amplifier failure modes

[east electronics]

as simple as it can be .... People discuss with uncle Charly his SOA calculations ( troyan threaD ) ... i discuss with ostriper his schematic that was inspired from the koda amplifier .

QUESTION IS SIMPLE AS THAT ...IN TWO STAGES ...

one from input to outpout and two from amplifier it shelf

---one from input to output
what will happen if you take a ( example ) troyan or mongrel amplifier doesnt really matter and you overdrive it ???
or give it signal from a bad source (assymetrical )
or with very low/high frequency included,
what will happen if in this situations VAs area is a bit out of order ? ( like a bit more temperature than expected? )
what will happen if drivers get a bit warmer than expected
what will happen if the load is crappy ??? piezos or terrible xover ???
what will happen if cables are too thin or too capacitive or too long ??
what will happen if you have more than one of these conditions active ??

can we simulate or learn from any other method the behaviour of amplifier when and if something goes a bit wrong compaired to the simuation ???

fell free to add your opinion

--------------------------------------------------------------------
it seems that now days we have plenty of big schematics arround the forum but also seems that designers focus mostly on quality and power and miss the ABUSE factor that is a fact in amplifiers that are rated at this power

1000W amplifier will obviously perform in PA application... i can imagine some forum members that have amps of 1K for home use but i wonder how often they use 100% of this power ...

Yet again it seems that only half of the job is done and sorry to say you cant get away with it by saying "I dont design idiot proof amplifiers "like Dr Bora said ... i will respect and agree with with Dr Bora but still amplifiers of 1KW power actually have to be designed as idiot proof ...

 

[east electronics]

it seems that there is no interest ...

 

[sreten]

Hi,

Designing for abuse and protection against failure modes does not get
discussed much but its perfectly valid, especially regarding overload.

One trick is to design gain for 0.5V in for maximum output and then
put parallel diodes across ther inputs to prevent internal saturation.
(needs some series input impedance.)

Clipping behaviour of an amplifier should be thoroughly analysed.

/Sreten.

 

[ostripper]

can we simulate or learn from any other method the behaviour of amplifier when and if something goes a bit wrong compaired to the simuation ???

The simulations mean nothing here. They just reflect ideal situations. One can simulate gross amounts of power supply ripple to plot PSRR and come up with numbers and pretty graphs , but the REAL amp will always be different.

As a example , when I did all this on my "supersym" 13 months ago , I was worried that I would have crosstalk between channels using a single PS. Upon actually hooking it up and "blasting" away , I was surprised to hear absolutely nothing in an unused channel while the other was straining the loudspeaker.

I have had 2 very close calls after 13 months. One was in testing for initial Vbias ... my probe slipped !! This did not blow the 8A fuse ,but did fry one NJW0281 ( small pinhole on middle let out the magic smoke). i thought I blew all the other 3 0281's but they were good.

After 6 months I had a new year eve's party , got sloshed .. shorted that same channel out at the output terminals. thought my party was over (have a 2nd DIY amp). Looking inside, all E-B junctions were .550-.600 , no C-E shorts. Put in another 8A fuse .. it played

So , you must LIVE with a real amp for any amount of time , even have a few accidents to really see if/ how they fail.

If I rented out my amp , I absolutely would want some insurance that my OP devices would come back as semiconductors .. not melted sand. But In the home with my ASKA style amp (18 months+) and my big one (8 OP device symasym - 13 months) , I see no need for any type of protection.
OS

 

[ostripper]

Hi,

Designing for abuse and protection against failure modes does not get
discussed much but its perfectly valid, especially regarding overload.

One trick is to design gain for 0.5V in for maximum output and then
put parallel diodes across ther inputs to prevent internal saturation.
(needs some series input impedance.)

Clipping behaviour of an amplifier should be thoroughly analysed.

/Sreten.

That is a good idea .. clip before the amp. I was thinking .7V ???
OS

 

[wg_ski]

Hi,
One trick is to design gain for 0.5V in for maximum output and then
put parallel diodes across ther inputs to prevent internal saturation.
(needs some series input impedance.)

Regardless of the gain, one can always put an anti-parallel diode pair directly across the diff inputs. The open loop input voltage will almost always be well below 0.5V p-p. If you're worried that one diode drop isn't enough, use two in series. Even if you didn't protect the inputs the usual "failure" mode is just the diff pair getting out of balance from driving Vbe into reverse breakdown. It won't blow anything up if the current is limited, but it will cause parameteric shifts over time.

 

[apexaudio]

Diodes on iput for limit I use in my amp:
http://www.diyaudio.com/forums/solid-state/162081-dc-servo-mosfet-amplifier-4.html
If amp need 0,7v RMS for full power that will be 1,98Vpp, and for 1v that will be 2,8Vpp. With 2,4V zeners in oposite you have 2,4V+0,6V about 3V peak to peak limit level. In this thread you can find PSU with short-circuit protect, I use this circuits in PA over 15 years.
Regards

 

[fotios]

The issue is how you can save your speakers from any abuse, either caused from clipping (clipping by 99% is caused from excess driving of amplifier from a pre-amp device, like mixer etc.) or from a malfunction of amplifier itself. The cost of repairing an amplifier is very small compared to the cost of replacing a burned speaker. That is a good guide to start thinking priorities and methods of protection ...of what first?

 

[Th3 uN1Qu3]

The issue is how you can save your speakers from any abuse, either caused from clipping (clipping by 99% is caused from excess driving of amplifier from a pre-amp device, like mixer etc.) or from a malfunction of amplifier itself.

For the amp malfunctions you can add protections. IMO protection against DC at the output is mandatory in any high power amp. As about overdrive, use a limiter. A release time of 700ms to 1 second will assure that when you hit the limiter, no matter how high you turn the gain knob afterwards, the signal going to the amp will *not* exceed the level set by the limiter. Sure, it'll kill some of the dynamics, but at any party a limiter is a very welcome device.