Help with the Series Notch Circuit Block!
I not sure I understand the values that need to be input in the Series Circuit Block.
notch frequency = frequency of the peak
determines notch width = the width of the peak?
notch depth = the number of decibels to reduce by?
I'm modelling a Dayton MB1025-8 midbass driver that has a peak at its top end at about 3200 Hz. I'd like to flatten it by 6 dB or so between about 2800 Hz to 3500 Hz. Should I put in the following values in the circuit block?:
notch frequency = 3200 [frequently referred to as "f"?]
determines notch width = 700 (3500 Hz - 2800 Hz)? ["f2" - "f1"?]
notch depth = 6?
Part of the confusion for me is that some sites provide info and calculators for a "Parallel Notch Filter" etc. and Dickason in the Loudspeaker Design Cookbook calls it a "Parallel Trap Circuit". The circuit topology seems to be the same for all.
I love that Dayton provides the FRD and ZMA files for their drivers! I actually purchased my drivers before coming across XSim.
Thanks!
I not sure I understand the values that need to be input in the Series Circuit Block.
notch frequency = frequency of the peak
determines notch width = the width of the peak?
notch depth = the number of decibels to reduce by?
I'm modelling a Dayton MB1025-8 midbass driver that has a peak at its top end at about 3200 Hz. I'd like to flatten it by 6 dB or so between about 2800 Hz to 3500 Hz. Should I put in the following values in the circuit block?:
notch frequency = 3200 [frequently referred to as "f"?]
determines notch width = 700 (3500 Hz - 2800 Hz)? ["f2" - "f1"?]
notch depth = 6?
Part of the confusion for me is that some sites provide info and calculators for a "Parallel Notch Filter" etc. and Dickason in the Loudspeaker Design Cookbook calls it a "Parallel Trap Circuit". The circuit topology seems to be the same for all.
I love that Dayton provides the FRD and ZMA files for their drivers! I actually purchased my drivers before coming across XSim.
Thanks!
XSim is a general purpose circuit simulator. You need to feed in the impedance (magnitude AND phase response) of your driver as the load for the notch filter. Then use the formulas to get initial values based on the impedance magnitude that corresponds to the SPL peak in the frequency response. That gets you to an approximate design. By varying one of the components you can get an appreciation for how it affects the peak. Then adjust accordingly. You may need to vary components by 10% or more to achieve a good match to the driver response you are trying to flatten.
As discussed below, I'd use 3160 as "fnotch".
From the FRD data, we are –3dB down from the peak at about 2980Hz and 3300Hz. That makes for a bandwidth of 320Hz. The Q is fpeak/BW = 3160/700 = 4.5 or so (moderately sharp as expected).
As discussed later, I'd suggest using at least 7.
The critical part is that the circuit topology is the same for all. The terminology can vary a bit from one text to another, as you've found. The physics of the equations doesn't.
They are very helpful to DIYers in that regard and should be applauded. It does help a lot to ensure that the simulations are as accurate as possible. The only better approach would be to do all the required measurements for yourself with your particular drivers.
Below is a plot of the Dayton Dayton MB1025-8 midbass drive. I can see the peak that you refer to, the one at approximately 3200Hz. I'd guesstimate it's closer to 3100Hz. As luck would have it, Dayton Audio provides an FRD file, which gives the peak response of 102.19dB at 3160.6045Hz. Oh, well, @holdent, your estimate was better than mine!
If I can make a recommendation, I would like to suggest that when using the notch filter designer, that you use the peak frequency of 3160Hz. The peak is quite sharp (moderately high Q), so it's important to get the center frequency of the notch very set quite close to the position of the peak. This will help to ensure that the notch works to drag down the peak.
It was also suggested that a notch of 6dB is being aimed for. I'd suggest at least 7dB, and maybe even 8dB. That way the effect of the notch will blend in with the rolling off in the driver's natural response. That might be an important consideration to try and keep the notched off-axis response as smooth as possible.
That peak that is visually bothering you MAY not be a problem, the program content in that spectrum may be low.
Look for a better speaker element without that peak. Complicated crossovers may help magnitude response but suck on transients.
With a good mid-tweeter that peak could alternatively be attenuated with a lower crossover frequency, and have little effect on the results.
Look for a better speaker element without that peak. Complicated crossovers may help magnitude response but suck on transients.
With a good mid-tweeter that peak could alternatively be attenuated with a lower crossover frequency, and have little effect on the results.
I understand what you're saying, but I don't believe crossover complexity is to blame.
Some speaker issues respond well to being fixed and others don't. Choose the right ones and the complexity of the crossover makes the speaker output more simple.
Complicated crossovers may help magnitude response but suck on transients. Really? And how would that work?
If a complicated network equalizes driver irregularities, the transients will not suffer but improve.
In case you are referring to first order networks: those rarely work with real drivers, because of the inherent limitations of in-box loudspeaker behaviour.
If a complicated network equalizes driver irregularities, the transients will not suffer but improve.
In case you are referring to first order networks: those rarely work with real drivers, because of the inherent limitations of in-box loudspeaker behaviour.
Last edited:
Thanks to Witwald - I initially didn't get the Q right and was getting some crazy results.
To recap - I used the Series Notch circuit block in XSim to tame the peak at approx. 3200Hz in the MB1025-8 midbass. I strayed very little from the initial Xsim calculated values only to fit parts available. I wound up with LCR values of 0.1mH, 30uF, and 9 ohms. That was for an LR2 filter at 2000 Hz. I was able to eliminate it totally using an LR4 filter. Xsim is great because it allows you to see what happens when you change the component values - very helpful in this case.
To recap - I used the Series Notch circuit block in XSim to tame the peak at approx. 3200Hz in the MB1025-8 midbass. I strayed very little from the initial Xsim calculated values only to fit parts available. I wound up with LCR values of 0.1mH, 30uF, and 9 ohms. That was for an LR2 filter at 2000 Hz. I was able to eliminate it totally using an LR4 filter. Xsim is great because it allows you to see what happens when you change the component values - very helpful in this case.
Anyone have files for the following drivers to import to XSIM?
https://www.madisoundspeakerstore.c...evelator-22w/8857t-00-8-woofer-aluminum-cone/
https://www.parts-express.com/Peerl...l-Range-Paper-Cone-Woofer-264-1062?quantity=1
https://www.madisoundspeakerstore.c...evelator-22w/8857t-00-8-woofer-aluminum-cone/
https://www.parts-express.com/Peerl...l-Range-Paper-Cone-Woofer-264-1062?quantity=1
"A compromise should consider everything that matters, even at the expense of what doesn't."
Maybe you can check how your VM handles mouse events and if you can alter the behaviour in settings.
Member
Joined 2003
Vituixcad is the continuation of Xsim, it does everything that Xsim can do plus 100x more, and mouse interaction for connecting crossover components is what you’d expect.
For exact same result of Xsim within Vituixcad, load single response for each driver in driver tab, enter delay in driver tab, and leave driver coordinates on crossover at 0,0,0. That’s it. Includes trace tool and delay solving function for usb mic users in aux > time align.
For exact same result of Xsim within Vituixcad, load single response for each driver in driver tab, enter delay in driver tab, and leave driver coordinates on crossover at 0,0,0. That’s it. Includes trace tool and delay solving function for usb mic users in aux > time align.
I've used it for years, works very well most of the time. I was using something before that much more manual and tedious.
Update. The VM author fixed this two weeks after I noticed the problem. Xsim is again working as new.