No such device seemed to exist, so the "Tuesday Evening Coffee Club" (aka Paul and Nigel at Winchester hospital radio) were asked for help.
Thus I got around to designing a Dead-Air Detector. As of Sept 99 the design has not even been prototyped, just designed... however, we are willing to help anyone who would like to build one.
The schematics are available for download and online viewing. It is
assumed that you have an ability to build circuits from schematics, and
that we don't need to tell you where to attach power to the logic chips
What follows is a description of the circuit, and should be read whilst
you view the circuit diagram.
The Detect Low compares the input signal with a relatively small
voltage, so that a bit of low noise isn't detected as valid input. This
gives out pulses for as long as there is audio. These keep re-triggering
the Timer output , so that it stays "on". However, if the pulses
disappear, then after a length of time (settable by component change
between 5 seconds and 60 seconds) the output will go "off".
The on-to-off transition causes Pulse to output a pulse, which will
"set" the Latch to indicate that silence has been detected.
Back to the top again: the Detect High bit causes pulses when the
input is "relatively high" (and certainly higher than the "relatively
small" signal used in the Detect Low). If the input is not high
enough, then the output will stay "high", but when there is high enough
audio present there will be low pulses.
The idea is that if silence is detected, but an audio signal re-appears,
then the unit will automatically reset itself. However, these pulses then
pass through a jumper (not shown on the block diagram): if the jumper is
not fitted, then the unit does NOT automaticall reset.
The signal then passes through the + combiner, to reset the
Latch.
The Reset input, if low, will also pass through the + combiner
to reset the Latch. If the jumper above is not fitted, then
momentarily grounding this input will reset the unit (however, see note
below).
In addition, if this input is HELD low (by a switch, rather than by a
push-button) then the silence detector is disabled.
The output of the Latch, along with a repeat of the Reset
input signal are then used to drive whatever indicators etc are needed (see
below for more details)
High-res Circuit in GIF format (36k)
The next stage is a peak rectification circuit: this converts the normal
sound waveform into positive-going pulses (effectively removing the
negative peaks) of variable height... the height depending on the
instantaneous input level.
Next are a pair of comparators: these each compare the DC level to
different preset levels.
The "LOW" comparator sets the level which defines "silence"... normally,
with incoming audio this will be producing regular positive pulses, caused
by the input exceeding the "low reference" level. If the audio input goes
below that level (= dead air!), then there will be no more pulses to keep
re-triggering the monostable (see below).
The "HIGH" comparator is used to reset the circuit after dead-air has been
detected. The "high reference" level is higher than the "low reference", to
make sure we get some noticible audio so we can be happy that "the audio
is back again". Again, this prduces pulses, but these are negative pulses,
and when the input disappears, the input stays high.
This signal, via a jumper, is combined with an active low "reset" or
"disable" input, and is used to (constantly) reset the "dead air detected"
latch (see below). If the link (lk1) is fitted, then the detector will
auto-reset when audio returns, and the detector may be disabled by keeping
the "reset" input low (thus the "reset" input is also a "disable" input).
On the other hand, if the link is omitted, then the detector must be reset
manually, by momentarily pulsing the "reset" input low. Again, keeping the
"reset" input low will disable the detector. (you may want to do this if
you know that there will be a planned silence... eg on Armistace Day)
[At this point, I must make an important note - in keeping the design simple,
there is a minor "limitation": if you use the "reset" to clear the "silence
detected" state, then a further period of silence will not be detected
until SOME audio input (enough to reach the "low" level) has occurred. I
don't see this to be a big issue, as if someone has manually reset the
system, they will then be "in control of the situation", and will be
reinstating the audio in any case!]
The next bit of the circuit is a "monostable" (U3a): whenever the "LOW"
level is pulsing, then the monostable will be constantly triggered, and the
output of the monostable will remain high. If the pulses stop, then the
output will go low after a several seconds (typically this would be
something between 10 and 60 seconds): if that happens, then we have
detected "dead air".
The "dead air length" is determined by R14, R15 and C5. R14 and C5 are
fixed, but the length can be adjusted by R15: 0 ohms (ie a wire link) sets
about 5 seconds, 150k gives about 20 seconds, and 390k gives about a
minute. If you need a larger time, then you can increase the value of R15
further, to 1 meg max. You could also increase the size of C5. [Beware of
the capacitor leakage current if you use a particularly large value for
R15]
The second monostable takes this low-going output, and generates a short
(several milliseconds) pulse, which sets the latch.
The "dead air detected" latch (U4A) is the bit that holds the current "Dead
Air" state: normally this is low ("ok"), but is set high ("alarm" state) by
the monostable circuit above.
It gets reset as described above, by either the audio returning or by the
reset input being held low.
There are 2 outputs from the latch, one is HIGH when silence ("silence
det"), the other is LOW when silence ("/silence det"). These 2 ouputs are
used to drive one (or more) output circuits.
It may be that you would also like an output to indicate when the unit is
disabled (eg to light a big warning light!): one is high, the other low,
when silence detection is disabled.
However, you may well want to have one of the following outputs instead, or
as well, on the latch output or the disabled output:
Whilst the current used from the dual supply is minimal (< 10 mA per rail),
the current requirement for the +5v supply is larger, and depends greatly
on how many LEDs and relays you use. As a good guesstimate, assume that the
electronics uses 50 to 100 ma, then add on the current used by the leds
(typ 5mA each) and relays (a 500 ohm relay uses 10 mA when on, whilst a 200
ohm unit uses 25 mA).
Whilst it is probably OK to use a switch-mode PSU, it is probably best
avoided, not only to prevent PSU noise from affecting other audio
equipment, but to prevent noise causing incorrect operation of the audio
levels. If you are having trouble finding a suitable PSU, then refer to
Article #3 in this series, which deals with Power Supplies.
(plb, rev 2.1, Jan 2000)
Block Diagram
Block Diagram Description
The Input Buffer take the stereo input, buffers it, and makes it
into mono. It also allows us to adjust the input sensitivity. The
Rectifier then gets rid of the negative bits of the signal, leaving
just a load of pulses (of course, if there is silence, the pulses will be
very small, or completely absent!).
Let me download the circuit details!
Low-res Circuit in GIF format (11k)Detailed Circuit Description
The first section is the input buffer: this reduces the load on the
external circuit being monitored, and allows you to set the gain to
match your setup (ie if you have a low output level, then you'll need
more gain, and if you have a high output level, then you will need a lower
gain). The gain is set by the feedback resistor (actualkly a preset pot).
What do I do with the outputs?
If you go & look at Article 7 you'll
learn all about the Remote Start circuit: to work, that circuit needs a
low signal to start, and a high signal to stop. Its input
may therefore be directly connected to the silence det output (pin
5) of the latch. This will allow you to start something when dead air is
detected, and stop it when normal audio returns.
In that case, go take a look at Article 8
which details the (very simple) circuits needed. Most need to be driven
from the silence det output (pin 5) of the latch (or the
disabled output), with the single exception of the "+5v signal"
which needs to be driven by the /silence det or /disabled
outputs (ie the active LOW outputs).
Power Supply
The circuit requires a +5v supply for all the logic, and a dual supply for
the audio input stage. The dual supply would normally be a regulated +/- 12
volt supply (quite sensible if the silence detector is part of some other
piece of equipment, such as a station output switch). However, a +/- 5 volt
regulated supply would be ok.... you could actually use the logic supply
for the +5v.
And Finally....
Finally, if you need any help with the circuit in any way, then feel
free to email me with your questions.