Impedances what's it all about.

[Deleted]

Nov 27, 2010 23:10:02 GMT FauDrei said:

Hmm... can you elaborate on speed/bandwidth? I'm not sure I understand.

My take is that if you bridge two f.e. 100V/us amps you get a bridged 200V/us amp.

Valter
I found the attached quote elsewhere. It seems to make sense to me. Also, noise will be higher, and damping factor reduced.
Alex

The rise per us is double, but so is the swing, so the end result is the same.

[XTRProf]

Nov 26, 2010 11:28:52 GMT @solderdude said:

Nov 26, 2010 9:53:36 GMT XTRProf said:

Yeah, balancing also means we have to have more components just to give a negative half of the positive signal for cancelling noise and distortion. Thru this, it doubles the voltage swing, if I'm not wrong. However, to really cancel noise and distortion, both signal halfs must be really well matched or else there wouldn't be any good cancellation. In other words, both the positive and negative halfs must be a good mirror image of the signal. A perfect amplifier must be built!

Also, more components also means more noise and distortion although they cancels out. So why build something with higher noise and distortion that cancels out instead of building something single ended and having lower noise and distortion? Keep it KISS will always be better sounding. In a home environment do we really run long lengths of signal ICs? No right? So the noise is minimally picked up unless your setup is not connected properly. Balanced is more for the professional AV scene like rock concerts, speeches, discos, DJs, etc with long signal leads that can act like antenae to pick up the noise and distortion.

Correct me if I'm wrong as I'm no expert in this but knowledgeable enough.

Driving balanced is effectively using 2 similar amplifiers with opposite signals that both drive the same driver.
For stereo in essence 4 amplifiers are needed instead of 2.
So not positive or negative halves of sinewaves, like in the output stage of an amplifiers complentary output stage, but both amps have the same full signal only turned 180 degrees in phase.
So when the output of one amp goes positive (or negative) the other amplifier goes negative (or positive).

To make it clearer I need pictures but do not have the time for it now.

NO noise and no distortion is cancelled.
In fact distortion caused by the amps is even ADDED instead of cancelled.
The noise generated by the amplifiers don't cancel and doesn't add as noise is random and the noise might be added (if in counterphase) at a certain moment or cancelled (if in phase) at another moment.
Effectively the noise floor of the headphone amplifier itself is hardly any problem (in most SS designs) and well below the hearing treshold unless you have a really crappy design.
The noise floor in the recording which is often several hundreds of times higher (30 to 40 dB) is amplified by both amps and does not cancel out nor add and is reproduced like the music signal in which it is included.

The only thing that is cancelled are COMMON mode induced (unwanted) signals in (very) low level signal lines which would NOT be cancelled/filtered out when induced in unbalanced lines (which a normal RCA cord is).

They also do not have to be in perfect balance for this trick to work.
It works better IF they are closely matched in levels but if not matched exactly it will still work but with a few dB less attenuation of these unwanted COMMON mode signals.

The reduction of COMMON mode induced signals is the ONLY reason balanced lines are used and, indeed, only needed for microphone, instruments and long line level lines to amplifiers on stage or in the studio.

Common mode signals is 50/60Hz hum and their harmonics or Radio Frequency signals that are picked up, like an antenna does, and is in equal phase (and mostly also amplitude, if twisted wires are used) and since these signals are equal in phase and amplitude this does not induce a voltage in the balanced load as this reacts ONLY to signals that are NOT equal in phase (180 degrees).

balanced driving of headphones is mostly used in studios as there are balanced signals there anyway and a lot of equipment with (long) cables lying around in places where they sometimes shouldn't be.
NOT to get the ultimate High-end experience.

Excellent explanation Prof Fran! All the major technical issues on balanced are there although without pictures!

[XTRProf]

Nov 27, 2010 19:12:43 GMT FauDrei said:

...so that makes me snobby A-hole with double impedance double "speed" amp...

Alright, double A-hole with double the "impedance" but without the double "speed". Smooth operator? Hmm, would that be nice and fun? Heh, heh, heh ...............

Ok, joke only for this time of the year for Merry Making .........

[Deleted]

Nov 27, 2010 23:40:14 GMT @sandyk said:

Nov 27, 2010 23:10:02 GMT FauDrei said:

Hmm... can you elaborate on speed/bandwidth? I'm not sure I understand.

My take is that if you bridge two f.e. 100V/us amps you get a bridged 200V/us amp.

Valter
I found the attached quote elsewhere. It seems to make sense to me. Also, noise will be higher, and damping factor reduced.
Alex

The rise per us is double, but so is the swing, so the end result is the same.

Correct,

If an amplifier has 1MHz bandwith (-3dB = 0.707x drop in voltage compared to 1kHz) and you take 2 of them the bandwith will still be 1MHz.
Since the risetime of the whole amplifier (V/us) is determined by the HF feedback of each single amplifier this doesn't change.
So speed remains the same, bandwith remains the same.

Except for the noise part.
This has to be nuanced, as Alex is partly right.

Let me explain...

Noise, meaning background noise of the amp itself, NOT the noise in the audiosignal that is amplified is random by nature.
This means that when both amplifiers are emitting random noise the amplitude and phase of both noises (being random) at some point in time may be in phase or in counterphase and may have equal amplitude or different one.
The fun part of this is that when noises sources are 'coupled' the noise level drops the more of these noise sources are used (to a certain degree)
This is why in some of the earlier designs in MC and microphone input stages where noise is a real problem they often lowered the noise floor by paralleling input transistors 4 or 8 times to gain some S/N ratio.

Now if the random noises are in counterphase at a specific moment and have (about) equal amplitude this means DOUBLE the noise voltage (hence Alex is right) BUT at another moment the phase might be the same and thus cancel out and thus be gone..
In effect the noise 'peak' level increases BUT the average level does not increase.
Since the output voltage is always in counterphase and is double that of a single amplifier the S/N ratio improves in this case when looked at from maximum power.
When referenced to the same SPL or a specific SPL (say 90 dB) the average level is about the same but since the peak value has increased the perceived noise level will be somewhat higher but not by 6dB

[kasper]

I never understood what Balanced phones really is about! Even balanced cabled phones dont carry more than 4 wires in the cable, so it couldn´t be balanced in the same way ic are??

With my VERY limeted understanding ic have a signal wire, a wire with the signal turned 180 and a screen wire.

Could the improved sound people experience with balanced have to do with just an exstra ground wire??

Normal jack plugs force the ground of both left and right into the same connection - that must give room for some spill over from side to side???

Sorry for my ignorance

[Deleted]

The technical differences between balanced and unbalanced is the complete separation of the L and R signals because there is NO common wire (ground) present.

This can also (almost) be achieved by having a HP cord with 2 wires going to each ear piece and connect the common wire in the plug instead of 3 wires and then splice them in the headphone.
In the amp itself there is always a single wire/connection/socket and therefore this can never be a true dual mono construction as the common is always connected somewhere.
Symmetric headphone amps that are connected with normal asymmetric RCA plugs are still connected at one point; the common/screen/ground but NOT at a common point in the headphone wiring.

Wanna know if the signals from the left earpiece enter the right one (this is called crosstalk and is NOT the same as crossfeed)
Here is an easy test:

connect the HP to your amp.
Disconnect the L or R input of this amp and preferably put a shortcircuited input plug in it's place.

Drive 1 earpiece pretty loud and cover that earpiece (DON't drive it into distortion, it must sound loud AND undistorted !).
Now ONLY listen to the earpiece that is NOT driven.

I need to say that IF you hear something it is more likely due to a bad separation between the L and R signal of the amp then the actual wires in the headphone.
If you hear distortion products or tinny or bassy sounds it's the amps fault (nasty crosstalk, don't do this with Indeed thingies that have not been crosstalk modded for instance).
If it is the 'normal' signal.. but just very soft the common cable MIGHT be at fault or the amp simply has nice 'crosstalk'.

For symmetrical HP driving there is NO extra shield/ground cable nor is is needed because of the signal levels and low impedant circuitry.
Also there are only 4 connections and 6 would be needed if individual screens were connected.
For low level symmetrical signal transport a screened cable is highly recommended though NOT needed IF the symmetrical signals are NOT referenced to ground (with coupling transformers)

If the perceived differences are real will always remain un answerred as you can't AB this.
at least not with a simple switch because the levels will have to be lowered by 6 dB for a real comparison.

[jeffc]

Frans,

Terrific insight, as usual . This seems to explain very well why a cheapie shielded twin cable that is only connected at the 3.5mm mini-jack has worked so nicely with my HD681's. Unfortunately due to using USB to DAC and mini-jack connenction to amps I can't test crosstalk as suggested. However, as stated in my comments on SQ improvements following the cable change, it was very much worth the effort.

A pic of the twin cable
i337.photobucket.com/albums/n390/jeffc_pics/PB271445.jpg

cheers..jeffc

[Deleted]

a 3 wire cable that is rather long with a low Ohmic headphone should translate into audio being in counterphase with the other channel attenuated about 50dB.

The difference between a 3 and 4 wire cable will be mostly in the stereo image which will be slightly more un-naturally wider with a 3 wire compared to a 4 wire.
The effect is subtle as only 0.5 % of the left signal is substracted from the right signal and vice versa.

The effect can be studied (greatly exaggerated) by mounting a series resistor of 10 Ohms in the common line of the headphone which is possible if your headphone amp has 3 wires coming from a board to the headphone socket.

The higher Ohmic the headphone the smaller this effect will be.
In a 300 Ohm headphone the level will be dropped to 65 dB already (0.05%).

[rowuk]

Kind of proof that much of what we imagine to hear is simply in the frequency domain and can therefore with relatively inexpensive modifications be cleaned up. Now we just need audiophiles that like flat frequency response.........................

Fletcher-Munson also need to be considered, unfortunately impedance does not change in the right direction to boost bass and treble when playing more softly.

I think the killer HA needs at least 2 parametric equalizer channels and adjustable impedance. State of the art headphones need adjustable earpads to match the geometry of the headphone to the head. My AKG701s came to life when I compressed the earpads about 2mm. The bass picked up and everything became much smoother. DT770 benefited from some cardboard spacers (1mm) that made the earpads thicker reducing bass and one noteness. Remember that there is a resonant space in front of and behind the headphone driver. They will also alter the perceived SQ.

[Deleted]

My evil twin (NwAvGuy) wrote an article about the power levels of headphones.
nwavguy.blogspot.com/2011/09/more-power.html

I also had been working on a similar subject already but from a totally different angle with different info.
The articles complement each other IMO.

here it is... hope it is found usefull.
www.mediafire.com/?jdpaj1r17x1sos5

[Deleted]

Downloaded and reading it right now, I was really waiting for this

[funk1969]

Thanks...

[Deleted]

My evil twin (NwAvGuy) wrote an article about the power levels of headphones.

;D ;D

[JohnnyBlue]

Thanks, Frans.

[pcourtney1]

some great info there, thanks Frans (you will always be our technical hero)

if a headphone has a peak input power of 9v and 200mW, and is 400ohm

what output power does the headphone amp need to be to drive it to an SPL value of 110

nb I am trying to work this out, an example is the X-CAN, which I think has an output of 1000mW, 47 ohm, but not sure about the output volts the x-can puts out ?

[PinkFloyd]

Nice article Frans, many thanks for highlighting what we have been saying for YEARS

[Deleted]

After mediafire removed my article I decided to rewrite it (as it was thrown together quickly) and include some enlightening pics.

www.mediafire.com/?82kf0r5kdckdcer

[Chris53]

Nov 14, 2010 19:45:32 GMT dicky said:

Me too! That has spurred me on to have a go changing the o/p resistors - see what I can do with my DT770s. (Which I think I like :

I plonked in some 75ohm resistors - 'cos that's what Mike did with Panda 1. But now I have a little knowledge mmmwwwahahaha.

Thanks very much for your post

Ahh so THAT's the value of the resistors in Panda One. Thanks! I've been trying to find that on here. If that's the case I should try to change that to 120 ohm for my AKG 601's.