Class D amps and compression

[R.G.]

@Teemuk: Yes. there is a similar ongoing discussion in the high-accuracy amplifier hifi world about soft clipping, where the problem is steering between the harsh clipping of 0.0001db over the power supply limit of the amplifier, and the limits of accuracy if you don't use soft clipping. The camps are (or were, when I last read about that) loudly divided over whether it's better to have soft clipping to hide the hard clipping when you run out of amplitude, or just build monster amplifiers that never -ever- get over the amp's clipping limit. So you have a fractional-kW amp to put out 1-2W of actual music and not clip. Scylla and Charybdis here.

@loudthud: There are several ways this is dealt with. Some of them have zero signal be active 50% duty cycle, constantly. So they never collapse in the sense I think you mean.

[Mark Hammer]

It seems we have overlooked the fact that the Rumble is a bass amp. Bass amps typically require a heftier output that guitar amps, simply because a) the bass generally needs to be cleaner than many of the other instruments in an ensemble, and b) bass transients can tax even the best of amps. So the inclusion of a compressor/limiting circuit is not so much to provide an "effect" as it is to reduce the demands on the power stage of the amp. Use of a class D circuit is part and parcel of having a bass amp that can be efficient (hence lightweight) and remain reasonably clean despite typical transients and a modestly-rated power section.

[tedmich]

FYI Sale on IcePower 1000ASP
https://www.parts-express.com/icepow...1000w--326-206
~$0.25/watt (if it actually delivered 1kW...

[Jazz P Bass]

"(if it actually delivered 1kW."
That was the point of my original post.
They do not deliver the stated output.
Maybe 'burst power' it does. (whatever the heck that is.)
If you read between the lines of there datasheet, they kind of fess up to it.

[teemuk]

Just glanced at the ICE Power datasheet: At least 1000ASP has an internal soft clip circuit as well, which can be turned on and off. Even when off, the integrated protection circuits - which there are many in this design - will activate the soft clipping.

The datasheet also states that the module is designed for music reproduction and output power will never be continuous. "If the average power exceeds 85W @4 ohm ... for a long time at 25 degree C ambient temperature, ...the temperature protection will be activated". Limit is only 40 watts at 50 degrees C!


There are two temperature protection circuits: First one limits output signal voltage, second one shuts down the amplifier.

BTW, Rumble 100 V3 DOES NOT employ an ICE Power module like higher powered Rumble amps, 500 and 200. It has Fender's own class D circuit based on TDA8950 chip.
The chip datasheet does not mention any soft limiting feature but there is a similar two-stage "Thermal FoldBack" protection with gradual output signal limiting and shutdown.

[teemuk]

... 'burst power' it does. (whatever the heck that is.)
If you read between the lines of there datasheet, they kind of fess up to it.

Well, read between the lines they have designed the modules for music reproduction and assume a crest factor of about 8:1 (transient peaks eight times higher in amplitude than the average signal).

You can calculate the difference of assumed peak power and average power and the latter one will be a pretty low figure in comparison to first. The modules will produce the peak power for a duration of several milliseconds, often even seconds, which is well enough to cover all frequencies within audible bandwidth and definitely enough for a bass guitar.

We don't run motors with such class D modules and therefore there is no need for continuous output power delivery. Clean transient reproduction for brief period of time should be sufficient for the intented applications.

As said, given all protection methods, measuring output power is not trivial. Point is, one should expect from the beginning that the rating is not for continuous output. What these modules can handle is a factor of many things, efficient cooling and dissipation probably having high importance.

The Fender schematics, BTW, state measured output power (at 100 Hz and at the onset of limiter) and I trust these to be quite accurate regardless of chip or module capability/incapability:
40: (TDA8950) 36.9 watts to 8 ohms
100: (TDA8950) 99.4 watts to 8 ohms
200: (50ASX2) 155W to 4 ohms
500: (125ASX2) 400W to 4 ohms

Without pre-heating...
50ASX2 is rated to sustain 170 watts to 4 ohms for 120 seconds
125ASX2 is rated to sustain 400 watts to 4 ohms for 80 seconds
1000AS is rated to sustain 1000 watts to 4 ohms for 15 seconds

[R.G.]

Teemuk is correct here - the ciritical thing in thermal limit is how much heat is generated inside the module and how effectively it's removed from the module. Class D amps generate far less heat inside the power amp itself than Class AB. Class AB typically wastes ... um. where is that number now that I need it?? I'll look it up, but about a third of the total power running through the output stage. Switching schemes are much better, typically less than 10% and in many cases under 5%. So - less heat generated in the module to heat it.
Still, you have to get the heat out. A Class AB 100W amp might well overheat at a watt or two with no heat sinking. The Class D needs cooling in the same way, just less of it because the Class D scheme, whatever its other flaws, generates much less heat inside it. So the same approach as with Class AB can be used - attach a better heat sink to the Class D module. Better cooling means it will not reach thermal shutoff as quickly.

[Enzo]

I learned that lesson a long time back when repairing my first Carver PM1.5. I repaired all that was wrong, and put it on the burn-in bench. Put it on loads and fed it a sine wave. Worked great...for a time, then went into thermal. Did this consistently. I wasted a lot of time trying to tweak bias or whatever. FInally called Carver for advice. The Carver guy basically said "You did WHAT? Oh my lord, the amp would never stand that. It is working correctly. It was designed for music, not sine waves."

[R.G.]

It's sobering to realize that no matter how small the power dissipated in an object is, its temperature will rise >>without limit<< until the heat flow out equals the heat being generated inside. Remember grain-of-wheat bulbs? Those things run on a fraction of a watt. But the gas and glass surrounding the filament insulate it so well that the temperature rises until the filament is at a yellow-white glow so the heat can radiate out as light to balance the heat leaving with the heat generated. A half-watt resistor is only different in that it can get heat out of itself much more efficiently, so the temperature where it reaches power balance is much lower.