Amp-Crossover-Speaker connection quiz...

[Bemused]

I was never good with the maths around electrical stuff. It explains why I struggle to digest beyond the first chapter of the Art of Electronics. But I have some intuitive sense.

If the pipes that carry the audio to the drivers have a thin return pipe that is shared, then the relatively higher pressure or volume of electrons could arguably go up the 'wrong' way on the return pipes of the smaller drivers or affect their 'flow' or pressure.

But if there is a big FAT return pipe that is bigger than the sum of all the pipes feeding the drivers from the filters, then logic would suggest they would drain freely and not go the 'wrong' way or affect the upstream pressure in any meaningful way. A simple way to test this hypothesis is to significantly decrease the DCR of the 0V cable that takes the ground from a common (star) point on the crossover PCB back to the amp to see if that yields the same improvement, as others have posted about in moving the crossover closer to the amp.

The other possible effect at play is modification of the transfer function with the added DCR downstream of the crossover, which affects the LCR seen by the respective filter sections.

Conversely adding say 0R22 to the amp return should make a negative impact.

 

[PigletsDad]

All I am saying is that any effects are linear - there is no intermodulation.

I would need to think through if star grounding matters in a crossover - it isn't something I have analysed. If it matters at all, it would be due to an effect on the frequency response(s), which should be very small - we are talking about tiny fractions of an Ohm versus 8(ish) Ohms.

Star grounding in amplifiers matters because there are currents (e.g. power supply returns) which are not simple linear replicas of the signals, and it matters greatly that these do not get into the small input signals.

 

[martin clark]

I had a think about this on the train home, and without close analysis - I think it's less important, not only for the reasons PD teases apart, but also because it is only one part of the impedances appearing in series, in driving the crossover.

Qualitatively:
Suppose you use a typical amp - I'll use Naim amp as remains a popular theme here: output Z is a discrete 0R22, plus an effective microhenry or so *; but choices are free right up to wilful amps with several ohms - SE valves, for example.

Suppose you have maybe 3m of chunky cable to the crossover, perhaps 2.5sq.mm cores: 0.02ohm one way, 0.04ohm round-trip - and again, up to 1uH per metre (ballpark intrinsic inductance of shotgun pair, even closely-paired)

You can see by inspection that star wiring - separate 0v return to amp for bass and treble from crossover back to amp - will only halve a figure that is already only a tenth of the complex output impedance of the (typical /good amplifier). As a minimum it is a drive-voltage effect 20dB below the effective amplifier's contribution. For 'better' amps the effect is not going to be much different; and for 'wilful' amps it will make no damn difference. And then this voltage division is finessed by the superposition effect PD describes - a big bass transient error voltage 3 octaves below a tweeter on a -12dB/oct high pass is knocked down by a further 36dB - not that a tweeter will have useful output in the bottom handful of octaves... or that effects on tweeter output near-60db down will be audible (below own %thd floor, I think, for most)

One run of 4sq.mm puts it well beyond contention, I think. Much bigger gains to be had elsewhere if you like to DIY.


*Naim's classic power amp design is a feedback-defined amp so output impedance increases with frequency even without a discrete output inductor; other amps have lower DC figures, but and addition 2-3uH in series by design instead; a modern 'Blameless' style is somewhere round 0.05R+3uH for example; swings and roundabouts.

 

[James]

Thanks, Martin. You have confirmed for me why I cannot fathom why shifting the crossover network closer to the amplifier should yield the benefits observed by others. There may be other factors at play, including psychoacoustics, but I guess until I've tried it myself, I should not be too dismissive.

 

[martin clark]

I don't disagree James,and as I said - my suggestion is a bit back-of-envelope/handwavy. Probably repays a more detailed review (montecarlo sim) but a worthwhile gain doesn't immediately jump out at me.



[I am not to be listened to on this anyway - I use Deltec amps with their four-wire/kelvin-sensing arrangement off the speaker terminals, so all the cable is inside the feedback loop - as if it matters ]

 

[mega lord]

I don't understand all this theory but i do understand improved SQ and when the music moves me more the hairs on the back of my neck tell me so.

For me the difference was very noticeable, not just a slight improvement. If i am imagining it then its time to end it all

 

[Chris]

I think the mistake was to place the OP in the DIY section which is frequented by many who need to understand and approve something before trying it for themselves. For those of us who already had external crossovers and are no-nos as regards theory it was a piece of cake to accept and try out LesW`s first appraisal that the earlier in the chain different frequency groups are separated the clearer and fundamentally better it sounds. The contradiction being that the only people with confidence in their ability to extract the crossover from their beloved speakers without damaging anything are precisely those that normally visit AND UNDERSTAND the DIY contributions. So it´s Catch 22. Others may think this is only for the Ergo speakers or certain amps but I swear it worked for a bog standard A21a/Wd20T combo.
I say suck it and see if you can be bothered. After all, it just appears like another reason for biwiring and none of us believe in that, do we ?

 

[andrewsutton]

I have tried this and all though I have about 15 inches of cable from amp to crossover I think this has made a surprising difference.
Regards Andy.

 

[Bemused]

If it wasn't for the fast approach of summer I'd have a good play with a Spice job on this, I find LTSpice helps me to overcome my lack of underpinning electronics knowledge and certainly gives me the picture that the clever guys just happen on.

One for the darkening nights of autumn then?

 

[mega lord]

Tony, i wish you would or somebody would. Everyone who has tried it on this thread thinks there is an improvement. I was not expecting the level of improvement that i have found. It makes me want to play music constantly discover new aspects of tracks i thought i new really well.

Andy, would you care to share what specific improvements you have found ? I am sure that i am not hearing things.

Mr Tibbs has been very quiet about this seeing though he started the thread. I seem to recall he remembered a couple of years ago on another thread that he blew a NCC 200 board and the speaker too but that might not be directly linked to this approach.

I have been monitoring the temperature of my NCC200 heatsinks and there is no difference to before doing the change - i can hardly tell they are heating up but i don't play too loud.

Stu

http://www.pinkfishmedia.net/forum/showthread.php?t=38385

http://www.pinkfishmedia.net/forum/showthread.php?t=70397







[/URL]

 

[Chris]

I´m not even so sure re. the importance of the extreme shortness of the cable between amp and crossover. In the case of the Sugden, all we did was switch the cables round and use the pieces previously used from crossover to speaker ( say 30 cms) in the place previously occupied by the 4m or so which were placed from crossover to speakers and vice-versa. This weekend we´ll try shortening them. Can it get even better ? Noooooooo, I hear.

 

[andrewsutton]

Hi Stu, improvements include clearer deeper base, improved image and a cleaner overall sound.

I was playing Songbird and there were little bits and pieces I had never heard before. Particularly a few faint strums on an acoustic guitar became apparent.

I was wondering how many people hear some of these details.

Additionally I was looking at upgrading my rather shabby collection of banana plugs terminating my cables as there will be 10 pairs involved altogether.

What plugs have been recommended lately?

Cheers Andy.

 

[Bemused]

Tony, i wish you would or somebody would. Everyone who has tried it on this thread thinks there is an improvement. I was not expecting the level of improvement that i have found. It makes me want to play music constantly discover new aspects of tracks i thought i new really well.
[/URL]

Similar to the dual mono Buffalo DAC thing as many say it should have no effect but I rigged an AB test between dual mono and single buffalo where many agreed dual mono had a distinct benefit after listening.

Sadly I am not so nearly as bright with electronics as I would like, often being very wrong.
Its taken me a few days to under stand PDs comment about

A tweeter signal at say 3 kHz can travel down the wires, and neither knowns no cares what happens at 70Hz.

I was stuck in picturing two separate frequency signals whereas its just one complex signal.

Anyway I just did a very crude spice and could not come up with anything that I expected I may see. Not entirely sure if I have a good approximation of the crossover. I could not get the SIM to model some things I would have liked as I am not great with LT.

After a little play I just wondered if it was a phase shift thing, screen shot below seems to show increased phase shifting.

 

[Bemused]

aHH I goofed
This looks better.

Running a step param 1 to 10R in the woofer and tweeter.

Running a step param 1 to 10R in the amp cable.

Trying to show a bass transient.

 

[Chris]

Very hypnotic, but what does it all mean, Basil ?

 

[Bemused]

Not at all sure, not even sure if the sim is valid, but the last one seems to show a woofer transient messing with the tweeter response a little.

 

[martin clark]

Intresting, Tony, thats sort of what I'd expect; look at the effect of cable or amp output resistance for a start -
http://www.acoustica.org.uk/t/spkrcable.html

- so I need to develop this further, as you have done, and to higher freqs.

I've never said 'noooo' I just think it's a small effect. Maybe not for some combinations, just like say, swapping tweeter caps in a crossover. (And I do really like my amps subjectively; maybe because they bring the cable inside the f/back loop...)

 

[Bemused]

Intresting, Tony, thats sort of what I'd expect; look at the effect of cable or amp output resistance for a start -
http://www.acoustica.org.uk/t/spkrcable.html

- so I need to develop this further, as you have done, and to higher freqs.

I've never said 'noooo' I just think it's a small effect. Maybe not for some combinations, just like say, swapping tweeter caps in a crossover. (And I do really like my amps subjectively; maybe because they bring the cable inside the f/back loop...)

Yes looks to be something there, maybe its the phase shifting. A small effect but right in an important place. Guessing that would affect how the tweeter was presented. Of course most of us are three way so were probably looking at how the midrange comes on line.
Now mostly gone beyond my abilities, crossovers are what other guys do

Seen your page on output R effects, I have about 4R in series with my out door speakers mostly to offer some short circuit protection to the HankerNAPs as cables get attacked by creatures of the nigh but it does add a lift for the "party thing"

I have always been very envious of your 100S, they stand out as something interesting.

Tony

 

[PigletsDad]

OK, I have thought a bit about the bi-wire question.

What this does is segregated the currents, at least for conventional crossover designs. If you look at a typical low pass section, it has an inductor as the first components, so its input impedance rises out side its band, it draws little high frequency current. The high pass section in front of the tweeter will have an input capacitor, and draws very little low frequency current.

We can model biwire versus shared (or mono) wire by lumping all the resistance together. In the biwire case, there is a separate resistor in the path to each filter. In the shared case, there is a single resistor. This makes a small difference to the response.

In the shared case, the level at the woofer is slightly reduced by the input impedance of the high impedance section; this will cause some extra rolloff in the upper woofer range. Likewise, the level at the tweeter is reduced by the input impedance of teh woofer network, especially at the bottom end of the tweeter response.

With biwiring, you will get a small amount of extra level either side of the crossover.

How big is the effect? Lets say the loop resistance is 0.1Ohm, and the impedance of the sections just out of band is say 10Ohms, then you will get a 1% change in level, this corresponds to 0.086dB, which is small enough to be hard to hear, but just possibly noticeable.

If the wire is lower resistance, or the input impedance of the sections is higher, the effect will be smaller. But using something like DNM single strand, I am sure the effect would be clearly noticeable.

 

[Chris]

Can´t help thinking we are barking up the wrong tree. This is much more than a "just possibly noticeable 1% improvement. Subjectively I´d say more like 15- 20% if I might be so bold. Certainly the biggest and most fundamental upgrade I´ve ever heard.

Might the puny 20w Sugden A21a have something to do with it ? In what way does said amp resemble the NCC thingies that Mega Lord has ? Or the thick Van den Hul CS122 hybrid silver coated OFC cables ? Or the weather or the oncoming of Spring ?