fuses on amp power rails

I fuse things almost habitually.
If you size them correctly, then by the time the fuse opens the sand is all blowed up anyway and the fuse just prevents smoke becoming flame, they also serve as an easy way to identify roughly where the fault is....

One place people often miss for fuses is the transformer secondary, where they protect the transformer from a diode failure (Not as uncommon as would be nice), but note that the RMS/Average ratio is massive here, because there is only current for a short period at the top of each half cycle, so you need fuses much larger then you would expect, the current here is also harmonically rich, so you need to pay attention to wire dress.

Speaker protection is its own can of wyrms, and I still like the back to back mosfets with opto isolated gate drive for this, much nicer then under rated relays, and mercury displacement relays have gone rather out of fashion.
One neat approach to speaker protection I have seen is a power switching mosfet in each supply rail and just turn off the DC feeds if something goes wrong! It is outside the feedback loop so essentially blameless, but you do need to watch how much energy is stored in the caps on the amp board, because that energy is winding up in your voice coils if it all goes sideways.
 
  • Like
Reactions: anatech
I always fuse the amplifier rails with ‘T’ fuses

5A for class A up to 15W RMS
8A for 100-120W class AB
8A for 60W class A
10A for 200-250 W class AB

I always include reverse bias diodes downstream from the fuses.

The fuses are there to prevent fire in case of a catastrophic failure.

YMMV 🙂
 
  • Like
Reactions: anatech
This long thread tell us how important fusing is. But many of us don't pay much attention to this.
There are other things that should need attention in DIY that are rarely looked at. In designing protection you normally start from a "single failure". What if "that" goes wrong? What happens? Things like when you short the output, short a single output device, one of the filter capacitors, any of the tantalums some like to use, a recifier diode, a wire comes loose... and make sure it stays "safe". Maybe not your speakers but surely your house.

Certainly here where much overspec'ed power supplies are used this gets important. 600VA torodial transformers, 40000µf/50V is a serious amount of power when it goes wrong. That welds copper wires several mm² with a short.
 
The challenge for a circuit designer is picking a fuse (amperage & time response) that always trips if there is a failure, but never blows in daily operation.
Fuses in amps are similar to breakers in residential installations. They are there to protect the "installation" such as PCB, cables, transformer etc and avoid catastrophic failure burning everything. Fuses are not supposed to protect the semiconductors, capacitors or resistors.

Component protection is provided by the circuit design, which will limit current, temperature and even voltage.

Amplifiers must have current limiter circuit and you can choose fuses to handle, for example, 20 or 30% above the maximum current scenario.
It's not that difficult.
Regardless of what the user do, fuses must not blow unless there is a component failure.

Current limiter can act when the impedance load is below the minimum and/or when there is excessive clipping leading the amp to provide a square wave to the output increasing normal maximum current.
You can also choose to shut down or limit power when overcurrent occurs. There are a lot of good protection resources to implement.
 
Hi Ray,
In Canada, this would invalidate your policy if they could raise enough suspicion your DIY project caused the fire.

I did work for insurance adjusters to investigate some consumer equipment. I found one case of an amplifier actually causing five fires in the same room. Four speakers, and the amplifier in the rack. The amp still had firefighting powder in it. A Quebec based product called a "Proximity".

If any of you ever run into a Proximity amplifier, do not use it. It is good for a chassis and maybe power transformer only. One of the most unsafe designs I have ever seen.
 
Worst amplifire (pun intended) I ever encountered was in a properly fused Flame Linear. That is where they get the name. Urban legend is that people died in a club in NY when the same failure occurred due to mains overvoltage. The other was caused by a QSC. She was using pioneer home speakers as vocal wedges. The coil spit fire and it caught the carpet. Lots of smoke too.
 
Yep, most folks are worried about electronics supporting combustion. Nope, some speakers burn very well.

Home speakers as vocal wedges?? How cheap can you be? Not sure why carpet caught fire so easily though.

Phase Linear does have a reputation for burning up, and I saw an Amcron (power base) shoot flames out all the vents once. Someone installed the amp on a patch bay. An input was plugged into an output and the result was pretty much instantaneous.

Looking at an SAE P500. I'm very, very uneasy about it since the output transistors are not rated for the supply voltages. I'm blown away they all haven't died, every single one.
 
SAE P500 interesting stamp:

1738590180625.png
 
Hi Ron,
Canadian market amplifier, shipped to your country. I see those stickers often, it cost the distributor $10 per unit for the sticker and a mains examination.

That approval only covers the mains circuit. It isn't CSA or UL approved. CSA approval is expensive - and sometimes very silly.
 
I’ve seen a lot of amplifiers where the outputs aren’t rated for the supply. Even ones with a reputation for being tough, but just as many that die if you look at them sideways. Some of those SAE’s use stacked series outputs. It works, but has limitations. Pull too much current at high vce and the base current gets high enough to upset the voltage divider used to bias the uppers. Then they just go into primary breakdown. Even worse in that damn Tigersaurus where the output stage has gain on top of everything. MJ802’s were POS’s to begin with (with second breakdown as bad as most SWITCHING types, in addition to the low Vceo) - you couldn’t ask for more trouble if you screamed it from the mountaintops.
 
  • Like
Reactions: anatech
This Tigersaurus has an interesting circuit, although a bit cumbersome. 🙂
I didn't know about this topology. I've just found the circuit from the Radio-Electronics 1973.

They made a lot of workarounds to overcome the transistors VCE limitations (100V) at that time.
Ouput stage with gain, common emitter and stacked transistors.

Instead of 70V rails, they could have buiit a traditional class AB topology with 35V rails and bridge 2 amplifiers - much simpler and same power.

I see a lot of modern commercial amplifiers just playing roullet with SOA of BJT, specially with the popular 2SD1943 / 2SC5200.
Good transistors, but poor SOA which needs much more parallel transistors than what is actually implemented.
 
9 out of 10 cats are talking fuses on the DC rails, but that seems so foreign to me. The reservoir caps are where the amp drinks from, not the transformer. The transformer is the meandering stream that fills the reservoir. We don't want fuses between the amp and it's reservoir. They interfere with transfer, and there are parts up stream that need protection to. We put the fuses in the AC secondary leads, to protect everything after them, but not slow transfer from the reservoir to the amp.

If anything, fuses in the AC can lower ripple. Not on the downward fall, but slowing the recovery. The overall voltage is falling, but ripple is lowered. Well sized caps should rarely be playing test tones to achieve this. However, a sudden pulse corrected over a few half cycles, is a lower magnitude ripple, per cycle. The full recovery not an audible frequency.
I believe this is why some amps have two sets of caps, with resistance between them. Someone recently pointed to a book that details the cap thing, that might give a more educated opinion.

These are reservoir caps I'm talking of. Not smoothing caps placed close enough to call decoupling. Which if we are talking silver wire, we should probably be thinking about.

Fuses on the AC are not meant to replace proper management. They are there for the times things go wrong. So should protect everything after the transformer, while a fuse before the transformer, protects that. The fuse at the wall, just protects the lead. If you have a fuse at the wall..
 
There are ways of making stacked transistors work (almost) perfectly. Hardly anyone did it back then. It goes without saying on those old designs that rail fuses were essential.

Stacking can mitigate the high voltage SOA problems of devices like C5200, but you still need enough vceo to handle the full rail for events like start up, slewing, or heavy reactive loading (any of which can put higher than normal voltage on one of the pair for a short time). It won’t go into second breakdown in 10 microseconds, but if you avalanche it — it’s game over.
 
  • Like
Reactions: anatech
The $64000 question is what testing did they do? For years and years in the 70’s all manner of folks used “selected” 2N3055’s, 2N3772’s and 2N3773’s at higher voltages. Most of the time they put a house number on them, and testing was at the semiconductor fab. If they were just the stock 2N number, somebody had better have tested them individually. Even then that’s no guarantee - if Motorola or RCA didn’t test it themselves they’d never sanction the practice “no warranty of fitness for purpose sold” or something to that effect. Of course when somebody goes to service it decades later and doesn’t know any better, it could go south at first power up. Sort of like using ECG parts.
 
Absolutely!!!

Sort of like using ECG parts.
That's another kettle of fish. Sylvania didn't use one part number for their ECG type. They performed poorly and some were leaky. I know because I was forced to use them at one point, working for a store that sold radio-TV parts (ECG among others). I was getting failures (and blamed for them) so I started sticking the leads through the bag and actually measuring the transistors with a Heathkit IT-18 (mine, I brought it in to work). I found them leaky, low beta and some just simply went BANG when used. I fought and switched to real JEDEC numbers from authorized distributors and my returns stopped. These were a lot less expensive as well.

Back then, no internet, everything was a secret and house numbered parts. No FAX, just teletype. You bought real data books and became good at looking at the circuit to figure out what transistor would survive. Eventually you stocked the few that would cover the lower powered ones and matched when you needed to. Marantz training means you learned to match outputs, drivers and diff pairs.

When Japanese transistors became more popular in equipment, you figured out the pinout differences (Pro-Electron was another mess). You bought data books from each market, they were your bible. The ECG cross-reference? "The book of lies". NTE copied that book exactly.
 
  • Like
Reactions: Mooly
TD fuses on secondary dual windings that connects to power supply - If an appropriate value is selected for + and - winding, can this be a sufficient alternative and protection when not using DC rail fast blow fuses?

Also, I have a board that is not completed yet but it has Soft start, DC protect and speaker relay on the one PCB. when it detects DC at speaker out it will disconnect the line at soft start relay.. Surprised I don't see this more!
 
  • Like
Reactions: sonicles