Audio and Video over Twisted Pair

audio over twisted pair

I want to send the signal from a sat receiver to other rooms by Twisted pair.Searched the net and found this lovely amplifier **broken link removed** for video signals, needed to know if anyone has anything similar but now for audio signal.
Or can this schematic be used for audio sending too, of course with some mods?

50 ohm twisted pair

For audio frequency I would rather use a set of audio frequency transformers.
The impedance of a twisted pair is in the 100+Ω region, so a 150Ω audio transformer should fit probably into this equation.
These transformers can transmit signals in the range of 30Hz to ≥25kHz.
If you have one receiver, then another 150Ω transformer will be the match for the line; if you have 4 receivers, then each of the receivers can have a 600Ω transformer, and so on ..
And, because transformers provide a complete isolation, you don have these .. brrrrr .. loops ..
Regards,
IanP

audio twisted pair driver

Thx for the response.
Didn't understand the last part fully.
The receiver states "Output Level L/R 0,5 Vss on 600 Ohm"
So what i need to do is to put the primary of a transformer with 600ohm on the output of the audio signal and then the secondary with 50 ohm (this is the impedance of twisted pair), and then on the receiver end a transformer with 50 ohm impedance on primary connected to the twisted pair and then the secondary with 150ohm (normal audio impedance i think) to feed the tv right ?
I found this nice page about transformer impedance matching too
https://farside.ph.utexas.edu/teaching/302l/lectures/node91.html

video to cat 5 impedance matching

>I want to send the signal from a sat receiver to other rooms by Twisted pair.

>Output Level L/R 0,5 Vss on 600 Ohm
If you want to send stereo you will need a pair for each channel.

>For audio frequency I would rather use a set of audio frequency transformers.
That is what I would use but I would use transformers to avoid ground loops and
common mode noise, not for impedance matching.

>The impedance of a twisted pair is in the 100+Ω region
This does not matter at audio frequencys unless you are driving thousands
of meters of cable. If you are using cat5/5e/6 cable then the resistance
of the cable has far more effect than transmission line effects.

>so a 150Ω audio transformer should fit probably into this equation.

Modern audio systems are not impedance matched. Maximum power transfer is not
required. Line level audio ouputs are lowish impedance and drive highish
impedance impedance inputs.

With long cables it would be better to use a transformer that increases the impedance
and voltage so the cable resistance has less effect, with a step down at the
endpoints.

If you put the transformers close to sources of magnetic fields such as mains
transformers, CRT televions or loudspeakers they may pick up hum, interference or
cause feedback.

spice model for cat5 cable

Microemission said:

Thx for the response.
Didn't understand the last part fully.
The receiver states "Output Level L/R 0,5 Vss on 600 Ohm"
So what i need to do is to put the primary of a transformer with 600ohm on the output of the audio signal and then the secondary with 50 ohm (this is the impedance of twisted pair), and then on the receiver end a transformer with 50 ohm impedance on primary connected to the twisted pair and then the secondary with 150ohm (normal audio impedance i think) to feed the tv right ?
I found this nice page about transformer impedance matching too
h**p://farside.ph.utexas.edu/teaching/302l/lectures/node91.html

Click to expand...

all cable have complex impedance for audo frequency range

I try calculate for long time ago on CAT-5 cable using model with sender output using 1 kOhm impedance and receiver input using 10 kOhm impedance resistive and give:

-----------------------------------------------------------------------
| length | 100 Hz | 22000 Hz|
| meter |
|CAT-5 |
-----------------------------------------------------------------------

10 -0.830 dB -0.847 dB
20 -0.831 dB -0.899 dB
50 -0.836 dB -1,247 dB
100 -0.843 dB -2.3 dB
500 -0.903 dB -11.49 dB

distance < 50 meters in practical not need compensate or matching.


If possibly to mounting capacitance parallell over output resistor (1 kOhm) on sender side (and driver is strong enough and stable) you can compensate high frequency attenuate depend of cable-capacitance:

-----------------------------------------------------------------------
| length |value |
| meter | capacitance|
| CAT-5|
-----------------------------------------------------------------------

10 4.9 nF
20 10.0 nF
50 24.9 nF
100 49.8 nF
500 298.6 nF

and 20000 Hz attenuation compare to 20 Hz is lower than 0.1 dB difference

(yes - value is pretty high for fully compensate and if try using smaller value typical 0.1-1 nF for 10 meter, give much worse result compare using without capacitance. try this on spice-simulate if not belive... )


my simulate model for CAT-5 cable is:


length resistance R = 188 Ohm/Km (DC)
(376 Ohm/km @ 1 MHz depend of skin effect)

parallell capacitance C = 52 nF/Km (calculated 49.75 nF/Km @ 1 MHz)

length inductance L = 0.4975 mH/Km ( calculated)

Parallell conductance G = 1E-7 Siemens/Km (audio frequency and loss angle at 0.0003 for plastic insulation)
(calculate 0.008455 Siemens/Km @ 1 Mhz)




And now possible to using telegraph equation:

complex attenuate constant = gamma = alpha + jbeta

alpha is attenuate constant in neper (1 neper = 8.686 dB)

and

beta is phase constant in radians


gamma = sqrt( (r+jwl) * ( g + jwc) )

and

character impedance Z = sqrt( ( r+jwl) / (g+jwc) )

(w = omega = 2 * PI * frequency)

---

if calculate Z, result is capacitive complex impedance for low audio frequency.

ex. for 1000 Hz:

Z = sqrt( (r+jwc) / (g+jwc) )

w = 2 * PI * f = 2 * PI * 1000 = 6283

Z = sqrt( (188+j(6283 * 0.0004975) ) / (1E-7+j(6283 * 52E-9) ) )

= 540-j532 Ohm or 758.6 Ohm |_ -44.5 degree characteric impedance
for 1000 Hz


for 100 Hz give 1700-j1692 Ohm or 2398 Ohm |_ -44.9 degree characteric
impedance

and for 10000 Hz give 184.3-j156.1 Ohm or 241 Ohm |_ -40 degree characteric
impedance


ie. CAT-5 cables characteric impedance varying very much on audio frequency, but not so intresting if wave length of LF signal is very much longer than cable distance between sender and reciver. - but for telephone company and long distance is very importent parameters...

-----------------------------------------------------------------------------


calculate with complex argument "r+jwl" is easy if you have
complex capable calculator simular hp42s, hp48xx etc.

you can download 'free42' from

http://home.planet.nl/~demun000/thomas_projects/free42/

twisted pair cable impedance 50 ohm

Interesting.

I have done speech and music over hundreds of meters of cat5 in buildings several times. I have not had a problem with high frequency loss. It sounds ok to me.
Last time I checked I can hear sounds up to about 15KHz. 10dB loss at 22KHz is not a problem.

Next time I will try a compensation capacitor.

I have found that screened cable gives fewer problem with picking up signals from other cables such as bleeping from fire alarm systems. Sometimes the screen needs to be conneced just at one end.

>but for telephone company and long distance is very importent parameters...
Telephones use bidirectional audio with hybrid circuit so impedance matching is more important. It helps that telephones only pass audio up to about 4KHz.

audio impedance and distance

Very nicely explained xxargs!
Where can i get the Spice model of the Cat5 cable?
Or you did it yourself with the parameters u gave?

schematics video on twisted pair

throwaway18 said:

Interesting.

I have done speech and music over hundreds of meters of cat5 in buildings several times. I have not had a problem with high frequency loss. It sounds ok to me.
Last time I checked I can hear sounds up to about 15KHz. 10dB loss at 22KHz is not a problem.

Next time I will try a compensation capacitor.

I have found that screened cable gives fewer problem with picking up signals from other cables such as bleeping from fire alarm systems. Sometimes the screen needs to be conneced just at one end.

>but for telephone company and long distance is very importent parameters...
Telephones use bidirectional audio with hybrid circuit so impedance matching is more important. It helps that telephones only pass audio up to about 4KHz.

Click to expand...

you notice high attenuate on high frequency and going worse you longer distance
of cable with exponent function of distance. ie impossible to make very long distance call even use amplifiers (depend gain of high frequency noise more then low frequency ie. S/N goes down more quickly on high frequency compare to low frequency)

This is depend of miss of length inductance in cable and for much parallell capacitance on low frequency.


low educated Mr Pulpin invented if insert extra length inductance on line to compensate cable capacitanse (and blamed very much from scientific socity for this 'stupide idea' - you make a low pass filter!!!, and missing whole point) and decrace attenuate very much on low frequency, but cost more attenuate on high frequency depend of cable capacitance is a distrubed and inserted extra inductanse is discrete points.

in low frequence see very many of 'inserted inductance' in wavelength and accept as distrubed to give real impedances and no extra attenuate on little higher frequency

but high frequency see wery few 'inserted inductance' as dicrete points in wavelength and make daisy chain-coupled low pass filter. Pulpin balanced this to give very low and flat attenuate up to around 4000 Hz and frequecy above attenuate very quickly - enougt to make usable long distance phone call .

this invent make phone line usable from only 10-50 km to >1000 km without amplifier and very, very importent invent to make phone worldwide before new HF-technology take over.

Added after 34 minutes:

Microemission said:

Very nicely explained xxargs!
Where can i get the Spice model of the Cat5 cable?
Or you did it yourself with the parameters u gave?

Click to expand...

I digging up my old news article written years ago, but written in swedish
and not usable for most of meatings reader here and i not time or good english skill to rewritten this, but if still intresting and see possible usable ASCII-chart, picture and formula you can look:
https://groups.google.se/groups?q=t...=hRQ6b.27574$dP1.73396@newsc.telia.net&rnum=1

(i fix up mostly of links to meaningsful picture now)

remember mostly of picture using simple model of CAT-5 as RC-links
but checked with transmision line mode, but tok so much simulator time (hour) is not make picture on it and more or less only differencer is a output signal is time delayed compare to simple RC-models.

twisted pair audio frequencies

Just to confirm if I understood right, i should send L+Gnd over one twisted pair and R+GND over another both with a capacitor in parallel on the source of the signal, depending on what the distance is like you refered, correct?
No need for a capacitor on the receiver end as the one of the source is compensating for all the loss right ?

sending video over twisted pair

Microemission said:

Just to confirm if I understood right, i should send L+Gnd over one twisted pair and R+GND over another both with a capacitor in parallel on the source of the signal, depending on what the distance is like you refered, correct?
No need for a capacitor on the receiver end as the one of the source is compensating for all the loss right ?

Click to expand...

Yes i have R+GND and L+GND in separate pairs - never using common gound if possible if you want best possible isolation between channel.

you want so low capacitance as possible, more capacitanse - more driver problem on sender side depend of mostly OP-amp cannot give unlimited current to load
in fast changing - and make possibly stability problem and cross over distorsion with high cap load depend of slow response of feedback in time to switch over push or pull driver in OP-amp (I have bad experences of cheap LM 324 on cap loads on line...)

for longer distance (> 500 meters) you need different strategy (and matching) depend of current limit of source - this solution works only on short distance.

----

My schedule using 1 kOhm to output and 10kOhm to input, but if you using different
fraction betwen output and input impedance you also need adjust compensation capasitance for new value - and suprising, compencate capacitance incrase in value if output resistanse goes down (and load still 10 kOhm) ie fraction between source and load incrase...

(oscillocope probe have opposite situation ie source is a high impedance and load have low impedanse and result parallell capacitanse of source(probe) is 1/10
of cable and load capacitance)

remember you have resistive voltage divider and capacitive voltage divider and
fraction between source and load must be hold equal both on resistance and capacitance load.

value desciribe above have ten time more compencate capacitanse compare
to sum of cable capacitance for 1:10 fraction as resistive voltage divider 1 k source and 10 k load

try this on spicesimulator with different value and see itself, try also using 0.2 -1 nF on 10 meters and see ugly result of compensation,
and try also to make balanced 'RC-ladder' and see what happend:

Unbalanced version schematic:

**broken link removed**

(each RC-link 0.188Ohm/52pF reprecent 1 meters of CAT-5 cable )

and for 100 meters without compencation( C1 = 0):

**broken link removed**

and 100 m with compensation (C1 = 51.97 nF) :

**broken link removed**

C1:s '51.97' nF is around ten times more cap than totale cable capacitance
on 100 meters cable (5.2 nF).


and 500 meters...

uncompencated:
**broken link removed**
compencated:
**broken link removed**

and explain why 11200 baud RS232 not working on long lines....
her is only 10 kHz square wave...


---

Home stereo is unbalanced system unfortly, but I handled same way i a balanced system ie on long lines transform unbalanced system to balanced and different pairs have not same referense as 'GND' in each end depend of inductanse coupling and resistanse in cable itself. If you have third ground loop outside signal cable (even current in cable screening folie) is easy to make 'hum' and noise from house power grid in unbalanced system

balanced system is best soloution do attenuate sourrounding noises
and possible to make from unbalanced system with 1:1 transformer for galvanic insulation.

/Xxargs

calculator twisted shielded pair impedance

I for a very similar application, I have used a simple circuit using the DRV13/INA134, two chips available from TI specifically designed for pushing audio through balanced lines. For a short run of 20 or so metres, the simple schematics shown on the datasheets work well, but I add a bit more rail decoupling to the driver chip to cut down distortion.

h**p://focus.ti.com/docs/prod/folders/print/drv134.html

h**p://focus.ti.com/docs/prod/folders/print/ina134.html

Once I work out how this board works,( ), I could upload a schematic if you want.

calculate twisted shielded pair impedance

For low cost solutions you may look at:
https://www.edn.com/archives/1995/081795/17di3.htm
**broken link removed**
https://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1009,C1044,D4146

audio video over phone line

xxargs said:

low educated Mr Pulpin invented if insert extra length inductance on line to compensate cable capacitanse (and blamed very much from scientific socity for this 'stupide idea' - you make a low pass filter!!!, and missing whole point) and decrace attenuate very much on low frequency, but cost more attenuate on high frequency depend of cable capacitance is a distrubed and inserted extra inductanse is discrete points.

in low frequence see very many of 'inserted inductance' in wavelength and accept as distrubed to give real impedances and no extra attenuate on little higher frequency

but high frequency see wery few 'inserted inductance' as dicrete points in wavelength and make daisy chain-coupled low pass filter. Pulpin balanced this to give very low and flat attenuate up to around 4000 Hz and frequecy above attenuate very quickly - enougt to make usable long distance phone call .

Click to expand...

Possibly little offtopic now, but for illustrate above i make two simulation chart to point out how importent mr Pulpins Invent for the old telephone industries:

network terminated to 600 Ohm in both ends.

Notice: Pulpin coil is _very_ essential for transporting Audio LF over long line and easy to gain 20 dB with amplifier to compensate loss everey 100 km phone line, but without pulpin coil you need gain 100 -120 dB every 100 km and practical impossible depend of noisfloor in amplifiers...

simple low frequency audio compensating projects

Just a few corrections. Inventor of pupin coils was Dr. Mihailo Pupin and not Pulpin.
He was well educated man. Few lines from his biography:
Michael (Mihailo Idvorsky) Pupin was born in Idvor, Banat, October 9, 1854. Received his academic education at Columbia College, New York, and his scientific education at the universities of Cambridge, England,and Berlin, Germany. Obtained his Ph.D. degree at the University of Berlin, and returned to Columbia University in 1889, to take a position as instructor in theoretical electrical engineering.
More about him and his patents you can find here:
**broken link removed**
I wish you to be well educated Mr. xxargs.