Low distortion Class D amplifiers? What to expect?

Jump to Latest
C

cheater

Member
Joined 2004
#1
Hi, I was wondering what sorts of distortion numbers people have measured in their Class D (and related) amplifiers, measured at full power? Both DIY and commercial, in the ~300-500W range. I'm looking over the docs for my Bryston electronics and I'm thinking of an upgrade at least for the LF and LFX drivers, and Class D was my first thought. The Brystons achieve around 0.0015% across the frequency range. That's roughly -40dB (on a power scale). The Brystons (3B, 4B, 7B, from the SST 2 series) are a bit of an older design (Class AB) and I guess modern Class D/T/etc designs can be much more efficient and more linear, and I wonder what performance numbers I should be looking for when replacing them. I want something good but also something that's reasonable.

The numbers measured during final checkout as supplied on the certificate by Bryston for my units are as follows:

Code: 
Distortion percentages at full rated output power

ID                     20 Hz       200 Hz      2K Hz       20K Hz      Rated power
3BSST2 CH1 (HF L)      0.00177%    0.00161%    0.00153%    0.00498%    150W/channel
3BSST2 CH2 (HF R)      0.00232%    0.00193%    0.00187%    0.00414%    150W/channel
4BSST2 CH1 (MF L)      0.00138%    0.00132%    0.00153%    0.00485%    300W/channel
4BSST2 CH2 (MF R)      0.00191%    0.00177%    0.00139%    0.00493%    300W/channel
7BSST2 unit #1 (LF L)  0.00125%    0.00105%    0.00144%    0.00444%    700W/unit (mono)
7BSST2 unit #2 (LF R)  0.00117%    0.00098%    0.00138%    0.00438%    700W/unit (mono)


Relative levels of distortion to signal
(calculated using Y=20 log_20(X) with X from above, eg 0.00177% -> X = 0.0000177)

ID                     20 Hz       200 Hz      2K Hz       20K Hz
3BSST2 CH1 (HF L)      -73.05 dB   -73.68 dB   -74.02 dB   -66.14 dB
3BSST2 CH2 (HF R)      -71.24 dB   -72.47 dB   -72.68 dB   -67.38 dB
4BSST2 CH1 (MF L)      -74.71 dB   -75.01 dB   -74.02 dB   -66.32 dB
4BSST2 CH2 (MF R)      -72.54 dB   -73.05 dB   -74.66 dB   -66.21 dB
7BSST2 unit #1 (LF L)  -75.37 dB   -76.54 dB   -74.43 dB   -66.91 dB
7BSST2 unit #2 (LF R)  -75.81 dB   -77.00 dB   -74.71 dB   -67.00 dB

noise levels for balanced input are all -116 dB to -113 dB

So I'm primarily looking to get a lower distortion unit while keeping or improving the power handling (especially on transients) and keeping or improving the noise levels.

How do those numbers compare to more modern designs? Alternatively - what would you say are the lowest distortion and noise amps you've tried that I can look up these numbers for?

One thing - I'm not really interested in talking about how you should be listening to the amplifiers rather than looking at numbers etc - I appreciate this train of thought and find it's valuable but this is not what this thread is for. I'm looking for measured numbers to have an objective comparison, long before I start doing all sorts of listening tests. Always feel free to start your own thread and I'll be glad to contribute 🙂 Thanks
 
Last edited:
#4
I do not think it is a bad idea. But I am not in the boat of the golden ear crowd who believe in their unlimited diagnostic capabilities. Concerning audibility, below a THD of, let us say 0.01%, it is not worth big efforts. As a pure technical challenge - it may be fun as well😉
 
#5
Thanks. I want this setup for a mastering studio, so low distortion is pretty important here. Whether I can directly hear it or not, a signal at -50dB to 40dB will skew my perception of the audio. The speakers I am using are PMC MB2 XBD if that makes any difference. I don't think they can easily be improved acoustically, but better electronics seem like they could be an "easy win" here. However, you are saying that Class D and related are not going to improve distortion that much. This means to me I'll have to research quite a bit to find something better. Thanks.

If anyone else has any suggestions what amplifiers to look at - DIY projects or not - let me know please.
 
#9
Normally no-one specifies THD as a power ratio. Normally it's the ratio of the RMS value of all the harmonics combined (excluding DC and the fundamental) to the RMS value of the fundamental multiplied by 100 %. That is, it's a voltage or current ratio and no power ratio.
 
#12
cheater said:
BTW, haven't noticed that before.

The THD numbers are percentage of power. Power decibels are calculated as 10 * log base 10. Voltage decibels are calculated as 20 * log base 20. Are you saying I used this incorrectly?
Click to expand...

there's no log20, that would be vicesibels, not decibels, assuming I've found the right latin word (!)

power is proportional to amplitude squared, and the log of a square is double the log of the value.


Basic law of logs: log(a) + log(b) = log (ab).
 
#14
The dB's of distortion to signal ratio work out to the same number whether you do it as power or voltage.

But you cannot do power decibels unless you square the voltage of the signal and also square the voltages of all the harmonics. Also, one is not then doing RMS summing of the harmonics, but MS summing of the harmonics (no square root)

Then 10*log(ratio) is the right answer. It will be the same answer you get if you RMS sum the harmonic voltages and take THD as a 20*log(ratio) of sum(RMS harmonics)/signal voltage.

Example: 50W power into 8 ohms, with 1 mV RMS total harmonic distortion.

As a Voltage ratio:
Signal Vrms = sqrt(50W*8ohms) = 20 Vrms
Distortion %= 100*(0.001 Vrms / 20 Vrms) = 0.005%
THD ratio = 20 * log(0.001/20) =-86.02 dB​

As a Power ratio
Signal Power = 50W
Distortion Power = (1mV)^2 / 8 ohms = 125 nW
THD ratio = 10*log(125nW/50W) = -86.02 dB​
 
You must log in or register to reply here.
Top Bottom