Tuesday, January 16, 2024

makePlaylist (mplay) version 0.79 released

I've talked about it, posted a video of it, and now it's available for download.

https://sourceforge.net/projects/makeplaylist

This allows you to automatically generate random playlists.  Now you can preface or mix in newer files in various ways, combine folders with weighting factors, and mix in favorites.  It makes for very infinite and very interesting scripting possibilities, with examples given.  One program included figures out how much of the last playlist you actually played and allows you to reset the history for those files that weren't actually played. 

Works immediately on older Macs* or Linux, on other systems Tcl must be installed first as all the programs are written in Tcl.  The example scripts are written in (linux/mac) bash.

(*Since I use an old Mac running 10.13.6 myself, I'm not sure about the status of Tcl in newer Mac releases.  I think it may not have been included by default with M1 macs, but not sure.  It's still available for nearly everything because Tcl is a very lightweight programming language, lightweight but powerful.  But you may need to download and install it first, and I have no recent experience with that myself because it was always included on Macs I have used.)

UPDATE: since I first made this post a few days ago, mplay has already been updated to version 0.79 which fixes a lot of things that had been bugging me



Sunday, January 14, 2024

New EQ, Day 3

 After messing with the "Background" subwoofer EQ (again, last time was a few months ago) in the right channel, which I had been re-EQ-ing because of the new speaker, but the subwoofer hadn't actually change...

I decided I first needed to test and make similar changes on the left.  Measurements at the doorway and the table showed overly bulging response at 31.5 Hz.  It looked better cut back to zero as with the right channel.

But listening to Grouse, We Want To Be Loved, I decided this all made the deep bass line, which reaches down to 31.5 Hz, a bit less easy to follow.  So I restored boosting at 31.5 Hz in both channels in background to about +2dB.  (In the Serious Listening EQ, it's boosted 5dB because of suckout at listening position.)

Also, I could not confirm with pink noise any serious issue at 20 Hz, so I decided to reduce the cut there from -4dB to -2dB.

Current settings:

Left (last one adjusted)

20 Hz    -2dB

25 Hz    +2dB

31.5 Hz +1.5 db

Right

20 Hz   -2dB

25 Hz    0dB

31.5 Hz  +1.5dB

I think I might make both channels more like the Left settings, mainly boosting 25 Hz at +2.  Otherwise, the response IS falling at 25 Hz relative to 31.5.  Even with that +2 boost it is falling at 25 and 28 Hz relative to 31.5, but it is falling only ever so slightly at 28, which is a good improvement.

But it was too late to go any further.  Actually, once again, for best results I need to sweep this with an oscillator, or use a LF tuned keyboard (as I have done before).

Just like a down sloping high end, required because of the higher side reflectivity of small rooms, a slightly tilted up bass is generally required, it's just a question of how much and exactly where.  Finer adjustment of the hinge points may mean less boosting is needed to retain musical coherency.  There is probably some perceptual compensation for room gain, and the total room system output should invert that compensation.

Now, morning after, taking a look at the graphs confirms my memory and shows some other interesting things.  First, what it looked like with the old boosting (which I think was +5db, maybe only +3dB), in the kitchen doorway:

Left 2+2 with +5dB deep boost at kitchen doorway

Reducing the 31.5 Hz boost to 0dB but the 25 Hz boost at +2dB, it looked like a much flatter high plateau in the deepest bass (though I later re-raised 3.15 part way to +1.5 for musical reasons, not measured).



+2dB at 25 Hz, 0dB at 31.5

One thing overlooked here last night was the incredible broad depression from about 80 Hz to 300 Hz.

It's almost as if (and I have tried that at times) attempting to compensate for the lack of 125 Hz in the right channel by giving it some extra oomph in the other, which only "partly" works.

That looks like the next thing which needs to be investigated.  What the hell is going on at 125 Hz.?

With the background EQ, in the doorway, it's nice looking now in the right channel but part of a long depression in the left.

Saturday, January 13, 2024

EQ'ing new speaker, part 2

 In the last post, I purported to show how removing most of the boost I had given the Acoustats a few years ago, +3dB and 1/3 octave at 1013 and 853 Hz, fixed the balance problem between new and old speakers.  It seems like the newer speaker doesn't need that boost, perhaps because of tighter diaphragm.  My initial approach was to cut those two boosts by 1-2dB to +1.5dB at 1013 and +2 at 853.  But there was still a troubling difference around 500Hz, the newer speaker having a suck out at that point.  At first I thought it was because the 853 Hz boost wasn't wide enough.

Fixing that was very problematical.  Boosting 500 Hz, with +3dB in a 1/6 octave GEQ boost, produces a peak around 600Hz that's even more alarming than the depression it replaced.


So the issue is we need to push 500 Hz up, but 600 Hz down.  It looked to me like the down part was actually the most important, and I found a 1/6 octave cut at 563 Hz worked pretty well.  After doing that, the peak seemed to move up to 800 Hz so it was clear I needed to remove any boosting from 853 Hz at all, so that as of January 13 at 4am the newer speaker EQ now looks like this:



Which measures (with the iPhone 8s held about 1 foot from the back of the chair, something I'm now trying to be more consistent with)

New Left 2+2 with EQ updates above

Even the Analyzer app thinks I've nailed the EQ at 1kHz, not that I trust such things.  But it's reasonably uniform from about 200 Hz to 2kHz where it starts an intended downturn pretty much as intended.

I'd still be interested in pushing down the region from about 600 to 800 Hz slightly, I don't believe that should be louder than 1kHz.  And possibly broadening the EQ right around 1kHz which is slightly depressed.  But I've already gone beyond the level of fine tuning which should be done with RTA.  I need to switch over to doing oscillator sweeps so I nail the critical frequencies where change is actually needed.  RTA is simply too approximate, you can't really see what is going on, though it's useful for getting the big picture.

But I think this is already pretty darn good and far better matching the right speaker than when I started.

I also took a brief look at the funky dips at 40 and 70 Hz.  Those are because I have been using the "Background listening" Bass (subwoofer) EQ for these measurements which has a suck-out at the listening position but sounds better (less boomy) everywhere else in the house.  

Measurements roughly confirmed this.  I decided for now upon two different places to measure the background EQ, the doorway to the kitchen (which is about equivalent to the couch) and the kitchen table.

At the doorway, the response rises with decreasing frequency below 180 Hz, peaking around 38 Hz.  I've needed to roll off 20 Hz in the background response to prevent the whole house mode that occurs around 20 Hz, which I can feel, for example, at the sliding glass door (which perhaps I should also measure...).

Kitchen Doorway with pre-existing 25-31.5 Hz boost

This is pleasant actually, gives a kind of loudness compensation for background listening, and requires little change from the 'flat' response at the listing position with the Listening Position EQ setting.

But I felt the rise including 31.5 Hz was just too much.  

It was pointless to boost 31.5 and 25 Hz at all, since that boosting was actually causing them to stand above everything else so much.  So I dialed those boost back to zero in the right channel of the Background EQ.  Even with 0dB boost at 31.5kHz, there is still a bulge there, dropping lower (by design) at 25 and 20 Hz.

Kitchen Doorway with no boost 21-31.5 Hz

That was fine at the dining room table position too, showing the slightest peaking around 31.5 Hz even with no boosting.  But the dining room table shows a huge notch at 56 Hz, a peak at 125, and another smaller notch around 180 Hz which need further exploration.

Dining Room Table

I do most background listening from here, so it should be better than this.  And certainly the recent removal of boosting 31.5 Hz is only to the good as well.  20 Hz is doing fine suggesting the cut there is of no harm here.  Save for 56, 125, 180, and 720 or so other, Background EQ'ing looks all to the good.

Thursday, January 11, 2024

The White and Black 2+2's

My blog photo now represents the new living room system reality.

I'm using one white Acoustat 2+2 from my original pair of 2+2's that I bought in 2019.

It's mate had an issue I was long aware of, but mostly thought I'd solved.  I don't believe I had.

On some recordings, there was a terrible resonance with some combination of bass frequencies.  I first heard it on the Grouse album 'We Want to be Loved' with the track 'Suicide kills me'.

I pretty much eliminated the issue with my 125 Hz 8th order crossover, which also resolved what seemed to be other issues resulting from rear-reflection and Acoustat resonances.

When testing the new black 2+2 I just bought in December 2023 before purchase, I noticed a similar issue with 'You Know Too Much About Flying Saucers' on Hello Waveforms by William Orbit, on one of the new speakers.

Well that same new speaker had a continuous but almost inaudible arcing problem I could only hear with my ear right up to the speaker.  I could not smell anything and believed the ozone emanation was trivial, not unlike with a small motor perhaps.  So I figured it was fine in the well ventilated but rarely used living room, but perhaps not so fine in the bedroom.

But rapid ABX switching of the speakers, and/or leaving the speakers connected to the running amp while plugging it in, or some other experimental mistake, I blew out my Hafler 9300 in one channel.

I don't really know that the arcing speaker was at fault.  Perhaps the ABX switching itself is unsafe (but it never hurt the Hafler or Aragon before, only the Krell).  But the coincidence of having an amp fail within a month after connecting to this arcing speaker makes me wonder.

(Also, the guy I bought it from, I well knew was having endless weird problems related to the 2+2's if not other things.)

Now perhaps the problem only occurs when the speaker was being charged up (though once again, that was never a problem before).  I'm resolved now to plug and unplug acoustats from AC power ONLY WITH THE SPEAKER LEADS DISCONNECTED.  You want the HV to stabilize before connecting an amp to it, especially if there is an occasional shorting issue.

But that kind of thing can't be prevented, for example, if there is a temporary power failure, when the power is restored once again the interfaces with be charging, mostly likely with the speaker leads connected.

So, I consider the second black 2+2 with arcing to be unsuitable for use except with "expendable" amplifiers.  Perhaps things like one of my two HCA-1000A's.  Not that special.  AND MOST CERTAINLY NOT WITH MY ARAGON 8008 BB !!!  WHICH HAS UNOBTANIUM TRANSISTORS!

And then I decided also that on complex music, the arcing 2+2, even if the arcing wasn't audible, still made for it sounding fatiguing after awhile, which I determined might well be traced to the arcing causing unstable membrane voltage.

Now the faulty black 2+2 has been moved off to my bedroom so it can be easily accessed for repairs.  I've decided it's the first 2+2 I want to fix, though it might be harder to fix.  I want to end up with two of the black ones, because I think they are better somehow, tighter membranes perhaps.  They were much later production, most likely made in Arizona instead of Florida, though not as far as I can tell using the later improved coating.  Perhaps other construction details had been improved.  Or perhaps they were just less used.

I can now mostly rule out the issue being a difference in interfaces, because I have been using one of my old Red Medallion interfaces with the new 2+2's, and my white 2+2's had a C mod done on them by a well regarded Acoustat refurbisher.

But it is true, though I often forget and doubt it means much, that the white 2+2 I am still using still has the non-Medallion transformers.  I plan to swap it with the other large-interface I have which has the a recently made Medallion transformer I installed in my first attempt to repair it, before I noticed that the issue was that the 'crossover' board had been connected incorrectly.  I then didn't bother to take out the Medallion transformer, I figured it was better anyway, even though there was now a mismatch with Medallion HF on one side and not the other, which I could never tell that any difference resulted from that.

THEN, I will have Medallion HF on both sides.  There will still be a lack of Medallion LF on one side, but that is said to be even less important for the sound.  It seems one of the primary purposes of the Medallion upgrade was to prevent transformer failure with improved insulation.  It may have also resulted in tighter coupling, and the improved sound was hyped, but never taken by anyone as a kind of 'night and day' difference, the old non-Medallion transformers sounded great too, and just doing the "C" mod alone is supposed to make a far bigger sonic difference than the Medallion transformers, though they're great if you got 'em.

Anyway, even right now there's still a difference between my good white and black 2+2's, despite both having "C" mod, and probably not just because one has Medallion's and the other not.

Now in these response I'm going to show, ignore the bass below 125 Hz, that's may look bad because of the new 'background' EQ that was operating in the bass, which produces a suckout around 45 Hz at the listening position.

(Update, I've discovered an error in the sub 1kHz eq, the left and right don't match in the GEQ section, it looks as if I've simply failed to dial in the Right channel, which is possible, back after my chairside EQ died.  It's unlikely the two channels previously needed lots of eq in the left channel and none in the right, in fact the right white speaker, now removed, was very problematic wrt EQ.  That explains some of the weirdness below 800 Hz in the right channel.)

I've always turned the HF level on the white acoustats to about 12 o clock, when I now understand 'flat' is supposed to be 3 oclock.  So what would 12 o'clock be?  I'd guess around -2.  I previously thought 12 o'clock sounded best and was supposed to be flattest.  (I'm applying a vast amount of HF downwards EQ above 2kHz too, without that and the HF adjustment it would be about flat to 17kHz.)  I have a smallish room which probably requires downward HF EQ.  It's not a particularly lively room, just about balanced perfectly for conversation.

To match it I've dialed the interface on the black units down to the lowest HF level the compact interface permits.  It's a very limited range of the 16 ohm resistor.  I'm guessing that level is around 2dB also.  And lo with these settings the frequencies 10kHz and above are fairly well matched.)

Also note that EQ from before IS being applied, as well as the 8th-order phase corrected crossover at 17kHz for my supertweeters when I no longer have supertweeters, but I haven't gotten around to taking it out and think it might improve the sound anyway because of the phase correction.

That EQ was dialed in by ear with the original white 2+2's, which seemed to need a 3dB (!) boost at 1kHz to sound good, also at 850 Hz at 3dB.  Otherwise, the midrange sounded weak, perhaps not just with regards to it's immediate surroundings.  Prior to EQ 2kHz measured higher than 1kHz, even with 2kHz dialed down.

Well, it seems the 1kHz boost is simply not needed with the good black 2+2.

Taking that out, or even just cutting it back to 1dB was sufficient to fix the 'balance' problem I had fixed by ear the day before by cutting the left channel 2dB in my Tact.  (But if I dialed it back to 0dB, I'd then need to cut the right side to get it to balance, so I'm now using 1dB boost at 1kHz for the black 2+2, pending further investigations.)

Now, it would be hard to believe that the Acoustats were originally voiced with a 3dB suckout at 1kHz.  I think that correction may have been needed because of membrane aging on the white 2+2's.

The result is that images are nicely behind the speakers as they should be, much clearer images than I've ever had before thanks to the speaker swap, removing the supertweeter towers, and possibly other recent changes.

Anyway, prior to adjusting the 1kHz EQ down in the right channel, here is how the white and black speakers now on the left and right sides respectively measured:

Left Channel with White 2+2

Right Channel with Black 2+2

You might notice as I just did that the actual level of the 1kHz bulge in the right channel doesn't actually reach quite as high as the narrower bulge in the right channel.  True, but the above measurements are WITH a -2dB reduction in the right channel level, as I needed to achieve left to right balance.

Don't pay much attention to irregularities above 2kHz as the measurement device is only being held in approximately the same position each time.

After taming the midrange bulge by tamping down the +3dB EQ at 1kHz to +0dB, it's looking a little different:

Black 2+2 with 0dB boost at 1kHz

But now 1kHz is lower than 800kHz, so I decided to up 1kHz to +1dB and cut the +3dB boost at 800Hz to 1.5dB, resulting in this:

Black 2+2 with less EQ at both 800 and 1000 Hz

That's the best yet in terms of flatness through the 700-1500 Hz range, and with these changes it no longer needs a -2dB overall level reduction in this channel.  Apparently the image shift was caused just by the boosted midrange.

However, noting that there's still a measured boost in the left channel, I'm wondering whether THAT is correct too.  But exploring that will require subjective testing, the way I arrived at the need for a boost there in the first place.

Sound level measurements with RTA are never definitive, only a guide to what's probably right.

But it occurs to me now that perhaps I boosted the midrange mostly because of issues with the right channel white 2+2 that has now been replaced.  It might have had quite a suckout in the midrange which I was compensating for by boosting both channels.

Meanwhile, I should also address the sub 1000 Hz issues that are apparent in these graphs.