Welcome the The HB Channel, my name is Hans Beekhuyzen and in this show we’ll look at the connection between a digital source and the d/a-converter and what we should know to avoid problems. D/a-converters might have a variety of inputs. Most common are SPDIF and USB, but some come with AES/EBU or I²S. I am often asked what interface is the best. To be honest, there is no definitive answer. All interfaces theoretically work well, it is rather the way they are implemented that makes the difference. The sending device needs to deliver a very clean digital signal, the digital interconnect should not alter the signal in any way and the receiving device should capture the signal perfectly. When this is the case, there will – at least in theory – be no difference between all the interfaces. Unfortunately, it’s not easy and often not cheap to achieve this. There are a number of things you as consumer can do: first choose the best matching source and d/a-converter combination you can find. Read reviews, like on theHBchannel.com, visit forums, ask friends; in other words, get informed on products. Or visit a trusted dealer and have him demonstrate a good combination. Or do all of the above. Always use a cable that matches the interface you use. A cable that just has the right connectors isn’t necessarily the right cable, as we will see later on. In all cases there are not only wrong and right cables, there are also good cables, excellent cables and bad cables. A good audio dealer will sell you an appropriate cable under the condition you can return it if it’s not satisfactory – undamaged and originally packed, of course. Don’t be fooled by people that say cable technology has no secrets and that all well made cables sound the same. I wish it were. On the other hand, not every expensive or fancy looking cable does sound right. If you want to know at least a number of reasons why both analogue and digital cables can make a great difference: I wrote an entire chapter on this subject in my e-book File Based Audio aka Streaming Audio that is available world wide at Amazon and Apple’s iBook stores. It’s only about 7 euro’s and helps me to produce more videos. Time to look at the digital interfaces, to start with the most popular ones. The AES Three standard, as incorporated in the IEC 60958 standard, comes in four shapes: SPDIF on RCA, SPDIF on Toslink, balanced AES/EBU on XLR and unbalanced AES/EBU on BNC. The data patterns of all four are about the same, but both AES/EBU versions have a slightly different set of metadata, more tailored to professional use. Data wise all four versions can be used on d/a-converters without problems as long as the electrical properties are matched. An isochronous connection is used, meaning that the bits are send at the same speed as they are processed by the converter. The sending device holds the master clock the d/a-converter has to slave to. It is therefore of great importance the bits are sent at a very precise pace, any deviation will potentially produce distortion during the d/a-conversion. The biggest difference between the audio quality of digital sources lie here. Responsible for sending out the audio data at a very precise pace is the designated clock oscillator. Unfortunately that clock signal is easily disturbed by all kinds of interferences. Especially computers, that contain various clocks for various functions, normally are unable to provide a properly clocked digital audio signal. Streaming players and cd-players are small computers too, but well designed ones can produce a very well clocked signal as a result of good and costly engineering. We now have seen what all four AES Three variants have in common, time to look at the differences that mainly has to do with the electrical properties of the interfaces. Two are designed for the consumer market and are known as SPDIF, which stands for Sony Philips Digital InterFace. The first one is an electrical version that uses a rather low 0.5 to 0.6 volt over a 75 ohms coax cable, terminated by RCA or BNC connectors. The construction of RCA connectors prohibit a pure 75 ohms impedance over the required bandwidth, which officially is 128 times the sample frequency. Therefore the use of 75 ohms BNC connectors would be a better choice. These connectors are also used for video and offer 75 ohms impedance over a far greater bandwidth. By the way, there are also BNC connectors and BNC interlinks that have other impedances like 50 ohms, so always check when buying. Unfortunately you’ll usually find RCA’s on consumer equipment. It’s therefore even more important te make sure at least the cable used is 75 ohms. SPDIF can theoretically run lengths up to 10 meters. Some electric SPDIF outputs and inputs are fitted with special transformers that provide galvanic separation to prevent ground loops. We speak of a galvanic separation when there is no electrical connection between one device to another. With digital audio is normally achieved by a transformer where magnetic flux is used to send the signal from the primary winding to the secondary winding of a transformer or by optical cabling as we will see now. The optical version of SPDIF uses a plastic light conductor with either Toslink or Mini-Toslink connectors. It offers less bandwidth than coax cable and is often limited by hardware manufacturers to 96 kilohertz to ensure a reliable connection even under difficult conditions. As mentioned, the advantage of optical is that there is no galvanic connection between the sending and receiving device, eliminating the risk of a ground loop. Furthermore optical connections are not susceptible to radio frequency interference. The optical conductor usually is made of Plastic Optical Fibre, POF for short. Due to optical losses, the maximum usable length is considered to be between 5 to 10 meters. That has to do with the quality of the optical sender, the optical qualities of the cable and the optical receiver. They all can vary in performance and the product of that performance determines the total signal loss. Therefore try to avoid adaptors, from Toslink to Mini Toslink for instance, since any extra connection will create a loss. The more expensive Glass fibre or Silicon fibre cables have a lower loss per meter and therefore distort the digital signal less. That can lead to longer runs and/or lower jitter and they become more affordable by the day. A one meter Toslink to Toslink glass fibre cable can be found on the web at around fifty euros. Balanced AES/EBU is designed for professional use and can run lengths up to one hundred meters due to the 6 volts voltage. It is loosely based on the RS422 serial computer interface and uses a 110 ohm shielded symmetrical cable on XLR connectors. These cables look like microphone cables – as SPDIF cables look like analogue interlinks – but they do need to have the 110 ohms specification. Balanced AES/EBU is also often found on high end consumer equipment, although in most cases it uses only the electrical properties of the AES/EBU standard. The data pattern usually follows the SPDIF standard. For connecting to d/a-converters the SPDIF pattern is no problem while the physical qualities of symmetrical AES/EBU offer more robust connections. Within the professional world the AES/EBU standard has been criticized heavily, for 110 ohm cables were initially hard to find. Furthermore the broadcast industry is used to use 75 ohms coax for high frequency signals. That brought about the second professional version. Unbalanced AES/EBU only differs electrically from balanced AES/EBU. It uses the 75 ohms coax cabling the broadcast industry likes and a 1 to 1.2 volts signal, much like a video signal. AES/EBU unbalanced can run lengths up to a 1000 meters. In fact the physical connection is almost like the electrical SPDIF connection, it only uses a voltage that is twice as high. As far as I know, this interface is not used on consumer equipment. SPDIF on RCA uses 75 ohms cabling, RCA or BNC connectors, is able to do up to 384 kilohertz or higher and is limited to 10 meters length
SPDIF on Toslink and Mini Toslink uses optical cable and is normally limited to 96 kilohertz, in some cases 192 kilohertz, and can run 5 to 10 meters Balanced AES/EBU uses 110 ohm symmetrical cabling, XLR connectors, is officially limited to 192 kilohertz and can run one hundred meters Unbalanced AES/EBU uses 75 ohms cables, BNC connectors, is officially limited to 192 kilohertz and can run 1000 meters All AES 3 interfaces are robust and as long as no a/d or d/a conversion is involved, not very critical. On playback d/a-conversion is involved and since the clock signal is embedded in the data stream, very good cabling and interfacing is eminent if sound quality is your goal. Always use cabling of the right specification and of the highest quality you can afford. Try to avoid ground loops and keep interfering RF signals, like cell phones, wifi and bluetooth at a distance. Doubling the distance between an RF source and the digital interlink reduces the RF stray on the interlink by a factor of four! The USB bus is designed for the computer world and in standard mode is bidirectional. This has the advantage of Universal Plug and Play, UPnP for short, that identifies the device and automatically loads the appropriate driver, when all goes well. Which isn’t always the case with d/a-converters. Sometimes you need to install a proprietary driver first. But let’s start at the beginning. Initially there was an audio-over-USB standard that used isochronous data transport, more or less like SPDIF. This means that the d/a-converter has to slave to the computer’s clock. Since computers have many clocks that can interfere with the audio clock, it will almost never be precise enough to provide quality audio. The USB audio mode does guarantee access to sufficient bandwidth by having the USB interface work in a unidirectional mode. But when the receiving device detects errors, it can’t request a resend. The USB Audio standard was also limited to 48 kilohertz, two channels. When a new standard was released and called USB Audio Profile 2, the old USB Audio Standard was renamed to USB Audio Profile 1 and an asynchronous mode and 96 kilohertz was added. In asynchronous mode the receiver determines the clock frequency and the sender – the computer – frequently asks the receiver what data rate to send on. USB Audio Profile two also has the asynchronous mode and can handle any combination of channels, sample frequencies and bit depths that can be handled by the available total bitrate of the USB connection. Both USB Audio Profile 1 and 2 are supported by all operating systems, including iOS and Android but excluding all Windows versions that only support USB Audio Profile 1. I had hoped that Windows 10 would include Profile 2 support but much to my surprise it doesn’t. It looks like Microsoft does not like audio and video media, they also dropped Windows Media Center in Windows 10 without offering an alternative. Luckily all manufacturers of USB Audio Profile 2 compatible d/a-converters supply drivers for Windows. From 2010 to 2013 a number of d/a-converters were fitted with a proprietary USB protocol that offered an asynchronous data connection and sampling rates up to 192 kilohertz. In these cases support for Linux usually lacks and a proprietary driver is needed for Windows and Mac OS X. Sometimes even different drivers are needed for for specific versions of OS X. In general, when you intend to buy a d/a-converter to use over USB, it’s good practice to check whether it is compatible with the operating system of your computer. Especially older operating systems like Windows XP or OS X 10.6 and earlier, might no longer be supported. If you have a Windows computer, USB Audio Profile 1 works without a driver, USB Audio Profile 2 needs a driver, just like the proprietary async protocols. With Apple OS X no driver is needed for USB Audio Profile 1 and 2 but an OS specific driver is needed for the proprietary async protocols. For Linux: no driver is needed for USB Audio Profile 1 and 2 but a driver for the proprietary async protocols will seldom be available. USB cables will normally comply to the USB standard and have the appropriate 90 ohms impedance. The data channels use voltages up to 3.6 volts, but things are slightly more complex than with the other interfaces. It absolutely pays to use quality USB cables for audio applications. Not only will a quality cable be shielded better, it will also have the correct impedance over a wider spectrum for less distorted square waves. If you want to go geeky: USB also provides a 5 volt power to devices that need it, like thumb drive shaped USB d/a-converters. Usually this power is rather polluted and unless special filtering of that power is applied, it’s better to use another power supply through a special USB to USB connector that takes out the 5 volt power. Some people cut open the USB cable and cut the red power wire or hook it up to a five volt external power. There are many articles on the web on this subject. I²S stands for Inter IC Sound or IIC and is the standard for PCM audio data transport within digital equipment. It was never developed for use between two digital devices and therefore there is no standard connector. For connections between devices the bus uses three lines: the bit clock line, the word select line and a data line. The bit clock line gives one pulse for every audio bit sent, the word clock signals whether the bit sent is either for the left or the right channel. The most elementary connection form consists of three BNC connectors, sending the bit clock, word clock and data line over separate cables. Like the AES Three interfaces it’s an isochronous connection where the source pushes out data at it’s pace and the d/a-converter has to follow. But since a separate clock signal is available, it should work close to ideal, provided the implementation is done well. To prevent timing problems, the three cables should have fully identical properties, including length. Other connectors used are 8P8C – commonly named RJ45 network connectors – sub-D9 – like RS232- DIN and HDMI. I²S works at 2.4 volts but further interface standards are not available since the bus was never intended for use between two devices. The same potential problem as with other busses can occur, square waves can be distorted to a degree that errors occur. But since the clock signal is carried by a separate connection, the system is far more rigid and is considered the preferred interface by many high-end manufacturers. Always assure yourself a matching digital interconnect is available when mixing a source and d/a-converter of different brands. You can read the full article including links on theHBproject.com. More videos are on the way, so subscribe to this channel, follow my Facebook page or my twitter account if you want to remain informed. You’ll find the information in the description below. Questions can be posted below, on my Facebook of Google+ page or on the contact page on theHBproject.com. And if you have enjoyed this video, please give it a thumbs up and tell your friends about it. My name is Hans Beekhuyzen for the HB Channel, thank you for watching and see you in the next video or on theHBproject.com. And whatever you do, enjoy the music!