Annoying solid state sound... what to do?

[SimontY]

Now, for the guy wondering about his tweeter, I've got a few suggestions:

(1) You may want to have an active 1st order lowpass filter with -6dB at 20KHz in your system, which you can turn on with a switch. ISO 9613 describes the high frequency attenuation characteristics of air. Since microphone distances in modern recordings tend to be short, often less than a foot, you often have to add the effect of air yourself, unless the recording engineer knows his/her job.

(2) Move the inductor so that it is in paralell with only the tweeter, not the tweeter and inductor. Damping factor for a tweeter is essentially determined entirely by the crossover network. It may be realized by either (a) using an RLC filter across the tweeter to improve the damping factor at the resonance frequency, or (b) using an even order parallell crossover, or (c) using an odd order series crossover. Note also that any resistor used to reduce the tweeter level will have less reduction at resonance (due to the higher resistance of the tweeter at that frequency) than it has throughout the rest of the operating range. You mentioned that going from a 8.2uF cap and 0.82ohm resistor to a 8.0uF cap and 0.18ohm resistor improved things. That is likely because that modification raised the other parts of the spectrum, causing the resonance to be swamped out.

(3) The W17 has severe breakup at higher frequencies. As far as I could see, the posted crossover only had a second order lowpass filter for the woofer, and for some reason places the impedance correction circuit on the amp side of things, when it would probably do more good connected across the woofer instead. Try increasing the order of the woofer crossover, and use notch filters and proper impedance compensation. It is very common not to treat this driver with the extra care it needs, resulting in harsh sound.

(4) Presence and siblant harshness tends to stem from the 3-5KHz range, and many recordings have unduly high level in this range, probably due to the fact that many cheap 2-way systems have a hole in this range. A single parametric equalizer should do the trick, but make sure you can turn it off. Also note that the W17 breakup occurs exactly in this range.

Hope this helps.

Hi Angel,

Thank you very much for making these suggestions. You may notice that this thread is very old, and my system is quite different now. I am getting the best sound yet, after scrapping the x-over, and using the woofer full range, but with the LCR circuit. The tweeter is 2nd order, reversed polarity, coming it at 5khz, and with no attenuation. I measured my response and found an absolutely huge peak at 900hz, which is a standing wave from the cabinet width. I also found that bass was down, and the top end was very far down - hence the removal of the tweeter resistor. To keep the sound balanced and make the response more even, I increased baffle step correction to a full 6db.

There is still some harshness and glare at very loud volumes.

In response to your points:

1 - Interesting, what perceived difference might that make?

2 - I'm not sure what you mean here, but it probably doesn't apply anymore.

3 - "W17 has severe breakup" -Don't mean to presume, but perhaps you're thinking of a different driver. Full range this driver doesn't stand out at any frequency really, although there is some raggedness at around 4khz, as you suggest. How well does the published frequency response curve relate to the audible effects of breakup anyway?

4 - Again, this range could be where 'annoying' sound comes from.

I obviously prefer a 5khz x-over to a 2.5khz one. Should I try and cross the woofer over 2nd or 4th order at 5khz, even though that's where it naturally rolls off?

Thanks. S

 

[angel]

I am getting the best sound yet, after scrapping the x-over, and using the woofer full range, but with the LCR circuit.

How did you manage to tame the breakup of the magnesium cone without any series components?

The tweeter is 2nd order, reversed polarity, coming it at 5khz, and with no attenuation.

Sounds good. A regular 2nd or 4th order filter will ensure good damping at resonance. I'd still recommend a low DC resistance for the last inductor, though.

I measured my response and found an absolutely huge peak at 900hz, which is a standing wave from the cabinet width.

This can be ameliorated by introducing a notch filter, adding wool felt to the front and side baffles, placing the drivers off-center and not on a vertical line, etc... provided it's not actually from the drivers.

I also found that bass was down, and the top end was very far down - hence the removal of the tweeter resistor. To keep the sound balanced and make the response more even, I increased baffle step correction to a full 6db.

Placing them closer to a side wall should make a full 6dB BSC unneccessary. 3-4dB tends to do the trick.

How did you do the BSC, by the way?

There is still some harshness and glare at very loud volumes.

I'm thinking this will improve if you actually use a lowpass filter on the woofer, as well as using a 4th order crossover on the tweeter.

1 - Interesting, what perceived difference might that make?

It should improve the perceived distance to the event, as well as making the sound slightly less bright. Some recordings don't need this, especially ones where all the mikes are at sensible distances.

A similar, but better, effect can be acheived by modifying a tweeter so that the upper resonance of the system occurs at 20KHz with a Q of 0.5...

(Off topic: Did you know that the Vifa XT25 can actually reach 100KHz with a few modifications?)

2 - I'm not sure what you mean here, but it probably doesn't apply anymore.

On the crossover schematic someone posted, the circuit consisted of a series capacitor, a parallell inductor, and a series resistor, in that order. By making it series capacitor, series resistor, parallell inductor, you would dampen the resonance of the tweeter more.

3 - "W17 has severe breakup" -Don't mean to presume, but perhaps you're thinking of a different driver. Full range this driver doesn't stand out at any frequency really, although there is some raggedness at around 4khz, as you suggest. How well does the published frequency response curve relate to the audible effects of breakup anyway?

I'm thinking about the Seas Excel W17, a 6.5" woofer with a cast magnesium cone. If that's the one, it has some serious raggedness from 4KHz up, which needs to be brought down hard in the crossover filter to avoid harshness and glare, especially at loud volumes. If this is done properly, the driver is one of the cleanest sounding ones out there. Of course, this would require moving the tweeter crossover to 3KHz or less, and using at least 4th order acoustic filters.

4 - Again, this range could be where 'annoying' sound comes from.

I obviously prefer a 5khz x-over to a 2.5khz one. Should I try and cross the woofer over 2nd or 4th order at 5khz, even though that's where it naturally rolls off?

If your woofer has a very smooth rolloff and smooth off-axis response (not necessarily high, but smooth) to at least 4KHz, you can set the crossover to about 4KHz. For a good example of a response curve that is excellent in this respect, look at the Fostex FW127 datasheet (I think Madisound still sells that driver).

This is usually not the case for 6.5" woofers that start beaming at about 2KHz, and magnesium cone drivers have a strong breakup and should not be used above 3KHz.

Having a single driver reproduce the range from 150Hz to 12KHz is of course pretty ideal, since that covers the vocal range, but rarely practical.

A good compromise is to go as high as you can without getting excessive cone breakup or beaming. For the driver you mentioned, this would be in the 2-3KHz range.

Try using a 4th or 6th order acoustic filter with good impedance compensation. The Dynabel Euforia (Seas Excel W17 and Scan-Speak 99000) is an excellent example of how to use these drivers.

 

[SimontY]

How did you manage to tame the breakup of the magnesium cone without any series components?

I'm thinking about the Seas Excel W17, a 6.5" woofer with a cast magnesium cone. If that's the one, it has some serious raggedness from 4KHz up, which needs to be brought down hard in the crossover filter to avoid harshness and glare, especially at loud volumes.

Hi again,

My driver is Seas H571, otherwise sold as W17PPI. It is a coated paper cone, and not from the Excel product range. I thought there may have been a mix-up here, as this is definately not a driver with very messy breakup!

With this new information, would you recommend using a LP x-over as high as 5khz to 'tidy' the sound? I have tried a 0.1mh inductor, as a 1st order filter, which improves the sound in some aspects, but makes the overall presentation too dead with a lot of music. Perhaps I'd need to bring the tweeter further down.

 

[SimontY]

Sounds good. A regular 2nd or 4th order filter will ensure good damping at resonance. I'd still recommend a low DC resistance for the last inductor, though.

The inductor is 0.25 DCR, air-cored.

 

[SimontY]

Placing them closer to a side wall should make a full 6dB BSC unneccessary. 3-4dB tends to do the trick.

How did you do the BSC, by the way?

I found that this gives the most realistic overall tone, and should measure the flattest too. The speakers are close to walls, and in an ideal situation would not need as much as 6db, but in the system as a whole, it is probably best.

The BSC is from Rod Elliott at:
http://sound.westhost.com/bafflestep.htm

It sits inbetween the pre-amp output and power amp input on my integrated amp.

 

[angel]

Could you post a schematic of the crossover you use now?

 

[SimontY]

Sure, here she is.

 

[angel]

Okay. So you're using two Seas H571 (CB17RCY/P) woofers with one Seas H1148 (27TBC/GTV) tweeter, no filter on the woofer, and a second order electric filter on the tweeter?

If so, then your 900Hz peak most likely stems from the 900Hz resonance frequency of the tweeter, which I think can get quite pronounced.

With this new information, would you recommend using a LP x-over as high as 5khz to 'tidy' the sound? I have tried a 0.1mh inductor, as a 1st order filter, which improves the sound in some aspects, but makes the overall presentation too dead with a lot of music. Perhaps I'd need to bring the tweeter further down. [/B]

Where did you put the 0.1mH inductor? What kind of inductor? And what kind of 'deadness'?

Having looked at the dispersion for these two drivers, I'd recommend 2-3KHz for the crossover frequency.

The inductor is 0.25 DCR, air-cored.

For the tweeter, this should probably be okay. Only marginally affects the Q, about 5% raise or so, if I remember my math correctly..

I found that this gives the most realistic overall tone, and should measure the flattest too. The speakers are close to walls, and in an ideal situation would not need as much as 6db, but in the system as a whole, it is probably best.

Have you measured the response of your system? And what BSC frequency did you end up with? What is your baffle width?

 

[SimontY]

Okay. So you're using two Seas H571 (CB17RCY/P) woofers with one Seas H1148 (27TBC/GTV) tweeter, no filter on the woofer, and a second order electric filter on the tweeter?

If so, then your 900Hz peak most likely stems from the 900Hz resonance frequency of the tweeter, which I think can get quite pronounced.

The tweeter is not that exact model (27TBC/GTV) - mine is not shielded. But it obviously has a resonance of 900hz.

I doubt the giant peak is from the tweeter, as it is crossed in at 5khz, and at 12/db octave 900hz should be very far down, probably inaudible - I could check this out to be sure.

The wavelength of 900hz is exactly twice the distance of the cabinet's internal width. This must be where the peak is coming from, at least mainly.

 

[SimontY]

Where did you put the 0.1mH inductor? What kind of inductor? And what kind of 'deadness'?

Having looked at the dispersion for these two drivers, I'd recommend 2-3KHz for the crossover frequency.

Have you measured the response of your system? And what BSC frequency did you end up with? What is your baffle width?

The 0.1mH inductor was/is in series with both bass drivers.

2-3khz was the original x-over fr., as used in the kit these speakers started life as. This x-over sounded unpleasant - it was nasal, bright, tinny, and emphasised the lower end of the presence region in an unnatural way. The sound was 'disturbing' and very difficult to listen to at high levels. I have concluded that 2-3khz is probably too low for the tweeter, and with no BSC or consideration of other cabinet effects, the speakers sounded as previously mentioned.

I measured my response using an SPL meter, so it might not be very accurate, but it did show very considerable peaks and dips.

Based on my measurements I increased BSC from 3db (roughly) to 6db, and raised the tweeter level. I also filled in the bottom of the cabinet to reduce the standing wave effects. The BSC comes in around 400hz, and is based on the width of 22cm (iirc). I will change it to match the room/measurements soon.

I am yet to stuff the cabinets again to see if that reduces the 900hz peak more...

 

[angel]

I doubt the giant peak is from the tweeter, as it is crossed in at 5khz, and at 12/db octave 900hz should be very far down, probably inaudible - I could check this out to be sure.

The wavelength of 900hz is exactly twice the distance of the cabinet's internal width. This must be where the peak is coming from, at least mainly.

Perhaps. The tweeter resonance should theoretically be down by about 25dB on-axis. These tweeters have no ferrofluid, and their resonance is pretty pronounced, though.

Try measuring the system with only the woofer, then only the tweeter. That way, you can pinpoint it.

The 0.1mH inductor was/is in series with both bass drivers.

Before the zobel network, right?

I have concluded that 2-3khz is probably too low for the tweeter,

With a 2nd order filter, you would probably want to cross over at a higher frequency, yes. Higher order filters may be in order.