Appeared in the August 2004 issue
Choosing Opamps, PART-2
©2004 by Eddie Ciletti
Go to Choosing Opamps
and Capacitors, PART-1
There are no black and white answers to your upgrade questions.
Even identical circuits will behave differently depending on the construction
(circuit board layout) and components used. To paraphrase an old saying,
learn to fish and you’ll answer your own questions. I encourage experimentation,
but, you really need an oscilloscope to keep an eye on things.
For simplicity, opamps can be distilled into three application-specific
categories — high gain preamp, general-purpose buffer and line driver (output
amp). Opamps aren’t supposed to have "color," the output signal should
be the same as the input signal with the least amount of additional artifacts.
All opamps are designed to have low distortion, low noise and wide
bandwidth — some turn out better than others.
For our purposes, IC opamps typically come in two packages
— 8-pin "DIP" for single and dual opamps, 14-pin "DIP" for quad opamps
— that are easily socketed. Many chips are also available in the surface
mount (SMT) package. Because the pin-outs are pretty much standardized,
the interchangeability factor is a tempting tease that implies plug-and-play
upgrade potential but comes with the warning "proceed with caution." With
the Internet providing easy access to data sheets as well as a deluge of
opinions, the only question left is "how to choose?"
The specs are mind numbing at first. With names like Open-Loop
Gain, Bandwidth, Distortion, Noise, Slew Rate, DC Offset and Settling Time,
some are tangible and some are not. Gain is a variable that effects each
of these parameters and one more, Stability. Increased Gain and signal
levels raises distortion and awareness of noise while decreasing bandwidth.
What starts out as several mega-Hertz (mHz) might be reduced to the upper
kilo-Hertz (kHz) when the gain is 100 times the original signal.
TAMING THE COWBOYS
In this age of kissing digital zeroes for fear of losing
resolution, one short-term fix for what might seem like "bad analog sound"
is to back off, take advantage of headroom and let noise be your dither.
Seriously, no amplifier sounds its best when pushed toward the rails except
when you are specifically looking for distortion. Opamps are not internally
designed to emulate guitar amps.
It is nearly impossible to resist the temptation to compare
specs. Yes, a modern opamp boasting a slew rate of 20-volts-per-microsecond
(v/uS) is better than 30 year old, 2v/uS technology, but that doesn’t imply
that 2000v/uS is going to be as noticeable an improvement. Such demon-speed
may be a gift to those in search of "effortless sonics," but taking advantage
of video-agile ICs is not a plug-and-play upgrade. That’s the domain of
people with RF skills.
There are plenty of vintage consoles out there "limping"
along at 2v/uS that just happen to sound amazing for that old-fashioned
rock ‘n roll. You might prioritize your upgrade investigation by targeting
the preamp ICs or, if outboard preamps are plentiful, focus on the mix
buss and master module as Dan Kennedy did for his Trident 65 upgrade (featured
in a Tech’s Files column in 2000).
Despite the ease of ICs, some designers still choose discrete
circuitry. Rather than initially obsessing over specs, their goal is simply
to realize an idea in hardware. Each person has tests that weed out problems
and assist in maintaining high standards — a type of "headroom" that translates
to reliable performance in the field under adverse conditions. And yes,
many designers do not have engineering degrees. This should be encouraging
to all you "tweakers" out there. Now get to work! You’ve got some catching
up to do.
When I asked Crane Song’s Dave Hill for an example, his
discrete output amplifier had to pass an "acid" test of driving a 100kHz
square wave into a 75-ohm load. If it still looks like a square wave and
remains stable, the circuit is a winner. In this case, a square wave reveals
an amplifier’s ability to drive a capacitive load (a potential cause for
instability), which is what 1,000 feet of cable looks like to a piece of
Dan Kennedy is using discrete amplifiers in the AC signal
path and ICs in the DC servo loops for his new EQ-2NV, a vintage Neve inspired
Equalizer. DC offset and DC stability (settling time) are two issues that
can reach a critical mass in high Gain applications (in a mic preamp, for
example). Such idiosyncrasies reveal themselves in funny ways — like scratchy
pots and switches — and oscillations.
DC offsets are measured in milli-Volts (mV) and hopefully
micro-Volts (uV), a rather intangible specification compared to the brute
force transistor stages of the original single-ended Class-A Neve designs.
The output amps are biased to operate around 12-volts (half the power supply),
so blocking caps are used throughout. If a capacitor becomes leaky (develops
a parallel resistance) DC from one stage can re-bias another stage to the
point of compromising headroom or worse, nasty distortion. Similarly, a
transient pumped through a Neve amplifier stage will typically bump all
of the bias points. With an o’scope (wink-wink, nod-nod) you can easily
watch the recovery, a snails crawl compared to the needs of Digital audio
converters that rely on amplifiers with fast settling times (on the order
of a few hundred nanoseconds).
At the other end of the spectrum, Jim Williams of Audio
Upgrades typically chooses video op amps for audio applications. It is
a different philosophy aimed at improving existing equipment and not recommended
for your first DIY. I recently repaired a mic preamp in a Fostex portable
DAT recorder, substituting an OPA2604 for an NE5532. You might view this
as an upgrade, but I saw it — as some engineers also see it — as using
what was on hand. This is a somewhat less romantic than some would imagine
— Audio is not Fashion — designers know what works and what doesn’t and
where the laws of diminishing returns come into play. You should know that
the OPA2604 was totally unstable without additional de-coupling — a challenge
in a cramped, surface mount village.
Supply Voltage Range, not detailed in this article, is
one parameter that is self-explanatory. Here’s a short list of popular
modern IC opamps along with a few new chips at the bottom. For more details,
go to the manufacturer’s website, download the full data sheet and happy
Return To Eddie's Homepage