audio-talk

but to be perfectly honest, as Fourier transforms would indicate, the absolute phase is - to me - inaudible although possibly obstructed by tinnitus.

Yes, same for me, though I can imagine given a system with a combination of even order non linearity could make it audible.

questioning the conventional tunings, about which perhaps I might post in due course.

That would be interesting. I quote the wiki entry on Fifths

"In music theory, a perfect fifth is the musical interval corresponding to a pair of pitches with a frequency ratio of 3:2, or very nearly so. "

Emphasis mine.

On tuning of pianos I'm advocating systems which happen to be historic that use numbers of exact pure 3:2 fifths. This puts many notes of the scale exactly on the harmonics of lower strings, with numbers of near pure thirds or sweeter thirds than we're conventionally used to, and enables the instrument to give a stillness to the sound, and the music thereby, in contrast to the conventional tuning, as well as tonal colour differences between keys, which is what classical composers wrote for . . . as well as actually improving the sound of the instrument, making it less metallic, sweeter and more harmonious.

explains a little more


Best wishes

David P

I think that it is problematic to compare musical instrument tuning to phase anomalies in audio systems. An out of tune string has the fundamental and all harmonics intact. When we tune one string to another, everything moves more or less linearly. In audio where we have varying types of phase shift, we NEVER have this uniformity. A fundamental could be reproduced by the woofer, the first harmonics by the midrange and the higher harmonics by the tweeter.

I have played with phase in analog (with 6dB crossovers, you just change the high or low pass frequency) and digital crossovers (MiniDSP, Behringer) as well as physically aligning. I can say that all of the corrective approaches sound differently. The problem is obviously not only absolute phase. Knowing that phase is audible, the question is what can we do about it? Physically time align our speakers with a tool like Holm impulse? Create 5/6/7 way speakers with drivers only working in their sweet spot - allowing for minimal slope? Investigate point source schemes? Do we need to kill room reflections?

At the end of the day, how do we verify that what we did has “musical value”? Do we even have words to communicate these differences. Are we objective enough to isolate “Pride of Ownership” from actual performance?

I guess to get started, one would need some kind of reference. What could that be? Headphones? Which headphones are free from ear cavity caused phase shift? Do we need someone with Quad ESL63s? Do we need an approach like Romy the Cat with a spectrum analyser and phase tester? Do we need something like DIRAC? Do we just do a little bit and be content knowing that we did something - until someone else shoots holes into what we thought was the answer?

There are complications - especially for instrument tuners. We can play a trumpet tone 8MY INSTRUMENT) in perfect tune with a tuning device, but it sounds off pitch depending on how the player nuances it. I have noticed speakers that do a similar thing in certain ranges. The pitch measures correctly, but sounds too sharp or flat or even worse, the sense of pitch is not “strong”. Bass lines are not “tuneful” rather wooly.

I think that Holm Impulse is a great tool to help get a system at least physically aligned. It can at least help us visualize things that we at first do not understand. I consider this to be a start. We can compare phase vs amplitude. We can “see” problem areas deserving attention. It will not turn the deaf into golden eared audiophiles, but the results are documented in a way that we can share them. Arta is also a great tool.

I did some playing around with FIR and tweaking the phase response of a set of speakers yesterday. REW, RePhase, some Powersoft amps and a pair of Faital 10HX230 in ported boxes.

At one point, I had a flat phase (and frequency) response from 100Hz-20kHz, but that required a lot of FIR taps and a long delay time - certainly not useful for live music.
For playback, though, it was the most "correct" sound I think I've ever heard. There's a passive crossover in the speakers, so I made a custom curve that would undo the phase shifts from the crossover, and linearise the overall response.

I'd like to demo this at Owston, probably by EQing the speakers flat and then using the FIR processing to just fix the phase response. That way, it can be switched on & off easily.

Chris

Good work, Chris. That sounds very interesting.

It's just occurred to me that the old "holy grail" of reproducing a square wave using loudspeakers ought to be possible here. Since it's coaxial, it ought to work off-axis, too.

I'll have a play around tomorrow and see what I can get.

Chris

Last December Scott sent me this email with a attached filter modification for my Edingdale GT speakers:

Happy Christmas Colin!

Little something for you attached, just for fun. Optional addition for the Edingdale GT. No changes to the main filter, all exactly the same.

Basically, an impedance & phase correcting circuit strapped across the input terminals before the crossover. The GT is a fairly easy load as-designed. With this -see the attached before & after. Apart from the 1st impedance peak (largely irrelevant), the impedance and electrical phase are to all intents & purposes flat-line, unreactive 'ideal' resistive loads. Voltage source, SET, current source -it doesn't care. System distortion reduced, dynamic contrast enhanced. Two of the caps are large values, but are bipolar electrolytics so fairly cheap; one inductor is a big value, but an iron-core in a small winding gauge so not particularly expensive. Resistors we know about.

Partly inspired by Joe Rasmusen's work but where he's going into other slightly more esoteric fields, here I'm just focussed on a full-on strike at the load & electrical phase to enhance the main dynamic range / low distortion & clean FR objects we've always had with the Edingdale. I've drawn it separately, and as part of the full GT filter for grins. In the latter it looks at first glance like it's in the low pass, but the signal in fact passes this load linearising filter before it actually reaches either the low or high pass. Just a convenient way of drawing it.


I had been following this thread at the time but must admit to not trying to build the add-on filter with other things occupying me.

With the pandemic situation of what seems like continual lockdown and restrictions on movement since March it wasn't until recently that I cobbled together a pair the filters with components I had to hand and built them into a couple wooden boxes in similar fashion I did to the external main filters for these speakers. I did however later install the main XO'ers back inside the speakers so I had the original boxes that I modified for these impedance and phase correction filters.

Having completed the build and spoken to Scott regards what to expect he did advise that the filter would most likely be more suited my 300b SET amp rather than a SS amp. On listing this has proved the case.

At first when I added the filter fed from my 300b's after a initial 15 min listen I wasn't that convinced the change was for the better. There are some big value caps and inductors in the new filter so I decided to leave a CD on continuous play and see if things improved with running-in of the components. A couple hours later and wow what a transformation in sound quality and presence of the music. Big improvements in soundstage and listenability. The sound simply draws one into the music that I found really engaging.

I reckon Nick was and is on to something with the title of this thread. A big thumbs up from me. A pity no one can come round for a A-B comparison of with/without the Impedance/phase compensating filter. It really does work and is something to be given serious consideration for anyone with a multiway speaker run by valve amp, especially if it is a SET.

Scott also sent a before and after simulation plot:
As is plot

and with filter


I can post up the filter if anyone is interested?

Interested to see the filter, Colin, but that appears to be about linearising the impedance and phase curve seen by the amplifier.

FWIW, I've done similar electrical tricks to get some speakers to behave on the end of long & thin speaker cables. 12ohm nominal impedance, which obviously varied all over the place. With the potential divider between that and a few ohms of wire, the frequency response was all over the place.
Strapped a 10ohm resistor across the speaker's terminals (ie, presenting an approx. 5.5ohm linear impedance), and things improved considerably. Obviously, that's a really wasteful way to do it (over half the supplied current goes through the resistor), but the amp's happy driving that impedance, and the resistor doesn't get particularly warm.

I expect Scott's filter has much more finesse than the above method, but the benefits are there when there's some series R in play (whether from the amp's inherent output impedance, or long cables), or if you're using an amp which prefers easy loads.

Chris

If you can post up the filter Colin, that would be great. I'm wondering if it's anything like the one Lynn used on the Ariels. However, given that the end result appears to be much better, I expect it's not the same ! Interesting stuff, anyway.

This is the filter that Scott designed. Must express that this was designed specific for my Edingdale GT speakers. Component values will have to be changed for any other speaker. Having listened more to this setup I am hearing better defined bass and low frequency sounds with a more 3D holographic soundstage. I wouldn't have believed we could improve on these speakers as they were, but this filter, as said, has added another degree of involved listening pleasure for me.

And here are a couple photos of the finished filter

I can certainly vouch for the results..I borrowed that approach from Andrea Cuiffoli when I did the Beyma AMT open baffles. I gave the crossovers to Steve Shiels but I'm not sure if he ever used them or just dismantled them.

Ah yes, the Zobels I did as a Christmas gift for Colin. Not used all that often in this way -certainly not unknown, Lynn had one as an option in his Ariels for flattening the impedance around the XO frequency, and there are a few others lurking around. I quite like them, they have their uses, especially if the speaker is a difficult load, straightening things out toward something more linear. The GTs are actually relatively benign, but every little helps and anything you can do to avoid reactive behaviour either with peaks or large capacative phase angles just makes life easier for the amplifier, which in turn should help clean things up.