Newly upgraded speakers now bright, harsh. Questions about how to determine what part of the upgrade caused this.

[distoga]

Is there a way to detect "harshness" with REW? Perhaps in impulse response or some other test? I'm trying to determine if/what I can do for crossover in the speaker or bi-amp, or if I'm our of luck because it's the natural behavior of an AMT tweeter or phase plug woofer.

I recently upgraded my beloved RBH 661-SE speakers to RBH's newer reference line. I was expecting their same signature sound but more clarify. The cabinet remained the same but tweeters, woofers, and crossovers were replaced (crossover stayed at 2.7k). The new sound does have more definition but they changed from warm to bright, which many people like but it's a bit much for me. I also noticed that there is a harshness at some frequencies. Violins and piano sound more detailed but acoustic guitars and male/female voices have lost a noticeable bit of their natural sound and are tinny, even my family/non audiophile noticed the harshness. I'm not sure if amt's trade natural sound for higher detail or spl?

The tweeters went from dome fabric to amt. If I put extreme toe on L/R and listen off-axis this reduces the harshness but I'm not sure I can do off-axis with the center channel well. The woofers are still aluminum but now use phase plugs. There's a very noticeable loss in low end sound/spl but rew didn't show a significant change in frequency response. For comparison, my side surrounds didn't get the phase plug woofer upgrade and those have more low end (even with new crossovers, same as the LCR's) and I can hear a bass guitar far, far better in the older woofer. What I can hear of the bass in the phase plug version is more detailed but very quiet and outshined by the brightness for other instruments.

I grabbed a pair of original 61-SE's I had for monitors (no modifications, stock original parts) for comparison and they are not harsh, guitar and voices are more natural, but it's very obvious they lack the detail the phase plug and amt's have. If I'm listening to music or drama movie I like the original components for 90% of the material, if it's an action movie then about 60% of the material. Changing the db/volume the speaker output had no effect on harshness/natural sound even from 50db to 90db, it was consistent.

I'm hoping there's a way to determine/evaluate the new components and see if there's something I can do, other then rebuild the speakers myself with the old parts I got back and sell/eat the new parts. Also to determine if I should look into the crossovers as the difference or the tweeter/woofers first for harshness and loss in bass spl.

 

[ddude003]

I would start with a description of your room and if you have any treatments... Post up REW frequency response sweep examples of RBH 661-SE before and after mods as a REW .mdat file so we can see what the differences are...

 

[distoga]

16x21x9, walls are 10" cement, ceiling 16" cement, floor 6" cement/dirt. All surfaces are polished/exposed concrete behind AT fabric façade. Ceilings 70% coverage of 4" 703, rear wall 50% coverage of 6" FR703, front wall fully covered 2" FR703. sides are 2" from floor to 6" above ear level. 2x2 triangle floor to ceiling bass traps on front wall corners. About 20sqft of quadratic diffusion above ear level on each rear/side walls, plus soffits (2'x1' soffits, heavily crowned for some diffusion near ceiling) around room. It took years to get a good sounding room which is just on the dead side as you can see in the rt60. Room has a few other quirks in that the ceiling is sloped 6" up front to back, front half room has a second level 10" down that's 12x8 (centered from 2' front front wall and side walls). Speakers behind AT screen and 10.5 from mic/MLP ear location.

Top two measurements were old R and L. Orange (lower spl) is new speaker. Old measurements do have 2 Rythmik 15" subs (located at 25% and 75% width front wall) included (I may have to dig on another composer for measurements without the subs if needed)

 

[Sonnie Parker]

If you were using subs in the old system and took those out, it could cause it to sound harsh.

There appears to be a lot more energy in the 4-10kHz range as well, which would cause it to sound brighter... especially with several other areas decreasing in amplitude.

 

[ddude003]

On first brush it looks like the old system has a nice room curve from say 40Hz all the way across to 15kHz... While the new system seems much flatter across the same frequency range... Most noticeable is the old system seems to roll off at almost double the rate from around 1.5Hz down to 15kHz... And again the new system is much flatter here...

I would imagine that the AMT would be a little flatter than a silk dome tweeter given the dispersion differences between the two...

Now the real question, what can you do about this? Got EQ? Try a house curve that mimics your old system?

 

[distoga]

I have a denon 8500 and their eq app so I can do some changes, but I'm not usually a fan of audyssey for music so if I can avoid that I'd prefer to. I'll play around with everything again tonight. I did do a quick eq drop of everything over 5k 6db and voices and other sounds still sounded 95%+ tinny so that didn't help IMO.

Considering that voice frequencies sound "tinny" I've wondered if it's some harmonic characteristic or issue of an AMT? I can't think of how a crossover could effect this specifically so that's why I'm focused on the amt, however, I'm sitting here remembering back and I remember more voice spl coming out of the amt than the mid. The crossover is at 2.7k and voice should be only from the mid? Unless voices have harmonics? When I put my ear close to the amt vs mid, the tinny sound was coming from the amt and not the mid. I'll REW again tonight but when I am I have a question...

Does anyone see something in the mdat file that might indicate harshness, phase, harmonic, or distortion issues for "tinny"? Or where/how should I be using REW to check for this? My still novice knowledge/review of data seems like the amt is has less distortion?

PS, I can hear the 'Tinny' effects even at low volume, but perhaps I should run the speaker through another test near the same spl as the old speakers and see if something stands our more?

 

[ddude003]

There are a lot of things to think about with voice/vocals... From an EQ point of view...

low-end starting around 90 Hz...
room noise around 250 Hz...
core vocals around 1 kHz to 2 kHz...
presence around 5 kHz...
sibilance around 5 kHz to 8 kHz...
harmonics out to 15 kHz or so...

Cutting or boosting, types of filters and filter slopes... So, cutting everything over 5k by 6db does not necessarily get you a nice roll off...

Don't know if they could sound better, smoothed out, crossed over a little lower... Also don't know if they need run in for a few hundred hours to sound smoother... I wonder how they sound in the horizontal position...

I noticed there seems to be a large difference between new and old in the area of distortion and noise floor... I am guessing your not measuring speakers in the same locations... Maybe try a near field measurement of a new and a old speaker with the each speaker in the same location and position of mic and same spl...

Disclaimer... I am not a studio engineer nor do I play one on TV... Although, I have some experience working with and around them... No studio engineers were harmed in the process...

 

[amail]

Having some 15 years experience with AMT as well as ribbon and quasi ribbon replacements for dome tweeter, it is almost certainly a problem of xover frequency. Even if on paper the AMT is happy to work from the same xover as the dome did, it sounds like that frequency is too low. AMT/quasi ribbons sound by nature somewhat more clinical than domes. To get the sweetness some of these units are capable of, they need to work well within their xover range, and then some. Try an approx. 3.5KHz xover point. Even a 4KHz one, if still not satisfied with the sound. Manufacturers of such units do like to quote the absolute lowest freq. these units can safely operate in order to appear more attractive. Don't be fooled by this. I don't know what slope your original xover employs, but this, as a secondary issue, may also influence the sound.

 

[JLM1948]

You can't just replace a tweeter with an other one and expect the response to be correct.
The cross-over's response depends very much on the load it is presented too. A ribbon tweeter has, by construction, a very different impedance curve than a dome tweeter. Even if they are rated at the same nominal impedance, the actual impedance is quite different. The ribbon tweeter may have a higher sensitivity, which would require to tame it with an attenuator, a quite difficult task, because an attenuator also changes the response of both the x-over and the tweeter.

 

[linearphase]

The "improved" speaker not only has more high frequency energy, it also has more variation in the the response over a large range. Additionally it has less bass extension due to the lack of a sub. All this assumes the new measurement was taken in exactly the same place. The problem is the crossover, drivers, cabinet integration.

Essentially you are redesigning the speaker so to make it work will require some effort.


Suggested TEST process:

It is only necessary to measure one speaker of each version.

First replace one speaker with the older version, careful note where you place it then remeasure.
Put one the new speakers in exactly the same position and orientation then remeasure.
Add a 2 ply tissue over the tweeter then remeasure.
Add a second 2 ply tissue then remeasure.
Toe out, from the existing orientation, the new speaker about 15 degrees, remeasure.
Repeat the above but without the tissue.
Definitely add the sub back in taking the time to carefully integrate.

Share the results of the above measurements.

From a room perspective you do have very dead room. But the deadness is relatively flat with frequency which is good. This deadness will accentuate the directional characteristics of the speakers as you hear mostly direct sound.
Also you have large reflections at 2 and 8mS as noted on the ETC. this is only on the NEW speaker. This causes the reposes aberrations in the 500, 250, 125 and 63Hz ranges.

If you wanted to try EQ and can manually tweak it then a high shelf filter with a 2.3 dB cut starting at 4500 will get you in the ballpark.

 

[Marzolino]

Searchin' a good sound

having a 3-ways multiamp system and a multi function equalizer I can use a range of regulations almost infinite and so I spent years searching the “good sound”, meaning a sound as possible bright, light and rich in details, but not fatiguing and offending my ears, I say more 'human' and less 'electronic'. That's the problem, if I well understand, posted above from distoga, in order to find what parameter can cause a harsh sound. The first I've tried to find the best equalization and the first tutor has been REW, simply wonderful and full of excellent modules. Following the REW's procedure 'Measure', and its windowed sweep, I got a good bass regulation and with a correct bass level also the mid-high frequencies went out definitely, improving the music scenario. However I have seen that the REW EQ gives, according with the theory of minimum phase, the filters only for low frequencies, but how adjust mid and high frequencies? I've found very useful the Moving Microphone Method well known also by the REW forum people. The best of the MMM is the repeatability of measurements: I walk around the room while the speakers play a pink noise signal, tracing half circles with a mic on the air and I do this recording for 60 sec. If I repeat it two or three times the resulting recorded files are surprising similar and the graphs fit together!

These measurements can be imported in REW (if someone is interested I'll explain the way) and from the graph you'll see how adjust the frequency response with an equalizer.

So use REW for bass equalization and the MMM for mid-high equalisation.

About the hardware I've tested and compared different speakers, tweeters and midranges, testing frequency response, distortion and,finally, listening to the music, comparing one vs another in real time.

What results? For whom interested here the bottom line:

• the harshness in sound is given by a top in the range 1.500-5.000 Hertz, that is the range of maximum sensibility of human ears (see the Fletcher Munson curve), so also +2 dB in that range make difference. In the distoga's graph we could see in the new tweeter a peak in the 5-7 KHz range - It could be more legible with a graph Var Smooth traced - where the old one had a dip, so it's not strange that the new one sounds more bright and harsh.
  
  Look at this compared frequency response of 4 midranges; the higher in SPL is an Acoustic Research AR 3a, original midrange from 70's, fully working, that is bright sounding, gives enthusiasm, presence, and its behaviour is well clear by the peak with centre at 2KHz and it must be well smoothed in the EQ setting. Other midrange in the same graph are more regular an sound quite smooth.
  
• You can measure with REW also the distortion both in regular measurements or with the Stepped Sine Measurement available in the RTA page. A low distortion speaker is obviously preferred
  
• a stereo system with linear response from 20 to 20.000 Hertz is terrible to ear ! Maybe that's because of the Fletcher Munson curve. Look at the graph “210420 APSX … “ of my room where we can see two curves: in the second dotted line I have reduced hi frequencies from high crossover potentiometer. That curve is not perfect but this preset sounds better!
  
• try to equalize only reducing the frequencies exceeding in the SPL level, avoid to increase the week frequencies
  
• the first rule for me is that the sound must be not irritating, a compromise is necessary if a bright,exciting sound after a time becomes fatiguing and irritating.

Now I realize that I've written a long post , be patient with a boring old audiophily.

By

 

[Mike-48]

I plotted both together, aligned at 1000 Hz, 1/6 smoothing. Here is what I see -- some mentioned by others:

1. Below about 50 Hz, low bass is missing in New. What happened?
2. Less energy in New in the broad range 250-1000 Hz. This is a hugely important range -- middle C being at about 262 Hz, and much of music's warmth and fundamental tones being in this range. This loss will make highs seem more prominent for sure.
3. The peak around 6 kHz may be annoying, too.

It might be you need to reposition the speakers closer to walls, for better bass response. The difference of over 10 dB around 300 Hz is going to be a major changer of the sound character. And as @Sonnie said, add the subs back in, or fix whatever else is causing the loss of very low frequencies. Any reductions < 1kHz will make the higher frequencies seem more prominent and irritating. If fixing the LF is not enough, EQ down the area around 6kHz, too -- around 8 kHz is where sibilance lives, and you don't want to exaggerate that.

One person's opinion.... I wish you good luck with it! I think that, with patience & experimentation, you will get where you want to be.