About 30 years ago I've built a pair of bookshelf speakers based on Audax drivers. Recently, I got back to diyaudio during Covid. I felt It was dull so I bypass R1 but it is too hot. I've used it like that for few years, now the tweeter is dead on one channel.
The original Audax tweeter is hard to find now but there are similar design from local manufacturers with 4ohms impedance. I'm thinking of just put an attenuating resistor in series with new tweeter. Maybe a 2 ohm resistor so about -3.5dB at 6 ohm instead of -6dB at 8 ohms in original design.
Would that work? Do I need to think about phases?
You 've got to analyze the two paths. If you want to change the tweeter, what you got? A cap, a R net, a RCL net. The 3 'blocks' are probably unsuitable for the new tweeter. At least, the first performs a bland high-pass, the second attenuates and the third, the notch filter is frequency-centered on the singular tweeter.
A couple issues:
1) assuming an 8 ohm load, the original L-pad looks like it was around 3 dB, not 6 dB. R1 and R2 work together to present a more constant impedance to the crossover. By varying them both, you can change the tweeter level without changing the impedance the crossover sees. If you just bypassed R1, I think you defeated part of the function of the attenuation circuit and shifted the cross point. As the circuit appears to me, anyway. Sometimes it's hard to draw a firm line between different circuit elements. It helps to model the crossover.
2) similarly, if you change from an 8 ohm tweeter to a 4 ohm, the cross point is going to shift some. If the L-pad is in the circuit and working correctly, the change may not be that significant. Or you could adjust the R1 and R2 values to compensate for the tweeter impedance change (and likely voltage sensitivity difference between the 4 and 8 ohm versions).
None of this means you can't do what you are proposing, just that you'll get a better result if you understand the circuit a bit better.
1) assuming an 8 ohm load, the original L-pad looks like it was around 3 dB, not 6 dB. R1 and R2 work together to present a more constant impedance to the crossover. By varying them both, you can change the tweeter level without changing the impedance the crossover sees. If you just bypassed R1, I think you defeated part of the function of the attenuation circuit and shifted the cross point. As the circuit appears to me, anyway. Sometimes it's hard to draw a firm line between different circuit elements. It helps to model the crossover.
2) similarly, if you change from an 8 ohm tweeter to a 4 ohm, the cross point is going to shift some. If the L-pad is in the circuit and working correctly, the change may not be that significant. Or you could adjust the R1 and R2 values to compensate for the tweeter impedance change (and likely voltage sensitivity difference between the 4 and 8 ohm versions).
None of this means you can't do what you are proposing, just that you'll get a better result if you understand the circuit a bit better.
Thanks all. Unfortunately, the crossover board was glued in and quite hard to access. From what I understand, I should keep impedance the same so I should use both parallel and serie resistors (using L-pad calculator). I probably can change value of R1 by parallel it with another resistor.
Attachments
You missed the main, principle and basic function of a crossover for the tweeter, which is to high pass from a certain frequency, and that is the f in the 1/2πRC = f
And that in order to blend with the woofer at the correct frequency
And that in order to blend with the woofer at the correct frequency
The circuit and the slope should be simple enough to hear most of the issue, and that's your saving grace here. The interaction is too complex to put in a sentence.. the impedance concerns, the frequency concerns, the 4 ohms unless you plan to measure.
Why not put in the new tweeters and try each resistor or both in setting the level. See if there is a midrange peak at the cross and adjust the capacitor etc..
Why not put in the new tweeters and try each resistor or both in setting the level. See if there is a midrange peak at the cross and adjust the capacitor etc..
Does it seem like hot glue or something like silicone, urethane, etc.?
If it's hot glue, you may be able to soften it with a heat gun and get it out.
Is this one available in your area? Looks pretty similar.
Visaton DTW72
https://www.parts-express.com/Visaton-DTW72-1-2-Polycarbonate-Dome-Tweeter-8-Ohm-292-676?quantity=1
Cutout Diameter 1.96
Tweeter Type Poly / Mylar Dome
Power Handling (RMS) 70 Watts
Impedance 8Ω
Frequency Response 2.200 to 23,000Hz
Sensitivity 91dB 2.83V/1m
Thiele-Small Parameters
Resonant Frequency (Fs) 3000Hz
Mounting Information
Overall Outside Diameter 2.83"
Depth 0.83"
That style tweeter was made for a long time in a few configurations, so you may be able to find an 8 ohm one somewhere with some digging. Though some used different style magnets or slightly different lenses, in my experience most of them sound pretty similar, even when made by different suppliers.
TW010F1 - 10mm Tweeter Audax - 8 ohm
High Energy encapsulated magnet
Compact and round light design
Ferrofluid cooled voice coil
Brand Audax
Series Tweeter
Magnet Material Neodymium
Diaphragm Material Polymer
Diaphragm Diameter [mm] 10
Ferrofluid Yes
Free Air Resonance [Hz] 3000
Impedance [Ohm] 8
Sensitivity [dB] 90
Power Handling [Watt] 25
The Visaton is not available in my area. There are few OEM Harman, LG-branded drivers. I plan to replace tweeters in both channels since the Audax is hard to come by. I don't want to mess with cabinet so I need a 3-inch drivers for drop in replacement.
I admitted that I'm not keen in crossover circuit design and mostly use chatGPT to help in design. The AI suggests 1.5 ohm resistor in parallel with R1 to use with 4 ohm tweeter driver with -1.9dB attenuation (compared to -2.7db in original 8 ohms design). I'm not sure AI calculation is correct or not since previously I use it on PSU design but their design is not entirely correct. Here's what the response....
### Recalculation for Design 5: New with 4 Ohm Tweeter and 1.5 Ohm Parallel Resistor
#### Tweeter Network Calculation
### Summary for Design 5
| Parameter | Value |
|----------------------------------|--------------------------|
| Tweeter Attenuation | Approx. -1.92 dB |
| Tweeter Network Impedance | Approx. 5.73 ohms |
| Combined Speaker Impedance | Approx. 6.5 + 2.96j ohms |
| Tweeter Crossover Frequency | Approx. 8 kHz |
| Woofer Crossover Frequency | Approx. 3.18 kHz |
This summary provides a clear picture of the key parameters for Design 5, ensuring it maintains balanced attenuation and good impedance matching while achieving the specified crossover frequencies.
I admitted that I'm not keen in crossover circuit design and mostly use chatGPT to help in design. The AI suggests 1.5 ohm resistor in parallel with R1 to use with 4 ohm tweeter driver with -1.9dB attenuation (compared to -2.7db in original 8 ohms design). I'm not sure AI calculation is correct or not since previously I use it on PSU design but their design is not entirely correct. Here's what the response....
### Recalculation for Design 5: New with 4 Ohm Tweeter and 1.5 Ohm Parallel Resistor
#### Tweeter Network Calculation
### Summary for Design 5
| Parameter | Value |
|----------------------------------|--------------------------|
| Tweeter Attenuation | Approx. -1.92 dB |
| Tweeter Network Impedance | Approx. 5.73 ohms |
| Combined Speaker Impedance | Approx. 6.5 + 2.96j ohms |
| Tweeter Crossover Frequency | Approx. 8 kHz |
| Woofer Crossover Frequency | Approx. 3.18 kHz |
This summary provides a clear picture of the key parameters for Design 5, ensuring it maintains balanced attenuation and good impedance matching while achieving the specified crossover frequencies.
I think the AI calculations are incorrect at many places. Lol.
Good thing that I think I found the spec of the new drivers and I can physically cut off R2, L2, R3, C2 from the tweeter network to simplify things.
The new tweeter has resonance frequency at 1600Hz, so I think I can cross at 3000Hz using simple first order design? Based on my calculation, I can keep L1 at 0.4mH and some how increase C1 value to 13uF? Can I've just parallel a 10uF cap to C1? The new tweeter SPL is at 91dB so I guess I can left R1 at 3 ohms for some attenuation?
Good thing that I think I found the spec of the new drivers and I can physically cut off R2, L2, R3, C2 from the tweeter network to simplify things.
The new tweeter has resonance frequency at 1600Hz, so I think I can cross at 3000Hz using simple first order design? Based on my calculation, I can keep L1 at 0.4mH and some how increase C1 value to 13uF? Can I've just parallel a 10uF cap to C1? The new tweeter SPL is at 91dB so I guess I can left R1 at 3 ohms for some attenuation?
Textbook crossovers are almost useless. The 6dB tweeter filter works by adding a capacitor that add more and more resistance the further down from 3 kHz you get (if the tweeter has a constant 4 ohm resistance.) If the tweeter has a peak at say 1600 Hz way higher than the resistance of the capacitor at that frequency the crossover will not attenuate.
I tried a philips 1"dome crossed over at 4 kHz (resonance at 1 kHz) The cap should reduce the output by 12 dB at 1 kHz in theory, in practice it was soemthing like 3 dB or so. If you want to use a first order network you must have a LCR trap for the tweeter peak!
I tried a philips 1"dome crossed over at 4 kHz (resonance at 1 kHz) The cap should reduce the output by 12 dB at 1 kHz in theory, in practice it was soemthing like 3 dB or so. If you want to use a first order network you must have a LCR trap for the tweeter peak!
OK, So I keep the L2 C2 R3 network. Resonant frequency should be 1 / [2π × √(L × C)]? The original network is around 8k?! I don't know why they design like that since the resonant frequency of original Audax TW010F1 should be around 3k...
For my new tweeter, if I keep LR and increase C, I should get 1.6k? Based on calculation I should use 55 uF in total (parallel another cap to C2 will work?)
For my new tweeter, if I keep LR and increase C, I should get 1.6k? Based on calculation I should use 55 uF in total (parallel another cap to C2 will work?)