Are you refering to the crossover in my first post?
This is how I interpret the R25A network from the limited view on the photos, I have removed the notch filter for clarity since this is confirmed.
HF PART;
I'm really not sure about the capacitor in the HF circuit, so I left it open. It is obviously connected to positive input, but it doesn't look like it is in series like I would have expected. What is the most likely position and role for this capacitor? I assume the green cap is bypass so I count them as one capacitor.
LF PART;
Again the capacitors is the mystery. I can only assume one is bypass and they are effectively one capacitor. Otherwise it looks like they would be in series, which wouldn't make much sense..?
I have tried simulating this in Vituixcad, but any other position than series for the HF cap gives very poor results.
The sketch I did was of the modded OBX crossover, but with one change cause we know there is a series resistor in the R25A tweeter.
Otherwise, the HP and LP are the same topology, and it's not so different from the Dali. I don't think it's worthwhile to try to see through mountains of glue - the OBX photos show they like to mess with stacking caps.
The LCR is one of three things: FS resonance notch for tweeter, midrange notch on woofers, or system impedance flattening for tube amps on woofers.
On OBX, it was not connected to the tweeter.
Otherwise, the HP and LP are the same topology, and it's not so different from the Dali. I don't think it's worthwhile to try to see through mountains of glue - the OBX photos show they like to mess with stacking caps.
The LCR is one of three things: FS resonance notch for tweeter, midrange notch on woofers, or system impedance flattening for tube amps on woofers.
On OBX, it was not connected to the tweeter.
Last edited:
Would there be any situation where it would be beneficial to use a tweeter cap in any other position than series?
(Besides in an impedance linearization circuit)
(Besides in an impedance linearization circuit)
@Lament
There's no way that ( in the LF area ) the capacitor junction ( that ultimately goes to ground ) is placed before the inductor. as you have it ( see my big Red-X ).
Doing so creates an untenable low impedance condition for the overall circuit.
Also, I believe you may be giving Kevin Scott ( of Living Voice ) too much credit ( for esoteric & expensive network design or complexity ).
OTOH, he certainly does get credit for knowing how to get the results that he's after, all on a budget.
FYI, I simply see a bunch of small SMP electrolytic caps ( undoubtedly bought for pennies ), arranged in series//parallel configurations and then brought up to final capacitance with another eBay special ( surplus Russian military capacitors or maybe a regular "store-bought" film type >> the white ones ).
Surplus Electrolytic SMD Capacitors
No matter what you think you may see, the schematic layout is common to all 2 pole x 2 pole layouts. It does have some resistors inline with most of the parallel devices ( to adjust their effective "Q" ).
Really, the value of this exercise ( really ought to be ) whether or not your ears prefer a lot of overlap between the two network sections ( as embodied in the original DALI 104 // LV design ) or do you ears prefer Troels network design philosophy ( of keeping the overlap at crossover to a minimum ). I note that your current network design has a lot of overlap ( & there are many fine designs out-there with this amount of overlap ).
I'll assume that you've read all of TG's Vifa C17 design pages. He states his design preference right off the bat in Vifa C17( all within the same paragraph mentioning LV ).
FWIW, when I mocked up the Dali 104( using your files and guesstimating many parameters ), I too get a lot of overlap in the crossover region ( I'm pretty sure LV considers that a "feature" not a fault ).
🙂
Last edited:
You are probably right Earl. I am in the process of learning, so I find it interesting to study how the "pro's" do things 🙂
Excellent post @EarlK, lots of good info! Thank you 🙂
When you mention my current design having alot of overlap, are you refering to my LV clone thread? Both HP & LP are 2nd order electrical, this is the same as Troels used with the Vifa C17.
And yes I have read many of his build descriptions including his work with Vifa C17, I find it very inspiring!
Regarding your mockup of the Dali 104, did you use the files I uploaded in my LV clone thread? Your mockup must be a combination of different networks, the Dali does not have a noth filter. Btw we both used the exact same values for the notch filter, thats interesting 🙂
I tried to replicate what you have done, and my results were quite different. I notice that you use the 30 degree off axis response for the tweeter, I assume this is to smooth out the nasty diffraction effects? I don't think it's possible to isolate off axis response to individual drivers in Vituixcad, only the system. I also notice you have only used one instance of the C17, is it scaled in any way in xsim to reflect two units?
This is what I get;
Last edited:
You're Welcome!
I used a single driver icon in the woofer circuit ( but representing the two woofers ) because you had already combined their impedances into one ( so I followed suit with the Freq response > I didn't see any other axis that I preferred, so stayed with 0 deg just cuz ).
I might have preferred to use 2 ( totally ) separate woofer files ( with individual FRD + ZMA's ) captured on the same axial plane as the tweeter's center >> but taken 30 deg off axis of the horizontal ( somewhat representing a possible listening axis ).
I don't know the values for the Dali's inductors ( the values that I threw down were simply conjecture that helped my cause ).
I have the Acoustic Center of the ( combined ) woofers as being .75" behind that of the AC belonging to the tweeters ( just a bad guess I suppose ) >but also chosen to see what's optimal for best phase alignment.
- I had to include the AC because all the .FRD files I previewed were essentially minimum phase types ( or, taken to T=0 from within REW ) and therefore lacking any TOF ( Time of Flight ) info.
I included a notch-filter on the tweeter ( aimed at it's primary resonance ) because even Troels eventually added one // and it seems LV uses one // in both cases employed to reduce some tweeter "grain" that was noticed at high playback levels.
- "In or Out", it doesn't effect the overall frequency response .
🙂
I used a single driver icon in the woofer circuit ( but representing the two woofers ) because you had already combined their impedances into one ( so I followed suit with the Freq response > I didn't see any other axis that I preferred, so stayed with 0 deg just cuz ).
I might have preferred to use 2 ( totally ) separate woofer files ( with individual FRD + ZMA's ) captured on the same axial plane as the tweeter's center >> but taken 30 deg off axis of the horizontal ( somewhat representing a possible listening axis ).
I don't know the values for the Dali's inductors ( the values that I threw down were simply conjecture that helped my cause ).
I have the Acoustic Center of the ( combined ) woofers as being .75" behind that of the AC belonging to the tweeters ( just a bad guess I suppose ) >but also chosen to see what's optimal for best phase alignment.
- I had to include the AC because all the .FRD files I previewed were essentially minimum phase types ( or, taken to T=0 from within REW ) and therefore lacking any TOF ( Time of Flight ) info.
I included a notch-filter on the tweeter ( aimed at it's primary resonance ) because even Troels eventually added one // and it seems LV uses one // in both cases employed to reduce some tweeter "grain" that was noticed at high playback levels.
- "In or Out", it doesn't effect the overall frequency response .
🙂
There is a scaling option for the impedance measurement in vcad, this is set to 2 to provide the correct combined impedance. I don't know if this is the same in xsim? Do you just add 6dB to the spl of the one vifa c17 instead of using two of them?
Still don't understand why our spl plots are so different if you added 6db to the woofer.
Do you generally measure like this, measurements for all drivers taken on the tweeter axis?
I guess that explains the very good phase tracking on your plots 🙂
I'm not sure I follow you, I thought I had removed time of flight (from baffle), ie 1000mm, from my measurements? But they should still provide info about relative z-distance between woofer and tweeter..
XSim doesn't have that scaling option.
Yes, I added at least @ 6db more gain to the woofer driver icon ( within XSim ) than I did to the tweeters driver icon.
- Real levels are all relative ( since these aren't my files ) and final tweeter level is made by the adjustment resistor ( R4 in my schematic )
I prefer to use the listening axis as my design axis. The listening axis is typically the same as the tweeters axis.
For an MTM, one might ( as Short-Hand ) take a single combined FR from the before mentioned position ( since it'll bake-in the combined FR of the twin woofers from that measurement position ).
You should review the advice that you received about a year ago ( in your other thread ) and see if it now makes more sense.
FWIW, these network simulation packages are incredibly accurate and give repeatable results // as long as one maintains the exact same measurement locations ( therefore, IMHO it's best to keep the workflow as simple as possible ).
🙂
Yes, I added at least @ 6db more gain to the woofer driver icon ( within XSim ) than I did to the tweeters driver icon.
- Real levels are all relative ( since these aren't my files ) and final tweeter level is made by the adjustment resistor ( R4 in my schematic )
I prefer to use the listening axis as my design axis. The listening axis is typically the same as the tweeters axis.
For an MTM, one might ( as Short-Hand ) take a single combined FR from the before mentioned position ( since it'll bake-in the combined FR of the twin woofers from that measurement position ).
You should review the advice that you received about a year ago ( in your other thread ) and see if it now makes more sense.
FWIW, these network simulation packages are incredibly accurate and give repeatable results // as long as one maintains the exact same measurement locations ( therefore, IMHO it's best to keep the workflow as simple as possible ).
🙂
Yes that is exactly what I have done recently, and you are right, it makes more sense now.
I will probably post some more questions related to phase tracking in the LV clone thread, as I'm having difficulties achieving satisfactory tracking with a 2nd order filter. Maybe an allpass filter to delay the tweeter is an option.. 🤔 I may @EarlK you there, as you have been very helpful 🙂