-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Clipping 101
- Thread starter DonH56
- Start date
Either the voltage swing limit (power supply rails) or current limiting/protection in the amp itself or in the SMPS. Plus, clipping behaviour of class D amplifiers is not that straightforward as of their linear brothers, near clipping the switching may tend to drop-out.
It is generally true that higher PS supply voltage will allow the amp to have higher output power (up to a max limit). Hypex doesn't provide that info in their amp module datasheet, but Purifi does. Higher PS supply voltage will allow higher amplifier output voltage, thus (usually) allow higher output power.
Yes, have seen this and makes sense when the supply voltage is different. What if it is same as when the amp clips despite plenty of power? Could be rail voltage diff if higher voltage (as you show above). Would have to be something else like current limiting as @pma suggests when rail voltage does not change.
Wonder if peak hold on my DMM can capture?
If current limit is exceeded, then the rail voltage will sag too. It is possible you can catch it with a DMM, if you are output continuous power for long enough time. However, peak hold holds the peak, not the sag. Some higher end DMM can measure min/max over a period of time.
restorer-john
Grand Contributor
The various Hypex/Purifi modules have internal clip detection and the way they detect and respond to that will vary from model to model. They can also control the attached SMPS power supply in certain configurations.
The earlier NC-xxx modules use a combination of detections- output current is monitored via an additional winding in the output inductor to determine clipping limits along with input level +/- clipping detection* all fed to an onboard micro (PIC) which can control muting, limiting etc. What the NCX modules have on board, I don't know.
*the amplifier has a fixed gain and that is known and a maximum input level for maximum output voltage is also known. A signal greater than the thresholds in either positive of negative swings of the waveform means the output would be overdriven, so the clip signal is sent to the micro prior to the output clipping event actually occurring.
A DMM is unsuitable for this purpose. An oscilloscope/DSO is required with differential probes/connection.
Clipping in Class D is a whole other ball game and the designers have done a great job ensuring the modules do not self destruct in adverse conditions.
As @pma said, the PSU rails need to be monitored for both current (via a very low value resistor-vdrop) and voltage, all whilst attached to a fast 'scope when driven to the required levels you are investigating.
The earlier NC-xxx modules use a combination of detections- output current is monitored via an additional winding in the output inductor to determine clipping limits along with input level +/- clipping detection* all fed to an onboard micro (PIC) which can control muting, limiting etc. What the NCX modules have on board, I don't know.
*the amplifier has a fixed gain and that is known and a maximum input level for maximum output voltage is also known. A signal greater than the thresholds in either positive of negative swings of the waveform means the output would be overdriven, so the clip signal is sent to the micro prior to the output clipping event actually occurring.
A DMM is unsuitable for this purpose. An oscilloscope/DSO is required with differential probes/connection.
Clipping in Class D is a whole other ball game and the designers have done a great job ensuring the modules do not self destruct in adverse conditions.
As @pma said, the PSU rails need to be monitored for both current (via a very low value resistor-vdrop) and voltage, all whilst attached to a fast 'scope when driven to the required levels you are investigating.
The scope measurements were taken quite long ago from two of my older power amp designs.
You need to go deeper into the issue and thus you need to see what is exactly happening. In case you are an amp designer it is a must and this should apply to OEM assemblers as well. ASR view stays on the surface, not going deep, it is oriented to general audience, not to expert members.
Response from Hypex support…
Clipping by definition is always caused by a limitation in the supply voltage, i.e the output voltage reaching the supply voltage rails.
At lower impedance loads the power will be limited by the current that the amp can deliver.
restorer-john
Grand Contributor
restorer-john
Grand Contributor
Is this behaviour / design used by other manufacturers besides NAD (and you) ? All power tests I've seen done with oscilloscope trace shows hard clipped sinewaves. NAD "soft clipping" excluded.
Clipping is a simple thing. The waveform runs into the limit of the available power supply rails at the time. The more you push past that point, the 'harder' the clipping appears. Some amplifiers behave very poorly at the onset on waveform clipping, others not so much.
D
Deleted member 48726
Guest
Yeah, and that's what I am fishing for. NAD's soft clipping seems (at least on the O-scope trace with sine signal) to work quite alright with softer shape format. And thus wondering if others have similar designs to mitigate too audible / obvious clipping but just don't use it as a feature in advertising.
Absolute clowns.
FInd another OEM. Fast.
I was hoping for more. Your answer was more specific.
I have a inexpensive DSO queued. Availability has been a challenge and so the price has been higher than normal. Is a Hantek DSO2D10 sufficient for audio work?
Last edited:
The "unpleasantness" of clean, hard clipping is vastly exaggerated (see below). So, there is no real need to prematurely clip with "soft clipping". The effects of clipping can be simulated with @pkane's distort.
Topping PA5 measurements )
There was an expensive preamp with some black epoxy potted "opamp" style gain stages, built around through hole components with flying leads embedded in the epoxy block. Can't remember the brand. A long time ago. Made reverse engineering impossible- until the schematic leaked... Burmeister had...
www.audiosciencereview.com
I would suggest getting a 4-ch scope, because you can use 2 channels for differential measurements (and use the "chA - chB" math function, so a 4-ch scope can make 2-ch differential measurements), instead of having to use differential probes.
[Edit] For more info, see: https://download.tek.com/document/3AW_19134_2_MR_Letter.pdf
Last edited:
KSTR
Major Contributor
More often than not those soft-clipping schemes should be better called clean-clipping as the transition still is quite kinky, producing lots of odd-order harmonics rather instantly. But at least the transitions into clipping as well as out of clipping are nicely shaped, no "rail-sticking" hangover and no oscillations.
To answer your question, most (if not all) ICEpower ASC series class-D amplifier modules used in some comercial amps show such a nice clean clipping when not totally overdriven. With a proper front end that cleanly clips (and recovers fast) just a bit above the module's clipping point the behavior can be nice and stable.
Perceptually benign soft-clipping must IMHO start quite at bit earlier to have enough headroom for a very round "knee", starting at about 1/2 max output voltage... but that severely reduces "distortion free" output power to about 1/4th. Therefore, we seldom see such behavior, notably not in solid-state amplifiers. It spoils specifications, a no-go. Yet it may sound considerably better in, say, the "student party" scenario: undersized amp cranked into clipping all the time.
All of the above refers regular voltage clipping, the amp running out of voltage. But when we have current clipping (amp running out of steam on low-impedance speakers) things tend to get nasty. Clean and soft over-current clipping is much harder to implement...
D
Deleted member 48726
Guest
So. Why is it that clipping due to current limitations isn't the same behaviour as with voltage? Is it to do with transformer saturation? Is it the behaviour of the output transistors operating out of the SOA? The rectifier? Or the filter caps not able to stay charged, causing noisy DC supply?
Similar threads
