Basic procedure for all online forums. Type your replies in notepad (or other editor) and then copy/paste into the window when ready to reply.
Dave.
Dave.
Yeah, u guys know... and I'm very fond of re-learning from (repeated) mistakes so I'll be more cautious with this site.
I did some maths on the "zobel" leg and also finally got around to plugging Medley's data into Win ISD to see what effect the small sealed enclosure is having.
This is the ISD plot:
Compare that to the test on Zaph's site:
Note the low-pass/zobel doesn’t match the reported Re and Le results for a pure zobel. Re is 3.09 and Le is 0.256mH resulting in a “standard” zobel of 17uF and 3.86R but this would result in a very low impedance speaker unsuitable for typical commercial applications. This low-pass/zobel attenuates the mid by 1.6dB @2.5Khz, 2.5dB @4Khz, and 4.4dB @11Khz. I haven't yet worked out if a series inductor would have the same attenuation.
Looking at the components, the series resistor is sufficient to balance the drivers’ output volume:
0.5R on a 3R tweeter is -0.67dB overall. 4.3uF with (3+0.5) 3.5R is -3dB at 10,580Hz and -6dB at 3900Hz. The stated crossover point is 2800Hz, which is 13.2R(cap) + 0.5R in series with the tweeter’s 3R, giving -7.5dB overall, and -7.8dB at 2500Hz.
On the Q100 with 0.68R and 3.9uF, -3dB is at 11,095Hz, -6dB at 4217Hz, -7.8dB at 2800Hz and -8.2dB at 2500Hz, the stated crossover frequency. Q200 is -7.8dB at 2500Hz.
Still, this is confusing since the Q200 appears to have a lower mathematical crossover frequency (3.9Khz vs 4.2Khz) but the stated frequency is higher (2.8Khz vs 2.5Khz).
However, the smaller cabinet gives the mid a slight output boost so the series resistor on the tweeter can be a little smaller, and is. So it seems the drivers in Q100 and Q200 are likely to be the same as makes no difference, yet something is amiss with crossover frequencies.
I'm waiting for parts so the project will pause, and I get some relaxed listening time 🙂
I did some maths on the "zobel" leg and also finally got around to plugging Medley's data into Win ISD to see what effect the small sealed enclosure is having.
This is the ISD plot:
Compare that to the test on Zaph's site:
Note the low-pass/zobel doesn’t match the reported Re and Le results for a pure zobel. Re is 3.09 and Le is 0.256mH resulting in a “standard” zobel of 17uF and 3.86R but this would result in a very low impedance speaker unsuitable for typical commercial applications. This low-pass/zobel attenuates the mid by 1.6dB @2.5Khz, 2.5dB @4Khz, and 4.4dB @11Khz. I haven't yet worked out if a series inductor would have the same attenuation.
Looking at the components, the series resistor is sufficient to balance the drivers’ output volume:
0.5R on a 3R tweeter is -0.67dB overall. 4.3uF with (3+0.5) 3.5R is -3dB at 10,580Hz and -6dB at 3900Hz. The stated crossover point is 2800Hz, which is 13.2R(cap) + 0.5R in series with the tweeter’s 3R, giving -7.5dB overall, and -7.8dB at 2500Hz.
On the Q100 with 0.68R and 3.9uF, -3dB is at 11,095Hz, -6dB at 4217Hz, -7.8dB at 2800Hz and -8.2dB at 2500Hz, the stated crossover frequency. Q200 is -7.8dB at 2500Hz.
Still, this is confusing since the Q200 appears to have a lower mathematical crossover frequency (3.9Khz vs 4.2Khz) but the stated frequency is higher (2.8Khz vs 2.5Khz).
However, the smaller cabinet gives the mid a slight output boost so the series resistor on the tweeter can be a little smaller, and is. So it seems the drivers in Q100 and Q200 are likely to be the same as makes no difference, yet something is amiss with crossover frequencies.
I'm waiting for parts so the project will pause, and I get some relaxed listening time 🙂
Last edited:
The impedance should be level with the low point above the resonance and upward, ie near 4 ohms. The inductor should bring this back up as far as the amp should be concerned.
Maybe unclear.. the RC on its own won't reduce the response (but that's not what you are saying?)
Is there a side by side comparison of the two? or is it just the cap like you mentioned in post #38?
Hi,
Yes, the inductor would raise this through impedance and a small amount of wire resistance, but 3.86R in parallel with 3R ... whoa.... many commercial amps are only spec'd /rated for 8 ohms, (my old Onkyo for example) so I think that's a reasonable comment?
RC shunts reduce hf output on full range drivers - I've done this a gazillion times on little full range PC speakers for my pals. And that doesn't feel like an exaggeration ;-) I've assumed it would on this too - is that not the case? The calculations are based on power dissipation so granted, not accurate at all, but intended to give me an analysis of what is happening in that circuit. It seemed to me that the mid was a damped 2nd order high pass followed by a "zobel" to make the high pass work consistently and attenuate HF a little too. I'm trying to understand, then replicate, then (pride before a fall) improve 🙂
It also seems that the SP1587 driver in Q100 is the same as SP1587.2 in Q200 so I cannot work out how a bigger cap can create a lower xo frequency. The fact the maths doesn't match Kef's numbers is probably the key, but I'm not experienced enough with xo's to understand why/how.
Anyway, thanks heaps for your help - much appreciated!
And yes, I coped to clipboard before posting. Old dog, new tricks. ;-)
BTw worth mentioning that I use a 3rd order pllxo before a cheap TA2020 amp as a temporary way to xo woofer & mid, and that knocks the mid down 3dB, but this is only as a way to find a sound that is pleasing so I can get to grips with the speaker. My big disappointment is the dimensions of the 2 litre box for the coaxial driver - it's almost a cube with 1/4 wave resonance around 400hz I think. I've damped this with egg box foam, but still.
Yes, the inductor would raise this through impedance and a small amount of wire resistance, but 3.86R in parallel with 3R ... whoa.... many commercial amps are only spec'd /rated for 8 ohms, (my old Onkyo for example) so I think that's a reasonable comment?
RC shunts reduce hf output on full range drivers - I've done this a gazillion times on little full range PC speakers for my pals. And that doesn't feel like an exaggeration ;-) I've assumed it would on this too - is that not the case? The calculations are based on power dissipation so granted, not accurate at all, but intended to give me an analysis of what is happening in that circuit. It seemed to me that the mid was a damped 2nd order high pass followed by a "zobel" to make the high pass work consistently and attenuate HF a little too. I'm trying to understand, then replicate, then (pride before a fall) improve 🙂
It also seems that the SP1587 driver in Q100 is the same as SP1587.2 in Q200 so I cannot work out how a bigger cap can create a lower xo frequency. The fact the maths doesn't match Kef's numbers is probably the key, but I'm not experienced enough with xo's to understand why/how.
Anyway, thanks heaps for your help - much appreciated!
And yes, I coped to clipboard before posting. Old dog, new tricks. ;-)
BTw worth mentioning that I use a 3rd order pllxo before a cheap TA2020 amp as a temporary way to xo woofer & mid, and that knocks the mid down 3dB, but this is only as a way to find a sound that is pleasing so I can get to grips with the speaker. My big disappointment is the dimensions of the 2 litre box for the coaxial driver - it's almost a cube with 1/4 wave resonance around 400hz I think. I've damped this with egg box foam, but still.
Last edited:
And here are the win isd plots with the low pass / high pass filters applied. It seems it is quite a narrow bandpass so that might explain why it's tricky to replicate.
An externally hosted image should be here but it was not working when we last tested it.
This is the worst case scenario. The 3 ohm voice coil resistance is also in series with the voice coil inductance, and the capacitor ensures it is only significantly shunted where the voice coil inductive reactance is significant. The resultant impedance that the amp will drive is primarily the crossover low pass inductor's reactance in series with the 3 or 4 ohms, if it was done right.
Creating a filter that has the same overall effect on the response without the RC can produce the easiest amp load (or, using an RC without sizing it for the goal of flattening the impedance. After all, the 'flatness' in this case isn't a virtue but merely attractive to the human that is designing it).
Yes, as would an inductor in parallel with a resistor all in series with the driver.
Thx, an LR-in-parallel filter put in series with the mid might be the way to go. V happy with my current filters: 5.5R and 3.7uF for the tweeter, 0.5mH for the mid, and 3rd order pllxo to bi-amp woofer and coaxial. Not perfect on upper female vocals but very good on all else.
Just a quick update - tried a 4th order Linkwitz-Riley active crossover for the woofer/coaxial instead of the passive and that has made it easier to get the tweeter-mid crossover sounding "right". And funny thing, the tweeter values are similar to Kef's. Measured at 4.36uF + 0.50R + shunt 20.7R + 0.50R for the tweeter; -2.7dB attenuation. Mid just has 0.5mH, no zobel. However, I'll keep tinkering... that's the fun of it !
An update.
J River Media Center 23 64 bits + iZotope RX 6 De hum (VST 64 bits plugin).
Before with only two harmonics and fixed 65 Hz. Now the sound is better. And LPF at 22050 Hz. (*)
With my foobar2000 only works the old RX 5.
(*) LPF at 22050 Hz. As I have been solving my VERY BIG problems with all kinds of noise, interference (mains + atmosphere), continue at mains ... I have been raising the cut frequency from 14800 Hz to 22050 Hz.
J River Media Center 23 64 bits + iZotope RX 6 De hum (VST 64 bits plugin).
Before with only two harmonics and fixed 65 Hz. Now the sound is better. And LPF at 22050 Hz. (*)
With my foobar2000 only works the old RX 5.
(*) LPF at 22050 Hz. As I have been solving my VERY BIG problems with all kinds of noise, interference (mains + atmosphere), continue at mains ... I have been raising the cut frequency from 14800 Hz to 22050 Hz.
Last edited:
Sorry to resurrect an old thread, but I'm looking at putting in a new crossover for my Q100 drivers.
Tentatively looking at an LR2 @ approx 3.6kHz, would the Zobel values (from above) at 17uF and 3.86R be the right ones across the woofer? I don't need to lower the rising response from the tweeter as I'm planning to listen 30 degrees off axis (which is close enough to flat for me - I can flatten later with EQ).
Tentatively looking at an LR2 @ approx 3.6kHz, would the Zobel values (from above) at 17uF and 3.86R be the right ones across the woofer? I don't need to lower the rising response from the tweeter as I'm planning to listen 30 degrees off axis (which is close enough to flat for me - I can flatten later with EQ).
Forgot to look at the directivity matching. Looks like 3 kHz 4th order is the way to go.
Still would like to know about the Zobel though.
- Status
- Not open for further replies.