OK
@SIY , I ran them all again. Some had got mixed up when converting from .bmp to .jpg en-masse and being pulled from my old AudioLab host PC.
All small to medium 2 ways, mostly 165mm. All bass reflex except one. Nothing really expensive, just fun little speakers.
B&W 602S2
View attachment 43395
Jamo E-800
View attachment 43396
Sony APM-121ES
View attachment 43397
Mirage FRX3
View attachment 43398
Wharfedale Diamond 9.1
View attachment 43399
MB Quart QL10C
View attachment 43400
Dali Zensor 1
View attachment 43401
Gale 3010S
View attachment 43402
Jamo E-855
View attachment 43403
Jamo Cornet2
View attachment 43404
I did this same thing during covid with roughly 35 speakers I just tested 1 of each and I was surprised to find that only 1 "dome" tweeter did not have that big 8 to 12KHz peak before rolling off to nothing. One more thing I found funny was how 1 set of Sony speakers had a big ad campaign for selling them was the incredibly high 30KHz range. That speaker tested pretty much worse than others with an enormous peak around 8KHz and then it started to roll off when it got to 20KHz it was -6db, the mic I used was easily showing a response up to 24KHz when testing electrostatic panels made by JansZen and RTR albeit they were slowly rolling off and -6db at that point. It was dependent on program used for testing not many give you the option to test past 20KHz.
Seeing your plots sent me back to that day of tests, the biggest issue I had was keeping tests labeled properly requiring me to retest many of them that I forgot to label and had so many similar looking speakers all tested in one sitting. Just constantly having to connect wires to them was a pain in its self since bookshelves will use all different connectors many barely fit 16ga wire and then finding a bad tweeter on some made it more confusing to think back if I didn't label it directly after testing. I put them all in a photo editing prog after I did another 10 tests after I retrieved more of them from storage and I used a few different programs on each speaker to provide real world comparison between so called valid results only to find that those tests are only good for comparing results per program and without having a second program that agreed and had exact or very similar results, to confirm which programs are believable, I pretty much scrapped the entire project.
I wouldn't doubt any programs being manipulated by speaker manufacturers to create programs that flatten that peak to help sell speakers (or more precisely, to stop people from checking out electrostats since they know they would never be able to sell another box speaker after that). Ive seen how electrostats are constantly misrepresented in forums and how "what determines a good speaker" in tests has changed to try and level the unlikely box speaker to even be a consideration when speaker shopping.
The one aspect I noted was the failure of some tweeters as an effective driver for high frequencies. The overwhelming peaks just before roll off (but within typical adult humans) was a built in trick that was used by almost every lower cost or entry level builder. I found far better plots with paper cone drivers that had much more linear response up to or near 20KHz. Those include budget Bose interaudio models (I remember because I was caught off guard) and a few JBLs that were very smooth sounding with real music but never would find their way into my system, that tells you just how awful the rest of them were compared to what I was used to hearing.
But if you consider what is most important to rate things, my choice of speakers would be far below the awful noise I heard from 35 or more other speakers that had a way better (but intolerable) off axis response, which for some reason has become so unimportantly important in ratings. Tests become meaningless once the test itself is investigated and without a second completely independent test written a completely different way, can corroborate the results, no tests are worth considering to be correct or usable unless comparing things only tested against each other within that one test session.
When I see tests done here corroborated with another sites tests I have to know if they used the same programs and equipment or can the results be the same using completely different set of tests and instruments. I would expect everyone here to have this same concern or you haven't thought this reality through.
I found no 2 tests that show exact or even nearly the exact results out of over 30 computer programs. I have saved them all and I can post them for others to try. They are all free to try and some are free to use forever. Some require older windows so and I wasn't able to get every one running, some just don't work, others are missing needed test equipment but there were at least 20 that did run and the final results had me believing that I could not trust any single program to be accurate. Without that I have abandoned computer RTA & FFT programs and began testing phone apps using a calibrated mic and then I found even more unreliable results by switching phones while using the same mic.
I have found 3 separate apps that do give very nearly exact results out of maybe 50 apps I've used over the last 8 or 9 years. I used to have a few eqs that had built in pink noise with mic inputs but at that time I lacked a real test mic so results were unusable and I never bothered. I still have 1 eq with that feature but it doesn't work properly after the fire at my place. So I can't use it as a tool that could bridge the gap between actual mechanical hardware made to do analog testing (which I would trust more than any program based on FFT's many unverified versions) and the results of programs.
Just as written in this very thread there's an ongoing conflict that's pointed out in the circuit Amire made and the simulated circuit. That point alone is key to exactly what I always say when it comes to test results. A real world hand held meter can and will give you actual real world numbers while a program will be inundated with constant scrutiny over how its implemented and how the numbers were derived. Just the values of each part are constantly being proven incorrect or changed constantly do to misinterpretation or not taking all consideration of how it will interact under a load or at certain frequencies is exactly why a meter is the best way to know values.
That makes me consider a mechanical testing machine that works with analog signals and a vu meter will always be a more trusted form of sound testing than anything that's converted to digital than back to analog through a series of guess work at some point assembled into 1 easy to use program that cannot be corroborated with any others and the scarcity of mechanical testing equipment available to find real world usable results makes it harder to know exactly what parts of programs can be trusted or if any are actually usable do to loss of linearity in the results. Some show humps at one frequency that are not there in other tests but it could just be a number used that produces skewed results when the formula might be correct in its method but how do we know which is correct when there are so many variables to consider in FFT tests?
One thing I figured out after it was all done was my computer sound card is complete garbage, unusable for anything via the analog outputs. I can hear my mouse moving if I connect my amps and speakers to my computer. As a source it's unusable and I ran all my tests through an external sound card via USB and my source signal was plstrc through an external device connected directly to a dual 31 band eq into an external crossover and 2 separate amps dedicated for woofers and for panels. I always use an external source no matter how I'm testing, phone app for computer program, that way I have 1 constant that I know will always be there. Never trust any tones or sound flles to contain perfectly leveled tones unless verified with an oscope first. But in reality none of this matters without corroboration.
I own many amplifiers and I'm a HiFi specialist with a lot of experience so you don't have to be above me. Everyone has their own opinions. I didn't write that it's bad but it's not a miracle for $190 either. The amplifier is quite suitable and cheap for the undemanding.
