I just Fried fairly expensive AVR which is NOT mine

[mhardy6647]

There are 8-ohm rated speakers that dip down to 2 ohms (and often 2 ohms reactive!) at certain frequencies

This -- I would opine -- is the fundamental problem.
Fundamental because it's utterly unnecessary, but this sort of thing is perversely pervasive in the loudspeaker business nowadays.

 

[restorer-john]

This -- I would opine -- is the fundamental problem.
Fundamental because it's utterly unnecessary, but this sort of thing is perversely pervasive in the loudspeaker business nowadays.

I call total BS. Seriously. How is a picture of Clint Eastwood remotely clever? Hint: it isn't.

Actually measure some random speakers you own made in the last 10-30 years yourself and show the results. I'll bet they aren't remotely difficult to drive.

Not only that, any reasonably competent amplifier made in the last 50 years can handle just about any speaker you throw at it.

 

[mhardy6647]

I call total BS. Seriously. How is a picture of Clint Eastwood remotely clever? Hint: it isn't.

Not meant to be clever at all. It's a well-known meme: in that movie he's a grumpy old man. Admittedly, a lethally dangerous grumpy one, but it should be pretty easy to suss out.
I guess I should've used my other signature grumpy old man image, so my apologies.
Henceforth, I'll use Mr. Rooney here, so as not to ruffle feathers, when I'm bein' curmudgeonly.

 

[TheBatsEar]

Nice to see @restorer-john and @mhardy6647 bonding:

I think we should continue our speculation about possible causes. How about a capacitor exploded? I haven't seen anything in the last years, but there was a time this was a serious problem:

Capacitor plague - Wikipedia

en.wikipedia.org

Did anyone see something like a large exploded capacitor as of late?

 

[H-713]

Not likely to be a cap in this case. It's much less of an issue now, doesn't manifest itself for a few years, and while heat is a factor, 3 minutes of very hot running won't do much to kill them.

The cap plague thing had everything to do with bad electrolyte formulations.

 

[Cbdb2]

It is much easier to protect against a short than an overload. With a short the current rises very quickly past the trip level and the protect circuit can react very quickly shutting down the output. I2t in the protected device is easily controlled. Not the case into an (especially inductive) overload where the current rise is realtively slow, and a lot of heat can be generated as it meanders up to and beyond the trip level until the circuit triggers.

With a 1 kHz signal the current and voltage will both be a 1khz sine. Thats not a slow current rise.

 

[levimax]

With a 1 kHz signal the current and voltage will both be a 1khz sine. Thats not a slow current rise.

When thinking about blowing up an amp you need to think about the amount of current being dissipated by the output transistors not how long the "signal" takes to reach a peak. A 2 ohm load is much different than a short circuit. With a short circuit the current will immediately spike and quickly and easily trigger protection circuits which are monitoring the peak current before anything is damaged. For a 2 ohm load the current will be much lower than a short circuit (but still too high but not high enough to trigger the protection circuit immediately) and will rapidly start heating up the output transistors. Output transistors tend to "run away" and get hotter faster the more they heat up. By the time the current limiting circuit is "tripped" it might be too late.

 

[AdamG]

 

[TheBatsEar]

For a 2 ohm load the current will be much lower than a short circuit (but still too high but not high enough to trigger the protection circuit immediately) and will rapidly start heating up the output transistors.

And why could we not have a microcontroller that measures actual current, looks up in a table how much for how long is acceptable and mutes the input signal accordingly?
I say it again, it's not rocket surgery.
I also say we don't have to accept cost as an argument, when Hypex does the same in their 150€ modules.

Also, the official soundtrack of this thread is now Senior Coconuts rendition of Beat it, it's spicy, heated and fast, not unlike the comments in this thread. ;-)

 

[Holmz]

Nice to see @restorer-john and @mhardy6647 bonding:
View attachment 254796



I think we should continue our speculation about possible causes. How about a capacitor exploded? I haven't seen anything in the last years, but there was a time this was a serious problem:

Capacitor plague - Wikipedia

en.wikipedia.org

Did anyone see something like a large exploded capacitor as of late?

Is that where the expression, ”Busted a cap in its ass“ comes from?

 

[antcollinet]

With a 1 kHz signal the current and voltage will both be a 1khz sine. Thats not a slow current rise.

Compared to the 2 or 3 microseconds of a short circuit rise time to well above the trip level, it is an eternity.

Then you have to remember that the problem is at the peak of the sinewave - where it is basically horizontal. Point is, it is trivially easy to detect a short circuit compared with an I2t type overload.

 

[H-713]

With a 1 kHz signal the current and voltage will both be a 1khz sine. Thats not a slow current rise.

Nothing about 1 kHz is fast. Even an Arduino Uno could implement a protection scheme fast enough to deal with this. Designing a good protection circuit that detects overcurrent in less than 50 us is trivial.

When thinking about blowing up an amp you need to think about the amount of current being dissipated by the output transistors not how long the "signal" takes to reach a peak. A 2 ohm load is much different than a short circuit. With a short circuit the current will immediately spike and quickly and easily trigger protection circuits which are monitoring the peak current before anything is damaged. For a 2 ohm load the current will be much lower than a short circuit (but still too high but not high enough to trigger the protection circuit immediately) and will rapidly start heating up the output transistors. Output transistors tend to "run away" and get hotter faster the more they heat up. By the time the current limiting circuit is "tripped" it might be too late.

Thermal runaway shouldn't be an issue in competently designed amplifiers. This issue was understood back in the 1970s, and by the 1990s amplifier manufacturers had thermal tracking down to a fine art. The Sanken transistors discussed earlier solve one of the most annoying parts of designing the thermal tracking scheme, which is trying to get the temperature of the bias spreader to track the temperature of the transistor dies.

I have a really hard time believing that a company like Yamaha would be stupid enough to sell an amplifier that only has short-circuit protection. VI limiters have been a solved issue for decades now, and they cost almost nothing to add to an amplifier. Given how many people (college kids especially) will parallel 4 sets of speakers off a receiver, not including this type of protection (along with thermal protection) seems unwise.

I haven't tried it, but there's no good reason that one couldn't design a pretty competent (and very cheap!) protection scheme using a microcontroller to monitor the output current, output voltage and transistor temperature, then disconnect the outputs if it decides the parameters involved indicate an overload.

 

[restorer-john]

All the Yamaha units made today have extremely sophisticated monitoring and protection systems. Even the stereo A-Sxx series amplifiers have A/D lines for three individual temperature sensors (L/R and TXF), DC, Over current sense, etc. The 16bit processor itself has 26 (yes) 10bit A/D converters for such inputs.

Previous protection events are stored for recall etc.

They are in a whole other league to traditional old-skool methods of protection. That being said, it is necessary because they are running these things on the ragged edge if you ask me. Output devices are not expensive and neither are press-fit fin aluminium heatsinks. I'd rather see more capable amplifier channels than ones that prematurely or catastrophically shutdown/fail.

 

[Cbdb2]

Compared to the 2 or 3 microseconds of a short circuit rise time to well above the trip level, it is an eternity.

Then you have to remember that the problem is at the peak of the sinewave - where it is basically horizontal. Point is, it is trivially easy to detect a short circuit compared with an I2t type overload.

Your saying the millisecond of rise time of a 1khz will overheat a component with only twice the current of a 4 ohm load which should not be a problem? At some point in the first 1/4 cycle, the over current protection should kick in and stay in. Not buying it.
And I know the protection can be usec fast, I'm counting on it.

 

[mhardy6647]

Is that where the expression, ”Busted a cap in its ass“ comes from?

that's a negative.

 

[levimax]

@thin bLue ... any chance of taking the cover off and looking around inside for the "smoke source" and posting a picture?

 

[amirm]

@thin bLue ... any chance of taking the cover off and looking around inside for the "smoke source" and posting a picture?

If he can't do that, a flash light can be used to see inside through the vents. One's nose is also a great diagnostic tool with respect to equipment that has smoked out.

 

[Holmz]

If he can't do that, a flash light can be used to see inside through the vents. One's nose is also a great diagnostic tool with respect to equipment that has smoked out.

Yeah the professor at uni would do that.

One day in electronics lab, the prof was helping Lutz with this breadboard, and I saw the waveform on the scope.
Dr. E. Had his nose on the board like a bloodhound on the scent.

I asked if I could help…
They said please…

And I slammed the box down as I flipped the power switch on, which made it go from passive to powered.
Dr E. said, “what did I you to make it work?”
I said, “The contacts get corroded and slamming it cleans them.”

After Lutz left, Dr. E said, “That was quite a magic trick, but what did you really do?”
I said, “I turned on the power switch.”
He said, “why did you slam it down?”
I said, “to obfuscate the causal mechanism So I coudl flip the switch up.”
He said, “well done Holmz… well done.“

 

[H-713]

All the Yamaha units made today have extremely sophisticated monitoring and protection systems. Even the stereo A-Sxx series amplifiers have A/D lines for three individual temperature sensors (L/R and TXF), DC, Over current sense, etc. The 16bit processor itself has 26 (yes) 10bit A/D converters for such inputs.

Previous protection events are stored for recall etc.

They are in a whole other league to traditional old-skool methods of protection. That being said, it is necessary because they are running these things on the ragged edge if you ask me. Output devices are not expensive and neither are press-fit fin aluminium heatsinks. I'd rather see more capable amplifier channels than ones that prematurely or catastrophically shutdown/fail.

Yeah, that's an interesting situation.

Still, I should point out that protection mechanisms are often a better solution than throwing lots of power transistors at the problem. MC2 amplifiers have relatively few output transistors for the power level (I think 5 pairs in the MC650, and 3 or 4 pairs in the MC450). By contrast, Crest amps (3301, 4801, etc) have far more output silicon than the MC2 amps do, but in my experience are less reliable. Crown is somewhat notorious for this too - the Macro-Tech amplifiers were NOT overbuilt in their output stages - the 3600VZ has 6 pairs per channel, and it will do 1800 W 2 ohms. Both Crown and MC2 used pretty sophisticated protection schemes that work really well, and it shows. The Crest CA6, on the other hand, has a much less sophisticated (but still pretty good) protection scheme it in it. They're no more reliable than the Macro-Tech or the MC2.

Of course, you do still need enough silicon to cope with normal loads - some companies cut things a little too close. That said, the number of transistors required doesn't really scale linearly with output power, because increasing output power requires higher rail voltages, and unfortunately, power transistors can't dissipate their maximum power under all conditions - secondary breakdown ensures that above a certain V_CE, their power dissipation starts to fall off a cliff. This is a big part of why large power amps tend to be class H.

The point is, part of designing an amplifier is making sure it can survive in the real world. Tube amps can get away with little to no protection circuits, aside from something to protect against open speaker terminals. Transistors are delicate little flowers, and they need babysitters to stay alive.

 

[antcollinet]

Your saying the millisecond of rise time of a 1khz will overheat a component with only twice the current of a 4 ohm load which should not be a problem? At some point in the first 1/4 cycle, the over current protection should kick in and stay in. Not buying it.
And I know the protection can be usec fast, I'm counting on it.

It's not about rise time of the current, it is about the amount of time spent in overload. (I2t - I squared t). You don't want the amp to trigger instantly the moment it goes (eg) 10% over rated full load - or you'll get nuisance tripping with transients in the music. So you have to implement a strategy that effectively simulates the heating in the device based on the level of the current and the time at that level. So either a complex modelling using analogue electronics, or accurately montior the current in a micro and implement in software. And it will only be an imperfect modelling - so may not capture all cases, especially if the amp is being run out of specification considered in the modelling.

Of course it can be done (and as pointed out by @restorer-john Yamaha (eg) are doing it. But my initial reply was to someone asking how can an amp protect shorts if it can't protect from overload. The answer is, short protection is trivially easy by comparison. you can set a trip level 2 or 3 times rated current, and use a simple comparator with instant shutdown. The current rise time into a short is so fast that even a current pulse much higher than rated won't cause damage in the time it takes to detect and shutdown.