Two quick answers you won't like: First, there are several WEB sites to do the calculations for you. Second, they will be wrong, so you will need to use trial and error to get it right. You might also go read the comments on ZAPH buried in one of his speaker designs on why you use a series or parallel filter and when.
BTW, you need to determine the Q of the peak. Careful measurements to find the -3dB points.
BTW, you need to determine the Q of the peak. Careful measurements to find the -3dB points.
TVR, First of all i do like your answer because of it's thruthful. How many times we use a so called calculator only to get it "wrong" So I appreciate the heads up so if I use a calculator it is only to be used as a starting point and i will not be surprised when I have to redo it. Thank you so very much. I can tell you I have already been to 2 websites and got 2 different results, just as you said.
So I will do some careful measurements to determine the Q and go to Zaphs site. Your assistance and honesty is greatly appreciated. I will let you know what I find.
So I will do some careful measurements to determine the Q and go to Zaphs site. Your assistance and honesty is greatly appreciated. I will let you know what I find.
Patience is your friend!
Thanks pano. I have patience and XO parts are relatively cheap so I know with persistence I should be able to attack itbut may take awhile. I'll do a before and after measurement and then modify as needed.
Thanks pano. I have patience and XO parts are relatively cheap so I know with persistence I should be able to attack itbut may take awhile. I'll do a before and after measurement and then modify as needed.
What design tools do you have on hand? Where do you get those 1150 numbers? Where do you want to put the notch filter, on woofer or on the input of crossover?
For 5-6 dB, it should be relatively a deep notch where for effective notch you want a LRC in series (in parallel with load).
At 1150Hz (middle of audio frequency), such notch filter will for sure kill some dynamics/sonic of the speaker. You need to pay attention to sonic versus distortion (due to peak) trade-off. Flattest response is usually not a smart goal here.
For 5 to 6 dB notch you will need 4.7 to 4 Ohm resistor.
For LC, simply use the basic resonance formula: f=1/(2*pi*V(L*C))
You will need L from 1mH to 1.8mH, depends on the Q or shallowness of the notch. 1.8mH will give the sharpest notch (high Q), 1mH will give shallower notch. Sharp notch is dangerous especially when not precise, because it may create an overshoot, augmenting frequency you actually want to dampen.
For 1150Hz, here is the LC combo (using common inductor size):
L=1.8mH, C=10.64uF
L=1.5mH, C=12.77uF
L=1mH, C=19.15uF
This is a theoretical notch on pure flat response curve on both ends. If you want to put the notch on woofer, you must know that the woofer response is decreasing (rolling-off) towards the high frequency end, and the lower frequency side is usually the "troublemaker". Because of this, it is usually better for sonic to use a shallower notch, or have bigger capacitance.
In this case, 1mH-20uF-4R would probably subjectively sound better than 1.5mH-12.77uF-4R.
Yeah, it is better to see real situation. A 6dB notch at 1150 with filter in parallel with the load is a real devil. But the other type of notch will have almost no effect if distortion is the object to be reduced.
OK I will get the -3db pts. As far as impeadance you don't mean the "published" driver impeadance at that frequency do you?
True, I am not sure if this is an artifact of BSC/crossover. i do not have any BSC.
Edge is very simple and easy to use, PCD has a different usage. If you have good frequency and impedance measurements you can use PCD to figure your notch filter. Knowing the cause of the peaks is important, there may be a better way to fix it than the notch filter.
As Jay points out, 1150Hz is going to be tough. You'll hear every little thing you do there, so you might want to make the notch broader and shallower than the FR indicates.
Listen, try, listen.
As Jay points out, 1150Hz is going to be tough. You'll hear every little thing you do there, so you might want to make the notch broader and shallower than the FR indicates.
Listen, try, listen.
Edge is very helpful. Great place to start, but the real world is never quite what you model. Build and measure. Think. Build and measure again. Look at the simulations and see if they can explain what you measure. They are a tool, not an answer.
Ignore the published nominal impedance. You should measure it in the range where the filter will be working. If you use a Zobel, that should be in place.
The reason I bring up BSC, is depending on baffle size, the 1K region is not atypical to be getting a big edge induced boost. Lowering the first pole may do a better job than adding a notch. Sometimes you can just increase the size of the first inductor in the crossover for the same basic effect.
Ignore the published nominal impedance. You should measure it in the range where the filter will be working. If you use a Zobel, that should be in place.
The reason I bring up BSC, is depending on baffle size, the 1K region is not atypical to be getting a big edge induced boost. Lowering the first pole may do a better job than adding a notch. Sometimes you can just increase the size of the first inductor in the crossover for the same basic effect.
Wow. Ok I have lots I want to communicate but in writing it is a bit much. Let me start with this. Someone on the PE forum offered to Sim my XO. The LP (on dayton rs270) is a 4th order. I questioned the 10mh L1. He said it was needed even though calculated values showed it to be 4.7mh. I changed it to 4.7 and the speaker sounds much better but, with the bump. Now i guess that is why it was such a large inductor. For those that are interested in the background info on the speaker and how it came about with all the gory details a link is provided. i hope this gives perspective.
Lambda Speaker Thread
Lambda Speaker Thread
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.