Tomi Engdahl’s ePanorama blog

I really like those electronics manufacturers that have the schematics of their devices on their data-sheets or manuals down-loadable from their site. Those schematics can aid in deciding if the equipment is well built, help in repair, designing possible modification and can be aid when you need to make your own designs that have similar features.

I saw the schematic of Rolls MB15b ProMatch Stereo Dual Converter mentioned at some Internet discussion.

I took a look at the schematic. At the first look of the device it seemed to be pretty normal design, but some closer look at the schematic revealed some details that do not look so well designed. So here are my findings on the schematic.

Here is the circuit section that converts balanced signal from XLR input to unbalanced RCA output:

The input section for balanced input looks pretty much standard design. It is a pretty well working classic design. The data sheet tells that XLR input impedance is 10 k Ohm and RCA output impedance is 100 ohms. Those are OK values if they are true. But the truth is that the technical data provided is not entirely correct.

The input impedance of the XLR input pin 3 to ground is 10 kohms as promised. The impedance from XLR pin 2 to ground seems to be around 20 kilo-ohms (two 10 kilo-ohm resistors in series; operational amplifier + input is pretty high impedance). So the input impedance the differential between pins 2 and 3 would see would be around 30 kilo-ohms.

But on the RCA output side the technical specifications are far more off from reality. The output impedance of the RCA output would be the promised 100 ohms only in two specific cases: The gain is sent to maximum (100 ohms resistor connected to operational amplifier output directly) or when there output is set to zero (100 ohm resistor connects directly to circuit ground). In all other cases the output impedance would be considerably higher! Turn the 100 kohms potentiometer a little bit off the extremes and the output impedance would be soon several kilo-ohms. Put the potentiometer to center (electrical center where there is same resistance from both ends to center), and the output impedance will be 25100 ohms!

This is for sure too much for a well designed RCA output and the cable capacitance will affect the sound on even quite short cables. As rule of thumb is to keep the RCA output impedance few kilo-ohms or less. A better design choice would have been to use a 10 kohms potentiometer on the output (would have given maximum 2600 ohms output impedance). Or a separate buffer opamp after the original 100 kohms potentiometer.

Here is the RCA input circuit:

RCA input floating ground strange if you look at it. The RCA connector grounds are not directly connected to circuit ground, instead they are connected to ground through 47 kohms resistor. The ground potential from RCA connector grounds is fed to differential operational amplifier circuit. It could be a good working circuit in the applications where both equipment connected to this adapter are grounded. In this case this quasi-differential input arrangement works as a ground loop problem solving circuit (similar approaches have been used on some old PC sound cards for audio signal that comes from CD-ROM).

The problem in this grounding arrangement is that if the equipment connected to RCA input is not grounded, all the leakage current on it needs to flow through the 47 ohms resistors to ground. A even small current (fractions of mA) can create quite considerable noise voltages. They can easily get to sound if the opamp differential circuit are not ideal. Also a leakage current higher than around 0.2 mA can create so high voltages over 47 kohms that the opamp input can’t handle them without distortion. So I don’t think this is an ideal input design.

In some other design version (older or newer version I don’t know) they use a different design.

The basic ideas used on the design are pretty good but the implementation lacks something….

Star-quad design is sometimes quite a bit of talked about cable type. When I saw Star-Quad Cables and Double-Blind Testing posting at Controlgeek blog I decided that it would be a good idea to write something about them.

Several companies use a quad configuration in their balanced cables instead of a twisted pair; Canare calls this Star Quad. The design uses four conductors twisted around each other; opposing pairs are connected at either end.

The star-quad design for microphone cables, where 4 conductors are used in pairs, with a short lay length, to cancel out the effects of external electromagnetic noise, is well known in demanding applications such as TV, video and theatre where thyristor lighting control systems are in use. Star quad, four conductors in square formation, in which opposite conductors are connected in parallel at both ends to form a single, magnetically-concentric balanced pair, in order to minimize magnetic induction. The truth is that it works. I have tested star-quad cable arrangement against normal twisted pair cable, and I have to admit that it rejects the noise better when I have tested that on lab environment (comparing several cable constructions).

A star-quad cable resembles kind of super-set of a normal twisted pair microphone cable. There are four cores in star-quad cable plus an overall screen. The screen helps to prevent electrostatic interference from reaching the signal cores, just as in a conventional cable. What’s special about star-quad cable? article tells that the four cores are very tightly twisted with each other and rotate over a relatively short length. The very short lay length of the core twisting also helps to make sure that any interference is induced equally into the the two paired sets of wires (although works quite well even if the twists are not very tight like in Gotham “Star-Quad”). This arrangement results in a magnetically coaxial structure and provides greatly improved rejection of electromagnetic interference compared to standard two-core mic cables.

The result of start-quad construction is better balance of interference and therefore better rejection. What’s special about star-quad cable? article mentions that the improvement is perhaps as much as 20dB. Up Your Audio: Time For Star Quad Microphone Cable article tells that star-quad cable is raising common mode rejection (CMRR) by a factor of 10, or about 20 dB.

The rejection is better especially for very close-source electromagnetically induced interference. You get this kind of interference when you lay a mic cable alongside something that radiates strong and nasty interference (mains power cables, dimmer cables). Star-quad cables are widely used today, particularly where there is a likelihood of having to lay mic cables alongside strong sources of interference.

Quad cable provides increased immunity to radio interference as well. The “Star-Quad” concept is known and recommended where the RF-rejection is the most important factor and where very long cable runs are needed. The star-quad configuration was invented for use in telephone cables back in the ’30s, and was adopted in the TV world in the ’60s and ’70s.

So how does the star-quad cable differ from normal microphone cable in technical properties? According to Up Your Audio: Time For Star Quad Microphone Cable article a star-quad mic cable contains four 24-gauge wires instead of the usual single 22-gauge pair. The four wires spiral together tightly so they alternate plus and minus, creating better “looping” that improves resistance to electromagnetic interference.

The four cores are arranged in a cross formation, and the opposite cores are wired in parallel to form a single balanced pair. The very short lay length of the core twisting also helps to make sure that any interference is induced equally into the the two paired sets of wires. Star-quad cable works very well with balanced audio signals.

Star quad construction reduces the inductive reactance of mic cable: Replacing a 22-gauge twisted pair with two pairs of 24-gauge conductors connected in parallel produces the same DC resistance, but about half the series inductance.

The downside is higher capacitance: with four wires in the cable, the core-core and core-screen capacitance will be higher than in a more conventional two-core cable.

If this provides better clarity without the need for EQ to boost the highs or causes loss of high frequencies due capacitance could depend on which kind of devices are connected to the cable. I think the effect could be in either case by case.

Up Your Audio: Time For Star Quad Microphone Cable article recommends to consider a new set of star quad microphone cables as the most affordable way to improve sound. According to that article nothing that costs so little can make such a big difference in the quality of a sound system as its mic cables and using phantom powered mics places a higher demand on mic cables than simple passive devices. According to Up Your Audio: Time For Star Quad Microphone Cable article claims that star quad cables can also improve sound quality. Most report clearer highs, less mud in the mid-lows, and a lift of the “haze” associated with previous mixes. I can’t verify those claims. I have not used that so much in real life audio systems that I could say does it sound better like one article claims. If you have noise problems then considering star-quad cable could be a reasonable idea. I don’t think that they could really make things sound better, but they sure can reduce interference.

Allen Avionics Audio Isolation Transformers page describes the typical ground loop problem situation with the following drawing:

If you want to do the ground loop elimination in audio path, you have to cut the galvanic connection but pass the whole audio range. The simplest and most common way to do the isolation is use audio transformer. High-quality audio transformers cover whole audio band with good response.

Usually when you install an audio galvanic isolation to almost any place in the audio cable connection is solves the ground loop caused humming problem nicely.

Xitel Ground Loop Isolator page gives a good example how to use a ground loop isolator (audio isolation transformer) to solve ground a typical ground loop problem:

To make an audio isolator you have to get two audio transformers which have 1:1 transformation ratio and greater than 1 kohm impedance (prefererably many kilo-ohms for consumer line level applications). There are high quality audio transformers in the markes that meet those specs.

There are several ways how to wire an audio isolation transformer. Here is a basic wirings for isolation transformers for unbalanced and balanced lines:

There are variations how ground are wired on different audio isolation transformers. There are various ways with their advantages and disadvantages. The followinf solution circuit diagram from Allen Avionics Audio Isolation Transformers page conbines all different variations to one device, so you can always easily select (with a switch) the one that works best on you specific application:

The 3-position switch will eliminate the need to cut pin “1″ to lift the audio ground and the connection to the shield. Switch Positions:

1. Chassis shield connected. Audio ground not connected. Acts as a balanced line transformer to break ground loops. Eliminates hum and buzz.

2. Chassis shield and audio ground connected. Corrects balance on audio lines.

3. Chassis shield and audio ground not connected. Breaks ground loops. Provides audio isolation.

When selecting a transformer for your project keep in mind the quality. Cheapest transformers for audio use have some problems on the sound quality, typically distorted bass response and attenuating in high-frequency response. Basically a poor transformer slows down high frequencies more than low frequencies (because of transformer leakage inductance). This allows the low frequencies to pass first, creating what we perceive as a “fat/warm” tone. Inadequate frequency response on the low end (roll off at like 20Hz), causes low frequencies to be “slowed”, allowing the upper frequencies to be heard first, this is perceived as “barky/ brittle”.

When installing audio isolation transformer to an audio system HIFI people usually ask does that transformer affect the sound in some way and if it does something how much effect does it have. The truth is that the effect of adding an audio audio isolation transformer to a system that does not have ground loop problem could be anything from unnoticeable to normal listener to well noticeable. How much the transformer affects depends on the transformer quality and properties of the equipment between which the transformer is connected. So if there is no humming on the system, the transformer usually has some slightly negative effect on sound quality.

In case an audio transformer is installed to a system with ground loop humming, the effect of it will for sure be possible. The annoying ground loop humming will be heard by everybody. When the isolation transformer get rid of that, the would quality is very much improved at that moment (typically the negative effects of the transformer to sound are much smaller than the annoying humming is). It is a good idea to keep an audio isolation transformer on your tool box when you work with audio/video systems, and use it in you get ground loop problems.

Ground loop can cause considerable currents to flow on all cables on the loop. It is not uncommon to have a voltage of 1V RMS between the earth connections of power outlets that are wired separately back to the switchboard. This small voltage, with a total resistance of perhaps 0.2-0.5 Ohm, will cause a loop current of 2 to 5 Amps, all of which flows in the shield of the interconnect. This is sufficient to cause a voltage difference across the interconnect, which the amplifier cannot differentiate from the wanted signal. An earth loop will typically inject either a 50Hz or 60Hz hum into the signal.

You might have seen the following kind of device marketed for ground loop problem solving for USA markets:

The product page boasts with the following features: The Hum X Exterminator removes unwanted voltage and current in the ground line that cause ground loop hum. This noise reduction adapter simultaneously maintains a solid, safe ground. You no longer have to run your audio signal through filtering that results in loss of volume, tone, or both. Some devices simply remove or float the ground, which is never safe! Hum X removes the ground loop while leaving the ground and your signal intact. Easy to use and completely effective!

The question what comes to my mind when I saw this is how this thing works. Harmony Central Ebtech Hum-X Review gives some interesting details how this device is expected to work:

After briefly analyzing the input to output ground connections I found this to be very similar to a galvanic isolator that is widely used in the boating communities for ship to shore power connections. It appears that there are two diodes and a 1Kohm impedance that are connected in parallel from input ground to output ground. The parallel diodes are configured back to back so that there is a 0.7volt drop in each direction. Since I could not look inside the unit the diodes are an assumption on my part? they may have used transistors configured as diodes. The potential safety hazard here, is that one or both of these diodes could fail open leaving the user with no safety ground.

Non-isolating isolator article gives information how those boat isolators work: When a boat plugs into shorepower, the shorepower cord will often make an electrical connection between the underwater metals on all the boats that are plugged in, creating a risk of galvanic corrosion. A galvanic isolator is designed to prevent this by blocking DC currents with voltages that reach as much as -1.2 volts DC. This is achieved by installing two sets of devices known as ‘diodes’, with one set installed in the opposite direction to the other. There are two types of galvanic isolator, one with a device known as a ‘capacitor’ wired around the diodes, and one without. Without a capacitor, if there is AC leakage on the shorepower ground circuit that has a voltage above 1.2 volts AC, this AC leakage will ‘bias’ the diodes into a conductive state.

Narrowboat AC Electrical systems article give the following application example for boat galvanic isolator use:

When the isolator is used for galvanic corrosion protection we want to block the DC and let the AC pass though, so the quite large capacitor in parallel with the diodes is a good idea. On audio systems ground loop protection we want to block low voltage AC, so the version without capacitor is the right one to use (there could be some very small capacitors used on audio isolators for RF protection and sometimes resistors to pass low leakage currents in case installed to system where there is no ground loop).

Elliott Sound Products article Earthing Your Hi-Fi – Tricks and Techniques article give some construction details how A High Current Safety Loop Breaker Circuit (pretty similar to one believed to be inside HumX) could be built:

This circuit example has the ground isolator between the mains power ground and audio part zero voltage line. Here the circuit the current loop breaking the loop is done with the 10 Ohm resistor, the current is now less than 200mA, and the voltage across the interconnect will be very much smaller, reducing the hum to the point where it should no longer be audible. This is how the circuits work when the potential difference over the 10 ohm resistor is lower than the voltage drop of the diodes on the rectifier bridge (around 1.2-1.4V). The capacitor will pass high frequencies (RFI protection).

In case there are is some serious ground leakage the diodes will start to conduct and pass through enough current needed to burn the mains fuse if needed without too much vooltage drop (no dangerous voltages over the isolator circuit). In the event of a major fault, one (or more) of the diodes in the bridge will possibly fail. Semiconductors (nearly) always fail as short circuit, and only become open circuited if the fault current continues and ‘blows’ the interconnecting wires. High current bridge rectifiers have very solid conductors throughout, and open circuit diodes are very rare. Use of the bridge means that there are two diodes in parallel for fault current of either polarity, so the likelihood of failure (to protect) is very small indeed.

If you plan to do any experimenting in this field, make sure that you find out the legal requirements in your country, and don’t do anything that places you at risk – either from electrocution or legal liability. Neither is likely to be a pleasant experience.

Electrical safety cannot be over emphasised. Hum is damn annoying, and everyone wants it gone. There is no good reason to sacrifice one for the other, since safety and hum-free operation can peacefully co-exist with care and the right techniques.

NEVER use a three prong to two prong AC adapter to fix a ground loop problem. These devices are meant to provide a safety ground (via the cover plate screw to a grounded outlet) in the event a three prong plug is used with a two prong outlet in USA. It is wrong and dangerous trying to use such adapter to break the safety ground connection. Also do not try to use any other adapter that breaks the ground connection (some travel adapters).