• 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 daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Why ADC converters do not use proper input anti-alias filters?

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,516
Likes
10,544
Location
Prague
Why ADC converters do not use proper anti-alias filters?

As we all know, the Nyquist theorem states a necessary condition of proper sampling - input signal must not contain frequencies above Fs/2, half of sampling frequency. If this condition is violated, we are experiencing alias frequencies that are not present in the input signal and the spectrum of the sampled signal is thus contaminated. The worst situation is when input signal frequency, Fin and sampling frequency Fs are related so that

Fin - N*Fs < Fs/2

This is especially critical when we measure class D amplifiers, as they always have considerable level of switching frequency residuals in their output spectrum. I will demonstrate this issue on my PMA-NC252MP amplifier measured through E1DA Cosmos ADC designed by @IVX .

I measured switching frequency residuals at NC252MP output, it was about 350 mVrms with frequency around 380 - 400 kHz, and this frequency fluctuates with output signal level.
Let's relate it to N*Fs, for sampling frequency 48kHz and N = 8 we get 384kHz and alas, 384kHz subtracted from switching frequency falls to the audio band. With respect to the fact that the switching frequency floats with signal level, the alias frequency will travel through the audio band. For 96kHz or 192kHz sampling frequency, we have similar issues.

Enough talking and let's see some measurements.

As a signal source, I have used my analog generator which operates in range of frequencies from 10Hz to 1MHz. It has distortion not better than -90dB, but for this test it is more than good enough. Output voltage level was set to 377mVrms, similar as level of switching residual from NC252MP. The generator frequency was set to ~400kHz, similar as NC252MP switching frequency. 4KHz signal of same level 377mV was used as well as a reference. ADC used was E1DA Cosmos ADC, set to 2.7V input range and provided with an additional input resistor divider that makes the ADC full scale range 0dBFS = 45.547V. As a comparison, the same measurement configuration was used, but with LC low pass filter added, which has flat response to 20kHz and then the slope -40dB/decade. This filter should attenuate 400kHz of some -46dB. The filter was described here in another thread:


Measurements

4kHz / 377 mV sine as a reference
GAG 4kHz_377mV.png

Signal level is -41.6 dBFS


400kHz / 377 mV sine (without LC low pass filter)
GAG 400kHz_377mV.png

We can see alias frequency near 10kHz with amplitude of -98.4 dBFS. So the alias is suppressed by the ADC alone of 56.8dB.


400kHz / 377 mV sine (with the LC low pass filter)
GAG 400kHz_377mV_LC.png

Now the alias amplitude is -140.1 dBFS. Additional suppression with the low pass filter is 41.7dB.

Now let's see the comparison in NC252MP intermodulation distortion measurement, 60Hz + 7kHz SMPTE IMD, 77W/4ohm. We can see the major contribution of the LC low pass filter that lies in the higher dynamic range of the measurement and noise floor decreased of about 12 dB.
PMA-NC252MP SMPTE LCfiltr anone.png


Conclusion

Though SOTA DAC converters are usually equipped with good reconstruction filters that effectively suppress the mirror images, the same cannot be said about SOTA audio ADC converters. By definition, sampled signal must be frequency limited to <Fs/2, however it does not happen effectively enough and ADCs usually have no input filters or simple filters, which does not prevent them from creation of alias frequencies that fall into the audio band and into converters usable dynamic range. I understand that it is not easy to provide the ADC with switchable input filters, however designers should consider this request that reflects definition of proper sampling. This is especially an issue when measuring class D amplifiers or components with noise shaped output. The situation becomes worse and worse as the level of the useful input audio signal is decreasing at the constant level of the switching residuals.
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
7,112
Likes
11,321
what Nyquist filter was your Cosmos ADC set to?
Screenshot 2022-12-18 192800.png
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,264
Likes
36,507
Some additional measurements of other ADCs.

 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
3,287
Likes
3,222
Location
.de
Under these circumstances, recording at fs = 48 kHz arguably isn't the smartest thing to do. The main AA filter generally is an 8X oversampling job (following fairly limited filtering during preceding decimation), so I wouldn't rely on it being very effective past about 7.5fs, particularly between about 7.5fs and 8.5fs - and in this case the offending signal is right there. Results at 88.2 or 96 kHz should be substantially better. I would also try whether 192 improves anything further. The Cosmos ADC has a response extending to 180+ kHz, so analog filtering alone can't be arbitrarily far down at 380-400 kHz, barely more than an octave higher.
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,516
Likes
10,544
Location
Prague
The situation is even more complicated for the reason that the HF spectrum at the Hypex module output, when amplifying input signal, is quite complex (switching frequency spectrum travels with amplitude of the input signal).

Under these circumstances, recording at fs = 48 kHz arguably isn't the smartest thing to do.
The result was not much better with Fs = 96kHz. The issue seems to be quite complex.

Below is the HF spectrum at NC252MP output when driven with 4kHz sine and the output level is 12Vrms. One can see the HF spectrum around the switching frequency that covers range of some 350 - 410 kHz. PWM modulator spectrum attenuated by the simple output filter of the amplifier.
8-bit scope is nothing special however enough to demonstrate the phenomenon.

NC252MP HF spectrum.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,516
Likes
10,544
Location
Prague
Further plots show measurements at NC252MP output with and without the LC filter. Signal source is my analog generator. It has quite high distortion, however the HF spectrum is clean. All the artifacts marked yellow are switching frequency aliases. Those aliases depend both on input signal frequency and amplitude. 2kHz base frequency has 13Vrms.

GAG+NC252MP+LC.png


GAG+NC252MP+nofilter.png


If the ADC had a proper built-in input anti-alias filter, the use of a bulky LC external filter would not be necessary.
 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
3,287
Likes
3,222
Location
.de
That looks like IMD to me rather than aliasing, in line with the level dependency you describe. You may be running into slew rate limiting in the input stage or something. Just because it's an ADC doesn't mean that the old analog problems will go away... (The Cosmos ADC is in good company if you consider that even the Audio Precision can be upset by ultrasonic noise, see e.g. Amir's Arturia Minifuse review.)

I would suggest using the signal generator to trace filter response starting from 20 kHz. You should be able to confirm (spot-check) the simulated responses as per ES9822PRO datasheet pp.17-19, accounting for fs differences and analog input filtering. (Note that responses are printed up to 4fs only, the spectrum above that should be the same mirrored about 4fs.)

Only if that turns up a systematic issue with filter ultimate rejection would I suspect deeper-rooted issues. High phase noise is known to limit filter performance, for example. (That being said, I have no reason to suspect that the Cosmos ADC would be suffering from such issues.)
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,603
Likes
5,855
Location
Berlin, Germany
Class D usually requires such LCR filter, for example, AP AUX 0025 for $1500(or $5 in DIY implementation).
An actually equivalent DIY one-off build will be more like $150... 2-channel, metal casing, connectors, PCB...
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,338
Likes
685
Recording in DSD should solve most of the issue of non-brickwall ADC filters in PCM mode... at least up to some decently high frequency.
I think I'm agreeing with that but not sure why-can you explain further?
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,603
Likes
5,855
Location
Berlin, Germany
All ADCs and DACs are Sigma-Delta type these day, internally, sampling at very high sample rate with just a single bit of resolution and rely on heavy noise-shaping. The requirements for the analog filter in front of an ADC is very relaxed.

Conversion to/from PCM format requires downsampling/upsampling with the associated anti-aliasing/-imaging digital filters. This filtering could be made perfect but for technical reasons most often these digital filters are not brickwall type.

Never leaving the Sigma-Delta domain avoids any need for filtering until the final "conversion" output to analog... the conversion only requires that very output filter, nothing else. No actual DAC proper, that is, but the demands on that 1-bit output stage and filter are extreme.

Actual practice thererfore is different, Sigma-Delta modulators are multibit these days, something like 5 or 6 bits. To get DSD 1-bit data from/to that, remodulation is required but usually the filter advantage remains, to some extend.
 
Top Bottom