That's why I just start with an amp friendly schematic in software, and spend tons of time measuring. I don't have the patience to dial in the offsets in software - I'd rather invest that time understanding the off axis stuff. Most, if not all, software uses a pistonic model to predict power response. Cone geometry and breakups make that model have limited value.
Also, next time you measure reverse null, move the mic a few degrees or inches in any direction. Be careful about optimizing for that null, you are actually optimizing to a listening position and you can lose sight of the big picture.
Noo.... More of a reverse null fetish. It leads to a narrow focus. I'm on my phone, I can expand later, but the biggest reason to focus on offset is to get a pretty reverse null in software. Minimizing iterations to a predicted result is secondary.
Do you mean using a zobel to flatten out the lower frequency resonance. Or do you mean the rst28a cone breakup resonance in the upper frequency region?
Had to open Winpcd where it is called a series notch filter. Even with a fairly steep acoustic slope, I felt using it with an xt25tg made a noticeable difference on certain voices but with the rst28a, the knee around its resonance does not appear as distinguished. It would simplify the crossover it it were not needed.
Only you can judge that for yourself I suppose, whether needed or not. Yes, a series-notch filter AKA conjugate filter, LCR, Fs comp, or resonance compensation. Some texts I've read have called series-notches zobels, but not very many of them do. I typically avoid them as such to eliminate confusion. To me and most, CRs are Zobels, and LCRs are series-notches or conjugates.
Surprised the sim came out this close to the actual crossover. 8 pcs. total parts count. I'm just going to do the other side and listen for a while w/out taking any other measurements. Just because.
Never heard the the original rs but the rsta doesn't sound all that bad. IMO, it does the lower to mid treble better than the xt25tg and most of the other economy priced tweeters I've had the money to play with, just not much on top where the finesse and reach of the xt leave the rsta behind. Just my $.02 of course.
Sibilance (snare area) is too hot for my taste. Adding a resistor across the tweeter pulls that area down about 2+db w/out pulling anything off 10k+. Will give that a listen tomorrow. Going to do this backwards, listen then tweak and when it sounds just about right, measure.
The graphs are from xsim, not actual measurements. It's going to be interesting to see how what I think I'm hearing, what xsim predicts, and actual measurements relate.
Light green is the starting point. It sounded hot in the lower treble w/voices pretty much buried waaay in back. The beat/bass of the music was also a little too driving.
Which lead to purple. Lowered R2 from 4 to 2ohm and added an 8ohm across the tweeter. Lowering R2 would have overall raised the tweeter but adding the 8ohm across the tweeter overall dropped and helped level its downward tilt at the top. Music is now driven by the beat/bass, voice still waaay too far back. Songs known to be hot in the treble now sound too laid back.
Which led to maroon. Raised R1 (across the tweeter) from 8 to 12.5ohm which raised the tweeter by ~1.5db keeping the desired tilt. Then dropped L2 from 7mh to 6mh (parts on hand). Although this last change only appears to be a ~1.5db change from 150 to 1.5k, it sounds like a much larger change. This last move made a big difference in the presentation of voices as they are now back up there where they should be.
First thoughts about the those large waves from 5 to 1.5k was my gating of the initial driver measurement used for simming, but that looks correct. Initial driver measurements were made inside and gating is very short. Thing is, I don't really hear anything that stands out in that area. Right now, I think they are close enough to leave alone and just listen to for a while with the idea that more time spent listening to them should help find other flaws and help fine-tune em'.
Anyhow, just my jumbled up process. What do ya' think?
Added what XSim calls a shunt notch to the woofer for that 5k bump. It seems to have helped.
Funny thing as we were talking about having a long-term reference to base our observations from in another thread. I've had at least one pair of NHT 2.5i circa 1995 and a pair of inexpensive Grado sr60 from the same time period. They are my base line.
For Christmas, boy got a pair of Seinnheiser hd6xx MassDrop cans, a Schiit head amp to go with his dac that I got like 4 years ago, and a portable Altoids cmoy amp. I've never heard his hd6xx until now. Quite a bit different than the Grado cans. The bass is deeper, maybe a touch bloated, mids are a little recessed, and the highs are definitely not hot. The Grado bass doesn't go as deep, but in no way bloated, the lower treble is a little hot and not as refined but has better treble extension. The Grado has voice dialed while the hd6xx does just about everything else better. The Grado are similar in sound to the NHT2.5i, and the sound of the hd6xx is sooo close to these speakers.
With the gating and the room/listening position having a deep null around 120hz. it took a while to realize just what was going on with the bass. There was a large +6 db hill centered around 120hz. leaving the lower bass trailing till ~30hz where it took a dive.
Since now having plenty of time ... performed an appendectomy and removed most of the swiss cheese brace and shortened the port. Now response drops like a rock after 37hz. but the bass is pretty even from there up.
And remember my skills with bondo? Almost half a gallon of plastic later and they're getting close to prime. Would be even closer until trying to make that small groove cut down the seams so they would never show. Ordered a small 1/4" 60* and 90* v router bit. Will try again when it arrives.
Comments
InDIYana Event Website
Wolfy's Photobucket
Also, next time you measure reverse null, move the mic a few degrees or inches in any direction. Be careful about optimizing for that null, you are actually optimizing to a listening position and you can lose sight of the big picture.
InDIYana Event Website
Wolfy's Photobucket