On Class D Amplifiers Measurements

[solderdude]

That's not what I call a powerful transmitter.

Interesting to follow 2 guys measuring the same types of devices and interpreting the measured results from different angles/viewpoints.

Let's agree there is not much RFI coming from the device measured by March Audio ?

 

[maty]

Today, with a very cheap class D amp, badly implemented. By Archimago.

MEASUREMENT: Yeeco TI TPA3116 Class D Amplifier (aka Nobsound NS-01G, Douk Audio F900S, Mogu F900S)

http://archimago.blogspot.com/2019/11/measurement-yeeco-ti-tpa3116-class-d.html

Another reason I wanted to measure this device is because this is a Class D amplifier. We are currently well into the ascent of Class D designs and I would not be surprised if by the end of the 2020's, almost all new devices are Class D by then for many good reasons! What we see in this device is a reminder of various limitations we should be aware of - potential low damping factor, phase variation, and ultrasonic noise...

 

[Juhazi]

^Wellwell, but who woud call that hifi device?

 

[oivavoi]

I really wish those who are critical of class d - or think that measurements are misleading - could provide some hard irrefutable evidence. What would that be? Simply to measure the output of actual speakers run with class d amplifiers, ncores or others, compared to top of the line ab amplifiers. Is there an actual difference? Or not? If there is an actual difference, I would like to know that, so that I'll go class AB if space allows (or perhaps choose class d amps that don't exhibit such behavior). If no difference can be seen at all, I'd like to know that as well, so that I don't have to worry about getting a class d amp that saves space and electricity and makes the wife happy.

EDIT: As was posted in the thread already, Archimago did just that with this cheap amplifier, and demonstrated that it actually affects the output on the speaker https://archimago.blogspot.com/2019/11/measurement-yeeco-ti-tpa3116-class-d.html. Cheap class d on this level is obviously not high-fidelity

 

[solderdude]

Most likely the distortion of the speakers, microphone, microphone pre-amps, slight positioning variances of for instance people present in the room and room reflections would drown out the subtle differences we would see between a good analog and class-D amp for certain.

One just needs to perform the same electrical measurements (using real loads ?) with all the possible measurements needed to evaluate both types of amplifiers.

 

[oivavoi]

Most likely the distortion of the speakers, microphone, microphone pre-amps, slight positioning variances of for instance people present in the room and room reflections would drown out the subtle differences we would see between a good analog and class-D amp for certain.

One just needs to perform the same electrical measurements (using real loads ?) with all the possible measurements needed to evaluate both types of amplifiers.

Good points. But I find it much more convincing when using real loads, rather than theoretical theorizing which goes over my head... This measurement here by Archimago, for example, tells me quite clearly that the high-frequency noise of this particular (cheap) class d amp actually does have an impact in the audible range around 1 khz:

 

[solderdude]

Yes, that appears bad. Most likely the average level of the noise is close to 0.
When measured through a very steep 20kHz filter it may not look that bad.

I agree on the real load thing. The problem I see here is that I am fairly certain any of the speakers in my house will measure / react very differently from the ones in your home and from those doing measurements as well as those of a 'standardized' semi realistic load.
That is the real difficulty in measuring power amplifiers.

 

[SIY]

This measurement here by Archimago, for example, tells me quite clearly that the high-frequency noise of this particular (cheap) class d amp actually does have an impact in the audible range around 1 khz...

How so? The noise there looks like it's ultrasonic and fairly low level (30mV RMS or so).

 

[Juhazi]

Coil/magnet motors in tweeters hardly ever have any response to signals over 40kHz. Some other types (electrostaic, magnetostatic) have some response above that too. Only some young humans can hear above 20kHz, when spl is high enough. So, for audio applications only interferences induced by supresonics might be heard.

We grey-haired members remember well what 450/900 NMT and early GSM phones could do for hifi!
https://en.wikipedia.org/wiki/Electromagnetic_interference
https://www.rfvenue.com/blog/2015/0...-phones-from-interfering-with-audio-equipment

 

[oivavoi]

How so? The noise there looks like it's ultrasonic and fairly low level (30mV RMS or so).

Hm... elaborate please? I'm a non-techie idiot so I need the "for dummies" version

 

[March Audio]

Good points. But I find it much more convincing when using real loads, rather than theoretical theorizing which goes over my head... This measurement here by Archimago, for example, tells me quite clearly that the high-frequency noise of this particular (cheap) class d amp actually does have an impact in the audible range around 1 khz:

No, the scope shows a wide bandwidth, at least 100MHz. So that includes and shows the switching frequency. This switching frequency is effectively subsequently completely removed because the speaker is a mechanical and electrical filter. The speaker has no response at 450kHz (or higher). It simply can't reproduce that high frequency signal you see riding on the square wave.

What is left and reproduced is the clean audio band signal.

 

[solderdude]

Hm... elaborate please? I'm a non-techie idiot so I need the "for dummies" version

The square-wave is 400mV tt so effectively a DC voltage of 200mV = 10mW output power.
There is about 80mVtt signal riding on the signal (with some peaks exceeding that), but this is most likely a sinewave as it passed through a coil.
This means the 'noise' is about 30mVrms.
IF the speaker would still be 4 Ohm at that frequency (which it isn't most likely) the noise would be 0.2mW (in the tweeter only) but will be lower due to a rising impedance (with dome tweeters)
Harmless and above the audible range.

 

[tw99]

As a user of a Hypex based amp, I like the fact that I have a small, energy efficient, relatively cheap box, that measures well on the main parameters, and drives my speakers very effectively to beyond comfortable listening volume. I'm not aware of any Class A/B device that combines all those attributes. So personally I'm less bothered by the nuances of corner case measurements as I'm happy with the tradeoff. Even if it is in some way worse measuring than a traditional Class A/B amp, it's not going to be audible to me.

 

[oivavoi]

No, the scope shows a wide bandwidth, at least 100MHz. So that includes and shows the switching frequency. This switching frequency is effectively subsequently completely removed because the speaker is a mechanical and electrical filter. The speaker has no response at 450kHz (or higher). It simply can't reproduce that high frequency signal you see riding on the square wave.

What is left and reproduced is the clean audio band signal.

Thanks. I may be a victim of FUD here.

 

[Matias]

Even if it is in some way worse measuring than a traditional Class A/B amp, it's not going to be audible to me.

The point is that they measure better where we can hear and measure worst where we can't hear. Smart.

 

[SIY]

Thanks. I may be a victim of FUD here.

There does seem to be a lot of it. Not sure why. I've had three different Class D amps cycle through here in the past month, and the thing that leaves me marveling is how much better they performed in the audio range than nearly any old school amp- which is what counts. Any weirdness in the ultrasonic or RF that would affect the audible range has mysteriously not visibly or audibly manifested itself.

 

[pma]

Facts matter? Sounds like religion, not science.

 

[tw99]

Don't think anyone is saying facts don't matter. Some facts are just more relevant than others.

 

[Juhazi]

Facts matter? Sounds like religion, not science.

???

 

[amirm]

At a measly 5 watts Amir. Let's not play games here.

What do you expect to happen to frequency response at higher levels?