I have an idea that I think is better than putting some sort of switch between the amplifier output and speaker.
Put the switches between the power supply's main voltage rails and the amplifier. This could use FETs or BJTs that would normally carry all the amplier's supply current, but only drop a fraction of a volt, thus have very little effect on normal operation. On the amplfier side would be the usual local decoupling, 25uF electrolytic and 0.1uf which would discharge quickly when a fault is detected and the protection transistors are switched off. This would keep the 'sacred' amplifier-speaker connection intact, but cut power virtually instantly when a fault is detected.
If course the fault-detect circuitry would run on the power supply side of the switches. And since the switches are there, there are failure modes that should be tested for, maybe one rail of the supply dies but the other doesn't. As a caution, it may be possible for the switch transistors to fail shorted. Perhaps a "speaker protection test" mode could be added to insure both switch transistors are working properly, only to be used with a resistive (non-damageable) load.
Feel free to tear the idea apart and say why it won't work.
Put the switches between the power supply's main voltage rails and the amplifier. This could use FETs or BJTs that would normally carry all the amplier's supply current, but only drop a fraction of a volt, thus have very little effect on normal operation. On the amplfier side would be the usual local decoupling, 25uF electrolytic and 0.1uf which would discharge quickly when a fault is detected and the protection transistors are switched off. This would keep the 'sacred' amplifier-speaker connection intact, but cut power virtually instantly when a fault is detected.
If course the fault-detect circuitry would run on the power supply side of the switches. And since the switches are there, there are failure modes that should be tested for, maybe one rail of the supply dies but the other doesn't. As a caution, it may be possible for the switch transistors to fail shorted. Perhaps a "speaker protection test" mode could be added to insure both switch transistors are working properly, only to be used with a resistive (non-damageable) load.
Feel free to tear the idea apart and say why it won't work.
I have an idea that I think is better than putting some sort of switch between the amplifier output and speaker.
Put the switches between the power supply's main voltage rails and the amplifier. This could use FETs or BJTs that would normally carry all the amplier's supply current, but only drop a fraction of a volt, thus have very little effect on normal operation. On the amplfier side would be the usual local decoupling, 25uF electrolytic and 0.1uf which would discharge quickly when a fault is detected and the protection transistors are switched off. This would keep the 'sacred' amplifier-speaker connection intact, but cut power virtually instantly when a fault is detected.
If course the fault-detect circuitry would run on the power supply side of the switches. And since the switches are there, there are failure modes that should be tested for, maybe one rail of the supply dies but the other doesn't. As a caution, it may be possible for the switch transistors to fail shorted. Perhaps a "speaker protection test" mode could be added to insure both switch transistors are working properly, only to be used with a resistive (non-damageable) load.
Feel free to tear the idea apart and say why it won't work.
The MOSFET "switch" is lower resistance than some speaker wires !
Read about the new "trenchfet" low RDS devices - milliohms !!
Shutting down the PS is also an option , but the stored charge in the capacitor
banks might do some damage (I have about 1.2Kuf on my output boards)..
Also, with just one "switch" ... non -critical faults (thermal/overload) can be
controlled at one point.
Your way ... (switching the PS -dadod ?) has been used on the forum , BTW.
PS - for the Sub , the "sacred" connection will be plain ol' relays , but
a sub is not as critical.
When I get a higher current 4R sub - I might upgrade to the MOSFET relay
for it.
OS
Oh, you mean a "crowbar" circuit - that does seem a bit extreme:
Crowbar (circuit) - Wikipedia, the free encyclopedia
I'm not sure how you mean "controlled at one point" but I was thinking both the positive and negative switches would be switched off by the same control line, controlled by whatever thermal and/or DC offset detect circuit you use.
dadod? "Total Posts: 2,479" I suppose I should start looking through all those posts...
Sub is a little lower rail voltage @ +/- 62V.
So , just 20V lower FET . 2 X 80 =160 .... should work. I already put the
relays in but would just have to unsolder 1.
OS
Nope , not a crowbar.
Just a bi-directional switch with < a milliohm ON resistance
and 10 ^ 7 ohm OFF resistance. A true "21st century" relay with infinite operational
life (cycles) in normal operation.
"Controlled at one point" .....
A thermal or current overload would not require a power supply disconnect -
reset and play some more.
DC fault is usually a sign of greater problems (device short) - best to
totally disconnect $$$ expensive loudspeakers.
Think of all possible fault scenario's -
output oscillates - even if you switched off the PS it would use the cap's power
for a while.
output offsets - output stage would still be to a rail until capacitance is used up.
You can't run most output stages without at least SOME local decoupling -
(after a PS switch).
To disconnect the speaker from any amplifier circuitry TOTALLY .... is the
only way.
PS - this control board also has the ability to monitor the rails ...BTW.
Also , I could use software to tell the control board to even run the power
supply through the soft-start resistor for a "hard fault" (like DC - short) !!
AND disconnect the speaker.
And "summer mode" ... I have a script that will tell one of the 3 extra
control board outputs to run a fan on "thermal" instead of shutting down.
OS
Hi,
I have been doing that for years. I shutdown the rails voltage to protect the speakers in case of a high voltage output eliminating the speaker output relays. My theory is that since I am regulating the rails voltage with the microprocessor the large capacitors bank are in the upstream of the regulator and used only one capacitor downstream so if the speaker current reached the set point the control voltage to the regulator will the be shut down and let the speaker load just discharge only one capacitor. I protected the speakers by using a hall effect current sensor that you can buy cheap as a module in Ebay. Just read the analog output with the micro compare it with the setting and shutdown the rail voltage to protect the speakers. No need the use of a mechanical or SSO relay. Attached it is a drawing of my design. It is a simple designed and you can regulate both rails voltages up to +/- 140 volts.
I have been doing that for years. I shutdown the rails voltage to protect the speakers in case of a high voltage output eliminating the speaker output relays. My theory is that since I am regulating the rails voltage with the microprocessor the large capacitors bank are in the upstream of the regulator and used only one capacitor downstream so if the speaker current reached the set point the control voltage to the regulator will the be shut down and let the speaker load just discharge only one capacitor. I protected the speakers by using a hall effect current sensor that you can buy cheap as a module in Ebay. Just read the analog output with the micro compare it with the setting and shutdown the rail voltage to protect the speakers. No need the use of a mechanical or SSO relay. Attached it is a drawing of my design. It is a simple designed and you can regulate both rails voltages up to +/- 140 volts.
I like the approach of shutting down the rails between the caps bank and the amp. In extreme cases, it can be used in parallel with disconnecting the speakers, making the whole system bullet-proof safe for the speakers
That would be easily done with the mosfet relays as well. A power relay board might be a good idea here. Board mount T90 relays could easily handle softstart and main transformer power.
I like the approach of shutting down the rails between the caps bank and the amp. In extreme cases, it can be used in parallel with disconnecting the speakers, making the whole system bullet-proof safe for the speakers
Think of a scenario where just one output starts to run away. Almost had that
happen to me ....
It would load one rail down (slightly). Might be hard to detect through offset
or any other alarm.
This is where a divided (proportional) zener protected analog input comes in.
Even if you were to just monitor this for another alarm to be used with
the present setup - better.
PS- I saw the rail sag slightly before I quickly shut down the PS.
OS
I've been thinking about the scenario too. I was thinking about an opto per pair of outputs. Makes for a pretty huge protection board though.
There are many other easier signs of "impending doom" that could more easily
be monitored analog.
With this present board (below - both work very well .. BTW
) ,there arestill a lot of options available through the script.
- You could power the amp through the inrush resistor .... Ha ha
!- the 2 extra outputs (D12/13) could actually be the PS MOSFET controls.
(leaving the unused "tubes" as well for fancy LED indication).
I have tested the AC fault ... had a "brownout" during an ice storm !!! 680pf
seems to work fine for the U7 decoupling.
The rails sense works , all regulators and IC run ice cold.
A 32Kuf cap bank takes <2 seconds to charge 0-60V through the inrush , main AC
snaps on after 4 seconds for 63V full rail .... PS control is perfect.
I have 4k ms/500ms/4K ms set for my start-up script.
Someone with a 100K+uf bank might need 5K/500/5K.
I'm just using a little 700ma wallwart trafo to power the board (in photo) 14V drops
to 13 with both large relays in "overlap" (both on).
I'm going to make a small adjustable 0-12V trimmable voltage source to test
exactly what alarm points the rest of the fault detectors trip at.
....Tidy up the dual LED script ( those bicolor LEDS are real bright !!)
Then my amp will be finally complete -
EXCELLENT work and design , Jeff (and Val)!!
OS
Attachments
There are many other easier signs of "impending doom" that could more easily
be monitored analog.
With this present board (below - both work very well .. BTW
still a lot of options available through the script.
- You could power the amp through the inrush resistor .... Ha ha
- the 2 extra outputs (D12/13) could actually be the PS MOSFET controls.
(leaving the unused "tubes" as well for fancy LED indication).
I have tested the AC fault ... had a "brownout" during an ice storm !!! 680pf
seems to work fine for the U7 decoupling.
The rails sense works , all regulators and IC run ice cold.
A 32Kuf cap bank takes <2 seconds to charge 0-60V through the inrush , main AC
snaps on after 4 seconds for 63V full rail .... PS control is perfect.
I have 4k ms/500ms/4K ms set for my start-up script.
Someone with a 100K+uf bank might need 5K/500/5K.
I'm just using a little 700ma wallwart trafo to power the board (in photo) 14V drops
to 13 with both large relays in "overlap" (both on).
I'm going to make a small adjustable 0-12V trimmable voltage source to test
exactly what alarm points the rest of the fault detectors trip at.
....Tidy up the dual LED script ( those bicolor LEDS are real bright !!)
Then my amp will be finally complete -
EXCELLENT work and design , Jeff (and Val)!!
OS
This is all Valery's design. I just reverse engineered it and converted some discrete sections to ICs.
The 680pF caps are good to know. I should hook up the scope and see how small I can go with them and still have acceptable voltage to the watchdog. It will affect reaction time of the circuit.
OK , Val rocks !!
I just realized I have 3 colors red/green/ and yellow !
ATmega pin 12 is "powerLED2" , pin 4 - "powerLED"
Turn both leds on full power = yellow.
Hmm , have to figure out what to designate for the new color ?
soft-start maybe ? fancy ... like an OEM !
Wow , so easy to determine status across the room - one color was bummer.
(no offense , Val - still "rock" !)
OS
I just realized I have 3 colors red/green/ and yellow !
ATmega pin 12 is "powerLED2" , pin 4 - "powerLED"
Turn both leds on full power = yellow.
Hmm , have to figure out what to designate for the new color ?
soft-start maybe ? fancy ... like an OEM !
Wow , so easy to determine status across the room - one color was bummer.
(no offense , Val - still "rock" !)
OS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.