Amp failure modes, preventing driver cooking

[Creative1]

I've read some about failure modes... and I know that some amps when they fail - can cook drivers because they pass the full rail voltage. I know some amps have DC protection and some don't.

I know that heat/voltage are factors in transistor failure, but if an output is going to fail open or shorted --- is that far more likely to fail when it's being pushed hard? In other words, if it's not being pushed very much....... are the transistors that much safer?

How often does an output transistor fail just in routine service, if it's not being stressed? From what I've read, open vs shorted failures seem to either be 50/50 or more likely shorted. Seems to depend on who you ask.

I've seen some DC protection circuits... I'm just trying to determine if I can use a couple amps with some more expensive drivers, or if I'm better looking for some new amps that have DC protection that might keep the drivers........ safer.

I realize my question may be a "duh, transistors can fail at any time... they are more likely to fail when being pushed hard..." - just curious if that was in fact reality or not.

 

[richie00boy]

Generally it's when they are pushed hard, as they are hotter. But if you short the output through a stray strand of speaker wire for example, then it can happen at low volume.

You can always build a DC protection add-on to go between your amp and speakers. Velleman do a kit, simple passive thing.

 

[ostripper]

There is the "idiot mode". Read about it here...

http://www.diyaudio.com/forums/solid-state/169590-mongrel-supersym-ii-12.html#post2310615


In short , what will happen with the ultimate disaster (the slipping screwdriver) , it blew into molten balls !!
Of course fuses will blow , they better or you and the family gets burned. On that board , it was interesting what semi's got burnt. The 2 closest to the fuses (njw0281/0302) and the positive flyback diode (5A). On the new one i'm about to make the fuses are right in the middle , absolute current sharing.... I learned from catastrophic failure mode. The amp has been absolutely abused for 3 months after and I am confident it would either take a lightning strike , screwdriver , or glass of water to bring it down. The MJE15032/33 drivers never broke a sweat , I doubt whether anything , including a OP short , could ever affect them. The loudspeakers were hooked to the amp at this point , besides a "thump" , no damage.

In normal to abusive operation, the drivers/OP's stay cool but the heatsinks become slightly hot with 2 pair 8R hooked in parallel (3R). Thermal transfer is optimal (no aluminum angle bracket for op's)

The secret is just thermal/electrical derating of all the components. For 75V rails , all caps are 100V , VAS/ drivers and outputs are at least 250 Vceo. Thermally , oversize .05C/w HS's for drivers/outputs and at least a small 10 - 20 sq. CM HS for the VAS. Everything should be just warm to the touch , even on a 37 C day. Something such as this would most likely last 20 - 30 years - even 24/7.
OS

 

[AndrewT]

Breaking a DC current fault is not as simple as opening a relay.
DC currents can strike an arc across the opening contacts and current continues to flow.
A fuse rated to break your worst case failure current may be a better solution.
A crow-bar to ensure the fuse blows quickly may also work better and may have a lower resistance than the arc and contacts in the relay, thus diverting the current flow from the relay and allowing the arc to extinguish..

 

[Creative1]

I'm familiar with the Velleman kit. It just gets expensive/cumbersome when trying to protect 8-10 outputs. I know that protecting from DC outputs is a heavily discussed topic in terms of most effective / least distortion.

I was most curious about how to prevent circumstances that would cook a driver.

I don't really want to risk cooking a $500 pair of woofers... so just trying to understand how to minimize risk.

I can't seem to grasp how often an amplifier cooks itself in regular usage, versus abusive usage (or straight up abuse like shorting outputs)

 

[indianajo]

protection

Music instrument resale shops are full of speakers with mismatched woofers, and amps with new shiny power transistors on the heat sink. Happens a lot in the trade.
My favorite amp http://www.diyaudio.com/forums/soli...120-can-beautiful.html?highlight=djoffe+st120 is capacitor coupled to the speakers, which provides pretty absolute protection for your speakers. It has wimpy heatsinks, so after a 3.5 hour rehearsal with the amp filling a church, it went up in flames (resistors to output trans) The speakers were fine.
Direct coupled amps are much scarier if you invested heavily in speakers. Before I trusted my new PA amp, a CS800S to touch my $600 (used) speaker pair, I found out it had 1. diac triac crowbar to short the output and blow fuses if the output gets too close to the rail (shorted transistor) 2. thermistor controlled fan on the heat sink. 3. current transformer around output to measure the AC current flowing (oscillation can kill a speaker, too) 4. microprocessor to measure the current going out and compare it to some limit 5. speaker relays to pull the output from the transistors if 4 finds anything wrong. I maintain item 4 can't be done by ordinary mortals at home. The australian kit has a microprocessor, but doesn't do that and doesn't hook up to a current transformer.
Peavey also has a budget line of amps without the protection, for musicians that believe it can't happen to them.

 

[Creative1]

So.... if I care about my speakers, am I safer to not use my old-school Crown Microtechs - and "upgrade" to some newer amps? I was thinking about some QSC amps.

It just seems like a number of newer amps, like QSC RMX, have DC protection standard.

 

[indianajo]

Crown, Peavey, QSC all make some amps with serious protection, others without. You'll have to look at the schematic and decide if it is good enough. The CS800S has DC and AC protection. Rather then build my own split supply amp, I bought one with serious protection and problems, and fixed the problems. Turned out to be a $.10 problem, two toasted input resistors.
CS800S was designed in 98 and CS800X was before that, so older amps can have serious protection. Older 80's peavey stuff only had the diac-triac DC crowbars and fuses.

 

[indianajo]

no protection

Well, I looked at the crown microtech manual available on eserviceinfo.com. That model apparently doesn't have any protection but an input fuse. I'm not the world's linear expert if someone else wants to take a look. I didn't see any current transformer or microprocessor chip or speaker relays, or diac-triac crowbar.

 

[Creative1]

Well, I looked at the crown microtech manual available on eserviceinfo.com. That model apparently doesn't have any protection but an input fuse. I'm not the world's linear expert if someone else wants to take a look. I didn't see any current transformer or microprocessor chip or speaker relays, or diac-triac crowbar.

I think if I understood the schematics too - I didn't see any real protection either.

So, other than use an add-on DC protect circuit - does anyone have any thoughts on how to best *prevent* a catastrophic failure that would cook drivers? Preemptively replace the output transistors, since it seems like those are the most failure prone - and most likely to result in rail voltage to the speakers? I guess that's $60 an amp ($2.50 x 12 per channel x 2 channels.... plus or minus)

I know ultimately newer amps simply have better protection. Now, I'm just curious about these older amps and keeping them alive..

I guess if it lacks output protection - what's to say that a wiring short wouldn't cause one of the output transistors to short...... and then result in the driver getting full rail DC.

 

[MJL21193]

So, other than use an add-on DC protect circuit - does anyone have any thoughts on how to best *prevent* a catastrophic failure that would cook drivers? Preemptively replace the output transistors, since it seems like those are the most failure prone.

Crowbar the output. Sized to suit your amps power. This WILL protect the speakers, is cheap and easy to build. It would likely kill off the output stage of the amp if it activates, but this is probably cheaper than new speakers.

 

[eyoung]

Sometimes input caps to divert DC at the inputs can be advantageous as well as rail fuses.Especially on sound reinforcement amps.Maybe a schematic for us to peruse for sugestions....

E

 

[audiomachines]

I think if I understood the schematics too - I didn't see any real protection either.

So, other than use an add-on DC protect circuit - does anyone have any thoughts on how to best *prevent* a catastrophic failure that would cook drivers? Preemptively replace the output transistors, since it seems like those are the most failure prone - and most likely to result in rail voltage to the speakers? I guess that's $60 an amp ($2.50 x 12 per channel x 2 channels.... plus or minus)

I know ultimately newer amps simply have better protection. Now, I'm just curious about these older amps and keeping them alive..

I guess if it lacks output protection - what's to say that a wiring short wouldn't cause one of the output transistors to short...... and then result in the driver getting full rail DC.

ODEP

 

[Creative1]

So ODEP prevents the speaker short from causing the outputs to be shorted.... but nothing to protect speakers from the outputs when they're shorted..

 

[indianajo]

aftermarket protection

Your output transistors don't age, they are destroyed by external factors like a speaker 1/4 phone plug pulled 1/2 way out (shorted) or a speaker spade lug pulled out and touching the chassis, or too many watts over too many hours heating up the heatsink. People tell me you can buy the Peavey diac-triac crowbar board fairly cheaply at dealers. There is a CS800B schematic uploaded here at diyaudio, it doesn't have the whole board pn but the diac is SB914 and the triac is SAC187. It is in the upper right of the schematic. You have to also put a fuse in ahead in the power rails before the power transistors, that can blow when the triac crowbars, before the transformer blows.
I'm working on a FET circuit to sense the heatsink temperator of my ST120 and stop current coming to the power rail based on a thermistor. Watch the ST120 thread listed above for updates. Blown up PC switcher power supplies are full of high wattage FET's and big heat sinks to mount them on.
Some people put speaker fuses in, but they are non-linear and other people say you can here them warming up, so I don't exactly recommend that.
I'll look at the Velleman board, now that I have the Velleman URL from another guy on organ forum. There is an add on protection circuit for sale on e-bay, but there is no schematic, and no visible current transformer or current sensor in the picture, nor no programming for what wattage you want it to blow at, so I told a guy that brought it up (on "my amplifier blew up and blew the speaker too" thread) that I thought it was snake-oil.

 

[MJL21193]

I'm working on a FET circuit to sense the heatsink temperator of my ST120 and stop current coming to the power rail based on a thermistor.

Easier: Thermal cutout
About $5

 

[audiomachines]

So ODEP prevents the speaker short from causing the outputs to be shorted.... but nothing to protect speakers from the outputs when they're shorted..

maybe I just love the macrotech's they have a different "kick on the bass notes" unlike the qsc's. If I were going to biamp a system I would use the qsc on the mid-hi's and the crown's on the lows. Crowbars are nasty, you'll get molten traces of pcbs and solders, some of them could false trigger especially when the triggering diac were place too close to the heatsinks (poor pcb layout), this is due to the increase of leakage Ig as the surrounding heats up. Well we cannot deny the existence of an MTBF of a product.

 

[indianajo]

Easier: Thermal cutout
About $5

Well, I downloaded this datasheet, and the device comes in 1 and 2 amp versions at 300V. I'm not sure the cantech device is a thermistor or a snap action thermostat. Snap action thermostats like AC, with it's zero crossing feature to quench the arc, and not DC. Amps like the crown microtech use more like 10 or 12 amps at the output collector feed rails. I have been looking at Tyco and Littlefuse positive temperature coefficient fuses (thermistors) but it seems you can break 7 A max at 16V, or 3 A max at 72V, neither of which is useful for 300 W/channel audio amplifiers like the crown microtech. PTC fuses are also resettable, and I'm designing a device that goes out, latches out, and lights a LED with the power supply still on so you know what the problem is.
You could take 24 quantity 3A 72V thermistor fuses and put one in series with each output transistor emmiter lead of the crown microtech 1000, which would protect the individual transistors, but that does seem like a lot of work, and easy to botch with a cold solder joint. PN RXEF300-2 Tyco for example. The emmiter resistors are .2 ohm, and the fuses are .1 ohm, so you would have to change 24 emmiter resistors to .1 ohm to stay original.

 

[MJL21193]

Snap action thermostats like AC, with it's zero crossing feature to quench the arc, and not DC. Amps like the crown microtech use more like 10 or 12 amps at the output collector feed rails.

Well, I wasn't suggesting that the cutout break rail voltage. It can be used in conjunction with something more capable, like a relay.

 

[indianajo]

Rail shutdown

Well, I wasn't suggesting that the cutout break rail voltage. It can be used in conjunction with something more capable, like a relay.

Well, fet's have so much gain that on DC they act like relays, and dead PC power supplies are chock full of them, so to me they are "free". NTC thermistors like the SK055E106ZAA are $1.40, and read 14K room temp and 1.5k in front of a hair dryer, useful for putting in series with a resistor as a fet trigger. The latching and LED display part is the tricky bit, if interested stay tuned to "ST 120 can be beautiful" thread where it is going first. The sT120 had a trip and latch protection circuit on its one power rail, but it was based on the characteristics of obsolete BJTransistors, and after burned up, the best I can do with 2 modern transistors on the standard board is trip at 6.5 amp current draw and self reset as soon as the excessive current to the output transistors disappears. I think the ST120 circuit tended to overheat and destroy the protection transistor, too, when protecting, since the heat sink was so wimpy. Mine was sure blown up. Fet's when off don't flow current and heat up.