Report back on stereo speaker voicing investigation

**Carl V** · 02 July 2016, 23:00 Saturday

interesting....
I recall The SL write up and the Harmon writings.
Some of their speakers do in fact of treble compensation
Switches. Curious about the slope & knee of their contour.

What speakers were you using the Ardents, B&C or your smaller two ways?
How close are you sitting? what sorta spl? Do you use 1st reflection acoustic trx?
Toole & SL are a bit agnostic about that aspect. (8+ years ago visited him at Corte madera
while en route to Healdsburg to look at his speaker builder's shop & then onto the Sea Ranch).

back in the day J Gordon Holt wrote "down with flat".
Von Scheirkert has a similar voicing idea....part of his reverse microphone notion.

**JonMarsh** · 02 July 2016, 23:58 Saturday

Hi Carl,

Listening was done with the Ardents, both at my normal couch seated position and in the dining room which starts at the back of the family room. Family room is about 25' X 25'; speakers are about 10 feet apart, centered in the room and about a 7 feet apart, centered on the far wall (the back wall is open to the dining room which is another 12-14 feet deep), and about 5 feet from the rear wall.

There are no acoustic treatments, but it is a high ceiling. Half the rear wall and all the right wall are glass. At the normal seating position, the speakers aren't near field, but they're much closer to the listening position than reflections from the side walls.

SPL was from about 70 dB to 90+, possibly close to 100 dB, becuase the shift in balance encourages listening a bit louder; so I was moving around a lot to judge the response from different locations and at different volume levels.

At the moment I'm sitting back in the dining room listening to Arvo Part Frateres (male vocal chorale) and it sounds pretty fabulous to me.

The electronics I'm using are what I call the midrange test setup-

SonicTransporter (Linux headless music player Ethernet output) (CAT7 cables)
Sonore microRendu (Ethernet input, USB output with low noise clocks and linear regulators)
Alpha USB to AES+S/PDIF converter
Cardas AES/EBU cable
AURALiC Vega DAC
AURALiC Taurus Pre preamplifier
Ayre balanced interconnects
NAD M22
Cardas Golden Reference

No re-clocking, no TotalDAC, non of the power conditioners, etc. Just basic good stuff.

**BobEllis** · 03 July 2016, 07:11 Sunday

I'll bring my comments from the other thread here. I approximated the version 2 curve using JRiver Media Player's DSP, setting a high shelf to drop 1.6 dB and a low shelf to boost, centered at 2KHz. I basically confirmed the curve by measurement but had a computer crash before saving. The speakers were generally flat before EQ. Equipment was a background music setup - PC, built in sound card, 15W T-amp, a pair of MTMs that were a Zalytron kit using Focal paper midwoofers and a Kevlar T120, generic cabling. No room treatments, wood floor. The only nods to good sound in the setup are a cathedral ceiling and wide room roughly Cardas placement, listening at about 10'.

Upon engaging the DSP it sounded a bit soft on top, but that's too be expected. Then I noticed that it held together much better at louder volumes than before. I'd always thought these got harsh at high volumes and sounded best loud from the next room. Now I could rock out in the sweet spot much longer. Definitely improved the sound of a basic system. Past tense because the computer had been on the same circuit as me when I had a lightning strike. It's cheap/old surge protector kept it alive, but once the weather turned warm the processor fried itself.

**ergo** · 03 July 2016, 08:02 Sunday

As Jon mentions I've listened to this mod with LspCad simulation

Spotify or JRiver --> RME soundcard --> digital loopback --> LspCAD --> RME SPDIF --> NAD C390DD digital power amp --> Minerva EE version.

Blue is Jon's curve of his analog filter simulatio and red is what LspCAD gets with "1st order bass boost" block.

What I wrote to Jon as my impressions

"What I hear as a difference is much more than I expected. One would of course expect a bit of a change in sound balance, but what I experience is something more. Somehow the in a soundfield something 'clicks' to place. Without the mod especially with complex music the imaging tended to smear and the depth layering was not quite good. With this voicing curve both of these aspects change. I experience better imaging and depth even with double bass etc. I tried this cautiously multiple times and with different time intervals spent on either setting. I am in end confident enough to say that, yes - I do hear and like this setting a lot.
A good sign is also that it is all in all very hard to remove the setting and I end up listening long just for fun each time as it sounds so good.
Anyhow. I share my thoughts first via e-mail. "

I also agree with observation that this curve allows or even encourages to turn up the volume. Seems to sound especially good with higher volumes.

**scottvalentin** · 03 July 2016, 09:18 Sunday

Very interesting results so far gentlemen. I look forward to hearing more, perhaps even one day doing a design that uses this FR as a reference!
Jon thank you for sharing and working on this. As always, I feel smarter this morning than i was yesterday thanks to you.

**sdl2112** · 03 July 2016, 10:16 Sunday

I want to say thanks to Jon and others who are participating. Great timing...I am eager to investigate this on my SSA-WG crossover design.

**ergo** · 03 July 2016, 13:40 Sunday

There is a headphone/artificial head topic going on the Innerfidely site but it touches the topic of Harman house curves also. For those who are interested in this and headphones as well.

Home Page

http://www.innerfidelity.com/content/warren-tenbrooks-summary-head-measurements-harman#eiUK2fQTMsr02eQA.97

Stereophile covers everything high-end and audiophile audio. Turntables and music servers, to solid-state and tube amplifiers and preamps, to loudspeakers.

Home Page

http://www.innerfidelity.com/content/first-crunched-data-harman-head-measurement-session#P0cwKTTRSPZvawuf.97

Stereophile covers everything high-end and audiophile audio. Turntables and music servers, to solid-state and tube amplifiers and preamps, to loudspeakers.

**augerpro** · 03 July 2016, 15:22 Sunday

So this tweak is in regards to the anechoic on-axis response? I figured something like this would have to be done on a dipole like the Orion, but it is surprising to see it works here too.

**JonMarsh** · 03 July 2016, 18:36 Sunday

I would simply suggest trying it. You just might be surprised...

Systems like the Ardents have solid and consistent off axis response to 45-50 degrees, which in my mind means they don't have as much down tilt in the power response as some traditional two ways with larger woofers.

**JonMarsh** · 03 July 2016, 23:20 Sunday

A few more data items...

First, for those curious, here is a voicing circuit to do the 1.8 dB curve - just changing the series resistor.

Note that R5 and R6, C3 and C4 are just modeling the input impedance of the amplifier, as previously.

In the actual built version, C1 and C2 are WIMA MKP2 polypropylene 250V 15nF capacitors, of a family/type that I use for the output filter in my own experimental Class D amplifier circuits, due to the very low impedance and ESL. They're the minimum coloration part I could think of on short notice for trying this out. Very good electrical characteristics, pure copper leads, stacked film low inductance construction.

I've ordered some SwitchCraft inline XLR male to female adapters to build a couple more pair of these for experimentation and test sharing.

**JonMarsh** · 03 July 2016, 23:57 Sunday

Last night I was listening to a Stereophile reference recording, a men's choral music performance, mostly accapello, and I found it most sublime.

And I may have had an ah hah moment, about understanding this performance of this configuration, because I realize what it reminded me the most of in reproduced sound I have previously heard is various models of the Quad ESL, including the original Quad ESL 57, which I have a lot of experience with due to selling them in the 70's and configuring Levinson HQD systems.

There's a perceived naturalness and ease that Quads have always had that I have rarely heard in other systems, and I think a voicing curve like this may be one aspect of that perception.

**dar47** · 04 July 2016, 00:27 Monday

Well let us know when they are ready to ship, this should be fun. I am thinking this is one of the reasons people stick to vinyl especially at louder levels.

**TEK** · 04 July 2016, 10:03 Monday

Originally posted by dar47

Well let us know when they are ready to ship, this should be fun. I am thinking this is one of the reasons people stick to vinyl especially at louder levels.

Could you elaborate a bit about the connection you see between vinyl usage and the voicing curve used here?

**JonMarsh** · 04 July 2016, 10:22 Monday

Originally posted by dar47

Well let us know when they are ready to ship, this should be fun. I am thinking this is one of the reasons people stick to vinyl especially at louder levels.

I think it should be- you may be surprised at how much "fun" you'll get. I ordered three more sets of the XLR adapter cases this morning, and as part of the original Wavecor Ardent crew, of course you'll get priority. :B

As mentioned, I am pondering adding this into the front end of the Class D receiver/control amplifier, as an option that can be turned on or off with a switch, and chose between the 3 dB curve and the 1.8 dB curve- this would be a pretty simple modification, done after the balanced input transformer, before the input differential active amplifier stage.

My thought is to continue designing speakers for "flat" response, and implement voicing tweaks like this where it is simple and very inexpensive compared to crossover adjustments, particularly when the drivers often have frequency ranges they cover that are not convenient for implementing the adjustable curves property. Simplifies speaker development, and makes the implementation cleaner.

**dar47** · 04 July 2016, 11:02 Monday

When playing laud complex passages with digital, sometimes a note will just ring my head and kind of distract me from the music braking my mode. I haven't experienced that with vinyl but it's not to say those passages are rendered wrong just maybe flat is heard harder in those examples. I may not like this as nice crisp highs at low levels is nice and I generally don't listen much above 80dbs anymore. If I am sitting for some music I always play low for at least a half hour and kind of warm myself up, then tend to bump up as I go but it sometimes is surprising how laud I end up at. If you sit and listen for a couple of hours at low levels your going to tune out or fall a sleep, haha.

**JonMarsh** · 04 July 2016, 13:51 Monday

Originally posted by dar47

When playing laud complex passages with digital, sometimes a note will just ring my head and kind of distract me from the music braking my mode. I haven't experienced that with vinyl but it's not to say those passages are rendered wrong just maybe flat is heard harder in those examples. I may not like this as nice crisp highs at low levels is nice and I generally don't listen much above 80dbs anymore. If I am sitting for some music I always play low for at least a half hour and kind of warm myself up, then tend to bump up as I go but it sometimes is surprising how laud I end up at. If you sit and listen for a couple of hours at low levels your going to tune out or fall a sleep, haha.

i don't want to go out on a limb here, but I think you'll find your listening habits changing if you plug this into the chain... :B

**NyxOne** · 04 July 2016, 15:19 Monday

Originally posted by scottvalentin

Very interesting results so far gentlemen. I look forward to hearing more, perhaps even one day doing a design that uses this FR as a reference!
Jon thank you for sharing and working on this. As always, I feel smarter this morning than i was yesterday thanks to you.

this ^

Thank you guys for sharing you're result! Now I know I need to try it too!

**Carl V** · 04 July 2016, 15:37 Monday

Dar...Others have made similar comparisons to clean & Capable tube amplification.
(((I recall back in the day getting Soundlab A3 and Wolcott amps singing.
But then Crown Macro Ref. and Big maggies made me smile))).

**meb46** · 05 July 2016, 02:19 Tuesday

Bob - Could you post a screen-shot of your JRiver setup/DSP? I use this also and would be a quick easy way to understand the details. Thanks

**PMazz** · 05 July 2016, 06:19 Tuesday

If anyone has insight to what a minidsp filter would be I'd give it a whirl.

**BobEllis** · 05 July 2016, 07:31 Tuesday

Here you go, MEB. There are two catches. Foremost you must be using the player to output music, not sending it to a DLNA renderer. Second is you have to have JRiver sending audio to a device other than the default playback device under windows. I assume you have an external DAC of some sort. Set Windows default playback device to the built in audio (you have to have something plugged into the jack, if just an unconnected plug) and click Player>Playback Options>Audio Device select the DAC or spdif device you use for music. This has the side benefit of keeping windows sounds out of your music playback if you do other things on the same box.

To get to the EQ click Player>DSP Studio

**sdl2112** · 05 July 2016, 16:46 Tuesday

I was curious of the effects of different amplifier input impedance so I ran three different values:

1) amp(Vo1) 200kohm balanced or 100k single ended (same as Jon's first example)
2) amp(Vo2) 75k balanced or 37.5k single ended
3) amp(Vo3) 20k balanced or 10k single ended

Markers on the graph show attenuation at approx. 20Hz, 2kHz and 20kHz. With decreasing amplifier impedance the insertion loss is larger and the delta between the shelves gets a little smaller. The transition region is similar.

One thing of note is that during the Stereophile review of the NAD M22 they measured a balanced input impedance of 75kohm, not the 200k in the spec sheet.

Response graph

Normalized graph

**BobEllis** · 05 July 2016, 18:09 Tuesday

I didn't think the range of normal input impedances would matter that much. I guess we'll have to simulate our own amps before building a dongle. Given the 1.8K input impedance of my Pass A75 the circuit as specified wouldn't have the desired effect.

**BobEllis** · 05 July 2016, 18:23 Tuesday

Originally posted by PMazz

If anyone has insight to what a minidsp filter would be I'd give it a whirl.

Pete,

I think this is pretty close - higher Q would be sharper corners, but I think .5 is right for this filter. Maybe a professional sparks and wires guy can verify.

Duplicate for the other channel, of course.

**JonMarsh** · 05 July 2016, 19:05 Tuesday

That looks like its more or less in the ball park, though if you could control the scaling of the display that would help! Of course, there always is that old standby, set it up, measure, adjust measure again, then at that point you're likely good to go. Key points might be looking at the 1kHz, 2kHz, and 10kHz levels.

Over a range of 200K to 50K balanced this tracks fairly well, just more attenuation overall, but below 50K I'd say give me your amp parameters, and let's customize for it.

I'll post a target plot with detailed grids and markers.

**BobEllis** · 05 July 2016, 20:13 Tuesday

I've got multiple single ended amps in the 47K range, and with a flip of the switch the A75 is a 75 kOhm single ender, too. I'll just build the filter myself, since I only have single ended sources for now. I've got a load of Panasonic and Wima stacked film caps and likely the correct value Dale RN60s in my parts bin. I also plan to build an Aleph-P, which will take care of SE to Balanced and I can put the eq in a higher impedance zone when driving the F5 or Aleph-J.

**JonMarsh** · 05 July 2016, 21:12 Tuesday

Sounds like you've got a solid plan, Bob! :T

**PMazz** · 06 July 2016, 05:02 Wednesday

Thanks Bob, I'll give it a whirl!

**JonMarsh** · 15 July 2016, 20:23 Friday

Moving the ball down the field, I've laid out a PCB to use for building this inside of the SwitchCraft balanced XLR connector tubes, and ordered some from my favorite past vendor in Colorado... call it nostalgia... :W

This gave me an excuse to setup and configure the new version of Altium Designer 16 and get the footprint libraries setup with it- not a big job, except that fitting it on the limited board space with conservative design rules for isolation wasn't as quick as I might have imagined.

Although it's technically laid out to use 0.1" headers for inputs and outputs, I doubt there is enough room inside the XLR tube to accommodate the vertical space they require, so handwriting the inputs will likely be required. It's just a two layer 1 oz copper board, so fairly inexpensive if you order a number of pieces. Any one interested in bare PCBs fairly cheap or with parts, send me a PM. Will take a bit over a week before I have them, considering lead time and ground shipping.

**JonMarsh** · 17 July 2016, 23:03 Sunday

Had some time late afternoon and evening to do some casual listening while working on things on the iPad Pro- just trying out some different things using Roon, (which I am really enjoying... Including what it does in selecting stuff from my own collection in a particular genre after an album I've already selected is over...)

Something I particularly enjoyed how it sounded was some soundtrack music I have... Which makes pretty logical, because it's often listened to as background music- but it made me actually stop and listen to it, and sucked me in more- particularly two favorites of this genre, one old, one new- "Firefly" sound track, and the Witcher 3 sound track (from the CD that comes with the delux edition of the game).

**jim1961** · 18 July 2016, 11:15 Monday

Ive played with a lot of FR curves and the attachment is where I have ended up (measured at the listening position). Looks a lot like what you guys are talking about.

I do like a bit of a bump in the high end and elevated bass though.

**JonMarsh** · 18 July 2016, 11:27 Monday

I do like a bit of a bump in the high end and elevated bass though.

Ah yes, what we used to call the old "boom-tinkle" phenomena! (when I was in the business) :W

you're following something fairly similar in the 200Hz to about 10kHz area, then deviating with about 5 dB lift in the bottom end (loudness compensation? Do you usually listen at 75 dB or less?)

My thinking at this time is that it's likely better to design the speaker for nominal flat response, and modify the curve with an adapter like this- because it's more easily changed, or can even be incorporated in the front end of an amplifier... :B

It's funny, depending on the source gear, that lift in the top may be fairly pleasant, or if there's any latent digititus in the system, somewhat annoying. the ScanSpeak 66020 tweeter (which I initially used in the early version of the Ardent) has that kind of profile sort of engineered in because of the short small waveguide in the relatively thick faceplate; I ended up deciding I really didn't like that, and much preferred the 6640, which doesn't have that lift.

One of the most common comments I get with regards to the Wavecor Ardent is that it doesn't sound like it has a metal tweeter... obviously, this coming from people that don't like the usual metal tweeter artifacts.

With this setup, that is even more the case.

**jim1961** · 18 July 2016, 11:43 Monday

Originally posted by JonMarsh

you're following something fairly similar in the 200Hz to about 10kHz area, then deviating with about 5 dB lift in the bottom end (loudness compensation? Do you usually listen at 75 dB or less?)

No, I usually listen in the low 90's SPLdb range. I just dont like thin sounding bass :lol:

My liking the high end lift maybe has as much to do with aging ears as anything.

**jim1961** · 18 July 2016, 11:46 Monday

Originally posted by JonMarsh

Ah yes, what we used to call the old "boom-tinkle" phenomena! (when I was in the business) :W

Many might call it a smiley face EQ :lol:

But in my case, id call it more a slight grin

**JonMarsh** · 18 July 2016, 15:55 Monday

the key is whatever works for you- the money and work you've spent should be pleasing to your ears!

**Renron** · 19 July 2016, 12:47 Tuesday

Originally posted by JonMarsh

Ah yes, what we used to call the old "boom-tinkle" phenomena!

With this setup, that is even more the case.

Doc, Awe man..., yer killin' me! :rofl:

Being honest here, I understand about 1/3 of the posts in this thread, but it's interesting none the less. I fear I'm learning something too.
Ron

**JonMarsh** · 15 August 2016, 10:44 Monday

Still haven't had time to build up some boards, and photograph them- it's on my to-do list, but work has been pretty crazy with some stupid stuff going on in more than one area...

Have gotten in more resistors, and will be building up both 3.2 dB boards and 1.8 dB boards.

I think this will be interesting to compare.

**JonMarsh** · 18 February 2017, 21:47 Saturday

Ok- Ergo is coming to town next weekend, and I also owe my friend a set of voicing modules, so they finally rose to the top of the priority list, along with making space in the man cave office/workshop to work on them (which took half of Saturday...)

They don't look too bad...

only problem is, they fit in the original Asian made XLR tubes, but not quite in the SwitchCraft. Good thing I'm sort of a wood worker and know how to sand things... I'll never live this one down!

**************************

Already sanded, and test fitted, but no pics- just take my word on this one, guys. I was also a little worried if the caps would be too tall, but it seems things are OK...

**************************

Company is all here, getting ready for Bday dinner for GF's son, with nieces in attendance from Seattle and Providence Rhode Island. Back to the voicing module build tomorrow morning...

**JonMarsh** · 19 February 2017, 14:28 Sunday

Both width and length appear to be an issue... good ole Milwaukee orbital sander to the rescue.

Now, the interesting question, is if this works, do I sand the rest of the boards I build like this, or go back and spend a few hundred dollars for a new PCB design set? I can shrink the resistor footprints, but OTOH, board fabrication with foil that close to the edge is not really cool. This size of a board can tend to be a problem- note that the foil angle brackets on the corner are supposed to be only on the OUTSIDE of the board, and gone after cutting to the right size. OTOH, I can shrink the resistor layouts to 0.4" from 0.5", and push some other things closer together.

I think I'll have to do a new layout, but the time for that and the expense for new boards will have to wait. For now, hand crafted will be the reality...

ops:

BTW, side comment, theses are XICON resistors specifically designed for Audio, and high grade WIMA polypropylene film caps, with very low dissipation factor. (means low ESR and low foil contact resistance)

**Steve Manning** · 19 February 2017, 14:35 Sunday

Originally posted by JonMarsh

Both width and length appear to be an issue... good ole Milwaukee orbital sander to the rescue.

Now, the interesting question, is if this works, do I sand the rest of the boards I build like this, or go back and spend a few hundred dollars for a new PCB design set? I can shrink the resistor footprints, but OTOH, board fabrication with foil that close to the edge is not really cool. This size of a board can tend to be a problem- note that the foil angle brackets on the corner are supposed to be only on the OUTSIDE of the board, and gone after cutting to the right size. OTOH, I can shrink the resistor layouts to 0.4" from 0.5", and push some other things closer together.

I think I'll have to do a new layout, but the time for that and the expense for new boards will have to wait. For now, hand crafted will be the reality...

ops:

BTW, side comment, theses are XICON resistors specifically designed for Audio, and high grade WIMA polypropylene film caps, with very low dissipation factor. (means low ESR and low foil contact resistance)

You knew from the get go you would go the "right" way and redo the boards.

Hand rework should certainly be good for initial evaluations though. :T

**Renron** · 19 February 2017, 14:53 Sunday

No way in hell you're going to fit Mundorf M-Resist Supremes in there. I don't think you would allow yourself to "get by" with sanding. Bob's right. Few hundred bucks sounds high thou. Must be somewhere that's cheaper.
Ron

**JonMarsh** · 14 July 2019, 19:41 Sunday

I still haven't figured out what to do for sure, but I think I'm just going to make another batch of boards... and build this into all my new designs!

**Dave Bullet** · 15 July 2019, 16:47 Monday

@Jon - thanks for the detailed reply in your new adventure thread. I had missed this post entirely. I will now read (and re-read). the dip I engineered on-axis to suit ELC listening measures flat on-axis at the listening position. I realise I am taking the room in here, but that's maybe why it sounds great up to the tweeter / treble. I feel I may need to take another 1dB-2dB off the tweeter to reach the right balance in my room for most of the "trash music" I like

without sacrificing the great masters / mixes. Right - sidetrack off.

**Dave Bullet** · 16 July 2019, 16:00 Tuesday

I'm using a Raspberry Pi with an IQAudio DAC mounted on top via PCIO pins (HAT) running a custom Archlinux distro called Archphile. It's a stripped down distribution with only MPD as a music server and WiFi networking for remote control.

I then installed the ALSA equaliser. This is s 15 (from memory) fixed band EQ and can change realtime.

I then wrote a script to monitor song changes and another to store and load a database of EQ settings for albums.

I pre-programmed the classic BBC dip and a few others to try.

The beauty with the above is a dynamic EQ that "remembers" a setting and re-loads next time that album is played. I've done album level (not song level) as usually a mix seems to be consistent across tracks (except for some compilations - but they are dubious mastering quality anyway - refer http://dr.loudness-war.info/ and check out your screechy albums

Anyway - I've found this a brilliant solution. The EQ is usually running (I can turn it off for bit-perfect playback if I really must - but I can't tell the difference TBH), and the EQ switches and remembers. I use "SSH button" on my phone to store the EQ setting for an album as I'm listening. I can switch between them to choose the right amount of attenuation if I like. If there is no stored EQ, it default to "flat".

All in all, this is a software based solution to achieve similar results. The only thing is I would need to loopback to see the actual curve being applied due to the fixed attenuation points. I'm pretty sure it won't be as "analogue / smooth" as Jon's solution above.

I agree with all the "it encourages you to turn it up". You finally get to use your mega-watt amplification and get those woofers moving

Report back on stereo speaker voicing investigation

Report back on stereo speaker voicing investigation

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment