does it make sense to test how amps behave when they clip?

[KSTR]

It's a pity that commercial HiFi amps seldom have full featured clipping indicators (with both voltage and current clipping detected). Smaller amps on low sensivity speakers clip in practice more often than not, and to avoid questions and support manufacturers tend to omit clipping indicators.
Soft (voltage) clippers ahead of hard-clipping amp cores are not very much in fashion, too, I suspect this is governed by the problem that a soft-clipper lowers 1% and 0.1% THD power specs and compromises measured THD vs. power curve, even though the amp may well sound actually "louder and cleaner" when on the edge. Plus, detectors and soft-clippers (let alone protective limiters) add to complexity and BOM cost.

And @pma, wow, brillant overdrive characteristic of this amplifier of yours!

 

[amirm]

Get a class D amp with a few hundred watts of power and then don't need to worry.

 

[March Audio]

Our P252 amp just edging into visible clipping. 8 ohm load. 160 watts (rated at 150watts into 8 ohms)

 

[RayDunzl]

Ok, Solid State clips abruptly.

Tubes clip gracefully.

Why?

I understand the SS side, No more volts.

Does the Tube side seem to clip more gracefully because it goes into compression before reaching no more volts?

What's the deal?

 

[March Audio]

@restorer-john for your info at 20kHz, same 160 watts, left it there for several minutes.

 

[March Audio]

Ok, Solid State clips abruptly.

Tubes clip gracefully.

Why?

I understand the SS side, No more volts.

Does the Tube side seem to clip more gracefully because it goes into compression before reaching no more volts?

What's the deal?

I think it just all adds to some peoples "perception" of tube v solid state. Its all quite explainable, there is no magic about tubes. They just distort differently and generally have wonky frequency response due to their high output impedance. Thats why they sound different.

 

[Blumlein 88]

It's a pity that commercial HiFi amps seldom have full featured clipping indicators (with both voltage and current clipping detected). Smaller amps on low sensivity speakers clip in practice more often than not, and to avoid questions and support manufacturers tend to omit clipping indicators.
Soft (voltage) clippers ahead of hard-clipping amp cores are not very much in fashion, too, I suspect this is governed by the problem that a soft-clipper lowers 1% and 0.1% THD power specs and compromises measured THD vs. power curve, even though the amp may well sound actually "louder and cleaner" when on the edge. Plus, detectors and soft-clippers (let alone protective limiters) add to complexity and BOM cost.

And @pma, wow, brillant overdrive characteristic of this amplifier of yours!

The McIntosh Power Guard circuit worked wonderfully. I forget whether it limited distortion to 1% or lower. An over would light the PowerGuard light which would reduce input to prevent clipping and distortion. In excess it eventually would act as a gentle compressor. It monitored both current and voltage. I had a little 752 amp with it. I purchased my first ESL, a pair of Acoustat 2's. This was the amp I had, and it played them pretty well, but if I got way up on volume the power guard lights would flicker regularly, but output remained clean. Probably not a bad idea even now.

Of course I tend to use Amir's solution. Buy class D with overkill power and don't worry about it. The speakers and my ears give up a bit prior to the amp being flustered.

 

[Blumlein 88]

Well, it would exclude my 2 watt SET amp, the best sounding amp I have.

Load the output and use it as a conditioner preamp to feed a big power amp. Have your cake and eat it too! This works, I've done it a few times.

 

[Blumlein 88]

Without much thinking about it, I've thought class D switching amps would naturally recover from overload much better than old school amps. The switching is happening at a high speed, and there doesn't seem to be the problematic ways that a conventional amp can have difficulty recovering from overload with its power supplies recharging at 60 hz.

It also seems a switching supply is a good idea even for old school amps for some of the same reasons. Some have been built, but they aren't common.

 

[pma]

Ok, Solid State clips abruptly.

Tubes clip gracefully.

Why?

I understand the SS side, No more volts.

Does the Tube side seem to clip more gracefully because it goes into compression before reaching no more volts?

What's the deal?

Because SS with usually strong global NFB usually keep the gain almost up to the maximum output swing. Then, when the output hits its maximum amplitude, global NFB suddenly stops working and now it depends on amp circuit design if there is a charge accumulated somewhere that must be taken away when amp return to linear mode and it may take time.

Tubes, on the other hand, never have that amount of NFB applied and they also have higher intrinsic impedance. They do not keep gain when loaded up to maximum swing.

However even SS do not need to clip abruptly. Abrupt ugly clipping is not good for tweeters, tweeter damage often occurs during abrupt clipping.

Clipping of the CFA amp already mentioned

 

[restorer-john]

Our P252 amp just edging into visible clipping. 8 ohm load. 160 watts (rated at 150watts into 8 ohms)

The 1KHz wave-form has a strange shape to it, almost concave on the rise and normal on the fall.

@restorer-john for your info at 20kHz, same 160 watts, left it there for several minutes.

Your 20KHz waveform shape is perfect however at the same level. Interesting.

 

[pma]

I concur, 1kHz looks strange even at light clipping. Like triangle wave through HP? Is it output LC filter? Or @March Audio has AC coupled scope input???

 

[RayDunzl]

The 1KHz wave-form has a strange shape to it, almost concave on the rise and fine on the fall.

 

[restorer-john]

View attachment 42972

Thanks Ray. Us scope jockeys can spot a weird looking sine from 20 paces, but I'm no good with Photoshop, mirror images and all that.

The fact that the 20KHz looks fine could mean he was loading the signal generator with a non-linear load, adapter, meter or something and the amplifier was just amplifying it. It's unlikely it'd look like that at 1KHz and then be fine at 20KHz. It'll be a simple explanation.

I remember years ago chasing down something similar I could hear, and it turned out to be non-active (turned off) line inputs of connected tape decks presenting a non-linear (unbiased input junctions) load and distorting the waveform. Turn on the tape decks and it all went away. Next amplifier had switchable/defeatable record outs and buffered tape outs...

Even if he had two amplifiers hooked up to one source, but only one actually turned on, it could do that.

 

[March Audio]

Ooh thats is odd.

Retested and seems OK now ???!!!

Shruggs shoulders.

 

[restorer-john]

Retested seems OK now ?

That looks fine. Sometimes even the x1/x10 switches on the probes can do strange stuff.

That reminds me, I need new good but inexpensive scope probes (maybe 6 identical ones 100MHz). Ones with decent hooks, switches and nice BNCs. What suggestions do you guys have?

 

[March Audio]

That looks fine. Sometimes even the x1/x10 switches on the probes can do strange stuff.

That reminds me, I need new good but inexpensive scope probes (maybe 6 identical ones 100MHz). Ones with decent hooks, switches and nice BNCs. What suggestions do you guys have?

Apart from the Agilent N2863, I do use this PMK.

https://au.element14.com/pmk/pmt211a/passive-probe-250mhz-10-1/dp/1552763

 

[pma]

That looks fine. Sometimes even the x1/x10 switches on the probes can do strange stuff.

Yes just in case that the probe 1:10 was not calibrated and trimmed with 1kHz calibrator square. I think it was the case and you probably hit the nail. Before ANY measurement with 1:10 probe switch position, calibration is necessary. We may have more than one scope ....... with the same probe used.

 

[amirm]

Ok, Solid State clips abruptly.

Tubes clip gracefully.

Why?

It doesn't have to do with tube or solid state. The latter usually uses a lot of feedback to get distortion low, the former does not. When the output starts to clip, the feedback pushes the input stage hard to correct. But we are already at max output and can't drive any higher. Doing so at the input stage only makes things worse. So hard clipping occurs.

Meanwhile, most power amplifiers have unregulated output voltages. They start to sag and feedback gets upset over that too and pushes harder yet. The whole thing becomes a mess of two feedback loops fighting each other causing severe distortion.

 

[restorer-john]

Meanwhile, most power amplifiers have unregulated output voltages. They start to sag and feedback gets upset over that too and pushes harder yet.

Regulated power supply amplifiers have little or no dynamic headroom, whereas unregulated can offer enormous reserves over rated continuous. The Hypex SMPSs are unregulated and they get a ton of short term undistorted power due to that. It's a perfectly valid approach.

There's very good arguments for each. Regulated and tight and no dynamic headroom- perfect (or close to it) voltage source behaviour. Or there is a "soft", unregulated supply that can naturally limit itself due to the rails sagging in continuous situations and thereby protecting the output stage from destruction. And, at the opposite end, you have a current source amplifier with it's high output impedance.

Basically, the entire thread is about what happens to amplifiers at the onset, or when driven past transient clipping events. As amplifier topologies and PSU topologies vary significantly and the various "protections" range from a simple fuse to a microcontroller monitoring everything, I don't think it is unreasonable to be looking hard at this, don't you?