The choke is not bad. I am building completely new crossovers on new mounting boards. The original crossovers have the autotransformers and capacitors potted in wax. I will leave them as-is in case I ever want to return the speakers to their original configuration
Yes, I see the paperwork says crossover 2kHz. The tweeter impedance will shift things some, and the series 8uFd/2.5mH at the tweeter tends to "notch" just above 1kHz, also the spec may be acoustic crossover which includes driver and horn drop-off. Since there is no filter at all on the woofer, the design is empirical worked from what the woofer gave and going from there.
Thank you.
I have seen the term "notch" used but never quite understood it. Is it possible to give a dumbed-down explanation for a newcomer?
I have seen the term "notch" used but never quite understood it. Is it possible to give a dumbed-down explanation for a newcomer?
PRR may know more about this particular speaker as far as horn and driver behavior.
By notch, he may mean a sharp peak or dip in frequency response.
The notch in this case may be caused by a peak in HF driver impedance. Compression drivers may have two or even three impedance peaks close to the operating range.
By notch, he may mean a sharp peak or dip in frequency response.
The notch in this case may be caused by a peak in HF driver impedance. Compression drivers may have two or even three impedance peaks close to the operating range.
Red is the simple 8uF+3mH C/L filter, a smooth roll-off.
Green adds the 8uF+2.5mH+5r notch filter, working in context of the 8uF+3mH filter. It sucks-out around 1.1kHz, and incidentally bumps-up 3kHz.
Since the woofer is likely to be a bit strong at 1kHz (cone diameter), this keeps the tweeter off-stage in a key part of the vocal band where the woofer is still strong.
Parameter 'rset' is '5' for as-designed value, much-much-much bigger (50000) to defeat the notch for comparison. (Easier than a simulated switch in this software.)
Green adds the 8uF+2.5mH+5r notch filter, working in context of the 8uF+3mH filter. It sucks-out around 1.1kHz, and incidentally bumps-up 3kHz.
Since the woofer is likely to be a bit strong at 1kHz (cone diameter), this keeps the tweeter off-stage in a key part of the vocal band where the woofer is still strong.
Parameter 'rset' is '5' for as-designed value, much-much-much bigger (50000) to defeat the notch for comparison. (Easier than a simulated switch in this software.)
Thank you PRR, I accept what you say in the cold light of day.
It is unexpected to see a transformer used that way. It's also unusual to see the avoidance of series resistance and I think it's something to question. I know that has been a trend, and perhaps I've seen it mentioned in the context of these crossovers although I'd have to confirm that.
Level in dB, here shown as a Voltage but it's sufficiently equivalent to sound pressure.
I would tend to expect what daqvin_carter has said. Impedance peaks tend toward response peaks so you remove them.
Here is a neat calculator for RLC traps. Can be used to smooth a resonance peak on a midrange or tweeter.
https://www.lautsprechershop.de/tools/t_schwingkreis_en.htm
Does not get used often but sometimes it is needed.
https://www.lautsprechershop.de/tools/t_schwingkreis_en.htm
Does not get used often but sometimes it is needed.
I believe the resistor switch and the autotransformer switch are to be used in tandem. I'll assume the tweeter is 8 ohms.
When connected to the autotransformer -0 db tap, the resistor switch should be open.
Since the autotransformer has no affect on the reflected impedance the load on the 8uf resistor is 8 ohms.
When connected to the -3 db tap, the resistor switch should connect the 20 ohm resistor. Since the autotransformer's reflected impedance is now 16 ohms, this parallel resistor brings the load on the 8uf resistor back to 8.9 ohms.
When connected to the -6 db tap, the resistor switch should connect the 10 ohm resistor. Since the autotransformer's reflected impedance is now 32 ohms, this parallel resistor brings the load on the 8uf resistor back to 7.6 ohms.
When connected to the autotransformer -0 db tap, the resistor switch should be open.
Since the autotransformer has no affect on the reflected impedance the load on the 8uf resistor is 8 ohms.
When connected to the -3 db tap, the resistor switch should connect the 20 ohm resistor. Since the autotransformer's reflected impedance is now 16 ohms, this parallel resistor brings the load on the 8uf resistor back to 8.9 ohms.
When connected to the -6 db tap, the resistor switch should connect the 10 ohm resistor. Since the autotransformer's reflected impedance is now 32 ohms, this parallel resistor brings the load on the 8uf resistor back to 7.6 ohms.
I modeled the circuit (without the 1125hz notch) in LTspice. The load on the capacitor does seem to be consistent.
Of course. Of course.
This remarkably simplified the problem. I thank you for computing the parallel effect.
Not sure what 8 microFarad resistor? Typo?
It is not unknown, or wrong, to put taps on a filter coil. RF designers do it all day and all night.
This box has several roots. Like all classic JBL there is a big toe in 1929 theater speaker tradition. And then a whole leg from the new guys, Locanthi and Eargle. In theater systems, power-wasting resistors were seen as a sin; however there is not a standard autotransformer. I think a lot of balancing was ad-hoc, also "good enough". (Later PA systems grew low-level EQ filters, before wide use of 27-band and digitals.) I think the post-1960 JBL geeks had one ear on the sound and another ear on the bottom line. The combo choke/level-set may just have been the penny cheaper. (But the DP switch and 20r or 10r resistors do smell like a bodge-- no 'modern' amplifier of that day would mind an impedance rise.)
Not only a typo, but some copy and paste of the typo.
Can't figure out how to edit that post 🙁
Can't figure out how to edit that post 🙁
PRR, I appreciate your input. Indeed, it was the parallel resistors that tipped me, that while they work, they seem a little out of place in the scheme of things.
The autotransformer is an inductor, and is part of the second order high pass to the tweeter.
Without the resistor, one could switch to a lower value capacitor with each tap. 8uf for -0 db, 4uf for -3db, and 2uf for -6db. Then, when you use the -0 db tap, the entire 3mh of the "inductor" is the shunt part of the filter. However, when you use the -3 db tap, only 70.8% of the inductor windings are used,, or 1.5mh. With the -6db tap, it's half the windings, or .75mh. Since the inductance is not increasing at the same ratio as the capacitance is decreasing, the transfer function differs.
For reasons I can't seem to figure out, the parallel resistor will actually keep the inductance close to the full inductance regardless of the tap used. That's why the curves above are so symmetrical.
I attached a graph of what changing the capacitance alone would look like.
Without the resistor, one could switch to a lower value capacitor with each tap. 8uf for -0 db, 4uf for -3db, and 2uf for -6db. Then, when you use the -0 db tap, the entire 3mh of the "inductor" is the shunt part of the filter. However, when you use the -3 db tap, only 70.8% of the inductor windings are used,, or 1.5mh. With the -6db tap, it's half the windings, or .75mh. Since the inductance is not increasing at the same ratio as the capacitance is decreasing, the transfer function differs.
For reasons I can't seem to figure out, the parallel resistor will actually keep the inductance close to the full inductance regardless of the tap used. That's why the curves above are so symmetrical.
I attached a graph of what changing the capacitance alone would look like.