Hi Ron or anyone who can guide me here,
If you would be so kind, please show me where I am messing up attempting to do a double differentiation on this file. I included the NF and Port Responses just as a reference but have attempted to double differentiate the Port without getting a reasonable result from what I can tell.
Thanks in advance.
Jay
If you would be so kind, please show me where I am messing up attempting to do a double differentiation on this file. I included the NF and Port Responses just as a reference but have attempted to double differentiate the Port without getting a reasonable result from what I can tell.
Thanks in advance.
Jay
Hi JMB,
If your data is in dB's, you need to first change it to a linear form and multiply it by f^2, then change it back to dB's. Use this in Excel;
=20*LOG10(POWER(10,B1/20)*A1^2)
where column A = Frequency, column B = Magnitude and column C = Phase.
Also, each time you differentiate the magnitude data, you also need to change the phase by 90 degrees. For double differentiated data, change the phase by -180 degrees.
Peter
If your data is in dB's, you need to first change it to a linear form and multiply it by f^2, then change it back to dB's. Use this in Excel;
=20*LOG10(POWER(10,B1/20)*A1^2)
where column A = Frequency, column B = Magnitude and column C = Phase.
Also, each time you differentiate the magnitude data, you also need to change the phase by 90 degrees. For double differentiated data, change the phase by -180 degrees.
Peter
Not sure what I am doing wrong but I keep getting a curve that does not seem to reflect a nearfield response combined with a port response. Any chance that someone might be able to just fill in a couple of rows for me in the excel sheet that I uploaded? I really appreciate your patience with me. I am trying to figure this out; it seems quite ingenious.
Thanks again,
Jay
Thanks again,
Jay
Hi JMB,
I hope the TS will excuse my digressing from the subject of his OP. To get back closer to the topic, I downloaded his plots shown in post #14 as shown in the attachment below. In this attachment, I show the NF port, NF driver and the sum of these two responses as originally posted.
I then exported the NF port data in .FRD format and imported it into Excel for processing as described in my post #25, after which I imported it back into LEAP as shown by the red curve in the attachment.
After adjusting the level of the red curve at a frequency of 40Hz to match the level of the TS’s summed response (green curve) at the same frequency, you will see that over the frequency range of 30Hz to 50Hz, the two responses are very well matched. However, above this frequency range, the response deviates considerably, because of the ports resonance at 850Hz, plus things like cross contamination from the driver since the driver’s output is increasing at frequencies above system resonance, while the port’s output is decreasing. The reverse is true at frequencies below system resonance where the port output decreases at a much slower rate compared to the drivers output, which I think is reflected in the close match between the red and green curves over this frequency range.
To verify that double differentiating the NF port output gives the same as the summed output of the NF port and driver responses, I simulated a 10” subwoofer in LEAP and applied the same process as described above on the idealized NF port response (no port non-linearity’s applied) and as shown in the second attachment, it gave an extremely close match to the summed outputs of the port and driver over all the frequency range that I was interested in (up to 500Hz).
So the idea of double differentiating the NF port response appears to be theoretically correct, but in practice, port resonance and crosstalk from the driver may significantly affect the response above system resonance, whereas double differentiating the internal box SPL does not seem to exhibit this problem.
Alf82,
with the above in mind, I would recommend (as Speakerdave said) that you either use the double differentiated internal box SPL or the impedance method to establish system resonance.
Peter
I hope the TS will excuse my digressing from the subject of his OP. To get back closer to the topic, I downloaded his plots shown in post #14 as shown in the attachment below. In this attachment, I show the NF port, NF driver and the sum of these two responses as originally posted.
I then exported the NF port data in .FRD format and imported it into Excel for processing as described in my post #25, after which I imported it back into LEAP as shown by the red curve in the attachment.
After adjusting the level of the red curve at a frequency of 40Hz to match the level of the TS’s summed response (green curve) at the same frequency, you will see that over the frequency range of 30Hz to 50Hz, the two responses are very well matched. However, above this frequency range, the response deviates considerably, because of the ports resonance at 850Hz, plus things like cross contamination from the driver since the driver’s output is increasing at frequencies above system resonance, while the port’s output is decreasing. The reverse is true at frequencies below system resonance where the port output decreases at a much slower rate compared to the drivers output, which I think is reflected in the close match between the red and green curves over this frequency range.
To verify that double differentiating the NF port output gives the same as the summed output of the NF port and driver responses, I simulated a 10” subwoofer in LEAP and applied the same process as described above on the idealized NF port response (no port non-linearity’s applied) and as shown in the second attachment, it gave an extremely close match to the summed outputs of the port and driver over all the frequency range that I was interested in (up to 500Hz).
So the idea of double differentiating the NF port response appears to be theoretically correct, but in practice, port resonance and crosstalk from the driver may significantly affect the response above system resonance, whereas double differentiating the internal box SPL does not seem to exhibit this problem.
Alf82,
with the above in mind, I would recommend (as Speakerdave said) that you either use the double differentiated internal box SPL or the impedance method to establish system resonance.
Peter
Thanks, PLB. So the formula you provided above is the double differentiation formula? I was attempting to use it on my original data and then again on the data that it provided. My results were as yours were, however, with a rising response instead of a relatively flat one, suggesting a lot of interaction between the driver and the port measurement. Obviously, my mic was not deep enough in the port.
I really appreciate you doing this. Again, thanks.
Jay
I really appreciate you doing this. Again, thanks.
Jay
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.