Noise at SMSL SA-50 (TDA7492) by low level

I note to loud noise by directly connection between Class-D amp inverted input without potentiometer and DSP controller analog output.
Have a look to Fig. 31 on page 24/32 under
http://www.st.com/web/en/resource/technical/document/datasheet/CD00205863.pdf
The input amp operates as inverted amp and Rf determine the input resistance of 60K. This isn't good for good results of signal to noise ratio - so I think:
The introduce of a potentiometer for use as volume control close at the inverted input provides better results.
What is here the right way for best value of signal to noise ratio in the low level area ?
Thank you for your advices.
This URL are also of interest:
http://www.trevormarshall.com/class-d-tutorial/
 
Last edited:
No information?
Under
QA400 Measurements on SURE 2x50W TDA7492 Class D Amplifier - QuantAsylum
I read follow (have a look to fig.8):

"Next, a small differential amp with a gain of 1/20 (0.05X) was constructed from an NE5532. The noise floor was measured (green) and the SURE amp output was measured differentially. Here we are showing the same good THD performance as achieved with the expensive diff probe, but notice too we have a substantially lower noise floor. The power out here is about 2W/"
Means that, that an additional front-end, consist of a NE5532 with voltage gain factor of 0,05 is the right solution?
 
I don't understand the question?

Just try 10k to GND at the input.
But I understand now the reason for the unwanted noise.
Check out fig. 5 under
Audio Designs With Opamps
and fig 1.9 under
http://www.analog.com/library/analogDialogue/archives/43-09/Edch 1 op amps.pdf
Thermal resistor noise of 60K (value for integrated Ri) is no longer negligible.
This means, the value for signal to noise ratio is not very good, particularly in the low level area.
Additional I note, the value of Ri isn't reduce from outside (no PIN available therefore).
Which means further, that the connection of a 10K resistor from input to GND or even input shorting reduce the noise not as strong as usual at non inverting input stages, because the noise of 60K value dominates in an inverting amp configuration.
I don't understand, why ST have selected such an input stage.
 
Last edited:
The S/N value in db isn't helpful in most cases, because in analog and digital communications, signal-to-noise ratio, often written S/N or SNR,
is a measure of signal strength relative to background noise.
Very helpful is the value (mV/uV) of the background noise (noise floor) itself, i. e. without signal, but this value isn't to find anywhere.
Otherwise nobody would be choice this class-D amp model or this IC for various projects.
Under
https://www.wpi.edu/Pubs/E-project/Available/E-project-041808-145643/unrestricted/report.pdf
from page 16 there are to read additional design hints.
For make a good choice this kind of measuring could be helpful - found under
http://www.prosoundweb.com/article/..._discussion_of_power_amplifier_specifications


EFFECTIVE RESOLUTION
The digital version of Dynamic Range (sometimes referred to as DNR) essentially indicates the effective “resolution” of a digital audio device. When used for amplifier measurement, it provides a number related to SNR in that the numbers should be pretty close (for example, 102 dB SNR and 100 dB DNR).

DNR is mostly used to measure digital devices, such as A/D (analog to digital) and D/A converters, and has recently become prevalent in the measurement of digital input amplifiers. It can be used to measure analog amps as well.

The typical method for measuring DNR is to input a small signal (-60 dB, referring to 0 dB for maximum output) and measure THD+N in dB, and then add 60 dB to the measurement. The -60 dB input is used to prevent auto turn-off of any signal stages that might contribute to noise.

If you notice SNR and DNR measurements that aren’t close, a trick is probably being used. The trick in this case is usually some form of gating, as mentioned previously. Gating results in an erroneous SNR figure, but doesn’t effect DNR, so you might notice a much higher SNR than DNR in this case. Exaggerated maximum power ratings also artificially inflate DNR and SNR figures due to increased headroom that’s not really there.
 
Last edited:
from http://www.diyaudio.com/forums/class-d/266561-class-d-elephant-room-hiss-3.html#post4615530
What loudspeakers are you using? If they are high sensitivity, sometimes you really do need to design the system carefully.
What means this exact in case of the TDA-7492 (SML-SA-50) ?? - go to
http://www.diyaudio.com/forums/class-d/286090-noise-smsl-sa-50-tda7492-low-level.html
A friend of me uses follow speakers with digital controller and three SML-SA-50 without potentiometer at the input
Bass: Klipsch horn
Klipsch Khorn V-Trac Upgrades - Restoration - "M" Khorns - Volti Audio
Midrange/hights: Jabo KH72B/BMS4590
http://gwhmm.de/eckhorn/pictures/800_pixel/eckhron_14_800.JPG
Overview
http://www.eckhorn.com/images/phoca...thumbs/phoca_thumb_l_eckhorn sota spiegel.jpg
First step is the introduce of an alps 6-ch attenuator at the inputs.
What is possible additional?
Try connect signal gnd to power gnd (at the PSU capacitors) with a thick wire.
In one occasion I managed to get hizz down to subjectively half the level.
Good advice - thank you therefore.
I will ask for additional advices at the manufacturer of the IC, STMicroelectronics
The most interesting question for me in the moment is follow:
The lowest possible noise floor of background noise in µV/mV at the output from TDA7492 without audio signal at the input and the associated necessary steps therefore (like e.g. emitter/source follower stage at the input).
 
My reading of the DS is that the input referred noise is 25uV typically (unweighted) in the 22Hz-20kHz band, therefore the output noise will be 270uV as the lowest gain setting is 21.6dB.
 

Attachments

  • teffTDA.png
    teffTDA.png
    9.1 KB · Views: 274
My reading of the DS is that the input referred noise is 25uV typically (unweighted) in the 22Hz-20kHz band, therefore the output noise will be 270uV as the lowest gain setting is 21.6dB.
This makes the difference to good known other power amp ICs (class A/B) and OPA's concerning the idle noisefloor:
TDA7492: 2 (max 5) µV (20Hz to 20KHz)
LM3886: 2 (max 10) µV (IHF—A-Weighting Filter, Rin = 600 Ω)
AD797: 1µV (10Hz to 1MHz)
even if one can not directly compare the values due different measurement conditions.

I note to loud noise by directly connection between Class-D amp inverted input without potentiometer and DSP controller analog output.
Have a look to Fig. 31 on page 24/32 under
http://www.st.com/web/en/resource/technical/document/datasheet/CD00205863.pdf
The input amp operates as inverted amp and Rf determine the input resistance of 60K. This isn't good for good results of signal to noise ratio - so I think:
The introduce of a potentiometer for use as volume control close at the inverted input provides better results.
What is here the right way for best value of signal to noise ratio in the low level area ?
Thank you for your advices.
This URL are also of interest:
http://www.trevormarshall.com/class-d-tutorial/
the marked term should read actually "this isn't good for low values of total input noise"
 
Last edited:

irribeo

Member
2013-04-01 2:46 pm
Yes, same question once more, you state in "table":
TDA7492: 2 (max 5) µV (20Hz to 20KHz)
LM3886: 2 (max 10) µV (IHF—A-Weighting Filter, Rin = 600 Ω)

How do you get from 20-35µV to 2 (max 5) µV for tda7492 ? Is there obvious formula for that, I don't see why you reduce the noise spec in that comparison, increasing it to 270µV for the comparison I would have found more logical considering earlier posts.
 
Yes, same question once more, you state in "table":
TDA7492: 2 (max 5) µV (20Hz to 20KHz)
LM3886: 2 (max 10) µV (IHF—A-Weighting Filter, Rin = 600 Ω)

How do you get from 20-35µV to 2 (max 5) µV for tda7492 ? Is there obvious formula for that, I don't see why you reduce the noise spec in that comparison, increasing it to 270µV for the comparison I would have found more logical considering earlier posts.
The comparison of the values of output noise is based on follow:
The result for the output noise is given by multiplication of the value of input noise voltage and the value of the gain factor.
Although I'm not sure, if that's true at all - but I see nothing at the moment that looks like a mistake.
The value of input noise at the TDA7492 is much more greater than the usual values.
According the datasheet on page 10/32 under
http://www2.st.com/content/ccc/reso...df/jcr:content/translations/en.CD00205863.pdf
the value of input noise is between 25 and 35uV.
The voltage gain is dependend of the configuration aproximately between 20 and 30db, i. e. a gain factor between 10 and 30. This means an output noise voltage value between 250uV (lowest possible value) and 1,05mV at shorted input.
This is too high for high efficiency loudspeakers like horn systems.
Which possible I overlooked for additional noise reduction ?
And a second question: Who can upload a schematic of the SMSL SA-50 ?
 
Last edited:
This makes the difference to good known other power amp ICs (class A/B) and OPA's concerning the idle noisefloor:
TDA7492: 2 (max 5) µV (20Hz to 20KHz)
LM3886: 2 (max 10) µV (IHF—A-Weighting Filter, Rin = 600 Ω)
AD797: 1µV (10Hz to 1MHz)
even if one can not directly compare the values due different measurement conditions.

So the comparsion is totally useless. ;)
 
Yes, same question once more, you state in "table":
TDA7492: 2 (max 5) µV (20Hz to 20KHz)
LM3886: 2 (max 10) µV (IHF—A-Weighting Filter, Rin = 600 Ω)

How do you get from 20-35µV to 2 (max 5) µV for tda7492 ? Is there obvious formula for that, I don't see why you reduce the noise spec in that comparison, increasing it to 270µV for the comparison I would have found more logical considering earlier posts.

This one is simply a mistake:
"TDA7492: 2 (max 5) µV (20Hz to 20KHz)" He ment TDA7294.
 
So the comparsion is totally useless. ;)
yes, because I still haven't a helpful step for remove hiss at low level.
The comparison of the values of output noise is based on follow:
The result for the output noise is given by multiplication of the value of input noise voltage and the value of the gain factor.
Although I'm not sure, if that's true at all - but I see nothing at the moment that looks like a mistake.
The value of input noise at the TDA7492 is much more greater than the usual values.
According the datasheet on page 10/32 under
http://www2.st.com/content/ccc/reso...df/jcr:content/translations/en.CD00205863.pdf
the value of input noise is between 25 and 35uV.
The voltage gain is dependend of the configuration aproximately between 20 and 30db, i. e. a gain factor between 10 and 30. This means an output noise voltage value between 250uV (lowest possible value) and 1,05mV at shorted input.
This is too high for high efficiency loudspeakers like horn systems.
Which possible I overlooked for additional noise reduction ?
And a second question: Who can upload a schematic of the SMSL SA-50 ?
Any news?
At the model "SA-S1" from the same manufacturer occurs obviously the same issue - go to
http://www.diyaudio.com/forums/class-d/230986-smsl-sa-s1-low-volume-noise-2.html
 
Last edited:
additional advices for noise reduction are to find here (post #266):
http://www.diyaudio.com/forums/class-d/275505-tda7492-amp-27.html
and here (post #25):
http://www.diyaudio.com/forums/class-d/266561-class-d-elephant-room-hiss-3.html

Nevertheless I want to have the genuine schematic.
Unfortunately until now I get no reply of an inquiry to
[email protected]
the address is follow:
8th Floor
B6A Building JunFeng industrial zone
Fuyong
BaoAn district
Shenzhen
China
Maybe a member of China can ask for me there.
Thank you for your efforts.
 
Last edited: