RCA to XLR cable off USB DAC -- reduce noise?

I could be wrong about this, but I believe that 'balanced' cables are only less subject to interference if the equipment at each end is also balanced.

Usually true.

Also, PCs are a harsh noise environment with noise both radiated through the air and coupled on ground and power lines. A lot of times the culprit is the ground line, or supply line if not filtered, from the PC to the DAC. The best cable in the world won;t help that, unfortunately, you need some way to isolate the PC's USB from the DAC. Some DACs are opto-isolated and use their own power supplies.

Other things you could try: ferrite beads (available from Radio Shack and I assume Monoprice, Part Express, etc.) that clamp on the cables to reduce HF noise coupling, and re-routing (separating) the cables so they are not in proximity and thus couple less. I would try moving cables around first to see if that might help. You could also try moving the sound card into another slot.

No, the presence of a balanced input is far more signficant than a balanced output.

There are three ways to drive (output to) a balanced line.

(1) Most ideal is a true active balanced output. One of its benefits is that it can apply twice the signal voltage to the cable as the other alternatives.

(2) Less ideal but surprisingly effective is the impedance balanced source. One wire of the balanced cable has a signal voltage applied to it with a given source impedance. The other wire is connected to a resistor with the same impedance, but the resistor just goes to ground, no signal.

(3) Least ideal but still more effective than an unbalanced input is connecting one wire to a signal source, and simply hooking the other wire to ground on the same piece of equipment.

In case 3 the balanced input still provides an advantage because it responds to the difference between the signal voltage and the source component. If there were no wire connected to ground at the source, the input of the equipment receiving the signal would respond to the difference between the signal voltage and ground at the receiving end. There can be a voltage drop across the ground side of the interconnect due to strong ground currents flowing in the wire of the ground side of the interconnect and it will add to the signal received at the receiving end.

This document sheds more light on this topic:

http://www.jensen-transformers.com/an/an003.pdf . Item 2.1 describes the approximately 30 dB (minimal) benefit of connecting a balanced input through an unbalanced cable - minimal, but beats nothing!

No, differential transmitters connected to differential receivers reduce common mode noise.

For those of us who just want simple answers, do commonly available RCA -> XLR adapters or cables provide either option 2 or 3 to us, or would we have to cobble something together on our own?

Whether you are right or wrong, I'd seriously go about fixing any "site wiring fault" situations before using my equipment.

no

I use RCA to XLR adaptors from time to time and they provide option 3 as long as the cable and input are XLR, and presuming that the XLR input is truly balanced, which is occasionally not true.

Actually, a transformer could isolate the grounds, up to the frequency at which the transformer ceases to isolate, but I am not sure the simple in-line units do that. Some DI boxes do that.

Nothing will help if the noise is inside the PC, coupling to the sound card, which was my original thought/fear. The two solutions I have used for that were an external sound card, or a high-end audio card that was shielded and isolated.

This is correct.

A balanced cable has no advantage unless it is connected to balanced circuitry at both ends.

The advantage of balanced circuitry is that the two signal wires carry signals that are 180 degrees out of phase with each other, and the ground/shield carries no signal.

Any noise that is induced into the signal wires is induced into them equally, with the same phase, which cancels and is nulled out when the signals reach the balanced circuitry at the receive end of the cable. The balanced circuit only sees and amplifies the out-of-phase signals, the DIFFERENTIAL signal, and is totally impervious to the noise, which is NOT a differential signal. The noise is what engineers refer to as a "common-mode" signal, which the balanced circuit essentially does not respond to.


This is why balanced cables can be run from a microphone, which often only puts out a few microvolts of signal, 100 feet or more to a balanced preamp/mixer unit, and have essentially zero noise. This is common practice in large auditoriums and similar recording venues.

If you tried that with an unbalanced cable, the noise might be louder than the microphone; bad hum and noise in most cases. 99% of the the signal voltage is on one wire, and this is easily contaminated with noise from many sources. From an engineering standpoint, an unbalanced cable is inherently a terrible design.

Unbalanced cables should always be kept under five feet if possible, and not used at all if possible.

Professional audio equipment always uses balanced circuits for this reason, and also top-quality home audio gear.
Edited by commsysman - 2/4/13 at 9:18pm

^^^ This.

and to add:

The cable in a balanced system needs to be twisted-pair, so that any noise induced into the cable is induced into both wires equally. That way, they are fully subtracted-out at the balanced receiver.

Many microphones inherently produce a balanced signal. So do transformers. So a single-ended signal can easily be converted to a balanced differential signal with a signal-transformer at the single-ended output. The inverse can be performed at the receiving end.

It is important to realize that an unbalanced cable is not connected to a TRUE ground at either end, but a SIGNAL ground, which is not at true ground. A close look at the internal circuit of unbalanced equipment will demonstrate this. The chassis may or may not be connected to a true ground, but the signal connectors never are (and never CAN be).

A close look at the circuit of electronic gear will show a resistor connected between true ground and the signal ground of the electronic circuit, which "floats" the signal ground of the circuit.

To have a signal, equal signal currents must flow in both the center conductor and the shield. One conductor does not make an electrical circuit, and cannot pass a signal.

If both ends of the unbalanced cable were TRULY grounded, no signal current could be induced in the shield, and no signal could pass through the cable.

This is why the unbalanced cable is so vulnerable to noise; the so-called shield is not at true ground and is therefore an ineffective one.

An interesting thing to consider is a telegraph circuit of the type developed in the 19th century. Only one wire is used to carry the DC current when the key is closed , it would seem on first glance. In reality, however, the earth has the proper resistance over a typical distance of 40 miles or so that the return current can be induced in it and the earth is connected to one side of the high-voltage telegraph battery at the sending end. The earth itself becomes the "signal ground" and carries the return current. This circuit is completed at the receiving end by hooking a receiver key between the telegraph wire and a ground rod.

Distances of over 40 miles or so have TOO MUCH earth resistance, however (as well as wire resistance), and the current becomes too weak. For this reason, telegraph messages initially had to be sent 40 miles and written down, then sent the next 40 miles and so on. Later on, they figured out how to mechanically tie a receiving key to a transmitting key so that a message could be passed from one 40-mile telegraph circuit to the next 40 mile circuit almost instantly, and so on for hundreds of miles if needed, passing through many circuits quickly.
Edited by commsysman - 2/4/13 at 9:22pm

If by simple in-line units you mean this:

http://www.radioshack.com/product/index.jsp?productId=2062214



Then it is something that has been on my test bench and also something that has received a fair amount of real world use. It is a very good example of what it appears to be.

It does well in the measurement department.

Besides my own "never fails" positive experiences a number of people on the AVS forum and usenet also report good results.

That is a false fear, or else all of the SOTA audio interfaces that live inside PCs like those from Lynx Studio, Apogee, Digidesign, and RME would not have the great industry acceptance that they have.

Heck even the better M-Audio and eMu cards are quiet.

These things claim > 110 dB dynamic range some near 120 dB and based on my personal experience and that of others, they meet spec. Many of these cards have no visible shielding at all.



Interestingly enough one sees shielding on wannabe consumer cards. I presume it is there to assuage consumer angst, not cover up a weak design.



It's all about mixed signal design which has to be used to even make the simplest DAC work well, since one side of the chip is analog audio and the other side, often as little as 1/8" away is sharp-edged TTL.

1. I was thinking of the small adapters that are basically inside the XLR housing; not all isolate ground but just short the XLR (-) to RCA ground and the shield. Very annoying. That one (your RS model) does provide isolation ("ground loop isolator" implies that) as do most pro units. Have you measured it, would you mind posting the results or a link? I am just curious.

2. I am not worried about SOTA cards, I agree those are fine (I said I went to better cards and that worked for me). I was thinking of the $10 cards... In any event, I did not look up what the OP has, looks like an external unit? I have seen ground and power noise coupled into external DACs when hooked into a PC, and read reviews of others. Seems like poor design to me, but cheaper. I do not know what the OP's DAC does for isolation.

The other issue with PCs is OS and application dependent; I have seen the USB signal "glitch" at times and that could cause clock jitter in the DAC. No idea how bad or how audible the effect might be; depending on the reviewer, devastating to totally inaudible.

As has been said, XLRs could help since they provide CM noise rejection (at least within the CMR range of the system) and a true signal shield. You could break the shield conduction path by attaching the XLR shield to chassis ground at only one end, probably the DAC end in this case since presumably that is the quieter side (probably opposite of what is usualy done, hooking the shield at hte source side).

I'd still try just separating and moving the cables around first.

No worries - Don

p.s. I designed DACs for a living, albeit for GHz systems and not audio, very familiar with their foibles. But, TTL? Dating yourself, Arny.

Back in 2006 did some bench tests of Radio Shack's "Ground Isolator" 270-054 using
test signals that maxed out around 2.5 v RMS. ZSource = 150 ohms, ZLoad = 5Kohms.

The measured performance was truely amazing for a pair
of transformers case and cables selling for only $16.65.

All IM, THD, and noise artifacts were at least 80 dB
down with most in the -100 dB range or better. Frequency response showed a
2 dB peak at 20 Hz and then 10 dB down at 10 Hz. There
was a 3 dB peak at about 51 KHz falling to about 10 dB
down around 100 KHz. +0.5 dB at 20 KHz.

I repeated the tests with the secondary loaded with
1.5K, and the peak at 51 Khz became well-damped with
only about 0.6 dB rise
Edited by arnyk - 2/5/13 at 1:17pm

Wow, thanks Arny! I am going to have to get a few of those to replace the old $150 jobbers wearing out around the church, plus a couple for my cable bag. - Don