I'm building a two way sealed bookshelf with the Vifa XG18 and Seas 27TBFC/G, using a .5 cu ft. enclosure from Parts Express. My first two crossovers were 2nd order electrical at 2khz and 2.2khz, but neither turned out well. In both cases I end up with a 4db dip at 2500 - 3000hz.
This is a gated measurement of each driver, taken on tweeter axis at 1m. I suspect the 2.5khz dip in tweeter response is caused by baffle diffraction, and it's the reason why I'm having trouble with the crossover.
Given these measurements, could someone with a bit more experience suggest a different crossover frequency/slope? I've spent hours in Passive Crossover Designer without finding any satisfying results.
*NOTE: I have an active baffle step compensation filter, so don't need to included it in the woofer's passive lowpass filter.
Dan
This is a gated measurement of each driver, taken on tweeter axis at 1m. I suspect the 2.5khz dip in tweeter response is caused by baffle diffraction, and it's the reason why I'm having trouble with the crossover.
An externally hosted image should be here but it was not working when we last tested it.
Given these measurements, could someone with a bit more experience suggest a different crossover frequency/slope? I've spent hours in Passive Crossover Designer without finding any satisfying results.
*NOTE: I have an active baffle step compensation filter, so don't need to included it in the woofer's passive lowpass filter.
Dan
I used that woofer and it did not look like that. I would look into the measurements first. There is a crossover for a two way, different box, on my site.
www.helarc.com
P
www.helarc.com
P
Keruskerfuerst said:
ZMA and FRD files
The ZMA files are just an SPL trace of measurements found at zaphaudio.com. The FRD files were created using Arta. I used only FRD consortium's passive crossover designer, nothing like LinearX or Lspcad.
Pjay said:
Measurements were taken using:
Behringer ECM8000 mic
Behrigner UB802 mixer
M-Audio Audiophile 2496 sound card, in 96khz/16 bit mode
The Arta software
The mic was not moved between taking measurements, and things were set up to avoid early reflections.
XG18 Impulse response and gate
An externally hosted image should be here but it was not working when we last tested it.
Seas H1212 impulse response and gate
An externally hosted image should be here but it was not working when we last tested it.
Dan
Keruskerfuerst said:Datasheets for both drivers?
You can get the datasheets from madisound.
Vifa XG18
http://www.madisound.com/catalog/pr...=1175&osCsid=0133b83fd8b700d82bd51dae79daef14
Seas 27TBFC/G
http://www.madisound.com/catalog/product_info.php?products_id=795
Dan
Hi owdi,
You mentioned you are using the 0.5 cubic foot enclosure from parts express. Where is the tweeter positioned on the baffle?
Suggest you use a baffle diffraction sim. program to confirm your suspicions around the 2.5KHz dip.
Two excellent examples are Svante's Edge and FRD consortiums BDS spreadshset (don't have the links handy).
Also- Zaph uses the same enclosure with the 27TBFCG in his L18 design. Have you taken a close look at the in enclosure tweeter curve? Note his tweeter driver offset on the baffle to smooth out the diffraction ripple. His design would be a very good starting point to tune the 27TBFCG. There's no notch or other shaping circuitry in Zaph's L18 design for the tweeter part of the circuit, so presume the 27TBFCG can provide a nice flat response - esp. since you are crossing higher than Zaph in the L18 design.
DAvid.
You mentioned you are using the 0.5 cubic foot enclosure from parts express. Where is the tweeter positioned on the baffle?
Suggest you use a baffle diffraction sim. program to confirm your suspicions around the 2.5KHz dip.
Two excellent examples are Svante's Edge and FRD consortiums BDS spreadshset (don't have the links handy).
Also- Zaph uses the same enclosure with the 27TBFCG in his L18 design. Have you taken a close look at the in enclosure tweeter curve? Note his tweeter driver offset on the baffle to smooth out the diffraction ripple. His design would be a very good starting point to tune the 27TBFCG. There's no notch or other shaping circuitry in Zaph's L18 design for the tweeter part of the circuit, so presume the 27TBFCG can provide a nice flat response - esp. since you are crossing higher than Zaph in the L18 design.
DAvid.
Dave Bullet said:
I looked very closely at the L18 project at zaphaudio.com... so closely that I started with a tweeter crossover using the same value cap and coil. I did perform a BDS simulation. I chose a slightly different configuration than John Kurtke because with my subwoofer stands, the front baffle will look more like a floorstander than bookshelf. Here's the sim:
An externally hosted image should be here but it was not working when we last tested it.
I really don't know where my measured irregularity in the 2.5khz area of the tweeter's response comes from. Maybe it is my uncalibrated measurement setup, and I'm chasing a ghost.
Dan
I think the BDS sim you have done highlights the irregularity you have found - ie. the dip between 2KHz and 5KHz of about 2dB - similar to your measured (the measurement graph has a finer scale of 1dB divisions hence looks worse than is?)
The only thing I can suggest is try using felt around the tweeter to see if it changes the response. At least this means you won't have to replace the baffle for a new tweeter place. in BDS it looks like your tweeter offset on the baffle is quite agreessive. Can you try a 60/40 (off centre) position? Golden ratio placement might help here if you have a tower, you might be able to offset the tweeter 1.618 from top and 0.618 from one side edge.
Dr. BDS (Svante) - if you are reading this, could you chime in with one of your expert opinions?
Cheers,
David.
The only thing I can suggest is try using felt around the tweeter to see if it changes the response. At least this means you won't have to replace the baffle for a new tweeter place. in BDS it looks like your tweeter offset on the baffle is quite agreessive. Can you try a 60/40 (off centre) position? Golden ratio placement might help here if you have a tower, you might be able to offset the tweeter 1.618 from top and 0.618 from one side edge.
Dr. BDS (Svante) - if you are reading this, could you chime in with one of your expert opinions?
Cheers,
David.
Thank you everyone for your advice. I'm going to try taking a second set of measurements. Besides the roughness in the 2.5-3khz range, the dip around 370hz is worrying me. The quarter wavelength of 370hz is 9.25 inches, which happens to be the distance from the woofer to the back wall. I'm going to experiment with adding a few blocks of MDF to the walls of the enclosure, in hopes of getting rid of this dip.
If my second measurements also show the 2.5-3khz dip, I'll probably buy new baffles from PartsExpress, and change the placement of the tweeter to match the Seas L18/27TBFC/G project at zaphaudio.com
Dan
If my second measurements also show the 2.5-3khz dip, I'll probably buy new baffles from PartsExpress, and change the placement of the tweeter to match the Seas L18/27TBFC/G project at zaphaudio.com
Dan
FYI, this is a bookshelf monitor design using XG18 and H1212 in case you're not aware of it:
http://www.lonesaguaro.com/speakers/XG12/XG12.htm
http://www.lonesaguaro.com/speakers/XG12/XG12.htm
Re: split resistors?
The resistor changes the crossover slope to something less than 2nd order, but more than 1st. This also changes the phase to something between a 2nd and a 1st order filter. The woofer is already out of phase with the tweeter, because it's acoustic center is further away from your listening position. By introducing a smaller phase shift in the electrical network on the woofer, it sums better with the tweeter. Hope that makes sense (and I hope it's right).
Dan
PeteMcK said:
The resistor changes the crossover slope to something less than 2nd order, but more than 1st. This also changes the phase to something between a 2nd and a 1st order filter. The woofer is already out of phase with the tweeter, because it's acoustic center is further away from your listening position. By introducing a smaller phase shift in the electrical network on the woofer, it sums better with the tweeter. Hope that makes sense (and I hope it's right).
Dan
PeteMcK said:
Passive crossover designer from FRD Consortium takes this into account. You can enter the offset, and see the effect on response.
FRD Constortium
Dan
Re: split resistors?
Splitting resistors like that is simply another way of L-padding. Although the main purpose of an L-pad is to reduce the tweeter level, many ways of L-padding (e.g., a single resistor before XO, a single resistor after XO, two resistors one of which is in series after XO and the other is shunted before the driver, two resistors one of which is in series before XO and the other in series after XO, and so on) all result in a bit different response shapes that cannot be exactly mimicked by each other through tweaking other crossover components. Choosing one of these ways may help a bit better phase alignment with a mid. But the effect of response shaping by a different way of L-padding is not drastic. We can always obtain a similar (but not exactly the same) frequency response by scaling capacitor and inductor values in the net.
This means that even if we used another way of L-padding for the above Lou's design, provided that inductor and capcitor values were scaled accordingly, we would by no means RUin the design although we might get a little bit worse phase alignment.
And there's no formula available for this kind of subtle effects. It's just a design know-how you may gain through experience.
PeteMcK said:
Splitting resistors like that is simply another way of L-padding. Although the main purpose of an L-pad is to reduce the tweeter level, many ways of L-padding (e.g., a single resistor before XO, a single resistor after XO, two resistors one of which is in series after XO and the other is shunted before the driver, two resistors one of which is in series before XO and the other in series after XO, and so on) all result in a bit different response shapes that cannot be exactly mimicked by each other through tweaking other crossover components. Choosing one of these ways may help a bit better phase alignment with a mid. But the effect of response shaping by a different way of L-padding is not drastic. We can always obtain a similar (but not exactly the same) frequency response by scaling capacitor and inductor values in the net.
This means that even if we used another way of L-padding for the above Lou's design, provided that inductor and capcitor values were scaled accordingly, we would by no means RUin the design although we might get a little bit worse phase alignment.
And there's no formula available for this kind of subtle effects. It's just a design know-how you may gain through experience.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.