John,
Before I made post #36 I actually tried it with my miniDSP to confirm.
It does work correctly with the settings I posted. (Q set equal to 0.8)
Quit running sims and analyzing poles/zeroes and actually program it into your miniDSP and run a real test on your bench to check it.
You're correct, this is not a Q issue per se.....that's why I put the "0.5" in quotes in post #38.
I'm not saying your analysis is incorrect......just that the miniDSP programming is confusing and can lead a user astray.
(As usual, with this sort of thing, there are multiple ways to achieve the desired response.)
You'll note that even the bi-quad filter spreadsheet (lowshelf tab) available on the miniDSP site defines "Q" differently than the "basic" setup within their miniDSP plugin.
Advanced Biquad programming | miniDSP
Cheers,
Dave.
Before I made post #36 I actually tried it with my miniDSP to confirm.
It does work correctly with the settings I posted. (Q set equal to 0.8)
Quit running sims and analyzing poles/zeroes and actually program it into your miniDSP and run a real test on your bench to check it.
You're correct, this is not a Q issue per se.....that's why I put the "0.5" in quotes in post #38.
I'm not saying your analysis is incorrect......just that the miniDSP programming is confusing and can lead a user astray.
(As usual, with this sort of thing, there are multiple ways to achieve the desired response.)
You'll note that even the bi-quad filter spreadsheet (lowshelf tab) available on the miniDSP site defines "Q" differently than the "basic" setup within their miniDSP plugin.
Advanced Biquad programming | miniDSP
Cheers,
Dave.
Yes, that's why I said "(on some) of the miniDSP plug-ins....." in post #36. Please read closely.
Your technique of inputting bi-quad coefficients and comparing them to known results is fine and dandy (and easily done by someone with your expertise) but it shouldn't be that difficult a process for the average/beginning user.
Cheers,
Dave.
In practice it may come out very close to the approach I discussed. But mathematically it can not be. You know how annal I get about this stuff.
The problem with what you are doing is that you just don't know what Q to use with what gain. What I discussed will alway work regardless because it is mathematically correct. It's no different than shifting the poles of the woofer with a shelf. Place a zero at the current woofer pole and a pole where you want it to be.Anal? You? No!
John,
What about folks who aren't using the miniDSP "advanced" plugins and don't have the option of computing/pasting digital coefficients to achieve exact mathematical results? They need to know how to adjust/massage the Q settings (regardless of how meaningful they are) of the "basic" plug-ins to achieve the correct result.
Cheers,
Dave.
John,
What about folks who aren't using the miniDSP "advanced" plugins and don't have the option of computing/pasting digital coefficients to achieve exact mathematical results? They need to know how to adjust/massage the Q settings (regardless of how meaningful they are) of the "basic" plug-ins to achieve the correct result.
Cheers,
Dave.
I'm sorry, but I warned you guys I was annal about this stuff.
The transfer function for a low shelf can be written as
T(w) = (Wz1 +jw) / (Wp1 + jw)
where Wz1 is the frequency where the zero is and Wp1 is the frequency for the pole. The transfer function is bilinear (the ratio of two linear (1st order)functions).
If you cascade two shelves,
T(w) = (Wz1 +jw) / (Wp1 + jw) x (Wz2 +jw) / (Wp2 + jw)
Now it should be pretty obvious that if Wz2 = Wp1 the resulting transfer function is
T(w) = (Wz1 +jw) / (Wp2 + jw)
which is a 1st order shelf with zero at Wz1 and pole at Wp2 and remains bilinear.
For any other values you get
T(w) =[(Wz1 x Wz2 -w^2) + jw(Wz1 + Wz2)] / [(Wp1 x Wp2 -w^2) + jw(Wp1 + Wp2)]
Or,
T(w) =[ Wz1 x Wz2 + jw(Wz1 + Wz2) -w^2] / [ Wp1 x Wp2 + jw(Wp1 + Wp2) -w^2]
which is biquadratic (the ratio of two quadratic (2nd order) functions).
The transfer function for a low shelf can be written as
T(w) = (Wz1 +jw) / (Wp1 + jw)
where Wz1 is the frequency where the zero is and Wp1 is the frequency for the pole. The transfer function is bilinear (the ratio of two linear (1st order)functions).
If you cascade two shelves,
T(w) = (Wz1 +jw) / (Wp1 + jw) x (Wz2 +jw) / (Wp2 + jw)
Now it should be pretty obvious that if Wz2 = Wp1 the resulting transfer function is
T(w) = (Wz1 +jw) / (Wp2 + jw)
which is a 1st order shelf with zero at Wz1 and pole at Wp2 and remains bilinear.
For any other values you get
T(w) =[(Wz1 x Wz2 -w^2) + jw(Wz1 + Wz2)] / [(Wp1 x Wp2 -w^2) + jw(Wp1 + Wp2)]
Or,
T(w) =[ Wz1 x Wz2 + jw(Wz1 + Wz2) -w^2] / [ Wp1 x Wp2 + jw(Wp1 + Wp2) -w^2]
which is biquadratic (the ratio of two quadratic (2nd order) functions).
Anal? You? No!
John,
What about folks who aren't using the miniDSP "advanced" plugins and don't have the option of computing/pasting digital coefficients to achieve exact mathematical results? They need to know how to adjust/massage the Q settings (regardless of how meaningful they are) of the "basic" plug-ins to achieve the correct result.
Cheers,
Dave.
They don't need to adjust the Q. They do need to know what Q is correct for a 1st order shelf. It is either 1 or 0.5 depending on the version. But once that is figured out they will always get the correct result doing it the way I (and SL,
) would.Isn't that what I just said?
They DO need to adjust "Q"....IF it's not correct for the plug-in version they're using. Your recommendation (above) for Rich's setup will work fine for his plug-in, but it won't work correctly in another.
Steve,
You can tell by inspection of the plug-in screen if programming a simple filter.
Here's an example where I selected a simple 12db shelving filter with 141Hz center frequency and Q=1 in both my 2-way Advanced plug-in and my 2x8 nanoDIGI plugin. (Note the difference.)
The 2-way advanced plugin is correct and would yield the proper first-order shelving action. However, the nanoDIGI plugin would need to be modified by reducing Q to 0.5 to achieve the equivalent result.
Actually, since these plugins both default to 0.5 it's the 2-way advanced plugin that needs to be modified and not the nanoDIGI plugin. Obviously I don't have all the miniDSP plugin's so you need to check the particular version you have.
Hope that helps.
Cheers,
Dave.
Attachments
Right I had a go at Combining Frequency Responce's out of .FRD files of the front and rear opening of the H-frame with the EQ applied.
Does this look right?
Does this look right?
An externally hosted image should be here but it was not working when we last tested it.
I think I did it wrong in the previous post,
My second attempt at summing two responses with delayed rear for H-frame after EQ was applied.
My attempt at summing two responses for the mid range baffle with one delayed, prior to importing the .FRD into ABC Dipole spreadsheet.
And a screen shot of Minidsp showing the EQfor the mid range as calculated in ABC Dipole.
And the speaker assembled for testing, not finished yet.
My second attempt at summing two responses with delayed rear for H-frame after EQ was applied.
An externally hosted image should be here but it was not working when we last tested it.
My attempt at summing two responses for the mid range baffle with one delayed, prior to importing the .FRD into ABC Dipole spreadsheet.
An externally hosted image should be here but it was not working when we last tested it.
And a screen shot of Minidsp showing the EQfor the mid range as calculated in ABC Dipole.
An externally hosted image should be here but it was not working when we last tested it.
And the speaker assembled for testing, not finished yet.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.