unbalanced to balanced

[ilfungo]

hello
I would do this:

"Just wire a 100 uf cap and a 47 ohm resistor to ground. Use the other end as the - output. There, you now have a asymetrical balanced output. It will output +21 dbu"


Can anyone help me a drawing?
I'm not sure of the connections
Thanks

 

[moamps]

here you go:

 

[ilfungo]

Thanks moamps!
C1 C2 r1 r2 value?
Thanks

 

[pucho812]

Thanks moamps!
C1 C2 r1 r2 value?
Thanks

didn't you give the cap and resistor values in your first post o the thread?

 

[Speedskater]

In the above drawing, connector 'X1' a 'NC3MAV' should not have pin #1 or the shell 'G' connected to circuit ground 'GNDA' they both should be connected only to the chassis very near the chassis.

 

[ilfungo]

Hi
I had found something like this, but I do not remember ...
Was defined Asymmetrical Balance

 

[Andy Peters]

Thanks moamps!
C1 C2 r1 r2 value?
Thanks

R1 and R2 set the line impedance; a common value might be 50 ohms or 25 ohms. They should be well matched.

The caps set the LF corner, since they're DC blocks. Do some math and see where they start to work. 47 uF and 100 uF are common values.

-a

 

[moamps]

@Speedskater

The body tab of the plastic XLR connector is electrically connected to the chassis via one mounting hole and screw.  So, the  pin "1" is connected to the chassis using shortest way possible.
Here is one more important question you are overlooked.

@ilfungo

The most common resistor's value for R1 and R2 is 47 ohms. Also, 22 and 75 ohms resistors are often used, 10 and 100 ohms rarely.
The most common capacitor's values for C1 and C2 are 47, 100 and 220uF.....   

 

[abbey road d enfer]

The most common resistor's value for R1 and R2 is 47 ohms. Also, 22 and 75 ohms resistors are often used, 10 and 100 ohms rarely.
The most common capacitor's values for C1 and C2 are 47, 100 and 220uF..... 

It's not a matter of habit. There's a scientific reason for the values used here.
Ther are two branches, one that includes the output impedance of the opamp. Contrary to basic theory, the opamp's output Z is not zero in practice and variable with frequency. So the resistor must be suffficiently high to make the opamp's output impedance negligible. Since the actual opamp's output Z is a fraction of an ohm, almost any value above 1 ohm will satisfy this condition.

So now, what governs the value of the resistor?
Primo, it must be low enough to provide a low source impedance to the load it feeds. It could be 300r if we wanted 600-ohm-matching, but today the concept is bridging, where the source impedance is << the load impedance. Typical values are indeed 100ohms into 10k. That's probably why 47r is the most common value.
Secundo, the resistor at the output of the opamp decouples the opamp from the capacitive load presented by the cable. For some reason (I suspect historic habits of RF engineers) 50 ohms is generally the value used in the application notes.

There are electrolytic capacitors, and we all know their actual tolerance is quite high, sometimes as high as -20/+50%, and varies with age. So the value of the capacitor must be made high enough to provide good LF response AND to make its impedance at the lowest frequency negligible compared to the resistor.

When capacitors are not matched, one can see that the CMRR gets degraded. Bill Whitlock has perfectly demonstrated that the CMRR of a balanced connection is governed by the actual impedance discrepancy between the two legs of the connection.
Assuming the values 47r and 100uF, with a 20k load impedance, CMRR at 50Hz degrades down to 70dB with only 10% difference between caps (with perfectly matched caps, CMRR is theoretically infinite). It is not uncommon to see elcaps value changing by -50% with age; in that case, CMRR@50Hz may fall down to 45dB.
Although 47uF satisfies the LF response condition, IMO it doesn't leave enough margin for tolerances and variations regarding CMRR.

 

[ricardo]

There's another important reason for 50R.  Good cable has a characteristic impedance of 45R. http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=464 has details and also 'Everything you wanted to know about Impedance Balanced outputs but were afraid to ask'.

'Afraid' cos I was hang, drawn & quartered for suggesting this heretical O/P stage 

 

[Speedskater]

I'm not at all sure just where you first read that good analog audio cable had a 'Radio Frequency Characteristic Impedance' of 45 Ohm.  Was that unbalanced coax or balanced twist pair?
If it's coax that you refer to, then you will find that 75 Ohm coax has a lower capacitance per foot/meter that is often considered a good thing, also it has lower cost per length and is more readily available.
If it's balanced twist pair (shielded or unshielded).  Nobody thought much about it's Characteristic Impedance until Cat3 to Cat6 and AES/EBU digital audio came along.

But in either case 'Radio Frequency Characteristic Impedance' has absolutely nothing to do with analog audio cables.

 

[Speedskater]

More on cables:

"Using Digital Audio Cable For Analog Audio"
Posted by: Steve Lampen on May 10, 2012
http://www.belden.com/blog/broadcastav/Using-Digital-Audio-Cable-For-Analog-Audio_301156.cfm

"50 Ohms: The Forgotten Impedance"
Posted by: Steve Lampen on August 27, 2012
http://www.belden.com/blog/broadcastav/50-Ohms-The-Forgotten-Impedance.cfm

 

[abbey road d enfer]

There's another important reason for 50R.  Good cable has a characteristic impedance of 45R.

We're talking about balanced outputs, then we should talk about balanced cable, and then we all know that good balanced cable has a characteristic impedance of 110 ohms, close enough to 50 ohms times two.
That said, we also know that this becomes relevant for cable lengths greater than 4km (2.5 miles) - one-quarter wavelength at 20kHz. When matching impedance becomes significant, the load impedance should also be 110 ohms.
In these conditions, I think I would think of something a little more elaborate than the proposed circuit. A standard opamp is not the ideal driver for such a load.

 

[JohnRoberts]

I'm not at all sure just where you first read that good analog audio cable had a 'Radio Frequency Characteristic Impedance' of 45 Ohm.  Was that unbalanced coax or balanced twist pair?
If it's coax that you refer to, then you will find that 75 Ohm coax has a lower capacitance per foot/meter that is often considered a good thing, also it has lower cost per length and is more readily available.
If it's balanced twist pair (shielded or unshielded).  Nobody thought much about it's Characteristic Impedance until Cat3 to Cat6 and AES/EBU digital audio came along.

But in either case 'Radio Frequency Characteristic Impedance' has absolutely nothing to do with analog audio cables.

Characteristic impedance is not a radio frequency thing, but admittedly termination with characteristic impedance is a little silly for short audio frequency cables that drive bridging (high impedance) loads.

My philosophy about this is I'm going to put a build out resistor there anyhow and it cost the same for a 20 ohm resistor as 200, so why not use the characteristic value? I've been using around 50 ohms for decades. While I do not claim any audio difference, it just makes me feel better about it.

I find in routine design there are many arbitrary value decisions that don't have to be completely arbitrary, even if only to amuse myself.

JR

.

 

[ricardo]

I'm not at all sure just where you first read that good analog audio cable had a 'Radio Frequency Characteristic Impedance' of 45 Ohm.  Was that unbalanced coax or balanced twist pair?

Post #4 of the link I posted shows the spec for Belden 9451.  Later in that thread are measurements of how terminating 400m of this affects transient response.

You may or may not regard this as necessary but the results are there for you to see and extrapolate to your own lengths.

Ask your pro-audio cable maker for more details.

If you are using Golden Pinnae cables, please ignore anything I say but ask for proof that your cable is hand carved from solid Unobtainium by virgins.

 

[abbey road d enfer]

There's another important reason for 50R.  Good cable has a characteristic impedance of 45R.

For complete transparency, you should mention that in that very thread, you've been corrected and you had to agree that the "balanced" impedance that must be taken in consideration is ca. 100R.
In that respect, the characteristic impedance that appears on the spec sheet should be referenced to the test conditions and method of evaluation. It is probably somewhere in Belden's vault.
The test made on 300ft of said Belden cable were done with 10kHz square-wave. The 'scope pix indicate the first visible effects are at about 3-400kHz, which is consistent with my estimation that transmission-line effects would be noticeable in the audio BW for cable length > 2.5 miles.

Please be assured that I'm not attacking you personally (although it may look like it) but my duty as moderator is to make sure that assumptions that may look controversial are presented in the most impartial and circumstanciated way.

 

[ricardo]

There's another important reason for 50R.  Good cable has a characteristic impedance of 45R.

For complete transparency, you should mention that in that very thread, you've been corrected and you had to agree that the "balanced" impedance that must be taken in consideration is ca. 100R.

I did mention I was hung drawn & quartered in that thread  :-[

That made it impossible to persuade Wayne to move his testing into more productive directions .. in particular, 'real life' RFI/EMI immunity of various output schemes.  Ro is an important aspect of this hugely important but little investigated field.

IMH & Jurassic experience, the various feared EN tests don't go anywhere near enough to predict 'real life' acceptability in a pro-audio environment.  But things may have changed in the 2 decades since this was of great importance to me.

I did get 2 valuable morsels from my traumatic experience thanks to Wayne's detailed work

• good performance of THAT 16xx series compared to its SSM & TI predecessors
• making Ro 45-100R will make certain measurements look better.  Alas, it is unlikely that I will be able to test if these result in better RFI/EMI performance this Millenium
  But using 45-100R makes life easier for my El Cheapo O/P stage ... even if it may not help EMI/RFI 

But I sorta expect people to do more than glance at the title of a thread I link to.  The Belden spec was on post #4  :

 

[JohnRoberts]

Since large scale MFG gets tested for the real world by the real world, in the real world,  ineffective practices get exposed and generally corrected. Some small manufacturers may get away with treating RFI/EMI as isolated events, while their names pop up from time to time in sundry forums.

JR

PS: Ricardo, there is never enough love to go around. The best you can do is lead the horse to water.

 

[abbey road d enfer]

• Ro is an important aspect of this hugely important but little investigated field.

  Agreed; Bill Whitlock inparticular demonstrated that the degradation in CMRR is directly related to the absolute error between the two legs. Thus lower nominal value means less absolute error.
  But not all noise is CM. Magnetically induced noise in the cable is constant whatever the source impedance. It would be easy to design a balanced output stage with a nominal 2 ohms impedance. Wheteher it would actually improve real-life performance is to be seen.
  Similar to your experiment, I "designed" the audio distribution for the pope's venue in Lyons eons ago. I used a Crown D75 in bridged mode to drive 14km of heavy-duty mic cable via a pair of 4.7 ohm 10W resistors. I didn't check the 10kHz square-wave response  but the FR was largely adequate. Nominal level was about +20dBu so noise performance was correct. Some kind of a steamroller approach.
  In conclusion, indeed, reducing the nominal Z provides better RFI/EMI rejection, but at the price of making the circuit more complex (a standard opamp cannot guarantee unconditional stability on capacitive loads with only a 5 ohms resistor). The mismatch between an actual 10ohms source Z and a 100ohms cable is a very remote concern for me.

  IMH & Jurassic experience, the various feared EN tests don't go anywhere near enough to predict 'real life' acceptability in a pro-audio environment. But things may have changed in the 2 decades since this was of great importance to me.

  Indeed things have changed. In 1976 nobody could have imagined everybody and their little sister would carry a 2-Watt jammer in their pocket. At the time we could get away with some LP filtering at the inputs and a little shielding. Today, the RFI/EMI principles are better known (although there is still some confusion) and we benefit from excellent ready-made building blocks.

  • good performance of THAT 16xx series compared to its SSM & TI predecessors

  Yes, they are members of the aforesaid building blocks.

But I sorta expect people to do more than glance at the title of a thread I link to  :

Do you mean the proaudiodesignforum link or the "Everything you wanted to know about Impedance Balanced outputs but were afraid to ask". The latter cannot be found with google...

 

[ricardo]

Since large scale MFG gets tested for the real world by the real world, in the real world,  ineffective practices get exposed and generally corrected. Some small manufacturers may get away with treating RFI/EMI as isolated events, while their names pop up from time to time in sundry forums.

There are very expensive famous name 'professional' active monitors that will loudly announce to all & sundry that you have received an SMS message.

Also old famous name mike makers whose current models suffer from classic Pin1 syndrome.

FWIW, JBL LSR25P active monitors and Schoeps/Calrec mikes are OK

It would be easy to design a balanced output stage with a nominal 2 ohms impedance. Wheteher it would actually improve real-life performance is to be seen.
Similar to your experiment, I "designed" the audio distribution for the pope's venue in Lyons eons ago. I used a Crown D75 in bridged mode to drive 14km of heavy-duty mic cable via a pair of 4.7 ohm 10W resistors. I didn't check the 10kHz square-wave response  but the FR was largely adequate. Nominal level was about +20dBu so noise performance was correct. Some kind of a steamroller approach.

I have no doubt a crude 'balanced' LoZ network works cos I did the same as you.

I used TDA2040 and drove my line directly without intervening resistors.  It is one of only 2 amps I know which are unconditionally stable with load without an output inductor network.  The TDA2040 protection made the whole system very robust.

When you have theatre lighting buzz problems, 1 ppzillion THD & transient response fades into insignificance.

What I don't know is whether similar RFI/EMI immunity is possible with THAT16xx etc & 'proper' line matching.  If I was more diplomatic and not so stupid, I might have persuaded Wayne to do the tests. 

Do you mean the proaudiodesignforum link or the "Everything you wanted to know about Impedance Balanced outputs but were afraid to ask". The latter cannot be found with google...

That was my tongue-in-cheek title for the proaudiodesignforum link.