Class D amps and compression

[Axtman]

I was playing my Fender Rumble 100 amp yesterday. I swear that it sounds like it has a little bit of compression even though it does not have a compression control. Am I imagining things or do Class D amps have a compressor to limit peaks?

[uneumann]

I glanced at the schematic and if it's accurate (attached) this is not a class D amp. It does have some diode circuits in the power amp that appear to be there to do soft clipping (or compression) at high signal peak levels.

There are many ways an amp can compress the waveform on it's way to your ears. Speakers, power amps, preamps... they can all be designed to compress waveforms. Even just simple clipping is a form of compression. Not all sources of compression feel or sound the same, so much depends on how you the listener/player perceive it. If you like it, it's a good form of compression... run with it.

Edit - attachment: fender_rumble_100_sch_rev-a.pdf

[Axtman]

Okay my bad. I forgot to mention that it is a Fender Rumble 100 Version 3. The schematic you posted I think is for Version 1. I think this is the correct schematic. http://www.fmicassets.com/Damroot/Or...s-Diagrams.pdf

[Jazz P Bass]

The power amp is a Bang and Olufsen SX ...whatever.
I have had these units on the bench and the power ratings are total BS.
Most probably what you are hearing as 'compression' is the power amp self limiting.

[Axtman]

Jazz P Bass,

Thank you for confirming that I am not crazy. I will let my therapist know.

Do all Class D amps have self limiters? Are limiters intrinsic to the design? Are they like cars where you can stomp on the accelerator all you want but still won't go any faster. Or do they have a limiter in order to prevent bad things from happening to the amp if it peaks out?

[J M Fahey]

Rumble amps all have limiters.
SS amps in general have them because of ugly buzzsaw distortion at clipping, Class D even more because they are uglier.

Each designer has his own pet ideas which he repeats over and over.

First is Rumble 100 V1 compressor using a now unavailable CA3080 OTA, second is V3 using an LM13700.

@ JPB: please,next time you have one on your bench, measure actual power and post bresults, so we know the proper "derating ratio" he he.

[Axtman]

It's kind of ironic how Fender puts a limiter on the amp to reduce the distortion yet put a distortion control on the amp.

[nosaj]

I believe drive and level will be in the preamp sections whereas the limiting you are referring to happens in the output section.
nosaj

[J M Fahey]

Yes..
The limiter (variable gain) circuit is in the preamp ... simply because here the power amp is a store bought unit not even them have the circuit for ,so they can´t add a limiter circuit, for all practical means it´s a sealed black box.
So next best is to monitor speaker out signal and compare that to rail voltage, when it reaches, say, 90% of that, sometimes closer, they generate a trigger signal and send that to the variable gain model up front.
*Some* Class D modules have clipping detectors and that signal is available at some pin , usually it triggers a clipping Led but can be used for limiting, but the first approach, checking peak output voltage is "universal".
problem is it always robs you of some possible voltage because it is not REALLY detecting clipping, just estimating it´s close, so tney play it safe, in any case 90W RMS and rated 100W RMS are close, at least by ear.

And yes, you will hear it "pumping", you slap and *everything else* is attenuated.

Just in this particular amp, they also have a "vintage" setting, where they use a poorly adjusted extra limiter (on purpose) to mimic an amp getting "out of breath" because of worn tubes. This causes distortion but less harsh than plain clipping, closer to having lossy caps wreaking havoc on "preamp tubes bias"

[Gaz]

What's wrong with using off the shelf power amps? They use them because only a few companies make them for MI use that work reliably, Icepower being one of them. Should Fender make their own tubes and capacitors too?

All these units can only take so much voltage at their input so they need a compressor or limiter at their input if the preamp outputs too much voltage. I don't really see how this is different than any other amp where the last stage clips first, except that they make the penultimate stage clip or compress enough first to avoid the nasty Class D clipping. My guess is that the Rumble's circuit is just not transparent and pumps in an 'unnatural' way. I've experimented with this and a diode clipper before the power amp sounds good and the most "tubey" (and costs about $.25).

Also the comment about the Overdrive control is ridiculous. A musical overdrive circuit in the preamp is not the same as preventing a class D amp from clipping. That said, the Rumble "overdrive" barely breaks up and sounds horrible. Whoever voiced this amp has never played bass before as far as I can tell.

[teemuk]

V1 and V3 OTA circuits are very different.

V1 has the usual Fender "Delta Comp" compressor that they have employed since 1980's: Distortion detected from differential input triggers the circuit and that diode-resistor-capacitor circuit introduces attack and decay characteristics. Gain is reduced throughout the entire signal and very little harmonic distortion is introduced by compression.

V3 circuit introduces instantenous gain compression, which produces soft "clipping" and plenty of harmonic distortion: OTA is driven with asymmetrically enveloped signal and gain reduces simultaneously (and instantenously) while signal amplitude increases. (This is similar mechanism to tube compression). Such instantenous gain reduction, proportional to input signal amplitude, produces soft peak clipping distortion as waveform peaks are gradually "squashed". Asymmetric drive of OTA' gain control produces asymmetric distortion: one half wave is gain compressed more than the other.

V3 series amps feature this soft clipping in both preamp and power amp. They also feature the traditional Delta Comp circuit.

Picture above actually depicts the "Vintage" mode circuit, which is basically just the asymmetric soft limiter described earlier and some treble cut filtering. "Overdrive" mode adds two more of the aforementioned limiter stages in cascade and a user adjustable gain control for controlling overdrive. Before power amp there's yet another one of these soft limiters leashed to control signal driving the PA, now configured to limit symmetrically instead. That same stage also incorporates the traditional "Dyna Comp" compressor for gradual amplitude reduction whenever signal is estimated to distort.

Yes, in both V1 and V3 the Dyna Comp circuit operates at fixed settings and is non-user adjustable.

[teemuk]

So to be more specific.... Rumble V3 has both soft clipping limiter and a compressor circuit to limit signal driving the power amp module. Both soft clipping and gain compression should be visible on scope. Obviously it's not trivial to measure average continupus output power at low THD from such setup.

Some Class-D modules may be optimized to give high burst power, which works fine for signals with high crest factors, such as music.

Nothing universal can't be said about class-D and clipping/limiting in general. It's always circuit dependent and class-D just indicates switching topology is used and nothing else. Some class-D driver circuits do have built-in limiters, and when they do they appear in various forms ranging from soft clipping circuits to plain gain reduction. One scheme modulates the carrier to introduce soft clipping whenever overmodulation is detected but not all class-D amps contain such sophisticated circuitry.

Like in Rumble's case, manufacturer may also choose to introduce their own external limiting circuits.

[R.G.]

This issue was originally a problem with simple solid state amps. Vox achieved an early and very usable solution. They contrived an adjustable limiter that did an early form of soft clipping of the preamp signal to the power amp, and a way to adjust it so the native clipping of the power amp was never heard.

[teemuk]

Indeed. And Warwick (Thomas Organ Vox parent company) patent circuit also specifically aims for "soft" clipping in order to reduce higher order harmonics of clipping distortion. The patent is also moderately specific about application being amplification of musical instruments, not amplification in general. Why? The drawback is that such "softness" steals theoretical headroom and reduces maximum output power.

Therefore sometimes amps feature peak clipping limiters that clip "hard" instead in order to maximize how high signal can swing within the power amp. This hard clipping isn't neccessarily as bad-sounding as people generally think and expect as long as clipping itself is "graceful" without any nasty side effects such as rail clamping, hysteresis or spurious oscillation, which produce much higher and more dissonant harmonics than mere "hard" edges of the clipped waveform.

In practice, the reduced headroom, resulting into earlier - yet soft - clipping might easily be more audible than harder clipping at higher headroom. Some amplification applications are very non-tolerant to any distortion.

...so if a class-D module aimed for universal use has a peak limiter it probably makes more sense for it to be "hard" instead of "soft".
In musical instrument amplification (excluding instruments such as flutes or synthesizers) softer clipping may be more beneficial. But musical instrument amps are niche market for vendors of class-D (or any other) power amp modules.

And if hard clipping was as horrible as claimed most tube power - and pre - amps would require soft limiters as well. ;-)

[loudthud]

I'd like to point out that Class D is a form of pulse width modulation, but you can't let the pulse width collapse to nothing when the amp clips. It would be next to impossible to filter out the RF noise if you let that happen.