I am soon going to reactivate speakers I built ~35 years ago, and I don't want to put them in danger: at the time, it took me lots of time, efforts and money to build these speakers and although they are probably crap by today's standards, I don't want to run any risk with them.
I have designed a stand alone protection circuit, broadly the equivalent of an ordinary fuse, but more flexible, more accurate and having negligible distortion. This will allow me to connect any amplifier without having to worry about internal protections, or the lack thereof.
A completely independent system is less sensitive to failures occurring in the amplifier itself: for example, a paper clip finding its way into an amp can at the same time blow the OP and defeat the protection circuit.
The circuit works by monitoring the voltage across the shunt, R19.
This voltage is amplified by a FET LTP and buffered/rectified by Q1/Q2.
When the resulting voltage exceeds the threshold of the schmitt trigger U1, the circuit locks itself up thanks to positive feedback from U4 and R9 and disables the oscillator generating the gate voltage of the power switching elements M1/M2, while enabling the warning flasher U3.
The fault can be cleared by the push-button, but only if the trip condition has disappeared.
The whole circuit operates at a very low power to allow battery operation, and the MOSfets are driven at ~16V to minimize the Rdson and make it almost purely linear to minimize the distortion.
The network between the shunt resistor and the LTP taylors the frequency response: in this example, the sensitivity is highest at DC and VLF, but still gives protection in the audio band, with a final lift at higher frequencies to better protect the tweeter.
I have designed a stand alone protection circuit, broadly the equivalent of an ordinary fuse, but more flexible, more accurate and having negligible distortion. This will allow me to connect any amplifier without having to worry about internal protections, or the lack thereof.
A completely independent system is less sensitive to failures occurring in the amplifier itself: for example, a paper clip finding its way into an amp can at the same time blow the OP and defeat the protection circuit.
The circuit works by monitoring the voltage across the shunt, R19.
This voltage is amplified by a FET LTP and buffered/rectified by Q1/Q2.
When the resulting voltage exceeds the threshold of the schmitt trigger U1, the circuit locks itself up thanks to positive feedback from U4 and R9 and disables the oscillator generating the gate voltage of the power switching elements M1/M2, while enabling the warning flasher U3.
The fault can be cleared by the push-button, but only if the trip condition has disappeared.
The whole circuit operates at a very low power to allow battery operation, and the MOSfets are driven at ~16V to minimize the Rdson and make it almost purely linear to minimize the distortion.
The network between the shunt resistor and the LTP taylors the frequency response: in this example, the sensitivity is highest at DC and VLF, but still gives protection in the audio band, with a final lift at higher frequencies to better protect the tweeter.
Attachments
Some more details on the operation:
The next pic shows the tripping action caused by a 2A/400Hz current.
The green trace is the gate drive: when the circuit operates, the interruption is very fast and clean thanks to the action of Q4 discharging the gates capacitances.
The following pic shows the frequency response (the whole circuit does not converge fast enough).
The CD4093 has a relatively important dispersion on the thresholds.
The average level can be adjusted by tweaking R3.
On some samples, the hysteresis may also be too large, meaning the circuit will be unable to return to one of its states, even if the threshold is optimally centered.
The cure is to artificially reduce the hysteresis with R20. If required, the value will be in the 10 megohm range.
The quiescent current is a bit disappointing: depending on the 4093 used, it is comprised between less than 200µA and more than 800µA. I had hoped for less than 50µA, but with its input in the linear range, U1 takes most of the current.
In the trip condition, the consumption is higher because of the LED, but thanks to the low duty cycle it remains at ~1.5mA average and normally the circuit will not stay in this condition for long periods.
The compensations are sufficient to allow operation between 6.5V and 10V without appreciable variation of the characteristics.
Additional timing could be inserted between Q1/Q2 and U1 if required.
The sensitivity can be tweaked by adding a parallel resistor between the gates of Q1 and Q2.
The range is chosen with R19: with 0.11 ohm, the DC sensitivity is 900mA, and the medium frequency AC sensitivity is 3.5A average.
If pure DC/LF response is desired, R8 can be zero, and C1 somewhat larger
The next pic shows the tripping action caused by a 2A/400Hz current.
The green trace is the gate drive: when the circuit operates, the interruption is very fast and clean thanks to the action of Q4 discharging the gates capacitances.
The following pic shows the frequency response (the whole circuit does not converge fast enough).
The CD4093 has a relatively important dispersion on the thresholds.
The average level can be adjusted by tweaking R3.
On some samples, the hysteresis may also be too large, meaning the circuit will be unable to return to one of its states, even if the threshold is optimally centered.
The cure is to artificially reduce the hysteresis with R20. If required, the value will be in the 10 megohm range.
The quiescent current is a bit disappointing: depending on the 4093 used, it is comprised between less than 200µA and more than 800µA. I had hoped for less than 50µA, but with its input in the linear range, U1 takes most of the current.
In the trip condition, the consumption is higher because of the LED, but thanks to the low duty cycle it remains at ~1.5mA average and normally the circuit will not stay in this condition for long periods.
The compensations are sufficient to allow operation between 6.5V and 10V without appreciable variation of the characteristics.
Additional timing could be inserted between Q1/Q2 and U1 if required.
The sensitivity can be tweaked by adding a parallel resistor between the gates of Q1 and Q2.
The range is chosen with R19: with 0.11 ohm, the DC sensitivity is 900mA, and the medium frequency AC sensitivity is 3.5A average.
If pure DC/LF response is desired, R8 can be zero, and C1 somewhat larger
Attachments
Elvee,
This is exactly what I need to build.
Should I build it 'as is', or after 4 years you came up
with some improvements/corrections since then?
Is R20 really 100M Ohm ?
Are R1/R2 really needed?
This is exactly what I need to build.
Should I build it 'as is', or after 4 years you came up
with some improvements/corrections since then?
Is R20 really 100M Ohm ?
Are R1/R2 really needed?
Sorry Minek,
I have forgotten to update this thread.
There is indeed a final version (thoroughly tested, but not yet installed in the speakers):
A.S.P. - Version 2.0
I have forgotten to update this thread.
There is indeed a final version (thoroughly tested, but not yet installed in the speakers):
A.S.P. - Version 2.0
- Status
- Not open for further replies.