How To Build An Audio Computer
By Timothy Smith
This Series of E-book/Articles will cover step by step the building of a computer used for audio recording. All information herein in the property of Recordinghound.com and may not be reproduced or used in any way without permission from Recordinghound.com. C 2016
This idea came about as the result of my own adventures in computer building for the home recording studio. I knew there were others who had never built a computer, but who might be willing to give it a go if they had the right information.
Even though this information is primarily concerned with building a computer for audio recording, the information herein can also serve as a guide to building a computer for video production. The hardware I have chosen to use for the computer build is suitable for most home studio work including the use of soft synthesizers and heavy use of plug-ins.
This series of articles will cover both the assembly of hardware and the setting up of the OS audio settings in the computer to allow for maximum performance of a desktop recording system or digital audio workstation.
The hardware and software used in the computer build were current at the time of this writing. As we all know, technology moves at a fast pace, it is likely that more recent and faster hardware will be introduced, even so, it is the opinion of this writer that these techniques and information will be valid for some time after this writing.
It’s always best to buy and build a computer with components that have been tried and proven themselves to be dependable. I strongly suggest, whenever possible buy hardware that has a proven track record. At the time I built this computer all hardware components had been out for long enough to determine that, other than the usual percentage of failure, they all held up well to constant use under heavy load situations.
Even though I chose certain hardware, I in no way particularly endorse any particular brand or type of hardware or software. Using my book as a guide, you may use any number of other manufacturers and vendors to build your own computer, keeping in mind that it always pays to research the hardware, particularly the failure rates of existing users.
I hope you have as much fun as I had in building your own computer!
Some basic understanding of computer hardware is essential to understand how your computer works. This is the first step to knowing how to diagnose any potential issues that might arise either during or after assembly.
The Hard Drive
First, let’s look at the hard drive. There are two types, HDD or hard disk drive and SSD or solid state drive. A computer needs at least one drive, but can have multiple drives. One is needed for the OS or operating system and the others can be used for additional storage.
The hard drive is where data is stored and accessed. The HDD depends on a rotating platter. The RPM or rotations per minute of the hard drive impacts both latency and access time. The higher the RPM the faster the access time. HDD drives are called “platter” drives since they use a rotating platter that spins at a designated RPM or rotation per minute. The higher the RPM the faster the drive. The most common drive rpms are 5400 and 7200rpm. For audio recording it is best to select the 7200 rpm HDD.
At the heart of an HDD is the platter that spins at a constant rate, usually on an oil filled bearing. Lightweight actuator arms ride on either side of the platter on a pivot assembly. Each arm is tipped with a read head. These heads never make contact with the platter. Instead they ride on a cushion of air made when the platter spins using a principle called Bernoulli’s principle.
The disk is made of glass or aluminum and is coated with a magnetic material. The write head inside the HDD writes data to the disk, the heads magnetize very small magnetic domains on the disk that can be saved or written over. In this way it’s possible to store huge amounts of data. Millions of 1s and 0s called bytes all written in a particular way stored on a semi- permanent medium. This is how the data is saved on an HDD. HDDs are mechanical devices with electronic control and therefore are susceptible to vibration damage and eventual wear. For this reason they need to be handled very carefully, kept away from moisture, static electricity and extreme vibration or jarring. Dropping a hard drive is likely to damage it. The tolerances of the internal mechanism are very small. Hard drives are manufactured beyond the precision of a fine watch. Marvels of engineering. The fact that they can be mass produced relatively inexpensively is a testament to the advances of modern manufacturing technology.
SSDs work much differently, are faster and generally physically smaller. Prices have started to fall for SSDs as of this writing. The SSD works by storing data in semiconductor chips similar to another part of the computer we’ll be learning about called RAM or random access memory. The difference between the two is that RAM only stores data during the time a computer is functioning. If you remove the power the RAM memory is lost. This is called volatile memory. HDDs and SDDs use a more permanent memory called non-volatile memory. These devices can retain memory after the computer is powered off. Both types of storage use 1s and 0s arranged into data strings called bytes.
The SSD uses rows of transistors arranged into rows and columns that have either on or off states depending on how the transistors are energized. This kind of memory is referred to as NAND flash memory.
In order for the operating system to recognize a storage device it needs to be formatted. HDDs are formatted in both tracks and sectors. If you can picture the disk platter as a pie, think of the disk cut into pieces like a pie. Tracks would look like concentric circles around the platter. A sector is the small section of the track in each piece or sector. Sectors are grouped together into clusters. When the operating system looks for or installs data it uses these areas as a reference.
The organization of these sections on the hard drive is called low level formatting and lays the basis for the OS to write bytes to the HDD . The SSD has the same allocations but they are written onto the transistors as NAND flash instead of onto a magnetic medium like the HDD.
Software formatting tools check all sectors and tracks to be sure they are in good working order usually before installing the OS and can be used to diagnose a hard drive if there are problems later on after the build. This will be covered in more detail in the trouble shooting section.
The hard drive is connected to the motherboard by a standardized cable. The most efficient hard drive connections are called SATA or Serial Advanced Technology Attachment. SATA is a serial bus interface that connects host bus adapters to mass storage devices. As the name implies SATA is a serial connection. This is an improvement over the older ATA standard in several ways. There are only seven connectors instead of 40 or 80. Faster data transfer rates are realized, and the cables are hot swappable, meaning you can plug and unplug them while the equipment is energized. I don’t recommend this though.
We will be using SATA drives for our build since these are the most widely available, the technology is proven and they are the most compatible with motherboards made today.
There have been different SATA standards developed over the years. The SATA 1 standard was developed in 2001 and allowed for 150 megabytes of data per second or 150 million bytes per second. Remember that bytes are the smallest forms of data written in 1s and 0s. The SATA 2 standard came along in 2004 and has double the transfer rate at 300 megabytes per second.
In 2009 the SATA 3 standard came out with a data transfer rate of 600 megabytes per second. In 2013 SATA 3.3 was developed, also called SATA express. This standard allows for almost 2000 megabytes per second transfer rate coming in at 1969 megabytes per second.
My motherboard supports another standard that allows for SSD installation of a different type. It uses a socket called an M.2 socket. Not to be confused with another standard called M.2×4. If you opted to use this instead of the SATA 3 connection, you will need an M.2 standard SSD drive. When I compared the two I found that the specs were essentially the same between the two with the MTBF or “Mean Time Between Failures” to be shorter than the drive I chose, meaning that I should get more mileage out of the drive I chose. The only advantage I can find in using the M.2 spec drive is that you don’t need a connecting cable. Some users of the M.2 standard were reporting problems loading their OS onto the drive and having it be recognized by the system. In all fairness, these could have been isolated cases. I didn’t think it was advantageous at the time to adopt a new standard. I have read reports from others that these M.2 drives are working wonderfully. This specification uses PCIe lanes as channels for SSD data transfer. M.2 was originally designed for laptops and has since been adapted for desktop motherboards.
Concerning the SATA 3.3 express standard, it also uses PCIe or Peripheral Interconnect Express Technology. PCIe uses a lower latency connection compared to standard PCI. It also uses a dedicated point to point protocol so there is no chance of cross interference between devices. PCIe can have up to 32 lanes going to and from the device. In some motherboard/cpu configurations using a SATA express or M.2 device will eliminate some throughput to other devices. For instance a 16x video card might only run at 8x if using a SATA express device. This is only limiting if you needed the extra capability somewhere else. In the case of the build I cover in this article I’m using an Intel 5820K cpu which has 26 lanes. Spending more money on a 5930K cpu will give you 40 lanes. Compare either of these two with the last generations of chips that only had 16 lanes. I had decided that the 5820K should be more than acceptable since I didn’t plan on adding any additional hardware that would need extra lanes. My video performance is excellent for an audio computer. The video card in this build is excellent. If you decide to build a video computer buying the Intel 5930Kcpu might work better for you if you plan to use multiple video cards. 40 lanes will give you a bit more throughput matched with the right motherboard. Motherboards and cpus will be covered later .
I opted to use a Samsung 850 EVO 500gb SSD for this computer build. It reads at 540 MBps sequential and writes data at 520 MBps sequential. Plenty fast enough for even taxing audio work. This SSD uses a 2.5 form factor. This is acceptable for both laptops and desktop computers. I used three of these SSDs in my build. One for my OS and two for storage.
Next on the list of things we need in order to build an audio computer is RAM or random access memory. Referred to as volatile memory because the memory isn’t retained after the computer is turned off.
Think of RAM as a temporary storage facility. A place to store data for immediate access. If there were no RAM computers would be incredibly slow and unable to run most programs. An audio computer is doing many things at once and each task takes memory. If all data were to be immediately taken from the hard drive there would be a serious bottleneck or restriction in the data flow. The computer needs a place to temporarily offload data so that when a program calls for a task that places a demand on the system the memory can take up the slack. Let’s look at it another way. The cpu sees the instructions on the hard drive and runs based on those instructions. Let’s say that it has a set of instructions to play a series of audio files inside of a software audio application. A sampled synthesizer is simply a long list of samples all recorded beforehand usually according to a given note or pitch and usually mapped to a midi keyboard so that when you play a key the computer calls up the samples for that key. In many cases the sampler is playing multiple samples for only one key. The more keys you play the more samples are also playing. Keep in mind the cpu is busy running the operating system. Many other background tasks are happening at the same time. Eventually the system becomes overwhelmed with too many tasks to complete.. The computer becomes starved because the data can’t all be pulled fast enough to perform all of these tasks at once. This is where RAM comes into play. The instructions or software determine that in order to keep all of the processes running it needs to preload some of the instructions into a temporary memory or RAM. It determines this in advance of when it actually needs it, yet it happens so fast we don’t notice it happening. In this way there is always a reserve of memory to pull from.
In the case of our sampled synthesizer, when a keyboard key is played, the computer looked for the samples on a secondary hard drive. The instructions loaded some of the samples into RAM so they would be available the instant they were needed. While the keyboard samples are playing the instructions continue to tell the cpu to pull additional sample data from the secondary hard drive. It has already calculated the shortfall and made up for the difference in RAM. Programs automatically put a certain amount of data into RAM to keep things all going along smoothly.
Most consumer computers use at least 8gb or gigabytes of RAM. I usually use at least 16gb for a recording computer. If you don’t use many sampled synthesizers you may be fine with 8gb. On the other end of the scale, if you use many multiple software synths you may need more than 16gb of memory. My recommendation is to start out with 16gb and if it isn’t enough you can go to 32gb or more. The motherboard I selected can be loaded with 128gb of memory if necessary. I will have a chart explaining the demands on a music system later on. Most home recording setups can have 24 tracks of audio and multiple soft synths all running fine with only 16gb of memory. Some might mistakenly think that the more memory they have the faster their machine will be. System speed is only partially dependent on memory since many other factors come into play. A person can have loads of RAM and still have a slow computer.
I decided on 16gb of GSKILLS Ripjaws memory. I decided to use 4 x 4 gb memory sticks to take advantage of the quad channel capabilities of my motherboard. You can buy memory in different sizes. I picked this memory mainly because it met the specs of my motherboard. There are many choices a builder can make when choosing memory. I’m sure I could have chosen from many other noted memory manufacturers and fared just as well.
Memory isn’t a one-size-fits-all kind of thing. The single most important thing starting out is to base your memory selection on your motherboard requirements. The basic requirements of my motherboard called for 288pin SDRAM DDR4 memory. Other factors include memory voltage, frequency and speed. In order to get the best performance I always pick the highest frequency memory I can find that will run on my motherboard. First select your motherboard based on the features you want and this will help to give you direction on a memory choice. Most memory will allow for several frequency and voltage configurations. If you don’t plan to overclock, then having multiple frequencies isn’t as crucial. In my build I planned to overclock my computer, so I selected memory known to work well in overclocking situations. The modification features in my motherboard bios allow changing the memory frequency and voltages. Don’t let this step scare you since the bios software can be set to make automatic corrections within the limits of the hardware to overclock your system without causing hardware failure.
In the Next Article– Motherboards and cpu