Hi,
I have been looking through a German DIY audio thread about making a pair of speakers similar to the Boenicke W5. One of the guys on there shows his crossover he uses for the drivers...
Please can someone explain the LCR network connected across the Fountek FE85 river.
He says...'
Using the calculator... https://www.diyaudioandvideo.com/Calculator/SeriesNotchFilter/
I enter:
Driver Re = 6 ohms (DC resistance from datasheet)
fs = 5200Hz (even though its resonant frequency is 125Hz in the datasheet)
Qes = 0.67
Qms = 2.13
This gives an answer of 7.26uF, 0.12mH and 7.88 ohms.
Any help on this would be much appreciated. I have a fairly good understanding of the basic crossover concept, but not this part of it.
I have been looking through a German DIY audio thread about making a pair of speakers similar to the Boenicke W5. One of the guys on there shows his crossover he uses for the drivers...
An externally hosted image should be here but it was not working when we last tested it.
Please can someone explain the LCR network connected across the Fountek FE85 river.
He says...'
LCR is "notch" filter, meaning it will squeeze just a certain frequency peak - and with these values that is around the 5.2khz, where the FE has some bad pick.'
I don't quite understand this as I though the series notch filter was to sort out the impedance spike at its resonant frequency and can't understand where the value for R comes from.
Using the calculator... https://www.diyaudioandvideo.com/Calculator/SeriesNotchFilter/
I enter:
Driver Re = 6 ohms (DC resistance from datasheet)
fs = 5200Hz (even though its resonant frequency is 125Hz in the datasheet)
Qes = 0.67
Qms = 2.13
This gives an answer of 7.26uF, 0.12mH and 7.88 ohms.
Any help on this would be much appreciated. I have a fairly good understanding of the basic crossover concept, but not this part of it.
Sorry, I don't understand which is your doubt. If you use those LRC values, or approximate, then it will short out this frequency, as you stated. The link you paste is unusable for me, doesn't conduct to anywhere.
Attachments
Last edited:
In case the picture isn't coming up...
I am not saying that this does not work, I just don't understand where the value of 2.7 comes from.
I am not saying that this does not work, I just don't understand where the value of 2.7 comes from.
OK, it crates a "dip" at the resonant frequency of L and C whose depth is adjusted by the value of R and the rest of the resistances in the circuit. Don't forget subtract from R, the DC resistance of the coil, if it is noticeable.
I think that's used to cancel the impedance rise at resonance rather than a frequency response notch - - so its an impedance notch filter
- the flattened impedance allows use of first or odd order passive highpass networks without the problem of interaction with the Z peak, which can "boost" LF output and shift the xover F way down.
The Crossover Design Cookbook Chapter 3: Speaker Motors and Crossovers
- the flattened impedance allows use of first or odd order passive highpass networks without the problem of interaction with the Z peak, which can "boost" LF output and shift the xover F way down.
The Crossover Design Cookbook Chapter 3: Speaker Motors and Crossovers
Last edited:
OK. So using the equation freq = 1 / (2 pi * root (LC)), the result is 5571Hz.
So, at this frequency the circuit conducts through the 2.7R, which is an easier route for the current to flow than through the speaker, so the speaker output is less at this frequency.
If the above is correct, by changing the value of R, you can adjust the amount of 'bump' in the speakers response?
So, at this frequency the circuit conducts through the 2.7R, which is an easier route for the current to flow than through the speaker, so the speaker output is less at this frequency.
If the above is correct, by changing the value of R, you can adjust the amount of 'bump' in the speakers response?
This works because the 10uF series capacitor still has about 3 Ohms impedance at 5.5kHz.
This allows the two branches of the circuit (RLC, driver) to interact. Otherwise they would
just be in parallel across the voltage source, with no interaction.
Last edited:
Yeah, that's what I originally thought as that's what it says on all the series notch online calculators. That's what I couldn't understand about the original post in the German website. He says it has a bad 'pick', I was thinking he meant frequency spike, but perhaps he meant impedance spike.
The assumption in my previous post was wrong. At 5.5KHz, the notch filter will allow current through it, but it would not reduce current through the speaker, as they are in parallel.
Ah, so at 5.2KHz, the 10uF capacitor has as impedance of 3 ohms...(from this website https://www.allaboutcircuits.com/tools/capacitor-impedance-calculator/). The impedance of the drive at 5.2KHz (from the datasheet is 10ohms), so 10 ohms in parallel with the 2.7ohms gives 2.1ohms.
Total impedance of that part of the circuit at 5.2KHz is 3ohms + 2.1ohms = 5.2ohms?
Is this anywhere near correct?
The capacitors' and the inductor's impedances also cause phase shifts and add as vectors,
so the result will in general be different than just adding the magnitudes of the impedances.
Also, driver impedance variation can affect the results, in addition to the driver's DC resistance.
It's easiest to simulate the crossover to see the results.
Last edited:
CircuitLab is pretty easy, so give it a try. It's also a web application.
You can use it for free without saving results.
Last edited:
I have LTspice already. Never knew you could use it for crossover design. I'll have a play with that.
Sure, it can use any electronic circuit, active or passive, even with tubes.
The results are only as good as the circuit model, though.
Be careful with a series LCR used as an in-bandwidth notch. This can cause a dip in impedance that could be at a very low magnitude. In essence you are paralleling the 2.7 ohm with that of the speaker's magnitude at the resonant frequency of the notch.
Parallel notches attenuate rather than short out, so they are safe on in-bandwidth notch impedance scenarios, but might not cut a metal cone resonance with enough gusto.
Later,
Wolf
Parallel notches attenuate rather than short out, so they are safe on in-bandwidth notch impedance scenarios, but might not cut a metal cone resonance with enough gusto.
Later,
Wolf
I can understand using the series notch for impedance matching, but surely a parallel notch is what is really needed to actually take out peaks on the drivers response.
Not necessarily, sometimes one network is better than the other. I had a metal flat cone once that I tried fixing with a parallel notch. Since the frequency of issue was only attenuated, it still got through the network and affected the response. I tried 3 different notches trying to get it to play nice. I then strapped an LCR across it dialed in for the right frequency, and the problem vanished. I had to watch my impedance since it was an in-bandwith issue, but it worked where the other did not. Sometimes shorting out the issue before the driver creates it is a good thing.
Later,
Wolf
Later,
Wolf
- Status
- Not open for further replies.