Netmux audio breakout Circuit Details: Difference between revisions

From YSTV Documentation Wiki
(Straight from the old website, needs files adding)
 
(No difference)

Revision as of 11:01, 6 October 2011

These are the individual sections of the unit, show as blocks on the NetMuxBreakoutBlockDiagram.

First a few numbers on levels:

  • Going from balenced to unbalenced with a "unity gain" differential amplifier (i.e. one that turns a difference of 1 into an output of 1) gives a doubling of the signal level. Why? because in a balenced setup the quoted level is on each line, and they are in antiphase, so the difference is double that.
  • +4dBu is 1.23V RMS
  • -10dBV is 0.316V RMS
  • going from +4dBu to -10dBV is attenuation by 3.9 times, or conversly going from -10dBV to +4dBu is a gain of 3.9. Remember the reference levels are different!
  • PC output line levels are often well below -10dBV, at 100mV RMS. So gain up to +4dBu is 12.3.
  • TVs and VCRs nominally need about 1V RMS of signal (historic, it's the same as the video sensitivity). So debalenced +4dBu is a bit too high, but -10dBV is too low.

Debalancer, Unity Gain

This is used to get from balenced +4dBu signals to an unbalenced signal at +8dBu (2.46V RMS). This is then suitable for passing through a gain control to either 1V RMS or -10 dBV equipment. The latter could use a lower gain debalancer, but it's easier to have them all the same.

debalencer.png

The CMRR is about 35dB with ordinary resistors, which is good enough for line-level cabling within the station. Variable Gain Buffer

This is used on the output of the debalencer for variable gain unbalenced outputs. The advantage over a passive potentiometer is that the output impedance is low and fixed (gain is constant with load variations). You can't do this and the debalencer in one unit without using a dual gang potentiometer, and even then it would need to be a very well matched one for the diff amp to work at all well.

vargain.png


Balancer, Unity Gain:

This again gets added to another block in order to convert a low-impedance unbalanced signal to a balanced one. It's just a unity gain inverting amplifier for the cold pin.

balencer.png Balancer, Variable Gain

Just a variable gain non-inverting amplifier followed by the balancer. Because it's a non-inverting amplifier minimum gain is unity, but for inputs that's not a problem. Maximum gain is 11 set by the ratio of VR1 to R1.

varbal.png


Balanced Summing Amplifier:

This makes a balanced stereo signal into a balanced mono one, at the same signal level assuming both inputs are driven. If only one is driven, then you get half level on the mono output.

balsum.png

This will degrade the CMRR of the following input stage if the hot and cold stages have different gains (because common mode signal gets converted into differential signal). The alternative is to debalance, sum and re-balance, which then limits the CMRR to the CMRR of the debalancers. It uses 4 opamps and restricts CMRR to about 35dB (as above). The CMRR reduction of the curcuit given depends on the differential gain of the subsequent input, but shouldn't be any worse. Power Supplies

None of the above diagrams show power supplies. Everything is meant to be run off +- 15V rails, and decoupling will be needed at each physical IC. AC / DC Coupling

No rude jokes please. All the circuits shown are DC coupled, which means they are prone to accumulating and amplifying DC offsets and not working. However, as I think everything else will be AC coupled into the system this shouldn't be a big problem. None of the circuits has a large gain, and so the output offsets should be little larger than the input offsets. The exception is the gain stage for the PC line inputs up to +4dBu. This also has very high Zin as show, so a 100nF capacitor and a 100k resistor to ground (providing the DC bias path) should work fine.

Other inputs are mostly around 10k Zin, and so would require much more troublesome 1uF capacitors. Outputs need bigger capacitors again, so if possible should be avoided. I can't see the net mux or the compressor being bothered by a few 100mV of DC offset, so only the outputs of the whole unit need looking at. Providing that the compressor removes the DC offset, then all the output signals are attenuated or at unity gain from this point, so no significant DC offsets will be introduced. The main output to LTC will probably have to be capacitor coupled as well as surge protected for the long cable run, so it's drawn separately on the NetContributionAudio page.

(X)Fig files for the diagrams:

[debalancer] [vargain] [balancer] [varbal] [balsum]