Troels' Jenzen Seas Build

[jim1961]

Hi Jim.

Please forgive my laziness as I have only had time to scan your posts including the link.......

Given that I don't have the full picture of issues & what appear to be rather robust remedies you have thus far implemented, I am curios to know if you have considered some form of DSP?

That is not a course I want to pursue. I'd rather leave things as is than put a processor in the signal path. But I do appreciate the thought, and it would be a good recommendation for most folks. I want to solve this acoustically. I am not that far off. If I can gain 2-3db in the null range specified, ill be a happy camper.

[wkhanna]

I fully appreciate your steadfast determination.
I also holey agree with by-passing any digital manipulation of your signal.
I too, am a stalwart of keeping the path unencumbered as possible.
Especially with regard to my vinyl playback, but also for my digital source.

While many are touting the efficacy of DSP, especially for LF, I do not deny that it can be a V good solution for some situations.
I recently heard the Legacy DSP products at the Capital Audiofest & it was impressive.

But I still prefer my coffee black, my scotch straight up, my women wholesome & my audio pure.:W

[jim1961]

I fully appreciate your steadfast determination.
I also holey agree with by-passing any digital manipulation of your signal.
I too, am a stalwart of keeping the path unencumbered as possible.
Especially with regard to my vinyl playback, but also for my digital source.

While many are touting the efficacy of DSP, especially for LF, I do not deny that it can be a V good solution for some situations.
I recently heard the Legacy DSP products at the Capital Audiofest & it was impressive.

But I still prefer my coffee black, my scotch straight up, my women wholesome & my audio pure.:W

I dont even use an active pre-amp. Coffee extra black please

I am a Crown Royal guy, but strait up is the only way!

I like the DSP's for guys that wont (lazy?) or cant (wife?) treat their room acoustically. They seem to do more than merely EQ.

If the case here was that the problem was in the range the sub operated, then I wouldn't have any qualms to putting a DSP type device on it. Its from 100hz up I don't want any signal degradation. My sub cross over at 40hz, way below the 170-250hz range where my problems lay.

[wkhanna]

One of OEM features of my 35+ yo modified Carver C-19 pre that I truly love is its by-pass function.
It allows zero gain in the output stage through the upgraded Alps volume pot that was installed.

[jim1961]

One of OEM features of my 35+ yo modified Carver C-19 pre that I truly love is its by-pass function.
It allows zero gain in the output stage through the upgraded Alps volume pot that was installed.

Out of curiosity, and off subject, if you don't need the gain, why not go full passive?

[jim1961]

Orange = yesterday
Green = today

(1/24th oct smoothing, mic pointed strait ahead, both channels driven)

Moved the speaker 6" closer together and gained about 1.5db @ 200hz. I dont think I should go further given other problems that are cropping up.

[wkhanna]

Out of curiosity, and off subject, if you don't need the gain, why not go full passive?

I guess I have few reasons actually……

The C-19 also has a built-in phono section that, since the entire unit’s modification 2 years ago, is one of the best I have ever heard in the < $2k category along with providing a selectable sub-sonic filter.

It has tubes which, up to this point, have proven quite helpful in taming the digital signals that are fed to it when I am not spinning vinyl.

It was a gift seven years ago from my V best friend, mentor, teacher, surrogate father & most extraordinary individual that I have ever met. Since Rod’s passing just over a year ago, I doubt even more that I will be replacing this most cherished object, ever.

Lastly, it is Black……
& it has had its LED converted to Red……
…….therefore, meeting the two most important requirements related to producing the most accurate reproduction earthly attainable. :W

[jim1961]

Is the dip centered around 225 Hz a function of the anechoic response or the room interaction? It's actually pretty wide for the latter, I'd think...

Other than that, looks pretty dang nice, bitty the drop in the 200 Hz area corresponds with what is usually referred to as the power region for classical symphonic music as well as key upper fundamentals in many other kinds of music- should be fairly audible. how's it sound to you, compared with reference grade headphones?

Have you asked Troels about this?

Jon, your reference to the anechoic response has some legs I think.

Leeuwarden, Your intuition has the same merit.




Looking at this again, it does seem the speaker response is part of it. The similarity to the near field my brother measured has some striking similarity.

[JonMarsh]

That makes more sense- and might be addressable.

[jim1961]

That makes more sense- and might be addressable.

As in changing the X-over?




This is what I am getting at the listening position.

Post 130 shows near field.

[jim1961]

Actually this is more right.

Looks to me like the biggest problem is the midrange suck-out 160-250hz. Looks like a room problem there. Real guesswork whether Xing the mid lower would fix anything. Maybe.

[jim1961]

Red = 46"
Black = LP (92")

Back to the drawing board. The mid/tweet curve looks fine at 46" distance.

[jim1961]

At LP

Here is a ETC (no smoothing) at 200hz .33 octave bandwidth (176-222hz) with just the mids tweets running (ignore blue line at top). If it was room interaction, wouldnt the ETC show a significant peak somewhere?

If the mid lull (160-250Hz) isnt caused by a reflection (SBIR), then what does that leave? Room modes? If its room modes, im sunk :cry:

[jim1961]

I removed a rather large absorber hanging over the speaker.

Orange = without absorber
Green = with absorber

I suppose:

1) An ETC isnt up to the task is seeing early low frequency reflections
2) I think ill stick with the Green one

[JonMarsh]

BTW, where is the woofer to mid crossover frequency? Is this dip possibly due to floor bounce effects?

[jim1961]

BTW, where is the woofer to mid crossover frequency? Is this dip possibly due to floor bounce effects?

The mid/woofer Xover is right in the middle of the null. Around 200-225hz. Floor bounce is a definite possibility. Even probable. But I already have so many rugs on the floor on top of each other that I nearly trip over them.

[Face]

Wouldn't the dip at 200hz be floor bounce?

I wish my speakers would measure this well at the listening position.

EDIT: Didn't see the last page before responding.

[jim1961]

Wouldn't the dip at 200hz be floor bounce?

Floor bounce was the direction I was heading yesterday so I did some experiementing:

1) Put a 24'x24" 2" OC703 panels at bounce spot
2) Put rather large 3/4" board angled to deflect reflection
3) Used solid board and OC703 panel to block reflection

In each case, the 200hz response went down not up. If the floor bounce is responsible, wouldn't one of the attempts above yielded MORE energy at 200hz, not less?

I wish my speakers would measure this well at the listening position.

Ive spent over a year treating my room to get this response.

[jim1961]

Red = Before
Black = Now

I added 34uf to the mid HP filter and tweaked a few other things. I think this is better. Thanks Jon for pointing out the value of these frequencies, for I probably wouldn't have taken things further otherwise.

[JonMarsh]

Jim,

Have you contacted Troels about what you've measured? I don't particularly want to suggest monkeying around with another guy's design, but I know from experience that getting the woofer to low mid crossover down just right isn't particularly easy- one of the many reasons three ways are harder than two ways. But my first impulse is to say, this should be fixable- or at least worth investigating to see if that's the case.

Troels has some pretty complete graphs and info for response above 200 Hz, but it seems to me that he treats below 300Hz as "There be dragons", as the old saying used on parts of the map where there is no information.

If you could measure the response and generate FRD files for the woofer, mid, and tweeter on axis at a listening position, with the speakers in a good position in the room to avoid too much in the way of boundary interaction, I'd be curious to try modeling the system with the published crossover and see if there is any straight forward way to move forward or not. I'm going to be real busy at work (probably 7 days a week) starting the end of next week for a while, but still I can find some time somewhere to squeeze that in, even if later in the fall or November.

Your call of course, if you want to investigate this some more. This particular response short coming is something I find fairly annoying and obvious- my GF says she almost feels sorry for the "ear" I have, both for acoustics and music (I'm very strict about being in key, and poorly tempered instruments drive me up the wall!) (I just gave her a new digital tuner with a built in microphone, if that tells you anything... :W )

Let me know what you think... It would be better philosophically if Troels could provide a solution, but he might not even have these sitting around.. Let me know what you think.

[jim1961]

If you look at my brothers near field reading (earlier in this thread), and my own unpublished ones, I think the speaker may have a null in the crossover region of the mid/woofer, but only a part of what I am seeing at the listening position. In other words, I can make the 220hz null either go away or move to a different frequency by moving the mic position. Post #192 clearly shows the midrange null goes away when reading at a closer distance.

So what we are up against is mostly a non perfect room (aren't they all?). Are your proposing we counter the room deficit with a proportion "hump" in the speaker response? If not, please explain what you are proposing.



This shows the L and R channels for the woofer/sub vs mid/tweeter responses. The sub and tweeter are inconsequential so including them doesn't tarnish the data. Even these responses are a bit different today than what these show per yesterday for I moved things a bit (again).

[JonMarsh]

I'm proposing that you do gated measurements of the woofer and mid woofer which will result in quasi-anehoic response to confirm if the existing crossover design is the issue or if room boundary and room modal effects are causing the issue. You can't separate out floor bounce- the design has to deal with that.

What is the window gate time on these measurements, and what kind of room position do you have for the speakers and measurement microphone? What position makes it worst, what position for the mic makes it "go away"?

Normally when I do in room measurements I do them at a minimum of 2 meters, and at ear listening height. I use the minimum window length with a Half Hamming window that gives me valid LF data down to at least 100 Hz, but preferably 50 Hz. Not many data points down there, and not very good resolution, but for a crossover in the 200-300 Hz area, it's desirable to have valid data points at least every 25-50 Hz. It's a balancing act for getting the system response without too much room effect. But then I also do measurements with 200 msec window, just to compare, and identify room issues and boundary effects that can mess up things in the crossover region.

Part of the system design, of course, is picking the driver positions relative to the floor boundary in the initial design and making sure that doesn't put problems in the crossover region, to where one could be fighting the crossover and the floor issues at the same time.

[jim1961]

I'm proposing that you do gated measurements of the woofer and mid woofer which will result in quasi-anehoic response to confirm if the existing crossover design is the issue or if room boundary and room modal effects are causing the issue. You can't separate out floor bounce- the design has to deal with that.

Posts #114 and #130 show the near field responses (made by Scott) and describe the measuring criteria.

Post #188 attempts to compare the near and far field responses.

What is the window gate time on these measurements, and what kind of room position do you have for the speakers and measurement microphone? What position makes it worst, what position for the mic makes it "go away"?

The Omnimic results I have been posting from the listening position are 250ms at 95" distance. Mic height is at ear height which is 1" higher than the exact center of the midranges (see my avatar for driver layout).



Black = Absorber
Orange = Diffuser
Gray = Reflector

This ^^^^ shows the basic speaker to mic orientation.

Post #192 shows when the null goes away and also describes at what distance (46"). Basically speaking, the closer the LP (mic) is to the speaker, the less the null. The closer the speakers are to each other, the less the null.

Normally when I do in room measurements I do them at a minimum of 2 meters, and at ear listening height. I use the minimum window length with a Half Hamming window that gives me valid LF data down to at least 100 Hz, but preferably 50 Hz. Not many data points down there, and not very good resolution, but for a crossover in the 200-300 Hz area, it's desirable to have valid data points at least every 25-50 Hz. It's a balancing act for getting the system response without too much room effect. But then I also do measurements with 200 msec window, just to compare, and identify room issues and boundary effects that can mess up things in the crossover region.

The Omnimic results I have been posting from the listening position are 250ms at 95" distance. Mic height is at ear height which is 1" higher than the exact center of the midranges (see my avatar for driver layout).

Omnimic (my measurement system) doesnt have a half hanning feature. So I would trust the measurements in post 114 and 130 for the near field readings.

Part of the system design, of course, is picking the driver positions relative to the floor boundary in the initial design and making sure that doesn't put problems in the crossover region, to where one could be fighting the crossover and the floor issues at the same time.

This is my latest measurement

[jim1961]

This is my best attempt to put the far and near field measurements on one graph.

I used the "blended" setting thats defined as: blends from the "only to" calculation at higher frequencies, to the "all" calculation at lower frequencies. In other words, this mode removes echoes when it can, and doesn't when it can't.

The "only to" in this case is a 5ms gate. "all" refers to no gate. The Green graph is my "blended" response at 1M distance at midrange height. The Red graph is the far field (listening position response, 2.4M distance 250ms gate)

[jim1961]

Here an an attempt to line up my far field with Scotts near field

[jim1961]

It breaks down this way at the listening position.

[jim1961]

This graph show how the L and R midranges/tweets measure at the LP. The fact that they are so different in the 150 - 250hz range seems to point to the room more than the crossover. I exchanged the R and L speakers to be certain they behaved the same when swapped and they did within +/- 1 db.

[Leeuwarden]

Jim I'm really curious how you find the sound if you forget about all the graphs...

[jim1961]

Jim I'm really curious how you find the sound if you forget about all the graphs...

I think it sounds fantastic. BUT, I have thought it sounded fantastic before also. Time and time again I have surpassed previous levels of perceived greatness (greatness in my own subject and biased perception ).

So as long as I keep surpassing previous milestones, I keep pushing on.

I think I need to add that despite the emphasis here on the FR, I actually find more bang for the buck in working on early high gain reflections. When you get ALL those down to -20db, -30db even better I hear things in the recording never experienced before.

I could achieve a better FR curve if I ignored the ETC data. So what is seen here is the lower priority aspect of my goals.



Getting this ^^^^ was much harder than getting the FR decent. As it turns out, removing early high gain reflections naturally makes the FR better given nulls and peaks caused by room interaction are minimized. The present course I am on is the icing on the cake

[jim1961]

Black = Mic at left ear position measuring left channel @ LP
Red = Mic at Right ear position measuring right channel @ LP
Green = Mic pointed strait ahead, both channels driven @ LP


If I can manage to keep this response, should I still pursue XO tweaks?

[Leeuwarden]

If I can manage to keep this response, should I still pursue XO tweaks?

I wouldn't. Now it's time to enjoy what you accomplished!

[jim1961]

Dark Blue -3db nulls
Light Blue -2db nulls
Yellow +2db peak
Orange +3db peak

Decided to look at things from a different vantage point. I only looked down to E3, so no data plotted below there (E3-C8 or 164.8hz - 4186hz).

I guess my D minors may sound a bit colored. :boohoo:

[wkhanna]

Wow!
I absolutely love that graphic!
First time I have seen that used.
It really shows the data in a manner that one can easily conceptualize.

[jim1961]

dbl post

[jim1961]

dbl post

[jim1961]

Black = Left Channel (1/24th oct smoothing)
Red = Right Channel (1/24th oct smoothing)

Orange = +3db peak
Cyan = -3db null
Blue = -4db null

[the bass (below E3) isnt considered given the intentional +3 - 4db]

Just looking for new ways to look at the data and put it in a context most easy to assimilate. What id really like to do is plot a graph right on the keys themselves. Havent figured that out yet.

[jim1961]

[ATTACH=CONFIG]22629[/ATTACH]

Green = L & R Summed
Black = Left
Red = Right