I am attempting to complete my speaker project but I can't figure out what I am doing wrong building the crossover.
It is a 2 way speaker with wide band driver for the Low Frequency and AMT Heil for HF drivers.
The HF transducer is 15db more efficient than the LF driver.
Both drivers are pure resistive (no incuctance or capacitive), the LF is 9 Ohms the HF is 2 Ohms.
I have tried many combinations (and components purchase) to try to crossover at 2Khz, but it did not do what I expected.
My last attempt has been to use the parts I had and try to cut the LF at 1.8Khz (12db/oct) and the HF around 2.4Khz (18db/oct)
LP filter: 1.36mH with a 3.9uF capacitor
HP filter: 6.8uF capacitor, 0.33mH and 22uF capacitor
Here is the schematic in spice:
Here the simulation result:
here what is the measured results (Red = LF driver supposedly 1.8K 12db and Green = HF driver supposedly 2.4K 18db)
As you can see it does not seem to do what I expect.
Below is the comparison of the LF driver in wide band (green) and the LF driver with the low pass filter (red) which I expected to cut at 1.8K 12db:
My question is what is the mistake I am making in my filter to not get the the 1.8K 12db?
It is a 2 way speaker with wide band driver for the Low Frequency and AMT Heil for HF drivers.
The HF transducer is 15db more efficient than the LF driver.
Both drivers are pure resistive (no incuctance or capacitive), the LF is 9 Ohms the HF is 2 Ohms.
I have tried many combinations (and components purchase) to try to crossover at 2Khz, but it did not do what I expected.
My last attempt has been to use the parts I had and try to cut the LF at 1.8Khz (12db/oct) and the HF around 2.4Khz (18db/oct)
LP filter: 1.36mH with a 3.9uF capacitor
HP filter: 6.8uF capacitor, 0.33mH and 22uF capacitor
Here is the schematic in spice:
An externally hosted image should be here but it was not working when we last tested it.
Here the simulation result:
An externally hosted image should be here but it was not working when we last tested it.
here what is the measured results (Red = LF driver supposedly 1.8K 12db and Green = HF driver supposedly 2.4K 18db)
An externally hosted image should be here but it was not working when we last tested it.
As you can see it does not seem to do what I expect.
Below is the comparison of the LF driver in wide band (green) and the LF driver with the low pass filter (red) which I expected to cut at 1.8K 12db:
An externally hosted image should be here but it was not working when we last tested it.
My question is what is the mistake I am making in my filter to not get the the 1.8K 12db?
Here are two suggestions:
1) Study the xo used in John Bau's Spica TC-50 at the spicaenthusiast.com site. The bass/mids he used in his famous 2-ways had a similar spike in the same range as yours and he figured out a way to mitigate that with a 'tuned' xo in matched pairs that had service codes assigned indicating the measured resp. peak and how much the peak was.
2) Look into Passive XO designer software. It's free and the proponents of it reside at parts experess tech talk forum. There is one thread there with many views and posts focused on using this software. In order to use it though, you need to generate data files that include impedance, phase and dB.
1) Study the xo used in John Bau's Spica TC-50 at the spicaenthusiast.com site. The bass/mids he used in his famous 2-ways had a similar spike in the same range as yours and he figured out a way to mitigate that with a 'tuned' xo in matched pairs that had service codes assigned indicating the measured resp. peak and how much the peak was.
2) Look into Passive XO designer software. It's free and the proponents of it reside at parts experess tech talk forum. There is one thread there with many views and posts focused on using this software. In order to use it though, you need to generate data files that include impedance, phase and dB.
From pictures the peak seems to be around 10db high so even with a 3rd order LPF you will need to cut lower than 2k.
When the peak is near the xo point adding a notch its a waste of components i believe. However its also important to note from the first measurement picture where do you want to set the HF level, at -20db (from the picture scale) or at average woofer level. More smoothing can help you just to see where you sit on tonal balance.
From what i see, your spice transfer function for the woofer should show a -15db ( maybe more, -18db) at 2kHz. So use a 18db/oct slope and choose a xo point around 1Khz. Simulate and measure, simulate and measure...
When the peak is near the xo point adding a notch its a waste of components i believe. However its also important to note from the first measurement picture where do you want to set the HF level, at -20db (from the picture scale) or at average woofer level. More smoothing can help you just to see where you sit on tonal balance.
From what i see, your spice transfer function for the woofer should show a -15db ( maybe more, -18db) at 2kHz. So use a 18db/oct slope and choose a xo point around 1Khz. Simulate and measure, simulate and measure...
Hi,
Perhaps it wasn't as big. Perhaps its suppressed on the listening axis.
Spica TC-50 loudspeaker | Stereophile.com
An externally hosted image should be here but it was not working when we last tested it.
rgds, sreten.
Thanks Sreten for posting the TC-50 xo schematic and pics of the speakers. I believe it was the R3 resistor that was varied to tune the hump for a flatter response .
Here is a link to the enthusiast page that addresses the service codes.
Spica Speaker Service Codes | Repair, by the Spica Enthusiast
The one example cited there had an 8 dB peak at 3.9 kHz.
Here is a link to the enthusiast page that addresses the service codes.
Spica Speaker Service Codes | Repair, by the Spica Enthusiast
The one example cited there had an 8 dB peak at 3.9 kHz.
Last edited:
Hi,
I'd say that is on the driver axis when tested, not the listening axis.
The driver filter fixes the off axis power response, but can't fix any
related distortion issue, but the unusual listening axis can / does ?
rgds, sreten.
Zaph covers distortion issues related to (metal) driver peaks a lot
on his site. Pretty sure he mentions somewhere distortion issues
can be less if not listening directly on axis.
I'd say that is on the driver axis when tested, not the listening axis.
The driver filter fixes the off axis power response, but can't fix any
related distortion issue, but the unusual listening axis can / does ?
rgds, sreten.
Zaph covers distortion issues related to (metal) driver peaks a lot
on his site. Pretty sure he mentions somewhere distortion issues
can be less if not listening directly on axis.
Next time I do some response measurements using HI, I'll be sure to check off axis 3HD compared to on axis.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.