Dis’ DIY D.I. I did

If you have read my previous posts, you know that I have been building much of the gear that is being used in the recording of the Solamors album. One piece is the D.I. box that is being used in the recording and re-amping of the guitars. This might seem like a trivial piece of the puzzle, but there is a bit more to it, as you will see.

What the heck is a D.I.?

If you are familiar with the function and usage of a D.I. box, you can skip to the next section, but for the uninitiated: read on. D.I. stands for ‘direct injection’ or ‘direct input’. Traditionally, a D.I. box was used as an impedance converter and format adapter to connect either high-impedance sources to low-impedance inputs, or to convert an unbalanced source (usually 1/4″ TS output) to a balanced input (usually an XLR input)… such as plugging an electric bass (which has an unbalanced output and requires a high-impedance input) into a mixer board’s input (which is a low-impedance, balanced input, intended for microphones).

A passive D.I. box containing a transformer provides galvanic isolation (isolating sections of an electrical system to allow the AC component of a signal through while blocking any type of metallic conduction). This has several important benefits including: blocking DC current in the case of leakage etc, breaking the common ground connection between two pieces of gear to prevent ground loop noise in the case of different ground potentials, and blocking other types of electrical noise, while optionally limiting bandwidth to the audio spectrum to block radio frequency noise.


SO why do I need to use one for guitar?

Ground loops are very common problems when recording guitars direct, as is electrical noise. Since it is often difficult to detect low-level noise in a clean guitar signal, using a D.I. is just good practice to avoid re-recording should a problem arise later on. A common practice in modern recording is to record the clean signal directly from the guitar, running it through an amplifier at a later stage. This adds an additional chance for noise and ground loop issues since the signal must once again pass through cables to another piece of gear. That being the case… better safe than sorry since discovering an unusable guitar take in the mixing phase after the amps and mics have been torn down is a disaster.

OK, then what is special about YOUR D.I.

Another common technique used in heavy guitar recording is to place a heavily bandwidth-limited overdrive pedal (like an Ibanez Tubescreamer) between the guitar signal and amplifier to tighten up (roll off) the low end before it hits the first stage of the amplifier. THis technique works well… BUT… it is not without it’s own set of problems. The first issue is noise. While guitar pickups are noisy little beasts… so are pedals, especially distortion pedals. Even though the tubescreamer is usually run with the gain most/all the way down, and used primarily for it’s EQ curve, it still adds noise to the signal. The second issue is headroom. I play heavy strings and hot, active pickups so that can push several volts of signal. The buffers on a tubescreamer (on my scope) seem to clip on signal peaks over about a half a volt, so that adds a coloration to the tone, limits the dynamics, makes hard single-notes sound compressed and prevents you from ever getting a perfectly clean signal.

That being the case (and in keeping with my philosophy of keeping the signal-chain as short as possible) I decided to experiment other places to achieve that low-end tone-sculpting. Initially, my thought was to build a tight bandpass EQ to use in place of a tubescreamer, and that would work fine, but since I would be using a D.I. in the signal-chain regardless, I turned to altering that link in the chain.

Most D.I’s have a flat response across the audio band, but what I was looking for was a sharp bass rolloff. That led me to good ‘ol Radio Shack. Since I would be recording into a high-impedance input, I didn’t need any impedance conversion… just isolation. So I grabbed a couple small transformers and tested them out. Here is the response of a tubescreamer at typical settings which I used as a reference:

tubescreamer noise sweep

Here is the response of Radio Shack 1:1 ratio audio transformer part number 273-1374:

noise sweep

You can see that the low end rolloff breakpoints are similar. Obviously, there is extreme high-end rolloff in the Tubescreamer, but that is easily duplicated with a passive RC network if that effect is required. I didn’t bother. Since I typically don’t change the strings on my guitars until they fall off, I was not interested in further dulling the sound.

Here is a composite graph of the tubescreamer and the transformer, laid on top of each other and aligned:

stacked sweeps

The fundamental frequencies of a guitar lie between (depending on tuning) 65Hz (drop C open) and 1.3kHz (high E, 24th fret). The crossover point in the graph above is around 300Hz. Since I will be using this D.I. box for both tracking AND reamping guitars, the end result of the EQ slope will be steeper and more closely resemble that of the Tubescreamer and the crossover point will be much lower… not a perfect match, but my intention is to tighten up the low end, not emulate a Tubescreamer.

An additional benefit of a transformer is that it has a few effects that are considered to be desirable: additional harmonics (to warm up an otherwise digital process) and other analog anomalies that are considered either desirable or not, depending on the application. In the case of transformers, one such effect is hysteresis… which is a natural effect of the system to hold it’s flux density relative to the force that is magnetizing it (in our case, the guitar’s signal). This trait is present in old-timey, transformer-input preamps, and is often one of the effects that is attempted to be emulated in digital plugins designed to impart an ‘analog sound’ to the signal.

Because of the nature of that curve, the distortion we expect from a transformer is additional odd-order, low-order harmonics. Here is a plot of the transformer being fed a 440Hz test tone, vs the raw preamp signal:


You can easily see that the 2nd and 4th harmonics are similar in amplitude with a higher 3rd harmonic and the presence of an additional 5th harmonic. That is typical of the ‘transformer sound’.

Here is a plot of the higher-order harmonics:

higher harmonics

These are more-or-less irrelevant since they will likely be swamped by fret/string/pick/finger noise and EQ’d down in the amplifier before heavy clipping (typically 2.2kHz-4kHz per-stage in modern high-gain amps).


The radio shack transformer is unshielded. Since transformers work on induction, they are susceptible to radiated signals and noise. I placed the transformer in a small mu-metal shield that I had scavenged, and then built that into an aluminum Hammond project box, which I place inside a make-shift Faraday cage built from an EMI shield that I also scavenged from a piece of computer gear. That is probably overkill, and I could probably do just as well dumping the whole thing in an old coffee can… but if something is worth doing, it is worth over-doing.


About alexkenis

Guitarist, philosopher, tinkerer
%d bloggers like this: