Buying Guide: Preamps Sidebar - Specs
As we stated in the Preamp Buying Guide, specs can play an important role in preamp selection, but they must be handled with care. While we discussed specs that provide a necessary basis for preamp selection, here we talk about the gray areas of specs. In many cases, different manufacturers will present the same specs in different forms, which makes it very difficult to compare them. You’ll also find specs that lack the proper qualifying information in order for them to have meaning. We’ve provided examples of how they should be expressed (with arbitrary numbers).
Specs That Don’t Predict Performance
Below are specs that play an objective role in design, but do not necessarily help you to predict a preamp’s performance, sound quality, or form a basis for objective comparison based on the way the numbers are presented.
Total Harmonic Distortion (THD)
As we discussed in previous sections, both tubes and transistors add a certain amount of harmonics to a signal. THD is the ratio of the total harmonics added to the referenced output level. The closer THD is to zero, the more “transparent” a device should sound. While it is said by some that this is where the “sound” of a preamp lives, in truth, this number cannot accurately predict the sound of a preamp. Two units with the same THD value can sound profoundly different.
Frequency Response or Bandwidth
This is another spec that while valuable may not predict sonic quality. Frequency response is a measure of the output amplitude (level or loudness) of a signal over a designated frequency range. Frequency response is measured in db and will usually be presented as follows: Frequency Response: +0/-1dB @ 35Hz-20kHz. The problem presented by the spec in this form is that it doesn’t tell you where in the range the variations occur. 10kHz may output at 80dB, whereas 1.8kHz might output at 83dB. While this is not a perfectly linear response, it may well please your ear. Frequency response is more critical in terms of selecting studio monitors. Since 3dB is actually a very significant and noticeable change in level, a 3dB boost or cut at certain frequencies in your speakers might cause mixing problems. Since our ears have difficulty hearing a 1dB change in volume, a spec that touts ±1dB @ 20Hz-20kHz is an example of a very linear system, and one that will most likely sound good, but not necessarily better or more transparent than one with a +0 /-3dB @ 4Hz-300kHz. Trust your ears.
Also a spec that can’t be used to accurately predict sound quality, phase response is a time-based measurement taken from input to output, that measures the angle or linearity of a signal across the stated frequency range. The idea of linearity means that from input to output, the integrity or shape of the waveform is maintained accurately. While this is a worthy goal to shoot for, it seems that there is a lot of conflicting opinion based on testing, as to whether we can perceive the variations in phase in a non-linear system. Some say yes, some no, but under normal listening conditions with real-world audio systems, there appears to be little significance. In short, a linear system is not necessarily better than a non-linear system, just different. From an engineering standpoint, phase coherence of a signal is a very appealing concept. If you are among those who are militant subscribers to the “transparent” school of audio recording, look for a very small variation over the given frequency range.
Many manufacturers brag about slew rate, particularly in the field of power amps, but its value to the consumer is negligible. Slewing is a measure of how fast a circuit responds to current passing over a node. While the idea of a faster slew rate providing better transient response may appear to make sense, it’s more of a marketing tool than a means of predicting performance. In truth, slew rate is something you’ve either designed properly or not. While it is debated that a faster slew rate may equate to greater ease in handling the transients of input signals, a unit with detectable difference (in the negative) should never have left the drawing board.
What Do The Specs Really Mean—What Aren’t They Telling You
Here we enter the realm of specs that create the illusion of good or competitive performance but don’t actually give you enough information to make valid comparisons between units. We’ve included some of the more prominent specs that are handled in this fashion.
Bandwidth or Frequency Response
This measures the range of frequencies that a unit will pass. All frequencies above and below this range are attenuated, and in some cases, quite severely. Within the specified range, we are measuring change in level across the frequency range. This tells us how linear the system is in its response. For example, if there is a –4db drop in level within the specified range, you’ve got a bad unit on your hands. If you measure an increase in level in a solid state device, you have an unstable system, since there should only be a decrease, never an increase in solid state designs. (This is why manufacturers set a +0 limit in the specs) An example of an improperly defined spec for frequency response would be as follows: Frequency Response: 20Hz20kHz. A correct listing would be 20Hz-20kHz @ +0/-0.5dB.
Dynamic range is a ratio between maximum output voltage and the output noise floor expressed in db. In order for the number to have significance, the manufacturer must tell you what the maximum level is. For example, if one company states a dynamic range 125dB and another’s is 120dB, before you settle on the higher number as best, you need to know 125dB in relation to what? You have no way to evaluate the number without also knowing what bandwidth it was measured across and whether or not weighting filters were used. Now, when you see the following: Dynamic Range: 126dB, you at least know that you are no better off than if you had ignored the spec all together. On the other hand, if you see the following: Dynamic Range: 122dB re +26 dBu, 22.5kHz BW, “you’ve got what to work with,” as they say in New York.
THD (Total Harmonic Distortion)
This is a type of distortion that is harmonically related to the input signal. The frequency components are tested at whole number multiples of the original input signal. To test for THD, a pure sine wave (no harmonic components) is input. The output is then tested to see what harmonic components have been added to the input signal. THD is the ratio of the harmonic’s RMS voltage to the fundamental tone’s rms. Since this is a ratio of voltages at various frequencies, the manufacturer must tell you the frequency and level of the sine wave, number of harmonics measured, frequency range tested, and the gain setting of the unit. Also, the testing should be done at the pro audio level of +4dBu. Be aware that a unit tested at a lower gain setting will show better THD results. Also, if a different number of harmonics are tested from one unit to another, a valid comparison cannot be made. Therefore, if you see a spec that reads, THD less than 0.01% and nothing else, you’ve learned exactly nothing. A correct expression would be, THD (5th-order) less than 0.01%, +4 dBu, 20-20kHz, unity gain.
THD + N (Total Harmonic Distortion plus Noise)
This measurement includes everything that comes out of a unit that isn’t the pure test signal (see THD above) including harmonic distortion, noise, RF interference, 60-cycle hum etc. The test conditions are the same as THD except all of the additional noise is added to the spec. Since this number is naturally going to be larger and therefore worse in appearance than THD alone, some manufacturers only offer the THD spec (without the + N). Properly expressed, THD + N should look something like this: THD + N less than 0.001%, +4dBu, 10Hz-20kHz bandwidth, 35dB Gain, +27dBu Out.
Input and Output Impedance
Input impedance is a measure of the load or resistance (not actual resistance but close enough for our terms) a preamp presents to a driving source such as a microphone. Output impedance measures the impedance that drives the next unit in the chain from the preamp. The only reference you really need to know is whether the impedance is balanced or unbalanced. The reason for this is because balanced impedances are almost always exactly twice that of unbalanced impedances. It is assumed that the impedances are constant for all signal levels and all frequencies within the unit's bandwidth (unless otherwise stated). When you see the following: Input Impedance: 1.5k ohms, you don’t have the complete picture. However, Input Impedance: 1.5k ohms, balanced line-to-line, does tell you something.
Maximum Input Level
This determines how hot a signal (in dB) can be accepted before clipping or causing a predetermined level of distortion. Usually, the standard is no more than 1%, however some manufacturers go with “visual clipping,” which can be as much as 10% distortion and gives the impression that the preamp can handle a hotter signal than it actually can. Correctly stated, maximum input level should be shown as follows: Maximum Input Level +16dBu, balanced, ≤1% THD.
Signal-To-Noise Ratio (S/N or SNR)
This spec tells you how noisy a unit is. Testing for S/N is done by setting all controls in a specific manner and measuring the output without any input signal present. The result is expressed against a specified reference signal. In order for this spec to make sense, you must know the level of the reference signal, measured bandwidth, and whether or not any weighting filters were used. For example, two units may show an SNR of 85dBU, however, one may be reference to a +4dBu level, while the other was +20dBu. The one tested at +4dBu will be quieter in reality. It follows, that if you see signal-to-noise expressed as S/N = 96dBu, once again, you’ve been told nothing.
EIN (Equivalent Input Noise)
This spec is for any unit that has a mic input. Basically it measures noise added to the input signal, with the principle noise generator being the microphone in response to the unit’s input impedance. Since output noise will increase with input gain and there’s no way to know what input gain will be used in a particular setting, the focus is on noise added to the input signal. The combination of the two is then amplified and measured at the output. Since the testing conditions are quite critical and more complicated to discuss than the scope of this article, suffice to say, EIN expressed EIN = -136dBu offers no valid basis for comparison, since bandwidth and gain are necessary components of this measurement. The use of weighting filters can also complicate matters as well. If an EIN spec appears to be overly impressive, it may be the result of a manufacturer using favorable testing conditions (and omission) to improve the truth.
The Best Judge Is Your Ears
As you can see, the art of spec writing has a tendency to cross over into areas of marketing more than we’d like to admit. Keep in mind that manufacturers aren’t necessarily lying to you when they eliminate qualifiers from specs. It’s actually more of a preemptive strike to possibly keep you from passing on a great-sounding unit that may not spec out as well as another. As we’ve stated time and again, certain objective specs have no influence over sound quality. Apart from buying the best preamp you can afford, the best advice we can give is twofold: 1. Read our Preamp Buying Guide in its entirety. 2. Check out the online forums and customer reviews on our website.