Circuit breaker for amplifier

[JonSnell Electronic]

Most manufacturers use anti surge fuses in the mains supply and fast blow for HT supplies after the main smoothing. This protects the output transistors/power output stage and then NAD used 2Amp thermal breakers in series with the loudspeaker connection, this helped to protect the loudspeaker from being cooked whilst having a time delay when overloaded.
That was quite successful but a relay and detecting circuit can protect against unwanted DC potentials.

[edbarx]

Originally Posted by JonSnell Electronic

Most manufacturers use anti surge fuses in the mains supply and fast blow for HT supplies after the main smoothing. This protects the output transistors/power output stage

I searched for quick blow SMD fuses with high current ratings like 10A and found they exist. The reason for choosing SMD, is because space is limited and using connecting wires to fuse carriers, would impact negatively the amplifier's performance. A narrow cut in the rails tracks should be enough.

For the loudspeaker there is already a fuse, but this is rated at 20A slow blow. Since this is huge, I would like to use a quick blow fuse with a rating of around 2A - 3A in series with the speaker. At 8Ω, output power should peak to 8*2^2 = 32W and 8*3^2 = 72W. For me, the latter seems enough.

I found the 10A quick blow SMD fuses at Farnell. Suggestions of other suppliers, and alternative fuse ratings but in SMD format, are most welcome.

AC mains input to the amplifier can be interrupted with an existing spring loaded 'fuse'. The only obvious condition when this fuse should activate, is to protect the large toroidal transformer against excessive current which is 10A per rail, if my estimates are correct. [This is basing on the fact the loadspeaker is already 'protected' with a 20A slow blow fuse. So, I am concluding, this implies 10A per rail.]

[amyhimeesamaa]

Originally Posted by edbarx

I would like to know whether ready made 0.5A to 5A presettable circuit breakers exist. The circuit breaker is required to trip immediately when a preset current limit is reached.

Since, this circuit breaker operates before the amplifier mains transformer, when it trips, the amplifier's large electrolytic reservoir capacitors will still contain stored charge. At 50V DC and 6600uF, the stored energy will be 8.25 Joules

[edbarx]

This amplifier in question is vastly modified, with only the power supply, the driver transistors, and the power transistors remaining connected as in the original circuit. The output stage itself has been modified by adding a 0.1Ω resistance to add some negative feedback and provide for thermal stability.

Fuse protection is currently provided as in the original circuit with a huge 20A slow blow fuse between the amplifier and 0V node. To tell the truth, I am NOT comfortable with this setup and would like to change it. The current fuse can be removed and the track shorted with a thick wire, so that, the 0V track always remains intact. Fuse protection can be implemented by breaking the +/- rails just after the smoothing stage and inserting two 10A fast blow fuses as is usually done. Had it not been, for the utter bizarre design of this amplifier, I would have preferred to repair the original circuit, instead of going the treacherous route of a complete circuit overhaul.

Datasheets for quick blow fuses is indicating they blow after a delay of around 10ms. This may be a long time for semiconductors, but it should offer protection against getting the entire power stage destroyed whenever a fault develops. The latter is an opinion, and I may be wrong. Amplifier service technicians know far better than me in this regard, and their advice is vastly appreciated.

[Mark Tillotson]

Silicon can fail much faster than 10ms, especially in a cascade where a power device shorts, puts rail-voltage on the driver output, which melts shorted, putting rail voltage on the pre-driver, which then pops, etc etc. Of course you can be lucky, but its probably a matter of luck whether the fuses protect the amp, or the amp protects the fuse! Its not an argument against having the fuse (it can save the pcb tracks for one thing).

[indianajo]

I captured the driver outputs, always below 1 v, with 3 v 3 W zener diodes to analog ground. Then I replaced a jumper going from driver to the 5 output transistor bases with a 5 amp automotive fuse. Hopefully the zeners will save the drivers predrivers op amps, 50 v rated capacitors, etc etc the next time the output transistors go. I put 124 parts in that amp. Learned a lot crawling through the circuit, though.
I drilled a #45 hole in the blades of the automotive fuses to solder in a component lead. Then they soldered right in the holes the jumper used to be in.

[dmills]

Fuses protect wiring, to protect the sand you need something very much faster.

I quite like Bob Cordells approach of over sized output stage and using diodes from the output to either side of the bias splitter so that a large over current event will develop sufficient voltage across the emitter resistors to cause the diodes to steal base current from the drivers.

SOA is an issue so you want something else to disconnect the load quick like, but a couple of small signal mosfets and a couple of those Vishay opto gate drivers would get it done. Going for the 'short the output to the ends of the VBE multiplier feels workable and does not need any high current doings.
VI limiting is the alternative but that tends to sound really gnarly if it ever activates.

[indianajo]

Originally Posted by dmills

Fuses protect wiring, to protect the sand you need something very much faster.

In my scheme if an output transistor lets current out the base line, the zener clamp conducts in nanoseconds, keeping high voltage off driver emitters. Max zener current is 487 ma; hopefully the die shorts across and blows the 5 a fuse before the bond wire goes. Cost about $.40/channel and required no circuit board mods.
TVS diodes take more current but the lowest voltage one I've found in stock is about 6 vac. EB junctions blow about 7 v backwards.

[PRR]

Originally Posted by edbarx

... ... ... Datasheets for quick blow fuses ... 10ms. This may be a long time for semiconductors, but it should offer protection against getting the entire power stage destroyed...

All your transistors, and even power diodes, CAN be toasted in 10mS.

Power transformers and large capacitors take *minutes* to be "crispy".

Since the large passives may be more than half the cost of a build, protecting them is wise, but also easy.

A complete set of transistors is what, 25% of the cost? Declining in recent years (you don't wanna know what I paid for my first TIP122). Try to add $2 of "protection" circuit, Volt-Amp. This only protects against expected faults, not The UnExpected, but is cheap. Then also put your first-repair power devices in your First build, double-up, so it has way more beef than "should" be needed. Even the for-profit companies now go with rows and rows of power devices.

[dmills]

If you do the VI limiter thing, make the stage really butch so it can be configured to never even come close to activation with program material (VI limiters sound NASTY when they activate), this is not the expense it once was when even at Mouser a 200W, 15A, 150V part with reasonable SOA and not too painful second breakdown can be had for a couple of bucks.

Do fuse the transformer secondary before the rectifiers, the objective here being the protection of the transformer from a failed rectifier bridge, fuses way bigger then the design RMS current are a good thing here, and remember that the current here has high RMS/Average ratio and is subject to the cap charging inrush.

Personally I would never trust a fuse to protect sand (It generally goes the other way), use electronics to protect the transistors, fuses to protect the wiring and large passives.
I might make an exception for the kind of UFF fuse sometimes used to protect butch thyristor packs, but that thing will cost you more then your transistors.