I would fuse just the 50Hz/60Hz AC power transformer primary with an adequately rated fuse rather than fuse the DC rails for safety purposes. Here's why: Take, say, a standard 3AG sized glass fuse. If it gets overloaded with AC, the element inside melts, and the plasma that forms inside the fuse and its holder extinguishes on the next zero crossing of the AC power. With DC, the current never goes to zero, and the plasma can keep burning, causing overheating and possibly a fire. That's why fusing the AC side is safer. Of course, if the DC voltage is low and power is low, then a fuse in the DC rail could work. In a high powered application, though, no.
It is dangerous not to fuse an amplifier supply rails. They are there to act as a last resort in case of a catastrophic failure on the amplifier module that could lead to a fire. Further, given the energy in the reservoir caps, there's a good chance tracks will be burnt off the PCB etc.
Modern fuses are designed to clear safely once thier I^2T limit is exceeded. I've never heard of a fuse not clearing after it has opened - that would be a serious design flaw in what is generally considered a critical safety application.
1. fuse each amplifier module
2. fuse the amplifier primary side
Modern fuses are designed to clear safely once thier I^2T limit is exceeded. I've never heard of a fuse not clearing after it has opened - that would be a serious design flaw in what is generally considered a critical safety application.
1. fuse each amplifier module
2. fuse the amplifier primary side
This while I experienced DC rail fuses to cause defective loudspeakers (instead of protecting anything) as the manufacturer was too cautious and a fuse blew at power on. It is also just 1 blowing and never both of course. Just a mains fuse and a loudspeaker protection/DC detection circuit and all is good. The semiconductors are already shot anyway when such a fuse blows, it makes no sense as no fuse is fast enough to do anything useful in that part of the circuit. Maybe if the loudspeaker/amplifier price ratio is not too far off it does 🙂
Last edited:
I had a fuse in the DC rail of a high power amplifier with+/-100Vdc rails go up like that. It was in a fuse holder which contained the plasma long enough for it to burn a bit before it extinguished. If the fuse were in an open-air holder, then the plasma would quickly diffuse away and no sustained burn could happen. I had to replace the burned fuse holder. In fuse testing, fuses are tested in open air conditions, not confined to an enclosed holder.
Witnessed very high current fuses to blow in energy distribution and high power DC systems. Spectacular as fuses then sometimes also make scary sounds.
Fuses have an interrupt rating, and in glass ones it can be surprisingly low (A few tens of amps in some cases!), the HRC ones which are ceramic cased and sand filled are usually much better in this regard, but it is important to read the datasheets and actually do the sums, particularly if you are the type to get carried away with bus capacitance.
While semiconductor protection fuses do in fact exist, they are really intended to protect hockey puck sized thyristor and IGBT modules in industrial motor drives, they will seldom protect an amplifier output stage, for that do current limit and VI limiters of various kinds exist.
Fuses usually do not protect electronics, they stop fires (and sometimes limit the damage to the electronics).
While semiconductor protection fuses do in fact exist, they are really intended to protect hockey puck sized thyristor and IGBT modules in industrial motor drives, they will seldom protect an amplifier output stage, for that do current limit and VI limiters of various kinds exist.
Fuses usually do not protect electronics, they stop fires (and sometimes limit the damage to the electronics).
Yes I know ultrafast fuses as used in high power electronics. The ones that make CLICK sound when blowing. In an audio amplifier the doom scenario is unlikely when the main fuse is chosen right. The doom scenario of phone chargers catching fire and cheap SMPS melting is way more realistic. 24/7 powered on mediocre stuff is what costs lives not an audio amplifier that is switched off after use. That is why real power switches should be used instead of standby circuits and/or leaving stuff powered on unattended.
Last edited:
I prefer all my amps have DC rail fuses but clips were not added to this particular board which also contains rectifier and caps.
It would be messy to try to modify and add DC rail fuses - I'll try for low amperage(that is possible) fuses on secondary windings.
Yes, I have soft start and DC protect on my amps, thanks!
It would be messy to try to modify and add DC rail fuses - I'll try for low amperage(that is possible) fuses on secondary windings.
Yes, I have soft start and DC protect on my amps, thanks!
Attachments
DC offset and short circuit protection is a must for amps including rail and primary side fuses IMV. If you have grossly oversized the OPS on low rails, a case can be made for not including OPS short circuit protection and just relying on the rail fuses - this is what I did on the sx and Kx2 Amps. I use ‘T’ fuses.
I do not like VI limiter protection and prefer to just oversize the OPS. Bob Cordell talks a bit about this in his book - he’s not a fan either.
VI limiters are tricky things, granted and they sound HORRIBLE if they activate, so you need enough output stage to avoid that for any normal situation. If however you have teenagers who like to party and are not too bothered by running the amp way into clipping, then a VI limiter will (hopefully) protect the sand from the abuse.
I generally avoid classic VI limiters, preferring a fixed current limit at maybe 4A amps per device pair which holds the short circuit current down for long enough to let a micro that is reading device voltage and current as well as heatsink temperature throttle the thing back if you start getting too close to the SOA limits. This has the advantage that the micro is an easy and cheap way to do the sums to calculate device dissipation and to do the integration to get to a junction temperature estimate, it lets you push things quite a bit harder then you otherwise could, and a 'subtractive' limiter at the input is almost blameless when not active. I commend the STM32F7 line to your attention, ADCs, DACs, OPAMPS, PWM IO and a floating point unit all in a QFN64, for a few euros each.
It is also worth noting that a series class G design will usually work with just a current limit, no need for VI limiting there as the output will only have maybe 20V across it worst case, SOA is something that starts to really matter when you have 50V++ rails.
Fuses are not generally for semiconductor (or speaker) protection, but they do prevent fires.
I generally avoid classic VI limiters, preferring a fixed current limit at maybe 4A amps per device pair which holds the short circuit current down for long enough to let a micro that is reading device voltage and current as well as heatsink temperature throttle the thing back if you start getting too close to the SOA limits. This has the advantage that the micro is an easy and cheap way to do the sums to calculate device dissipation and to do the integration to get to a junction temperature estimate, it lets you push things quite a bit harder then you otherwise could, and a 'subtractive' limiter at the input is almost blameless when not active. I commend the STM32F7 line to your attention, ADCs, DACs, OPAMPS, PWM IO and a floating point unit all in a QFN64, for a few euros each.
It is also worth noting that a series class G design will usually work with just a current limit, no need for VI limiting there as the output will only have maybe 20V across it worst case, SOA is something that starts to really matter when you have 50V++ rails.
Fuses are not generally for semiconductor (or speaker) protection, but they do prevent fires.
VI limiters (foldback current limiting) had two problems: soft knee and possible tripping with a reactive load.
I use fixed-current limiting with a sharp knee and fuses on the AC and DC. My amplifier should be able to survive a momentarily shorted output.
I notice that many hobby designs lack current limiting. I view that as a shortcoming.
Ed
I use fixed-current limiting with a sharp knee and fuses on the AC and DC. My amplifier should be able to survive a momentarily shorted output.
I notice that many hobby designs lack current limiting. I view that as a shortcoming.
Ed
Early Marantz protection circuits are worth looking at. Current limits were lower when the output voltage was lower.
VI limiters can be made to have a sharp knee. You need CFPs. Inductive load protection is simple, and you need that anyway. Fixed current limiting will only get you so far, when amps start getting big. 50 amps into a short circuit would be undesirable in many ways.
Build amplifiers with adequate output power for normal use and current will be limited by design.
lol!
Let's talk about sane home amplifiers driving reasonable loads. I think I'm on the edge driving PSB Stratus Gold's with a Yamaha PC2002 really loud (near clipping). Those are 4 R, 86 dB/watt speakers. That is well beyond where most home users are ('cause I'm crazy and love to feel the music).
To be honest, the designer of a speaker that is extremely reactive or has a very low impedance isn't too smart. No amplifier delivers it's best performance when it is unhappy, so if you want good sound, why would you buy such a thing?
Let's talk about sane home amplifiers driving reasonable loads. I think I'm on the edge driving PSB Stratus Gold's with a Yamaha PC2002 really loud (near clipping). Those are 4 R, 86 dB/watt speakers. That is well beyond where most home users are ('cause I'm crazy and love to feel the music).
To be honest, the designer of a speaker that is extremely reactive or has a very low impedance isn't too smart. No amplifier delivers it's best performance when it is unhappy, so if you want good sound, why would you buy such a thing?
Fuses have voltage ratings, and they are different for AC and DC. Ceramic- or sand-filled fuses are needed for higher voltages and currents. Some fuse datasheets even have graphs of current-breaking ability against voltage. The common 20mm glass cartridge fuse isn't always appropriate, but usually it is OK, and ceramic versions are available - for instance the datasheet for Littelfuse 487 series (ceramic 20x5mm) gives a max breaking current of 300A at 420Vdc.