This design was born of frustration with
the mic preamp in my Sony R91. It clips at ridiculously
low levels, basically it has 28mV of headroom to deal
with an input that could be as high as 1800mV (depending
on mic & volume), so this design attempts to cure that
by :
A: Providing a higher electret bias
voltage, 9V through a 10k resistor to give the mic's fet
more headroom. (The R91 provides 2.5V through a 6k8
resistor.)
B: Having much greater preamp headroom,
and bypassing the R91's mic preamp altogether, using a
hi-fi or low noise op-amp to go straight into the line
input.
The circuit needs to bring the signal
from an electret microphone up to a level suitable for a
line input. Based on measurement of the microphone
output under a range of conditions, I have chosen two
gain levels - 2 & 23. 23 is suitable for everyday
general recording. 2 is suitable for live music such as
a rock concert. There is a output pot to allow
fine-tuning of the level if necessary, but it is
intended that usually the wiper will be on the land.
When choosing an op-amp, it's important
to find out the open loop gain at 20kHz. The circuit has
roughly 27dB of gain, and at least 47dB at 20kHz (i.e.
20dB higher) is desirable. This is because op-amp
circuits are predicated on an ideally infinite open loop
gain, while in reality 10 times (20dB) higher than the
closed loop gain at highest frequency of interest is
acceptable. Otherwise the amplifier becomes
progressively non-linear. The TL072 at roughly 45dB just
makes it. The LM833 is good at �55dB and the NE5532
sparkles at �60+dB.
If you're going for ultra small size,
you can build it single supply rail, but an LM833 won't
run like that. A TL072 will, with reduced output swing,
it'll swing about 1.8VRMS off a single ended 9V supply,
which is adequate for the purpose. Otherwise two 9V
batteries will do the trick.
Fairly high quality components have been
used - 1% metal film resistors, solid aluminium caps,
though the output pot is a bit miserable.
Now about these solid aluminium
capacitors - in some respects very good capacitors, but
I think now I would use 1uF film capacitors as input and
output capacitors. They are likely to more closely match
the other channel (tighter tolerances), and the golden
ear brigade seem to like film capacitors better.
So in summary, it's supposed to have a
lot of headroom, be small, quiet, cheap, have a low part
count, somewhere between a battery box and a full
fledged professional box.
Schematic Of One Channel - Single Rail
Version (Slight Difference From Text)
The leftmost 10k resistor supplies
plug-in-power to the electret, forming part of the FET
amplifier in the electret capsule. This could be
anything from 2k to 10k, the higher the better the
stereo separation (another mic derives bias from the
same rail). Apparently higher values also lower
distortion, and the best bias power circuits involve
actually breaking a trace on the electret capsule to
allow the use of both a drain & source resistor, but I�m
not going that far.
The leftmost 2.2uF cap blocks the bias
voltage from the input. In conjunction with the
following 27k resistor it forms a high pass filter, but
cutoff is essentially near DC.
The input impedance is set by the two
27k resistors and the 10k resistor. The +ve rail is also
connected to ground as far as the AC signal is concerned
because of the power supply cap. So there are two 27k
resistors in parallel, making 13.5k, in parallel with
the 10k, making about 6k or so for the input impedance.
But if you're making it proper dual supply, you don't
need the upper 27k resistor, as the input doesn't have
to be biased mid rail anymore.
The feedback loop has two resistors 27k
& 1k5 from the inverting input to ground. When they are
both in circuit, the gain is a bit under 2
((28.5/33)+1). The 27k resistor can be bypassed with a
switch, then only the 1k5 sets the gain, to 23
((33/1.5)+1).
The 10uF cap in the bottom half of the
feedback loop reduces DC gain to ~1. The value isn't
very important. If any DC input offset were amplified it
would create a larger output offset, pushing the output
toward one of the rails and reducing headroom. (At a
gain of 23 with the expected input levels it probably
doesn't matter.)
The optional 2pF cap in relation to the
33k resistor sets the high frequency rolloff. The cutoff
frequency is in the 100�s of kHz. It has to go further
than 20kHz to keep the phase shift at audio frequencies
small, and also because output starts falling long
before cutoff. The op-amps cannot maintain enough gain
at these frequencies anyway and their output will
already be falling, but the cap makes the circuit more
stable, though it will probably work without it. There
will probably be 2pF of capacitance just from the PCB
traces, and op-amps tend to be fairly well compensated
these days so it�s really not needed. I think in
retrospect this cutoff frequency should be much lower,
say 30kHz-50kHz.
The 100ohm resistors are there partly to
limit current to protect the op-amp if the output is
shorted, but the op-amps have internal protection
anyway. They mainly allow the op-amp to drive capacitive
loads (long/cheap cables) without oscillation.
The 2.2uF cap on the output blocks DC
and the value is not specially important. It forms a
highpass filter with the 10k pot, the cutoff is
virtually at DC.
If you think you might accidentally
start connecting the battery the wrong way round, you'd
better put a diode in series with the battery clip, or
you'll smoke your ic. Put your ic in a socket too just
in case you do want/need to change it. You could try
several dual op-amps against each other, they're all
direct plug in replacements.
|