Skip to main content ↑ ↓
Topic: Crossover cascading (Read 18549 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Crossover cascading

Reply #50 –
The folks here are an excellent source of information, but we are having a difficult time understanding why this is so difficult for you, so I will try one more time.

I think the problem may be that you are under the impression that an 80 Hz low pass filter passes everything below 80 Hz and removes everything above that. While it is possible to design a filter that approximates that behavior, that is not the kind of filter we are talking about here.

Most likely this would be a filter that passes frequencies below 80 Hz almost 100%, but it also passes higher frequencies to a lesser extent. 160 Hz, for example, might be attenuated to 1/2, 320 Hz to 1/4, 640 Hz to 1/8, etc. Now if you combine that with a 500 Hz low pass filter, that 640 Hz is attenuated even more. It is still not eliminated entirely, and the difference may not be audible, but it is certainly measurable.

Crossover cascading

Reply #51 –
Perhaps where the confusion is coming from is that when I discussed this with my friend, he told me that in the same way that two crossover filters cascading (combining) if, for example the AVR and subwoofer low-pass filter were both engaged, would apply to speaker crossover and AVR. If the passive crossover in a speaker is isolated from the active crossover in the AVR then I just don't get how there could be any cascading.

When I argued that two LP filters combining would be a bad thing he's argument was that the passive crossovers in a speaker also cascade with an AVR so basically there would also be cancellation going on. I don't see this as the same thing. I wish someone would understand what I'm trying to say. I'm sure the effects are measurable. There will always be a measurable effect, but there is no actual cancellation going on at the kinds of frequencies we are discussing.

Crossover cascading

Reply #52 –
Quote
I think the problem may be that you are under the impression that an 80 Hz low pass filter passes everything below 80 Hz and removes everything above that. While it is possible to design a filter that approximates that behavior, that is not the kind of filter we are talking about here.


I am aware that a crossover is not a brick wall. It has a slope with a rate of attenuation. With low-pass filters it's usually 4th-order, but if it's 4th-order at 80 Hz, by the time it reaches the kinds of frequencies I'm talking about in a passive speaker crossover, it would be too far down to interact in any meaningful way. So while I agree there may be measurable interaction, however slight, there is no possible way there could be actual cancellation or any audible interaction given how far apart the crossover frequencies are.

In that, I believe my friend is COMPLETELY wrong. I hope I have explained myself more clearly.

Crossover cascading

Reply #53 –
Quote from: Yahzi on
I am aware that a crossover is not a brick wall. It has a slope with a rate of attenuation. With low-pass filters it's usually 4th-order, but if it's 4th-order at 80 Hz, by the time it reaches the kinds of frequencies I'm talking about in a passive speaker crossover, it would be too far down to interact in any meaningful way. So while I agree there may be measurable interaction, however slight, there is no possible way there could be actual cancellation or any audible interaction given how far apart the crossover frequencies are.


The filters will reinforce each other. It might be measurable, it might be audible - or not. They will still reinforce each other.

Quote
In that, I believe my friend is COMPLETELY wrong. I hope I have explained myself more clearly.


Clearly not, as it seems from the above that you are admitting that your friend is right.

Crossover cascading

Reply #54 –
Quote from: Yahzi on
But I don't understand how filters can combine if they are operating in two different ranges! How can the slopes combine if one is at 80 Hz, for eg, and another is at 500 Hz?


Because on starts a drop a bit above 500 Hz and goes down, and the other joins in a bit above 80 Hz. They both combine at, let's say, 25 Hz.

Quote
I refuse to believe my friend is right.


I think we get that.

Quote
Speaker crossovers CANNOT cascade with the amplifier crossover.


If you say so...

Quote
Goodness, I had a discussion with a few EE's in another thread and they agreed that the filters cannot cascade. Now you are saying my friend is correct. WTF?


Welcome To Finland.

Crossover cascading

Reply #56 –
Quote from: julf on
Because on starts a drop a bit above 500 Hz and goes down, and the other joins in a bit above 80 Hz. They both combine at, let's say, 25 Hz.

Please stop with the inaccurate to the point of being nonsensical filter responses!
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

Crossover cascading

Reply #58 –
Well greynol, what am I supposed to do? The EE is telling me one thing. You guys are telling me something else. Who am I supposed to trust?

Crossover cascading

Reply #59 –
I'm an EE (so is Arny and likely others).  I really don't care who you trust. Go back and read about how I feel about your trying to air your petty laundry over here if I haven't already made myself clear about it.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

Crossover cascading

Reply #60 –
Well thanks for the conflicting information. You really were helpful! I'm done with this thread. At least I know in future where NOT to go for any advice.

Crossover cascading

Reply #61 –
Just reading your posts you get conflicting information, so I'm not at all surprised.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

Crossover cascading

Reply #62 –
Quote from: greynol on
Please stop with the inaccurate to the point of being nonsensical filter responses!


I think it would be helpful if you could point out in what way my responses are inaccurate and/or nonsensical. Mostly I am just struggling with understanding the conflicting information / questions from the OP.

Crossover cascading

Reply #63 –
For one, you keep talking about filters that are seemingly highpass filters, whereas the OP has been asking about lowpass filters.

Then there are things like this:
Quote from: julf on
You have a 12 dB/oct filter at 600 Hz. Down at 80 Hz that filter will provide approximately 96 dB attenuation.
By my reckoning, the actual figure would be just above –48 dB. The filter should already have cut by 12 dB at the nominal cutoff frequency, and then it attenuates by another 12 dB for every octave below that. So we have three further rounds of 12 dB: one at 300 Hz, one at 150 Hz, and one (just below) at 75 Hz. So, 12 + 36 = 48 dB of attenuation. You should now see that the figures you gave correspond to the response of a filter with a slope of 24 dB/octave.

This is quite basic stuff, so it’s best not to get things like this mixed up whilst attempting to explain things to someone who is greatly confused about similarly elementary concepts… not to mention ever more obstinate as time goes on.

Crossover cascading

Reply #64 –
Perhaps you can provide either some math or a graphical representation on where you get 25 Hz in this latest example or how you arrived at 96 dB of attenuation in the previous example.  Until you do, I contend that these numbers make no sense (at least to me and most likely to the OP); and as such are not helping. IME these numbers are the only obvious source of contradictory information in the discussion that isn't coming from the OP. Considering everything else you've said was extremely helpful, I'm perplexed by this aspect of your contribution.

EDIT: Beaten by db1989.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

Crossover cascading

Reply #65 –
Quote from: db1989 on
For one, you keep talking about filters that are seemingly highpass filters, whereas the OP has been asking about lowpass filters.

Yes - seems I misread the posts from the OP with regards to the direction of the filters.

Quote
You should now see that the figures you gave correspond to the response of a filter with a slope of 24 dB/octave.

Indeed. Not sure why I was off by a factor of 2 - teaches me not to try to do numbers in my head without checking them.

The 25 Hz point was just any random point below the cut-off point of both filters (assuming, erroneously, they were high pass filters).

Quote
This is quite basic stuff, so it’s best not to get things like this mixed up whilst attempting to explain things to someone who is greatly confused about similarly elementary concepts… not to mention ever more obstinate as time goes on.

Have to agree. Should not post responses in the "I'll just quickly post something while waiting for the compile to finish" mode. My apologies. Thanks both to you and greynol for pointing out the mistakes.

Crossover cascading

Reply #66 –
Quote from: julf on
The 25 Hz point was just any random point below the cut-off point of both filters (assuming, erroneously, they were high pass filters).

Assuming they are high-pass filters and assuming that by "combine" you mean the point at which the slope changes, that would occur at 80Hz, not at some random point below 80Hz.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

Crossover cascading

Reply #67 –
Quote from: greynol on
Quote from: julf on
The 25 Hz point was just any random point below the cut-off point of both filters (assuming, erroneously, they were high pass filters).

Assuming they are high-pass filters and assuming that by "combine" you mean the point at which the slope changes, that would occur at 80Hz, not at some random point below 80Hz.


The point I was trying to make was that at that random point, below the cut-off point of both filters, the gain would definitely be affected by both filters - so the effects would "combine".

As English is my third language, I might not always manage to express myself as clearly as I would like. Let me try with a graph instead, and hopefully I have now understood the situation the OP described correctly, and got all the numbers right...

So, we have 2 low-pass filters. The one in the AVR has a cut-off frequency at 80 Hz, and the one in the speaker has a cut-off frequency of 500 Hz. I used scilab to plot the resulting frequency response curves - for both filters separately, and for the combination of both in series. Does this help?

Crossover cascading

Reply #68 –
Quote from: julf on
The point I was trying to make was that at that random point, below the cut-off point of both filters, the gain would definitely be affected by both filters - so the effects would "combine".

And the point that various members have being trying to make is that the gain will – equally “definitely” – be affected at any point that is reached by the slopes of both filters. Including yourself! As you said in your first reply, “the signal will be affected by the transfer functions of both filters. If the crossover frequencies are far apart, the interaction will be small, but it will be there nevertheless.”

Your latest invocation of a “random point” seems to refer to some point at which this interaction becomes noteworthy. I don’t wish to speak for greynol or anyone else, but I think that’s what is causing confusion/bemusement. The conversation is about defining the point at which the effects of any two filters combine, not some subjective definition of when that effect is significant. The filters will already be interacting at any frequencies where they are both applying attenuation, which will presumably be some point before the nominal cut-off of the higher LPF, to a degree depending upon the specific way in which it is implemented.

So, the “random point” is either well-defined for any two filters of known implementation, or it’s a subjective marker for when the (always present) interaction becomes worthy of discussion. If the latter, what are your criteria?

Crossover cascading

Reply #69 –
Quote from: db1989 on
And the point that various members have being trying to make is that the gain will – equally “definitely” – be affected at any point that is reached by the slopes of both filters. Including yourself!


And I still agree with that.

Quote
Your latest invocation of a “random point” seems to refer to some point at which this interaction becomes noteworthy. I don’t wish to speak for greynol or anyone else, but I think that’s what is causing confusion/bemusement. The conversation is about defining the point at which the effects of any two filters combine, not some subjective definition of when that effect is significant. The filters will already be interacting at any frequencies where they are both applying attenuation, which will presumably be some point before the nominal cut-off of the higher LPF, to a degree depending upon the specific way in which it is implemented.

So, the “random point” is either well-defined for any two filters of known implementation, or it’s a subjective marker for when the (always present) interaction becomes worthy of discussion. If the latter, what are your criteria?


My point was simply to show to the OP that there are some frequencies where the filters *very clearly* interact. Of course they interact to some degree at any frequency.

I think we keep flogging a rather dead horse. I agree with your view about the filters combining, and the 25 Hz point was a failed attempt at explaining the combinatorial effect. Hopefully the scilab plot explains it better.

Crossover cascading

Reply #70 –
Quote
Hopefully the scilab plot explains it better.
Very much so. A graph is worth a thousand words.  Thanks for taking the time to make it!
 

Crossover cascading

Reply #71 –
Quote from: db1989 on
A graph is worth a thousand words.  Thanks for taking the time to make it!


The least I could do after muddling up the thread so badly...
 
SimplePortal 1.0.0 RC1 © 2008-2018