|
The Tale Of The 1 kΩ Output Resistor
Introduction
Once upon a time, some folks felt the need to put 1 kΩ resistors on the outputs
of their modular synthesizer. The reason is lost in the mists of time, but since then
some enthusiastic DIYers followed suit, for their own reasons. Today, where cargo-cultism
trumps engineering, we find many modular synthesizer manufacturers putting 1 kΩ
resistors on their outputs, and when asked, the usual answer is along the lines of
"well, everyone else does it, don't they?".
Over the years,
engineers have tried to explain why it is not a good idea,
but to understand the answer requires hard-earned engineering knowledge, and who bothers
with that nowadays when you can just google up a schematic and become a millionaire over
night, eh?
Then one day one of the original DIYers, a chap by the name of Bernie, wrote
a piece on a forum
explaining why those early DIYers had added 1k output resistors.
What is interesting about this post is that it brings together history, engineering,
and - most importantly - context.
The Post
About those 1k resistors:
The 1k series output resistor was not really for protecting the outputting op-amp from damage (very unlikely),
nor was the automatic "mixing" (often touted) of much importance. [It did assure that something well-defined
happened contrary to an ill-designed chance occurrence of two op-amp outputs being directly connected! Also,
a somewhat similar moderate output impedance (600 ohms) was common in audio signal work. And it was obviously
never suggested for the lines carrying a main-control-voltage (volts /oct.) .]
Two useful things: First, the SDIY experimenter frequently (typically) has his/her "synthesizer" (finished modules)
driving an external breadboarded module under test (MUT). For example, a finished VCO might be modulating a new VCO MUT.
Things are not working - no surprise. Then to your horror you see that your good VCO has also now stopped - what have you done!
Well, if you had the 1k series resistor your finished VCO would be running happily regardless of what is going on
in the connected VCO MUT. (Op-amps generally drive anything if isolated by 1k). Without the 1k, a wiring error
(perhaps a breadboard short to ground) may well get back into your finished VCO, causing confusion as well as anxiety.
The second useful function of the 1k series resistors is that they "decouple" capacitive (typically shielded cables
at perhaps C=100 pfd/meter) loads from op-amps thus preventing high-frequency (MHz) oscillations WITH ASSOCIATED DC
SHIFTS. Connecting a cable (even to a scope) directly to an op-amp output forms an RC low-pass (R being the inherent
op-amp output resistance of perhaps 100 ohms). Such an oscillation is HF but low level (slew limited) and non-symmetric
(non-symmetric slew limiting). The result is a fuzzy looking scope trace (looks out of focus) and is only there when
a cable is attached, and as a VCO control can cause a small but noticeable pitch shift.
The oscillation occurs because the R is internal to the op-amp and the RC is INSIDE any feedback loop and contributes
excessive phase shift. With the 1k series the RC (R now 1k) is OUTSIDE the op-amp's feedback loop.
Fuzzy trace and pitch shift gone.
-Bernie
|
Analysis
That post packs a lot of information into a few paragraphs. The best way to
understand it is to pick it apart and understand each piece in turn.
- The 1k series output resistor was not really for protecting the outputting op-amp
from damage
- Pretty much all opamps used in electronic music synthesizers have some output
protection built-in. For example, the popular TL072 has 128 Ω resistor in the
output pin and 64 Ω emitter degeneration resistors, so for either driver transistor
there is 192 Ω in the output path, which is enough to protect the device from a
temporary short.
- nor was the automatic "mixing" (often touted) of much importance
- To do this mixing you need a way of connecting multiple cables together without any
intervening electronics. This is typically done with passive multiway adaptors
("mults"), originally
designed to allow one output to drive multiple inputs. It is possible to use them in
reverse, combining multiple outputs, but this has several pitfalls: the mixing depends
on all outputs having this 1 kΩ resistor, the mix ratios depend on the output
resistance of each source, which is not guaranteed to be exactly 1 kΩ so it is a bit of a
crap shoot as to what you end up with, and it results in significant currents flowing
around the patch cables, which can inductively couple into other patch cables.
That said, some folks still prefer to do this, although it must be taken with the
above caveats. And it will only work when all the connected outputs have the
1 kΩ output resistor -- the output with the lowest resistance
will dominate the mix.
- It did assure that something well-defined happened contrary to an ill-designed
chance occurrence of two op-amp outputs being directly connected!
- In simple terms: this is done at the bench during module development, it is not for
use in final products.
- And it was obviously never suggested for the lines carrying a main-control-voltage
(volts /oct.)
- Sadly there are many, many threads on various forums from users complaining that their
VCOs do not track that well across the keyboard. With 1 kΩ output resistors in the
most common uncompensated configuration feeding into 100 kΩ (typical) inputs this
is a problem designed into the modules: a -17.226 cent error to be precise. Because it
was assumed by the early pioneers that no one, beyond the test bench, would do this crazy
thing. Sadly, this is not so...
- First, the SDIY experimenter frequently (typically) has his/her "synthesizer"
(finished modules) driving an external breadboarded module under test (MUT)
- As described above, this 1 kΩ resistor is purely an aide during development.
It was never intended for finished products.
- The second useful function of the 1k series resistors is that they "decouple"
capacitive (typically shielded cables at perhaps C=100 pfd/meter) loads from op-amps
thus preventing high-frequency (MHz) oscillations
- Now, this is a perfectly valid reason to put some series resistance on the
opamp's output. But to solve that specific problem you only need around 50 - 100 Ω,
and for pitch control voltage outputs the damping resistor needs to be inside the
feedback loop so that its voltage drop is compensated for.
- oscillations WITH ASSOCIATED DC SHIFTS ... Such an oscillation is HF but low level
(slew limited) and non-symmetric (non-symmetric slew limiting). The result is a fuzzy
looking scope trace
- I see many builders using poor-quality test equipment. Those cheap LCD oscilloscopes,
or worse an iPhone, will not capture or display a 1 MHz oscillation sitting on a 1 V
control signal. You just won't see it. Or as someone once wrote:
"You need bench test equipment that HP or Tek made".
The DC shift, due to asymmetry in the HF noise, will then introduce further tuning errors
in pitch control voltages, or might be rectified by the input stages of connected opamps
and introduce audible artifacts.
Counter Arguments
Any argument of course will have its counter arguments. Sometimes there are valid reasons
for adding a high output resistor, under the right circumstances and by
experienced designers who know what they are doing and not simply doing what everyone else does.
To this end, here are some of those counter arguments put forward by some of those
highly-experienced designers:
- A very useful feature of the 1 kΩ output resistors, when applied to audio outputs,
is the user can quickly check the audio in a patch by inserting headphones into an output.
Without the series resistor the output opamp would go into current limit and severely
distort.
- Some designers are still strong adherents to passive audio mixing. However, there is
a very special caveat here: only for audio outputs, not for pitch outputs.
Passive mixing, and hence the need for those 1 kΩ output resistors, must never be used
on outputs that directly affect pitch. For example, control voltage mixers, MIDI-to-CV
interfaces, etc. Of course, you can also modulate CVs with audio signals, but in that
case the user is adjusting by ear and is not too concerned about the accuracy of the
modulation.
Solutions
AC Outputs
For signal outputs the only requirement is stability with patch leads plugged in.
To this end a small (50 - 100 Ω) resistor in series is sufficient.
It is also entirely the designer's choice to add extra resistance to support
passive mixing and/or headphone monitoring and any other application-specific need for
them (e.g., the ARP2500's bus-bar passive mixing system, although it is interesting
to note that some outputs had 1 kΩ resistors while others presented an effective 1.1 kΩ
output resistance, and the pulse and double-pulse outputs of the VCOs had asymmetric output
resistance).
Pitch Control Voltage Outputs
Where the voltage level directly affects the tuning of the instrument the standard
solution is to put a series resistor inside the feedback loop of the opamp.
This
article
provides a good overview of the problem and solution.
For inverting output configuration the following circuit is used:
Typical values might be R13 = 100 Ω, R11, R12 = 10 kΩ, C8 = 22 pF.
|
|