I noticed something missing from the straight-through amps I have built then realized my RCA Pro Series receiver (2004 vintage, I still have it) has DSP and spatial expansion functions that greatly enhance the audio ambience. There are other things to decode from CD's and maybe MP3's also, so unless I use my DIY amps as repair blocks for a system with that stuff built in, I will have plain "boring" audio straight from a CD player.
I found only a few enhancement IC's such as the hard to get TDA3810, JRC NJM2701, and QX2020 plus some more advanced chips that look to require some extensive DSP knowledge to work with.
The TDA3810 by the spec sheet causes as much as .1% distortion, NJM2701 either 3% or .3% I couldn't read it well, and nothing found on the QX2020.
Has anyone found and analog spatial expander IC's that are easy to use and cause little distortion ?
I found only a few enhancement IC's such as the hard to get TDA3810, JRC NJM2701, and QX2020 plus some more advanced chips that look to require some extensive DSP knowledge to work with.
The TDA3810 by the spec sheet causes as much as .1% distortion, NJM2701 either 3% or .3% I couldn't read it well, and nothing found on the QX2020.
Has anyone found and analog spatial expander IC's that are easy to use and cause little distortion ?
Last edited:
Unexpanded stereo sounds too narrow to me now that I am accustomed to spatial expanded audio. It's like 180° dispersion speakers. 3D processors do fairly well at the illusion of rear speakers.
I haven't found much of any discussion on stereo enhancement IC's here or anywhere.
I think they are essential.
I haven't found much of any discussion on stereo enhancement IC's here or anywhere.
I think they are essential.
There is a way to wire four speakers up for pseudo-enhancement. I will try that in a few days and have a hear.
I think the rear speakers are still the same source as the front but wired in reverse phase to imitate a delayed reflection.
It's been many years since that was the only trick available. Then came the bucket brigade IC.
I think the rear speakers are still the same source as the front but wired in reverse phase to imitate a delayed reflection.
It's been many years since that was the only trick available. Then came the bucket brigade IC.
Last edited:
The spatial expansion idea took two basic forms, starting in the early 1980s.
One was to cross-feed a bit of one channel into the opposite channel, delayed a tiny bit using a stack of all-pass networks corresponding approximately to ITD, and filtered to approximate HRTF. It kind of worked, but has a very small listening window, and you had to eliminate as many early acoustic reflections as possible. This was the Carver Sonic Holography idea which, when it worked, was pretty amazing, with very palpable 3D imaging. However, the effect was fragile, and often didn't work at all if the listener wasn't perfectly between speakers, or the room was asymmetrical acoustically.
The other general type was to take a difference between the channels, L-R, filter and delay ( or just filter) then mix it back into the opposite channel so that you got 2L-R or 2R-L, with of course the delay if used and filter. The filter was basically rolling off below about 250Hz, and above 5kHz or so. An L-R level control was usually included, and you'd vary the amount of effect as needed. If a delay was used, (it was a BBD back in the 80s), it would be about 100-600us, sometimes variable, again to approximate ITD. The results were generally more forgiving of early acoustic reflections, had a somewhat larger listening window, but also tended to "recover" a lot of ambience. One problem was that some recordings had a lot of random phase information that popped out strongly in L-R, and would tend to bury center-panned vocals, hence the level control. The amount of cross-feed also varied with pan position, with center-panned, in phase signals getting no cross-feed (L-R would be zero), which differs from the first technique which provided a fixed amount of cross-feed for even mono signals.
Modern DSP realizations could have been either technique, but I'd expect a variant on the first type, general HRTF cross-feed with ITD.
One was to cross-feed a bit of one channel into the opposite channel, delayed a tiny bit using a stack of all-pass networks corresponding approximately to ITD, and filtered to approximate HRTF. It kind of worked, but has a very small listening window, and you had to eliminate as many early acoustic reflections as possible. This was the Carver Sonic Holography idea which, when it worked, was pretty amazing, with very palpable 3D imaging. However, the effect was fragile, and often didn't work at all if the listener wasn't perfectly between speakers, or the room was asymmetrical acoustically.
The other general type was to take a difference between the channels, L-R, filter and delay ( or just filter) then mix it back into the opposite channel so that you got 2L-R or 2R-L, with of course the delay if used and filter. The filter was basically rolling off below about 250Hz, and above 5kHz or so. An L-R level control was usually included, and you'd vary the amount of effect as needed. If a delay was used, (it was a BBD back in the 80s), it would be about 100-600us, sometimes variable, again to approximate ITD. The results were generally more forgiving of early acoustic reflections, had a somewhat larger listening window, but also tended to "recover" a lot of ambience. One problem was that some recordings had a lot of random phase information that popped out strongly in L-R, and would tend to bury center-panned vocals, hence the level control. The amount of cross-feed also varied with pan position, with center-panned, in phase signals getting no cross-feed (L-R would be zero), which differs from the first technique which provided a fixed amount of cross-feed for even mono signals.
Modern DSP realizations could have been either technique, but I'd expect a variant on the first type, general HRTF cross-feed with ITD.
I just decided to try experimenting with a similarity to the Bose direct/reflecting speakers. I have gotten some nice dispersion effects by turning a harsh tweeter backwards and reflecting off the wall. It may look bad on the graphs but sounded good IF positioning is correct. It starts getting sort of like the coffee can sound if the reflection speaker is too close to the wall.
The rear ported bass cabinets, to me they mainly sound awful, in contrast to my experiments with tweeters the rear ported bass usually sounds best when very close to the wall.
The rear ported bass cabinets, to me they mainly sound awful, in contrast to my experiments with tweeters the rear ported bass usually sounds best when very close to the wall.
The direct/indirect speaker idea, similar to a bipole speaker in some respects, will create a sense of space and sort of expand the presentation, but it's adding random reflections to do that, which will confuse and blur the imaging characteristics. Sources won't be as clearly defined in location. Things like center-panned vocals will become unnaturally huge, but not located well. It actually is almost the inverse of what image enhancers/expanders do.
IIR, Polk made speakers with a built-in acoustic crosstalk cancellation function. He wired both channels to both speakers, and used a mid driver driven from the opposite channel, at a lower level and inverted to cancel acoustic crosstalk from the opposite speaker. The result was similar to the Carver method, but a bit more tolerant of acoustics. Again, impressive when it worked. Not sure if Polk still does that, they might.
Philips used the same principle in their crappy boomboxes for a while. And it sounded so bad a friend of mine actually called it "spastical" stereo. And that term covered the effect nicely: at times it sounded decent, at other times it sounded really bad. Bass for instance was terrible: initially you got this huge "punch" of bass, but it couldn't be sustained so it died out very quickly, leaving a very odd (hence the spastical) sound.
People here tend towards "Hi Fi" which means no effects boxes in the chains.
Yes. As I recall, it was named the Stereo-Dimensional-Array. The cross-channel delay cleverly being accomplished purely acoustically, simply by physically offsetting the cancellation drivers toward the outside edge of the baffle board. As you indicated, both Polk's acoustic solution, and the electronic solutions are based on the principle of inter-aural-cross-talk-cancellation.
If I'm not mistaken, solutions based on the stereo difference signal was patented decades ago by Joel Cohen of Sound Concepts. Utilizing the difference signal (L-R) and its inverse, (R-L) for the cancellation signal reduces the comb filtering effect which otherwise occurs with solutions simply utilizing the L and R . Which I believe the Carver Holography circuit does. The Cohen solution originally utilized a bucket-brigade style switched capacitor devices to achieve the required delay. I don't recall what Carver utilized. At any rate, as you stated, all-pass filer delay networks can be utilized. These produce the least noise and distortion.
While the Cohen and Carver approaches each has its own particular attributes and drawbacks, both share a common drawback, which is also shared by the Polk solution. Which is that while all three perform a first suppression of the cross-talk generated by the original L and R stereo signals, there is a second cross-talk effect stemming from the cancellation signals themselves. In effect, the cross-talk cancellation signal itself produces a second delayed cross-talk. Trying to cancel that second cross-talk then generates a third, and so on. I liken this to a fun-house hall-of-mirrors effect, where the same image is reflected, ping-pong style, into infinity.
On a separate but related note, a very natural surround field can by generated from a 2-channel stereo signal via an purely acoustically means, by an old, but seemingly forgotten, technique known as the (David) Hafler matrix. I suggest that anyone interested in the general topic 3D enhancement give it a try. You will need a pair of spare speakers. These can by inexpensive bookshelf types, as the signal they will be producing typically lacks deep bass. Perhaps, an unusued pair is on hand in the basement. Simply wire the two speakers in series with each other, and then across the positive output terminals of your amplifier. This works with the vast majority of amps, an exception being amps with floating outputs.
So, wire amp +Left to surround speaker A +terminal. Then, surround speaker A -terminal to surround speaker B -terminal. Finally, surround speaker B +terminal to amp +Right. Leave your main speakers wired normally. The surround speakers will play only the difference information contained on a stereo program, WITHOUT any active surround processor. The concert hall's ambient field information is largely contained in the stereo difference (L-R) information. After experimenting a bit with aiming the surround speakers, I felt the best effect was obtained with the speakers located adjacent to each main speaker, but angled to fire toward the listening end of the room, down along the nearest side wall.
They actually made their first significant appearance in 1979, and into the early to mid 1980s.
Oddly, London/Decca Phase 4 recordings were over by 1980 when stereo expanders (image enhancers) were just kicking in. Because of their hyped stereo mix they did expand well, even with passive matrix pseudo-quad systems like the Halfer circuit.
Remember, this was the era after Quad was pretty much pronounced dead, but anything resembling 5.1 (or 4.0) in the home hadn't begun. The interest was a sort of immersive audio effect without adding speakers.
Endless Groove - London Phase 4 Stereo
The Decca Phase 4 Stereo Concert Series | The Absolute Sound
Well, yes, sort of. The Sound Concepts device (I tested one extensively) had several controls and an extender cable that let you adjust the controls with the unit in your hand in the LP. It also had comb filtering issues, but progressively less as images were panned center, in phase. However, it had a less specific, more ambiguous localization effect, while providing a much more pronounced (if you whipped the controls up) expansion effect. Carver had no adjustments other than a switch or two, and depended on careful speaker placement to minimize early reflections. Its effect was more subtle, but the localization, when it worked, was surprising.
Correct. The BBD device was terrible fidelity unless you clocked it quite fast, which is why Cohen's application worked as well as it did. He didn't need much delay. Carver used all-pass networks, a lot of them, and filtering to shape the cancellation response. I probably still have FR plots of both in a file somewhere.
Correct, but the secondary crosstalk was lower in level, and re-filtered by HRTF, so it's not really much of a factor in practice. Early room reflections were a much bigger issue because they make crosstalk cancellation nearly impossible.
One issue the L-R technique suffered badly from was that vertical groove modulation contains somewhat higher levels of distortion. That distortion is amplified by employing the difference signal, so this technique did not sound as clean with some records, and some were really nasty. CDs completely mitigated the issue. The Carver topology did not have this problem at all.
The Hafler matrix is not forgotten at all, just made moot by today's surround processors. The problem with the simple difference signal is that while it tends to recover random phase or out of phase ambience, the actual channel separation between adjacent channels is quite poor. Worse, unless you provide some means of level control, there's a frustrating range of surround levels from different material. So, if you're going to perhaps drive the difference speakers from a separate stereo amp, you might also consider about 20ms of time delay too, which enhances the effect of separated ambience by introducing the precedence (Haas) effect. There will definitely need to be level control, but the additional forced separation is beneficial.
One side note: Joel Cohen/Sound Concepts ended up in a patent battle with Carver. I don't recall the precise outcome, but I think Carver may have lost (he filed the suit). To complicate matters, there were several other products that employed the L-R method, including one that I manufactured on a custom basis. The others were simple boxes with just a bypass switch. One was called the Honeybox, which was changed to HBX at some point. Think Cohen without time delay. My version payed more attention to the difference signal processing, and I experimented with various filter types, delays, and even dynamic processing. Mine found limited use in broadcast facilities, then just a short time later, I morphed the entire thing into the first home Dolby Stereo compatible 4.0 surround decoder. Interesting how similar the processes were.
Mitigating those concerns about the Hafler technique a bit are that wiring the two surround speakers in series attenuates the output of each by 6dB. Still a fixed output level, but it is a noticable amount of attenuation, which, as you indicate, usually sounds better than no attenuation. There number of variables here, though, including the relative sensitivity between the surrounds and the mains.
Helping on the issue of low channel seperation is that thre surround speaker output is somewhat decorrelated by wiring them so that they are in inverse phase with respect each other. This helps increase the perception of sound field size, countering the objective channel separation problem. Which is a problem with most in-home stereo reproduction in general. My recollection, which may be in error, is that there is acostically maybe about 10dB of channel separation at the listeners ears.
The final mitigating factor is that the ambient field of concert halls and rooms inherently features a natural delay, with respect to the direct sound it correlates with. The surround speaker output will inherently reproduce that captured hall delay, obviating the need for 20ms. electronic delay, which is long enough to require a digital delay line to achieve.
Of course, the exceedingly simple Hafler matrix is unlikely to find itself randomly optimum for a given room or program content, but is simple to try, natural sounding, and fun. Especially attractive for those who don't currently have an surround system, or who have an all analog system and are wanting to keep it that way. ☺
This circuit-
Stereo Width Controllers
Combined with the Surround Sound decoder circuit (line level hafler matrix)-
Simple Surround Sound Decoder
And the PT2399 delay circuit-
Digital Delay
Will make a very versatile and enjoyable preamp with a good range of features for those who like such controls. I had just such a rig going for a while, mainly for gaming and movies, but with the right recordings it sounds almost ethereal how well it sounds. Highly recommend these circuits for someone that likes the effect.
I eventually moved on to mostly tube gear, and can speak highly of the following circuit from john broskie for the l-r/l+r side of things-
The Tube CAD Journal,Design Idea: (L+R) & (L-R) in One Stage
A stereo width control is easily implemented with a simple differential amplifier with tubes as well-
The Tube CAD Journal: Out of Control: The missing sonic controls
Hi fi? Eh, maybe not, but for home theatre, or relaxing and sipping your favorite beverage such terms have little meaning, so long as it works the way you like it to. No harm in wanting what you like, or what you have grown accustomed to.
Also, the use of speakers with wide dispersion (such as the karlsonator) can really help give a very wide, natural soundstage. I use a pair and can't stop recommending them, they are fantastic for imaging, and giving a natural and non-artificial soundstage. I was blown away by how damn good they sound.
Stereo Width Controllers
Combined with the Surround Sound decoder circuit (line level hafler matrix)-
Simple Surround Sound Decoder
And the PT2399 delay circuit-
Digital Delay
Will make a very versatile and enjoyable preamp with a good range of features for those who like such controls. I had just such a rig going for a while, mainly for gaming and movies, but with the right recordings it sounds almost ethereal how well it sounds. Highly recommend these circuits for someone that likes the effect.
I eventually moved on to mostly tube gear, and can speak highly of the following circuit from john broskie for the l-r/l+r side of things-
The Tube CAD Journal,Design Idea: (L+R) & (L-R) in One Stage
A stereo width control is easily implemented with a simple differential amplifier with tubes as well-
The Tube CAD Journal: Out of Control: The missing sonic controls
Hi fi? Eh, maybe not, but for home theatre, or relaxing and sipping your favorite beverage such terms have little meaning, so long as it works the way you like it to. No harm in wanting what you like, or what you have grown accustomed to.
Also, the use of speakers with wide dispersion (such as the karlsonator) can really help give a very wide, natural soundstage. I use a pair and can't stop recommending them, they are fantastic for imaging, and giving a natural and non-artificial soundstage. I was blown away by how damn good they sound.
Thanks, for those interesting links.
I'd not before heard of the Karlsonator. I did a brief search for info. The uniform and wide polar pattern of such speakers seems particularly intriguing. What are some the physical details of the Karlsonators you use, and also please share your subjective assessment.
Last edited:
My karlsonator set use a pair each of the Faital pro 3FE25-16, 97db/w efficient, ~120° dispersion pattern. plans on the following thread, they can be scaled to work with many different drivers-
Mini Karlsonator (0.53X) with Dual TC9FD's
Here are my first set using a single vifa TC9FD per side, prototypes made of foamcore board, wood to come soon.
Honestly they are the most natural and transparent speakers I've ever heard, and the wide dispersion makes them perfect for music and home theatre. No other speakers I've heard give you the same "you're there" feeling.
Mini Karlsonator (0.53X) with Dual TC9FD's
Here are my first set using a single vifa TC9FD per side, prototypes made of foamcore board, wood to come soon.
Honestly they are the most natural and transparent speakers I've ever heard, and the wide dispersion makes them perfect for music and home theatre. No other speakers I've heard give you the same "you're there" feeling.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.