Protects up to six amplifier modules (six
single-ended or three bridged outputs)
Very simple, small in size and low in cost
Can operate from the same power supply
as the amplifiers (up to ±40V DC)
Disconnects the speaker(s) in 100ms for
full-rail DC fault <at> 30V
Provides a 1-2 second turn-on delay,
allowing amplifier outputs to settle
Insensitive to low-frequency
AC signals
Uses DPDT relays with contacts rated
to break 10A <at> 28V DC (repetitive)
Multi-Channel
Speaker Protector
If you’re driving a lot of speakers, you will need a matching compact speaker
protector to prevent driver destruction, should something go wrong. Our
Speaker Protector, when combined with our Hummingbird Amplifier module
(published last month), is excellent when driving stereo loudspeakers with
an active crossover or for surround sound systems where you have many
speakers to drive.
A
re your expensive speaker
drivers protected if the worst
happens, and an amplifier
module failure results in them having direct current applied? This very
simple and effective board matches
our Hummingbird Amplifier modules,
protecting between one and six channels with a switch-on delay in a PCB
measuring just 67 x 120mm for up to
six channels, or 67 x 91mm for the
four-channel version.
Over a few years of building Hi-Fi
and PA equipment, it would be fair to
say that this author has not destroyed
that many speaker drivers. But when
I have, it has always been expensive,
painful and inconvenient.
The experience of watching a 60W
amplifier deliver 40V DC to the voice
coil of a very expensive driver that represented months of savings is burnt in
my memory. This was a 250W driver
but it was no match for 40V DC! In a
matter of seconds, the voice coil turned
into smoke, much faster than a human
being could turn the power off—all for
the sake of a 50p insulator.
There have been two main destructors of my drivers:
1.
Over-excursion of drivers, particularly in vented enclosures below
resonance without appropriate
Practical Electronics | January | 2023
subsonic filtering. This is a surefire
way of killing a bass driver.(It can
be addressed in active crossovers by
including a subsonic filter.)
2. By DC from the output of an amplifier, either due to a failure in the
amplifier or finger trouble by the
builder. (Have you ever left a fuse
out or forgotten to connect a wire?)
This project solves #2. But, you might
ask, what about over-
p owering a
speaker? Won’t it blow up that way
too? In my experience, that takes a
heroic effort if your crossover is set
up correctly, so we leave the volume
control to your discretion.
The impetus
When I built an active crossover combined with six Hummingbird Amplifier modules, I found myself running
out of room. To fit this lot with power
supplies into a 330mm-deep 2RU case,
I had to move from beer mat sketches
to CAD and ‘the computer said’ that I
needed to make the speaker protector
small. So I did.
This device will protect your
speaker from most amplifier failures.
The modest investment will pay itself
By Phil Prosser
off the first time it activates, but we all
hope this is one project that you never
see ‘work’.
There are many ways of approaching
a speaker protector. This design aims
to keep it simple and small by keeping
the parts list to a minimum.
Circuit details
The circuit used is straightforward, as
shown in Fig.1, with three duplicated
stereo sections providing the six protection channels.
The main part of the Speaker Protector circuit is elegant, but it might not
be obvious how it works at first glance.
Its first job is to detect the presence
of DC at an amplifier output, connected
to one of the AMP x OUT terminals at
right. This is done by the 100kW input
resistor and 47μF bipolar capacitor,
which form an RC low-pass filter with
a −3dB point (corner frequency) of
0.25Hz. The output of this filter feeds
a DC detector that triggers at the Vbe
voltage of a transistor, around 0.6V.
So for regular operation, the amplifier must generate less than 0.6V at the
output of this filter. Choosing 10Hz as
a ‘safe’ low frequency limit and assuming an amplifier that can deliver 100W
into 4W, we can calculate that only
135mV would appear on the output
41
Fig.1: the Speaker Protector has three
identical sub-circuits handling two
channels each. Each input signal
passes through a simple RC lowpass filter and is applied to three
transistors. If a large DC voltage is
detected, those transistors switch
off the associated DPDT relay,
disconnecting the speaker from the
amplifier. A basic linear power supply
provides around 24V to drive the
relay coils and incorporates a switchon delay of around one second to
avoid thumps.
of the filter. So it won’t trigger during
regular amplifier use.
But say an amplifier goes faulty and
delivers its rail voltage of 40V DC (of
either polarity) to the output instead
of an AC waveform. In that case, after
100ms (0.1s), the low-pass filter output
will reach 0.84V, which will definitely
trigger the DC detection circuit that follows. This filter operates identically for
both positive and negative voltages.
With 40V across 8W for 100ms,
20J of energy will be delivered to the
voice coil (the impedance will drop
over time, approaching its DC resistance value, but this is a good enough
approximation). This will make a solid
thump and probably make you jump,
but it won’t cause anything to catch
on fire. Even better, if the fault exists
from switch-on, the speaker will simply never be connected.
The DC Detector comprises a total
of three transistors. For the top-most
section in Fig.1, these are Q2, Q3 and
Q4. Positive DC detection is handled
by Q2, which has its collector tied
directly to the 4.7kW load resistor. A
positive voltage from the filter of more
than about 0.6V will switch this transistor on and consequently pull the
base of Q1, an emitter follower, low
and thus turn off the relay.
Q3 and Q4 detect negative voltages. NPN transistor Q3 is connected
in a common-base configuration; its
base is tied to ground, and its emitter is the input. A negative input
voltage will pull current from 0V via
the base-emitter junction, causing its
collector to sink current. Because the
current it sinks at the collector goes
42
Multi-channel Speaker Protector
to the emitter, this current must be
kept low.
Hence, this tiny current is buffered
by Q4, a PNP device connected as an
emitter-follower. The emitter of Q4
connects to the same resistor as the
collector of Q1. So a negative DC voltage from the filter similarly pulls the
base of Q1 low, switching the relay off.
There is a balance in this circuit
between setting a low cut-off frequency and the minimum DC voltage
Practical Electronics | January | 2023
Parts List – Multi-Channel Speaker Protector
1 double-sided plated-through PCB coded 01101221, 67 x 121.5mm
available from the PE PCB Service
3 (2) 30V DC 10A contact, 24V DC coil DPDT PCB-mount/cradle relays
(RLY1-RLY3) [Altronics S4313, Jaycar SY4007]
8 (6) 2-way 5.08mm pitch mini terminal blocks (CON1-CON8)
1 44mm-tall, 16.5 x 10mm PCB-mount finned heatsink (HS1; for Q16)
[Altronics H0645]
1 TO-126 or TO-220 silicone insulating washer and insulating bush
[Altronics H7230, Jaycar HP1176]
1 M3 x 10mm panhead machine screw
1 M3 shakeproof washer
1 M3 hex nut
4 tapped spacers and 8 machine screws (to suit installation)
Semiconductors
16 (11) BC547B/C 50V 100mA NPN transistors, TO-92
(Q1-Q3, Q5, Q6, Q8-Q10, Q12, Q13, Q15, Q17-Q19, Q21, Q22)
6 (4) BC557B/C 50V 100mA PNP transistors, TO-92 [BC558-9B/C will also
work] (Q4, Q7, Q11, Q14, Q20, Q23)
1 BD139 80V 1.5A NPN transistor, TO-126 (Q16)
1 27V 1W zener diode (ZD1) [1N4750]
3 (2) 1N4004 400V 1A diodes (D1-D3)
Capacitors
6 (4) 47μF bipolar/non-polarised electrolytic [Jaycar RY6820]
3 47μF 50V electrolytic [Altronics R4807 or Jaycar RE6344]
Resistors (all 1/4W 1% metal film axial)
6 (4) 33k-100kW (see text; if unsure, use 100kW)
1 47kW
3 (2) 4.7kW
(n) for the four-channel version (PCB code 01101222, 67 x 91mm), the
quantities required are listed in red.
at which the circuit will switch the
relay off. 47μF is a reasonable maximum for the filter capacitor, so any
tweaking is best done by varying the
value of the input resistor(s).
We chose the 100kW value to guarantee no problems with false triggering for very high power, very low frequency applications. But if you are
not protecting a subwoofer, any value
greater than 33kW should be fine and,
as a bonus, lower values will provide
faster turn-off for fault conditions.
The 100kW resistors also affect the
lowest DC voltage that will cause the
detector to trigger. A fault in the amplifier front-end could cause a few volts
DC to be present at the output, and if
applied to a driver for long enough, it
could overheat and be damaged. So
ideally, we want to detect that condition too, not just a fault where it immediately pegs to one of the supply rails.
Assuming a minimum transistor hFE
of 120, and that the relay will switch
off with 20V across the 4.7kW resistor (leaving just a few volts across the
relay coil), the transistor base current
must be at least (20V ÷ 4.7kW) ÷ 120
= 35μA (or thereabouts) to switch the
relay off. This means the DC from the
amplifier must be at least 3.5V (35μA
x 100kW) to trip the relay off.
But this is with a worst-case hFE
value. We recommend using BC547B
Practical Electronics | January | 2023
or BC547C transistors, which have
higher guaranteed hFE figures and will
switch the relay off with about 1.5V
DC on the input. Lowering the input
resistors would reduce that trip voltage further.
The DC Speaker Protector disconnects the speaker any time that DC is
detected. The relays used are robust
and should be able to interrupt the
fault current that can be expected from
a Hummingbird Amplifier module or
similar. However, there is the possibility that upon disconnection, the voltage and current will be high enough to
form an arc between the relay contacts.
The normally closed contact of the
relay is used to shunt this current to
ground when the speaker is disconnected. So if an arc forms and current
continues to flow, the amplifier’s DC
fuse for that rail will blow, and the
arc will extinguish. You likely have a
failed output transistor already, so a
blown fuse won’t exactly be high on
your list of concerns.
We have put three sets of this circuitry on one board, allowing six standard amplifier channels to be monitored and protected. The relay selected
has a standard pinout and is available
from many suppliers. Make sure that
you get the correct version, though; we
are specifying 24V DC coils, though
you could use 12V provided you adjust
the DC regulator, and the BD139 can
handle its heat load (see below).
The circuit uses the power ground
pin as the ground reference. This connects to the earth of the power amplifiers being protected. Since the inputs
are already paired up, this Protector
would work well for DC protection in
a bridged amplifier.
The power supply
The power supply is a basic seriespass regulator generating about 25V
DC. The relays need 24V on their coils,
and this suits amplifiers with various
rail voltages. It can be adapted for
supplies below ±25V or above ±40V
(see below).
The power supply provides a
turn-on delay of about one second.
This is because the 47kW resistor
delays the charging of the 47μF capacitor at the base of Q18. This applies to
all channels protected by the board.
As you increase the supply voltage,
the turn-on delay decreases slightly
because the capacitor will charge
faster. You could compensate for
that by increasing the resistor value
if needed.
If you have an amplifier with rails
below ±25V, you have the option of
swapping the relays for 12V DC coil versions and make necessary adjustments
in the regulator (we expect a 15V zener
would work well for ZD4). Similarly,
if you have higher rail voltages, this
should be fine; just watch the sizing of
the heatsink. The specified Altronics
H0655 heatsink (or similar) should be
fine for any normal rail voltage.
Construction
Construction is straightforward. There
are two PCBs available; a six-channel version (coded 01101221, 67 x
120mm) and a four-channel version
(coded 01101222, 67 x 91mm), both
available from the PE PCB Service.
We have described the six-channel
version here; the four-channel version
is identical except that one relay and
its associated components are omitted, so the PCB is smaller. Refer to the
appropriate overlay diagram, Fig.2 or
Fig.3, during assembly.
Start with the resistors and diodes.
Make sure you get the diodes in the
right way around. Then mount the
two-way terminals for power, each
input/output pair and the earth terminal to prevent any arcing in the
relays (CON8).
Now it’s time to solder in the BC5XX
transistors. Try to mount them at the
same height so it looks neat.
Next come the capacitors. The three
47μF polarised capacitors need to
be rated at 50V DC, and all go in the
43
same way, with the longer positive
leads to the pads marked +. The six
47μF bipolar/non-polarised electrolytic capacitors do not need a high
voltage rating as they will never see
more than 0.6V – they mount to the
PCB in any orientation.
Now fit the BD139 to the heatsink
using an insulating washer, 10mm M3
machine screw, locking washer and
nut. Solder the heatsink to the PCB, but
remember it can be hard to get enough
heat into it to solder those big pins.
Finally, mount the relays. The PCB
has 1.5mm holes which are the minimum that this family of relay recommends – the devices from Altronics
and Jaycar leave a fair bit of room in
the holes. Solder them in well.
Testing
Now that you have all the parts
mounted, it is time to test it out. During
the initial tests, leave the amplifier terminals (CON1-CON6) disconnected.
First apply power and check for the
25V output of the regulator The end
of the closest 4.7kW resistor right near
Q11 is a convenient place to probe;
you can use the anode of any of diodes
D1-D3 as a ground reference. The reading should be between 24V and 26V
for an input above 32V DC.
If this is not present or correct, check
that ZD4 has about 27V across it; if not,
look to the 47kW resistor and transistors Q16 and Q18. Check that these two
transistors are the right types and soldered in correctly. Also check for short
circuits – is the BD139 getting hot?
After a second or two, the relays
ought to click in. If this does not happen, check the voltage on the bases of
Q1, Q8 and Q15 (the driver transistors
for the relays). Are these within a volt
or two of the 25V rail? If not, check
that they are the proper devices and
soldered in correctly.
Check the voltage at the output
of the RC filters. The voltage at both
ends of the 100kW resistors (or other
value you might have changed them
to) should be close to 0V. If not, check
that the BC547 and BC557 parts are
in the right places and they are oriented correctly.
Assuming the relays do switch on,
let’s check that they trip off correctly.
The easiest way to test it is to take
a 9V battery and connect the negative
end of the battery to the ground terminal on the DC protector. Then touch
a wire from the positive pin of the
battery to the ‘AMP’ terminal of each
installed channel of the DC protector.
The associated relay should switch out
almost instantly.
Repeat this for all channels, checking that the relays switch quickly.
44
Fig.2 and Fig.3:
build the smaller
board to protect
up to four
channels, or the
slightly larger
board for five
or six channels.
Assembly is
straightforward;
all components
are through-hole
types and can
be fitted in order
from shortest to
tallest (the latter
being the relays
and heatsink
for Q16).
The finished Speaker Protector board will look something like this. Note the
holes drilled into the board under the heatsink to allow convection to pull fresh
air up from underneath. The CON8 (GROUND) terminal block is missing on this
prototype version; you could leave it off, but it provides better protection for the
speakers if you wire it up to the amplifier earth.
Then repeat the test with the battery
the other way around (ie, positive
terminal to GND and negative to the
AMP terminals).
If a channel does not switch as
expected, measure the voltage at both
ends of the 100kW resistor. One should
be ±9V, the other ±0.6V.
If the ±9V end is not correct, there
is a short or open circuit somewhere
in that area. If the other end is not
close to +0.6V, check the two NPN
Practical Electronics | January | 2023
The MultiChannel Speaker
Protector comes in a
six-channel (pictured) and
smaller four-channel version.
The four-channel version would
be suited to a two-way stereo speaker
system with an active crossover or a
bridged stereo amplifier.
transistors on the DC detector (eg, Q9
and Q10), especially their orientations.
Check the associated PNP transistor
(eg, Q11) for a fault where you aren’t
registering −0.6V.
We don’t suggest you do this, but to
verify that the Protector does indeed
protect the driver, we connected a 4W
subwoofer to the DC Speaker Protector
along with a 6A limited, −34V power
supply to the ‘AMP’ input of the Protector. There was a solid thump and
click as the relay saved the sub from
inevitable destruction. We monitored
this with an oscilloscope, and the
result is shown in Scope 1.
We noted a flash of arcing as the
speaker was disconnected, which is
no surprise when breaking the very
high direct current flow. Please don’t
try this at home, as a speaker protector
does not make this sort of thing safe
for your speaker.
Application
The DC Speaker Protector needs to
be connected to the power amplifier
ground/earth via the provided terminal (CON8) and supplied with 30-40V
DC to the power connection (CON7).
If your amplifier has a higher positive
rail voltage than this, you can use a 5W
wirewound resistor to drop the supply voltage to the Protector. The six-
channel Protector draws about 100mA,
so a 100W 5W resistor will drop 10V
and dissipate about a watt.
Connect the terminals marked AMP
to the amplifier and the corresponding
SPKR terminal to your speaker outputs. It’s OK to leave some channels
unused; for example, if you have a 3or 5-channel amplifier. Once installed
in your amplifier, let’s hope that you
never hear those relays click unexpectedly. But if you do, you will be glad
they are there!
Practical Electronics | January | 2023
Remember to fit an insulating washer to the BD139
transistor (visible above) to prevent it from shorting
out on the heatsink. Also note the use of shakeproof
washers on all screws so they won’t loosen due to
vibration or movement.
Reproduced by arrangement with
SILICON CHIP magazine 2022.
www.siliconchip.com.au
Scope 1: the blue trace is the voltage across a 4W loudspeaker driver (zero volts
at top), while the yellow trace is −34V applied to the DC Speaker Protector from
a bench supply. The speaker is disconnected in less than 80ms. The AC voltage
generated by the cone movement due to back-EMF after the relay disconnected
the driver will not occur in this final version as long as CON8 is connected to
earth, as that will brake the cone movement.
Scope 2: this scope grab shows the response time of the Protector to a 20V
DC fault. The input voltage step is at t=0 and the output starts to drop before
t=60ms. It reaches 0V before t=80ms. The 80ms delay is due to the RC time
constant of the filter reaching 0.6V.
45
