Protecting expensive loudspeakers from being damaged by a faulty amplifier is important, so how about this alternative method?
Does one really need to use an electromechanical relay in the amplifiers loudspeaker output? Consider contact resistance of relay's, physical size, and minimum switching current.
If one has a blown power mosfet causing d.c or failed input decoupling capacitor, or other component faults, detect the d.c offset fault at startup, then if a fault is detected switch off the main amplifier power supply; find a main LLC power supply with enable.
One will need a second isolated and independant low power supply to power the d.c detection circuit, I have chosen a PSU that make the +/- 15 Volt rails to be separate, XP power, or Tracor, modules.
The +/- 15V is switched on first, the amplifier signal input is grounded, then the main power supply is switched on, if there is d.c offset the main power supply is immediately switched off, the +/- 15V remains on. d.c offset circuit uses a fast time constant when the amplifier switches ON, slow time constant when amplifier is running, consider >100V p-p at 5HZ must not trip the d.c detector.
Does one really need to use an electromechanical relay in the amplifiers loudspeaker output? Consider contact resistance of relay's, physical size, and minimum switching current.
If one has a blown power mosfet causing d.c or failed input decoupling capacitor, or other component faults, detect the d.c offset fault at startup, then if a fault is detected switch off the main amplifier power supply; find a main LLC power supply with enable.
One will need a second isolated and independant low power supply to power the d.c detection circuit, I have chosen a PSU that make the +/- 15 Volt rails to be separate, XP power, or Tracor, modules.
The +/- 15V is switched on first, the amplifier signal input is grounded, then the main power supply is switched on, if there is d.c offset the main power supply is immediately switched off, the +/- 15V remains on. d.c offset circuit uses a fast time constant when the amplifier switches ON, slow time constant when amplifier is running, consider >100V p-p at 5HZ must not trip the d.c detector.
Or you could use the age old Diac feeding the gate of a Triac sat across the output. As soon as DC flows in either direction, governed by an R C circuit for a few milliseconds delay, the triac turns on until the supply stops after a supply fuse has blown. It even works when the amplifier produces square wave under over driven conditions.
No contact to weld or arc over!
This has been used on high power amplifiers for years and saved countless speaker systems from frying.
No contact to weld or arc over!
This has been used on high power amplifiers for years and saved countless speaker systems from frying.
Hi, we use very reliable power MOSFET-based solid-state (SS) relays for disconnecting the speakers and optionally disconnecting the rails in case of trouble. Please see "Amp Control Boards" section on the website in my signature.
Attached is the schematic of the DC offset protection module, utilizing the SS relay for controlling the speaker. This is a part of the complex microcontroller-based amp control / protection system. We've got many options of the system for different requirements, including rather simple ones, and PCBs in stock - let me know if this may be of interest for you
Cheers,
Valery
Attached is the schematic of the DC offset protection module, utilizing the SS relay for controlling the speaker. This is a part of the complex microcontroller-based amp control / protection system. We've got many options of the system for different requirements, including rather simple ones, and PCBs in stock - let me know if this may be of interest for you
Cheers,
Valery
SSR with IPB025N10N3 is safely usable only for amps with Ub max. +-50V (Vds 100V).. Also clamping diodes to +-Ub at load side needed (inductive voltage spikes at fault conditions switch of )
look at geofex.com
They don't have a direct url for the R J Keene amp protection circuit using mosfet disconnect, but down toward the bottom of the index there is an article listed as "three amp protection circuits". Article title "A DC fault protection circuit for audio amplifiers".
The rail disconnect circuit may save 19 of your $5 each output transistors if one blows. The triac shorting circuit described in post 2 overstresses your already hot output transistors, and may not blow the breaker (mine didn't, it melted the lands off the board instead). I use panasonic apv1122 fet drivers to drive the fets, requires no power supply above the rail and has leads unlike some surface mount ICs used by others on here.
I reused the DC detector circuit in the PV-1.3k from the triac "crowbar" to instead latch a 74HC74 IC and stop driving the rail fets (from Q) and extinguish the green 'OKAY" led in series with the fet driver led. Qbar of the 74HC74 drives the red fault led. (through current booster pn2222 transistors, 74hc74 only produces 1 ma APV1122 needs 10 ma) You can see the triac crowbar on peavey PA amp schematics on eserviceinfo.com
They don't have a direct url for the R J Keene amp protection circuit using mosfet disconnect, but down toward the bottom of the index there is an article listed as "three amp protection circuits". Article title "A DC fault protection circuit for audio amplifiers".
The rail disconnect circuit may save 19 of your $5 each output transistors if one blows. The triac shorting circuit described in post 2 overstresses your already hot output transistors, and may not blow the breaker (mine didn't, it melted the lands off the board instead). I use panasonic apv1122 fet drivers to drive the fets, requires no power supply above the rail and has leads unlike some surface mount ICs used by others on here.
I reused the DC detector circuit in the PV-1.3k from the triac "crowbar" to instead latch a 74HC74 IC and stop driving the rail fets (from Q) and extinguish the green 'OKAY" led in series with the fet driver led. Qbar of the 74HC74 drives the red fault led. (through current booster pn2222 transistors, 74hc74 only produces 1 ma APV1122 needs 10 ma) You can see the triac crowbar on peavey PA amp schematics on eserviceinfo.com
Last edited:
Hi BV,
They never see anything higher than a single rail voltage in the worst case, so there is plenty of headroom even with +/-75V (we normally don't go higher than that).
Also, in normal operation both MOSFETs are open, so Vds is very low even at the high swing. As soon as protection triggers, we've got maximum a single rail potential at the MOSFETs. As far as we always shut down the amplifier in this case - those rails go down.
Clamping diodes - we assume they always present at the output of the amplifier (between the output and both rails).
Cheers,
Valery
There has been a lot of progress in nfets since the 2001 geofex article. I picked up a dead 50" LED TV which was chock full of IXTK62n25 nfets which are rated at 62 A 250 V in a TO264 package.
I don't like the TRIAC idea, it over stresses other healthy components way to much, fatigue, the amplifier may fail again after being repaired during a live concert.
I suppose the worst case set up is multiple amplifiers and active crossovers.
With 100 Volt D.C fault condition, How many milliseconds before the Bass speaker blows? compression driver blows? tweeter blows?
I would never use a tweeter without a series protection capacitor.
I suppose the worst case set up is multiple amplifiers and active crossovers.
With 100 Volt D.C fault condition, How many milliseconds before the Bass speaker blows? compression driver blows? tweeter blows?
I would never use a tweeter without a series protection capacitor.
Imagine one shorted output transistor (output at + or -Ub ), faulty current via load and than turn off (complex load !) in this situation. One MOSFET is exposed to inductive peak, if clamping diodes are used, this peak is limited to sum of the + -Ub . Without clamping diodes at load side it is "unlimited" . Clamping diodes at output side protects output transistors, not SSR. The second MOSFET is bypassed through the body diode. Try it in simulation .
Last edited:
I had the following test - not intentionally - on the working amplifier (+/-70V rails) with the speakers connected, I have accidentally shorted the output terminal to the positive rail with a screwdriver. It followed by the barely hearable click in the speaker and then a relatively loud click of the PSU relay, shutting down the amplifier. That means the speaker could only see a few volts before it's been disconnected.
In fact, protection was triggered not by the DC offset sensor, but the over-current sensor, that triggers much faster as its time constant is much lower.
However, it probably makes sense to consider the situation where protection is triggered at the full swing signal when the load is disconnected at the momentary potential close to the rail. I will try to simulate this one.
That's why it makes sense to use a complex protection system, watching not only the offset at the output. I normally watch the peak current through the output transistors (very fast reaction), AC failure (mains cord is disconnected unexpectedly) and overtemperature (the slowest one, but very useful sometimes, saving from some real troubles before they occur).
I think an electromechanical relay can take 20 milliseconds to open, this is too slow, and can be even slower if a diode is in parallel with the relay coil to protect the switching transistor.
I want a design without any electromechanical relays.
My circuit does the same and senses overcurrent with immediate shut down, just the energy stored in the secondary side PSU capacitors to consider.
I don't want to use a current sensor, use mosfet VGS to estimate current, check out the schematic I have uploaded that shows how this was done.
I want a design without any electromechanical relays.
My circuit does the same and senses overcurrent with immediate shut down, just the energy stored in the secondary side PSU capacitors to consider.
I don't want to use a current sensor, use mosfet VGS to estimate current, check out the schematic I have uploaded that shows how this was done.
For the best results, it makes sense to disconnect the amplifier from the tails at the same time you disconnect the speaker (SS relays as well).
If a mosfet blows short in service, the opposite rail mosfet will absorb the stored energy in the capacitor not the speaker.
If a mosfet has blown short on a previous power cycle, with my circuit the PSU will shut down before the capacitors have charge to +/- 15V; also with 100 KHZ LLC PSU the capacitors are smaller and E=0.5 C V^2
If.. And what if blows fuse only one in polarity PSU rails (they should be behind PSU capacitors), and opposite output transistor is shorted?? It is worst case, but it can happen. Protection should be safe and the best way is to disconnect speaker.
Last edited:
For the rails disconnect working properly, there must be no high-value capacitors on the OPS board. In case of trouble, SS relays disconnect the board from the PSU (and from the big filter cans). Thus, it doesn't matter what is shorted somewhere in the OPS.
I know a number of protection circuit systems working perfectly based on this principle.
I know a number of protection circuit systems working perfectly based on this principle.
Why use two SSR for PSU rails, if one reliable SSR in series with load can do the same "multipurpose" job for delayed turn on, DC fault, thermal protection....Depends only on driving circuit.
Rails "switch" must be able to whistand and disconnect PSU capacitors short circuit (faulty output transistors, crossconduction..) current limited only by Re resistors and parasitic impedance of PCB!
Short circuit protection should be (IMO) realized with dual or triple slope SOAR protection with dynamic delay.
Rails "switch" must be able to whistand and disconnect PSU capacitors short circuit (faulty output transistors, crossconduction..) current limited only by Re resistors and parasitic impedance of PCB!
Short circuit protection should be (IMO) realized with dual or triple slope SOAR protection with dynamic delay.
The SSR for the power supply requires only one mosfet per rail so it's pretty much the same parts count as a SSR on the output. I normally use both, rail switches and output relays. The rail switches can save a lot of collateral damage and burnt part stink when an output device fails. I don't bother with multi-slope output current protection myself, an extra set of output devices is fairly cheap. Rail fuses protect from dead shorts and simple emitter resistor voltage drop measurements triggering an optoisolator protect from the high current transients.
- Status
- Not open for further replies.