase and group delay plots tell nothing about whether I'm doing it right or not.
 
David McBean said:

Thanks for providing the additional information.

With any luck, it will not happen again.

Greets!

You're welcome! Thanks for the explanation.

No such luck. :( I 'added' a NEW RECORD of an existing design (attached), converted it to an EXP, clicked FIND to save and move on and instead got the same error message which of course closed HR when I clicked OK. Reloading HR shows that same as last time, it didn't save my new sim.

GM
 

Attachments

  • new record - p-audio fl-12lf 221.21-324.24 hz conic th.gif
    new record - p-audio fl-12lf 221.21-324.24 hz conic th.gif
    15.5 KB · Views: 483
GM said:
I 'added' a NEW RECORD of an existing design (attached), converted it to an EXP, clicked FIND to save and move on and instead got the same error message which of course closed HR when I clicked OK. Reloading HR shows that same as last time, it didn't save my new sim.

Hi GM,

Thanks for this. No matter what I try, I cannot generate the error you are seeing - it's quite frustrating, really :). I will have another close look at the code over the weekend to see if I can find anything that I think might be capable of causing the problem.

Has the error only appeared since the record capacity was increased to 9999?

If anyone else has experienced a problem similar to GM, could they please let me know. Thanks.

Kind regards,

David
 
Eva said:
The phase plot includes propagation delay, thus making it much harder to interpret due to wrappings.

Hi Eva,

Thanks for the comprehensive background information.

Just to clarify, the system phase response model used in Hornresp takes propagation delay into account intrinsically as part of the fundamental calculations - it is not something that is added in separately. I am not sure how it could be satisfactorily isolated out and removed to reduce wrapping without corrupting the overall results. Wouldn't this in effect be changing the system phase transfer function to something that it is not?

Kind regards,

David
 
I think that the easiest solution is to add a manual propagation delay compensation dialog or slider allowing the user to enter any value and then watch the resulting phase and group delay plots with this delay substracted.

By the way, this allows to find out the theoretically best delay compensation to be used with DSP crossovers (which may include some group delay correction in it too).
 
David McBean said:

No matter what I try, I cannot generate the error you are seeing - it's quite frustrating, really :).

Has the error only appeared since the record capacity was increased to 9999?

Greets!

Me neither since I posted it. Understood.

Correct.

GM

edit:

Hmm, I wonder...........I made it happen again as soon as I created a new (ADD) design from a pre 999 one, modded it, then clicked on FIND, but haven't been able to repeat it since the re-boot nor does copying a > 1000 design trigger it (so far). Seems strange that I could only do it once in at least a couple of dozen tries........
 
Eva said:
I think that the easiest solution is to add a manual propagation delay compensation dialog or slider allowing the user to enter any value and then watch the resulting phase and group delay plots with this delay subtracted.

Hi Eva,

Not sure what this would mean in the case of a tapped horn, where multiple acoustic paths exist. I think I will leave things just as they are for the time being, at least until Jean-Michel has had a chance to investigate the feasibility of un-wrapping phase.

Kind regards,

David
 
GM said:
I made it happen again as soon as I created a new (ADD) design from a pre 999 one, modded it, then clicked on FIND, but haven't been able to repeat it since the re-boot nor does copying a > 1000 design trigger it (so far). Seems strange that I could only do it once in at least a couple of dozen tries........

Hi GM,

I had another good look at the Hornresp code on Saturday but still couldn’t find anything wrong. Since I cannot generate the error, and because it only seems to occur intermittently for you, I have added some special error trapping code in an attempt to identify what statement in the program might be causing the problem.

Could you please download a copy of the latest release, and if you get a message similar to the following at any time, advise me of the error code and statement number.

'Subscript out of range' error 9 generated by statement A72.

(In the above example message the error code would be 9, and the statement number would be A72).

If you still get the original error message rather than the one above, and no one else reports a similar problem, then it probably means that the error is machine-specific, and that there is little I can do about it. Would your computer have an AMD (Advanced Micro Devices) processor, by any chance? Also, what operating system are you running?

Kind regards,

David
 
Greets!

Interesting, I started to ask if it might be machine specific and HR just a 'trigger'.

According to the Belarc Advisor software it's a 2.40 gigahertz Intel Pentium 4 running Windows XP Professional Service Pack 2 (build 2600).

Anyway, I've DLed the latest version and will let you know what, if any more problems/error messages pop up.

GM