| Pbassred |
I can understand balanced line: two wires carrying a signal. One has reversed phase. Any noise affects both equaily and when that signal wire is unreversed the noise is also flipped. the positive noise cancelles out the negative noise.
How does twisted pair work? |
|
|
| moamps |
| twisted pair = balanced line |
|
|
| Pbassred |
I wish it was that simple. I have balanced line cables that aren't twisted.
I think balanced is electrical noise cancellation whereas twisted is electromagnetic, but I don't understand the magnetic part. |
|
|
| audiofan |
A pair of parallel wire will make a single loop and works as an antenna if exposed to magnetic field and an electric signal proportional to flux change will be present in loop circuit.
A pair of twisted wire will make many loops each of them working as an antenna but electric signal from any loop will cancel signal from next loop.
This is why on a balance system each pair of wire must de twisted, have a look on LAN wire you will see how it is done 8 wires twisted to make 4 pairs. |
|
|
| owen |
Twisting a balanced pair reduces the amount of electromagnetic induction, as the helix reduces the amount of conductor normal to the field, and so reduces the amount of noise induced. In balanced operation, this is 'balanced' out by the negative conductor, but the benefits of twisted pairs isnt limited to balanced cabling.
Ok?
Owen |
|
|
| Pbassred |
| quote: | | helix reduces the amount of conductor normal to the field |
............... that's the bit I don't get. |
|
|
| jneutron |
| quote: | Originally posted by Pbassred
............... that's the bit I don't get. |
While the helix does indeed lower the total magnetic intercept cross section on a loop by loop basis, that is not why the twist works. For reaonable twist pitches, that reduction runs anywhere from 5% to 20% reduction.
As mentioned before, Faraday's law of induction applies. If you take a twisted pair, and project it's metal's "shadow" onto a surface, you will see the wire forms football (american, not the rest of the world) shaped openings. Note that the "positive conductor" in that pair swaps side to side, from one loop to the next.
When a changing magnetic field is going through those loops, each loop will produce a voltage, depending on the rate of change of the magnetic field, and the total area of each loop. The polarity of the voltage generated within each loop will depend on which side the "positive wire" is on. When it swaps side, the voltage polarity also swaps.
This causes the voltage that was generated loop to loop to be opposite sign, so they cancel.
Cheers, John
Had a pic, just couldn't find it...sorry.. |
|
|
| janneman |
Hey John,
That means we should have an even number of twists, no ? ;)
Jan Didden |
|
|
| audiofan |
Magnetic field will induce voltage in a loop of wire, the voltage is proprtional to the flux change in the loop ( farady,s law ) so for a flux change the voltage is proportional to the loop surface normal to the flux. Using twisted pair reduce surface and also give several surfaces opposite one to the other so several voltages oppotie one to the other this is why you get cancellation of the noise ( induced voltage )
Owen is right but you must use calculus to see how it is going when you read : the helix reduces the amount of conductor normal to the field |
|
|
| audiofan |
| A even number of loop is ideal but if you have an odd number noise will be limited to the noise of the the last loop if the setup is ideal , in practice it will not make any difference |
|
|
| jneutron |
| quote: | Originally posted by janneman
Hey John,
That means we should have an even number of twists, no ? ;)
Jan Didden |
Odd you should say that, No?
If the magnetic field were uniformly changing along the entire pair length, yes, even would be required for complete cancellation.
Since most sources in the audio range are not uniform, but change in space, the effect would be localized to one area of a cable.
So, I would expect a distribution of generated voltages loop to loop looking somewhat like a sin x over x construct, x being distance along the cable.
Cheers, John
The possibility that you were being tongue in cheek did cross my mind, but I've never been one to consider common sense when posting..:xeye: |
|
|
| EC8010 |
The two wires in the pair can only receive exactly the same noise (and therefore be cancelled) if they occupy the same point in space. A tight twist averages them to occupying the same point. If they were simply laid parallel, then one conductor would receive slightly more field (electrostatic or electromagnetic) than the other. Twisting alternates the positions so that they each receive equal interference.
In theory, that means you need an even number of twists (but see the previous post's comment). In practice, the wire is so short compared to the wavelength of the induced field (50Hz = 6000km) that it really doesn't matter. Even if the wire was 10km (not unheard of in the days of analogue telecommunications), then the slight imbalance caused by an extra twist that's a few inches long wouldn't matter. |
|
|
| jneutron |
| quote: | Originally posted by EC8010
The two wires in the pair can only receive exactly the same noise (and therefore be cancelled) if they occupy the same point in space. A tight twist averages them to occupying the same point. If they were simply laid parallel, then one conductor would receive slightly more field (electrostatic or electromagnetic) than the other. Twisting alternates the positions so that they each receive equal interference.
In theory, that means you need an even number of twists (but see the previous post's comment). In practice, the wire is so short compared to the wavelength of the induced field (50Hz = 6000km) that it really doesn't matter. Even if the wire was 10km (not unheard of in the days of analogue telecommunications), then the slight imbalance caused by an extra twist that's a few inches long wouldn't matter. |
Wavelength is not a consideration with twisted pairs for "space static" magnetic fields. Yes, it will be for fields which are propagating in space, as in radio.
I was speaking of fields which are not propagating per se, but are local phenomena, like the field generated by a transformer or degaussing coil or a length of romex.
We just gotta make sure we're talking apples and apples.
Cheers, John |
|
|
| EC8010 |
Yes, I knew you had local sources like leakage from transformers and the like in mind; I was thinking of the rather more diffuse fields caused by power distribution, and, as you say, radio. It all goes to show that twisted pair is jolly good stuff.
I know what an apple is, but what's romex? |
|
|
| jneutron |
| quote: | Originally posted by EC8010
Yes, I knew you had local sources like leakage from transformers and the like in mind; I was thinking of the rather more diffuse fields caused by power distribution, and, as you say, radio. It all goes to show that twisted pair is jolly good stuff.
I know what an apple is, but what's romex? |
Even the diffuse fields such as house wiring and power distribution can be picked up by small loop sizes. In fact, that is actually why hum exists at all.
Our enviro is such an e/m messy one. I'd prefer all ac distribution be carried out with coaxial wires to eliminate it. Course, that renders the handheld ac wire detectors useless, and I've been using one lately.. Boy, some people can really mess up house wiring...
Romex is what we use this side of the pond to carry power in the walls of the house.
Here's a pic of romex.. this is actually a piece of what I found in the wall of my house. The "installation" goofball was putting a GFI in the bathroom, and in drilling to snake the wire, did this to the feed cable to the junction box...
Sheesh..
Cheers, John
Edit: that is what I found in the wall years after the work was done..I shoulda said "when he put it in".. |
|
|
| EC8010 |
| Years ago, I was worried that I might manage the same trick, and I was alarmed that my power drill was not earthed. Technically, that was OK because the motor was double insulated, but the (metal) gearbox on the front would have been in very good contact with the cutting tip of the drill. I replaced the drill's cable and earthed the gearbox. I haven't yet managed to drill a cable, but if 30mA is needed to trip the RCD, I'd rather it went through a bit of wire rather than me. |
|
|
| Nordic |
On connecting a bipolar PSU... Do you twist together +V, -V and 0V, just +V and -V, or make two 0V leads to make +V and 0V, and -V and 0V leads?
Currenlty I use method A, but not based on any information other than readig about peolpe twisting cable, (heck I even googled and learned how to make braids with cables, never learned how in the more than 10 years I had long hair). |
|
|
| EC8010 |
It all depends on where the currents go. If you knew that the current only flowed from +ve to -ve (and never into 0V), then you'd twist +ve and -ve together, and ignore 0V. But the load (probably a loudspeaker) has one end connected to 0V, so there's load current in the 0V wire.
The braid versus two pairs of twisted wires is a good question. I think it's down to the class of output stage.
If you had a pure Class B output stage, you would know that load current flowed into the 0V wire from either the +ve or -ve, but never both simultaneously. In that case, the best solution would be to make a power supply with two transformer secondaries, two rectifiers, and two reservoir capacitors. You could then take a twisted pair from each and join the 0V from each supply at the loudspeaker GND terminal.
If you had a pure Class A output stage, you would know that load current supplied from one rail would be matched by an equal and opposite reduction in current in the other rail, so all three wires need to be twisted or plaited together.
In practice, power amplifiers are not usually pure Class A or B, but a mixture of both (Class AB) and the reason for twisting supply rails in a PSU is to reduce the interference it produces, rather than to protect it from interference. Twisting probably helps, but keeping the leads short and pushed into corners is likely to be more effective. On my rare forays into solid state, I twist or plait all three leads. |
|
|
| Nordic |
| For such a good answer you deserve 3 magic beans, to do with as you like. |
|
|
| EC8010 |
| quote: | Originally posted by Nordic
For such a good answer you deserve 3 magic beans, to do with as you like. |
Thank you, that's very kind. Sadly, I'm rubbish at gardening. No empathy at all. :cannotbe: |
|
|
| jneutron |
| quote: | Originally posted by Nordic
On connecting a bipolar PSU... Do you twist together +V, -V and 0V, just +V and -V, or make two 0V leads to make +V and 0V, and -V and 0V leads?
Currenlty I use method A, but not based on any information other than readig about peolpe twisting cable, (heck I even googled and learned how to make braids with cables, never learned how in the more than 10 years I had long hair). |
Use neither. Do something I have never seen used in power supply design.
Triaxial.
Take an insulated wire, whatever guage..
Cover it with the braid removed from a microphone balanced run cable..(first ya gotta remove it from the cable..)
ps..I purchased 1000 feet of balanced mike cable for this purpose..
Cover that braid with heatshrink tubing.
Cover that tubing with a second braid.
Cover that with a second heatshrink.
What you have duplicate is a triaxial cable.
Regardless of what current flows where, there will be net zero current flowing through this cable, hence ZERO external emf, at any frequency. This cable also will be impervious to induced voltages along it's length. This frees up the chassis layout design.
ALL connections to the power supply must be through this cable, do not form any chassis loops.
edit: it is of no concern which wire is assigned to which, but I would recommend ground as the outer shield to prevent electric field induced noise from a rail broadcasting into into hi-z circuitry.
Edit: as an added little bonus, the inductance of this cable is gonna be in the mud, you will not be able to measure it. It will run in the 20 to 30 nanohenries per foot range..
Cheers, John
I had considered working with an amp designer or two to put this design into public domain...but the cat is now outta the bag. Perhaps a new thread on this?
The pic is a quadaxial cable.. |
|
|
| EC8010 |
| Now that's a very interesting idea. And for those of you who don't fancy making the cable up, you can buy ready-made tri-ax from vendors who deal with broadcasters; tri-ax is the standard camera cable. |
|
|
| poobah |
In the case of a bipolar PSU... you twist all three wires. You need to study the current loops... keeps all currents flowing one way right next to the same current flowing the other way... and reduce or cancel the loop areas by twisting.
I once worked on an inverter installation (500 kW) where the electricians did not follow the mechanical layout for the wiring in plastic conduits under the a poured concrete floor... they thought it was silly and wasted wire. They didn't keep the currents together.
After running the invertor for about an hour one day while testing; the concrete under the unit exploded. A chunk of rebar in the concrete had gotten very hot from magnetic induction/eddy currents. While we don't deal with such high power in audio... the principles are the same.
;) |
|
|
| jneutron |
| quote: | Originally posted by EC8010
And for those of you who don't fancy making the cable up, you can buy ready-made tri-ax from vendors who deal with broadcasters; tri-ax is the standard camera cable. |
Hey, whatttsaa matta??? Ain't dis a DIY site?? C'mon..;) | quote: | Originally posted by poobah
In the case of a bipolar PSU... you twist all three wires. You need to study the current loops... keeps all currents flowing one way right next to the same current flowing the other way... and reduce or cancel the loop areas by twisting. |
That'll work well as long as the pitch of the three is tight enough to integrate the helical fields out to zero. I would worry about running the twisted tri near any low level signal paths or grounds. I prefer not considering the texture of the associated electronics.
Your solution also guarantees the inductance is in the 200 nH per foot region.
My solution avoids all that.
| quote: | Originally posted by poobah
I once worked on an inverter installation (500 kW) where the electricians did not follow the mechanical layout for the wiring in plastic conduits under the a poured concrete floor... they thought it was silly and wasted wire. They didn't keep the currents together.
After running the invertor for about an hour one day while testing; the concrete under the unit exploded. A chunk of rebar in the concrete had gotten very hot from magnetic induction/eddy currents. |
Don't you love anarchy when it comes to building high power stuff?
| quote: | Originally posted by poobah
While we don't deal with such high power in audio... the principles are the same.
;) |
High power? Speak for yourself..:smash:
(It don't mean a thing, if it ain't got that "sting".
Cheers, John
Edit: spellfix..Condisering...sheesh..sounds like eviscerating. |
|
|
| poobah |
John,
Heat shrink is spendy... I may go for that triaxial thing when I do my 350 kW diesel gas turbine powered bass guitar amp though...
;) |
|
|
| jneutron |
| quote: | Originally posted by poobah
John,
Heat shrink is spendy |
It is:confused:
| quote: | Originally posted by poobah
my 350 kW diesel gas turbine powered bass guitar amp
;) |
Sigh..I talk pie in the sky, you talk limitations...;)
Cheers, John |
|
|
| poobah |
Yeah John,
If you would BUY some of that heat shrink; instead of taking samples home so you can "work at night" you would know how 'spensive it is... jeeze.
I miss that... I am self employed now... I actually have to buy printer paper and stuff... sucks!
:mad: |
|
|
| jneutron |
| quote: | Originally posted by poobah
If you would BUY some of that heat shrink; instead of taking samples home so you can "work at night" you would know how 'spensive it is... jeeze. |
At work we use tefzel heatshrink, from raychem. That stuff is way too expensive, 2 to 4 dollars per foot. And, we have to buy minimums of about 200 or 250 foot spools. Plus, it's a pain to shrink as it requires way too high a temperature, it will start to melt the insulation of a regular pvc wire.
Nah, I was thinking of PE stuff, the 3M GP 2 to 1 tubing that is sold in 4 foot lengths priced from 1.25 at 1/16th, up to 2.76 for 4 feet of 1/12 inch.
So making a foot long triax would require about two feet of heatshrink, perhaps a buck thirty??
Cheers, John |
|
|
| poobah |
ok... ok... ok,
We still need to check with SY to see which ceramic colorants are acceptable though... |
|
|
| jneutron |
| quote: | Originally posted by poobah
ok... ok... ok,
We still need to check with SY to see which ceramic colorants are acceptable though... |
OH MAN!!!!!
I forgot all about that..
Nevermind..:D |
|
|
| bluebeard |
| I think my neighbors are a bi-polar twisted pair. |
|
|
| AndrewT |
Hi,
have you upset them in some obscure way? |
|
|
| jneutron |
| quote: | Originally posted by poobah
We still need to check with SY to see which ceramic colorants are acceptable though... |
Ok, so..now that I have the heatshrink, where's Sy??? I need to know which colors are ok.
The stinkin MSDS says exactly one thing..
Blue.
or, Yellow..green, red, black, white..
Now, I'm......:confused:
Hmmm..where did I put that 800 feet of mike cable... I got's me some braid to harvest..
Say, can anyone use a coupla hunnert feet of #22 stranded blue/white twisted pair, stripped from it's home??
Seriously..I'll make some triax and ele test the stuff for C and L.
Anyone interested in trying some of it for dual rail supply of a gainclone??
Cheers, John |
|
|
| poobah |
John,
Yellow or white will give the brighest sound; green, the most natural, brown or black, dark and veiled; red sounds warm, blue.... well... sad...
:cool: |
|
|
| jneutron |
| quote: | Originally posted by poobah
John,
Yellow or white will give the brighest sound; green, the most natural, brown or black, dark and veiled; red sounds warm, blue.... well... sad...
:cool: |
They didn't have brown.:bawling: :bawling: :bawling:
John |
|
|
| poobah |
should should have gone for the clear John... transparent sound!
:smash: |
|
|
| BrianDonegan |
You can also buy copper braid without having to disassemble existing cable. Even comes in various diameters.
[EDIT] I briad my trafo wiring, cause it looks nice. Just try to keep it as short as possible. |
|
|
| jneutron |
| quote: | Originally posted by BrianDonegan
You can also buy copper braid without having to disassemble existing cable. Even comes in various diameters.
[EDIT] I briad my trafo wiring, cause it looks nice. Just try to keep it as short as possible. |
Yup. It is amazing the variety that is out there.
Comes in handy when you want to make 1/4 to 1/4 low level cables..:D
OH, forgot..that's an eminence delta 12 in the back, with a selenium d205ti tweet.
John |
|
|
| jneutron |
Did the triax. It took about 50 minutes to make this four foot section.
Used what appears to be 12 guage orange for the center, and two pieces of shield I harvested from the aforementioned kilofoot mike cable spool.
The first layer of heatshrink is a light blue 4 foot piece of 3/16 that I just bought from PE.
The second shrink layer, yellow, turns out I was still able to use 3/16 shrink. I had purchased a mess of 3/8 and 3/4 in case I needed bigger, but didn't.
I'll mess with the meter another day, and measure the L and C of the cable..both measures with respect to the middle coax. Oh, and resistance of course..
Any amp builder feel like tryin the stuff for supply runs when I'm through with it??
Cheers, John
I expect this cable to be bright and sad....course, don't know what the orange will do..:confused:
edit: PS, Brian.. I am using what I have on the shelf at home. For an app I was doing here, I needed two and possibly 4 runs of balanced mike cable at 120 feet each, so bought the 1000 foot spool from parts E for 175 dollars... Such a deal.. so, I use it.. I note now that it has climbed to 199 dollars for the spool of 1000 feet. Still, not bad..
kinda wish the camera woulda focussed better.. |
|
|
| Hector Ruiz |
| Balanced cable with respect to externally induced noise rejection is only as good as the circuitry producing the desired signal signal (i.e. equal amplitude, but opposite polarity). On the far end a balanced differential receiver is necessary. The receiving circuit re-inverts the out-of-phase desired signal and the audio reinforces itself [+1 - (-1) = 2]. On the other hand external, undesired noise is induced equally and in-phase into both legs of the balanced transmission system. Recall that the noise is equal and in-phase on the balanced transmission lines, so when the receiving circuit inverts one leg of the noise signal, in-phase becomes out-of-phase and adds up to zero [+1 - (+1) = 0]. |
|
|
|
