No balanced input on the sound card.. I only measure 'half of the output', with ground as reference.
Do you think there will be some cancellation that I'm missing in the measurement?
Can you just connect some speakers and listen to some music? How does it sound and is there noise with inputs grounded?
What kind of power supply do you use?
Today there is a tendency of feeding smart phone signals into amps and complaining over "not sounding well". A smartphone is no high end audio device and uses all kinds of non audiophile parts and compression techniques.
A proper audio source is a CD player and a normal (not MP3!!!) Cd.
What kind of power supply do you use?
Today there is a tendency of feeding smart phone signals into amps and complaining over "not sounding well". A smartphone is no high end audio device and uses all kinds of non audiophile parts and compression techniques.
A proper audio source is a CD player and a normal (not MP3!!!) Cd.
So you are one of the "try first, read later" kind. This will get expensive and frustrating.
See, theory is the most practical thing there is. If theory tells you it will not work, you do not need to try it. Unless you are a total #§@&?=!
See, theory is the most practical thing there is. If theory tells you it will not work, you do not need to try it. Unless you are a total #§@&?=!
On the other hand I must admit, the (today) 1128 pages and 11286 posts of this thread are a bit much for the average reader.
In this thread any modification has been made and any board been described, testet and changed until it worked fine.
In this thread any modification has been made and any board been described, testet and changed until it worked fine.
Thank you for understanding 🙂
Yes, on 'side projects' like this, I just want to do a quick and dirty to get some bonus points from 'the boss at home' 😉
But on the other hand I like playing around with it a little bit, and get it to a standard where I'm not annoyed by it when I hear it. I think now it will be quiet (no audible noise) and sound good enough.
Thank you for your support!
I would not trust such measurement. Or measure the output with notebook running on batteries so that it can float against the amp ground.
As of the load - just add shunt diodes to Vcc and ground for each output https://www.diyaudio.com/forums/class-d/237086-tpa3116d2-amp.html#post3528020 (there was a schematics in the following post, which showed the shunt diodes).
No calibration done, I just installed it on a laptop here at work and ran the measurement with my USB sound card. (not so much work at the moment)
Calibration is just for the amplitude scaling of the graph (voltage) anyway, right? I never do that, because as soon as one volume slider (input or output) is touched, the calibration has to be redone anyway.
I usually check the output voltage to know the power level, and see how low the harmonics are compared to the test tone.
No, calibrating ARTA makes it linear, reduces distortion and noise to the minimum your sound card can give. You do not even need a special one, even a 25 Euro Sound Blaster can give you 112db an 0.0001% distortion.
If you do not calibrate it and use it as installed you measure what we see...
Maybe read the ARTA tutorial?
If you do not calibrate it and use it as installed you measure what we see...
Maybe read the ARTA tutorial?
Ok, I guess you're right. There could be some cancellation giving a better result with a differential measurement.
Diodes: I did not see any actual schematic, but as I mentioned before, the pwm is really clean, and that is without load. Only a small overshoot at the edge, and after that absolutely flat. No ringing with high spikes. I think I looked at some sanwu boards a long time ago, and the ringing was terrible.
I have never had a problem with uncalibrated Arta. The calibration in Arta does not reduce distortion, such procedure is a bit more complicated Digital Distortion Compensation for Measurement Setup 🙂
Calibrated Arta would show the same poor THD. Perhaps the input filter is limiting (not the ADC since the fundamental is below 0dBFS)?
Calibrated Arta would show the same poor THD. Perhaps the input filter is limiting (not the ADC since the fundamental is below 0dBFS)?
You measure with a notebook, run it on battery and use your USB soundcard single-ended inputs hooked to speaker outputs.
OK, it is just a precaution...
Sorry @phofman,
but you are telling something that is absolutely wrong.
If you do not match ARTA to the sound card, you can get distortion of any magnitude, 100% are no problem. You can only know it if you test the sound card and adjust levels.
Just installing and measuring will never give a good result.
Did you ever use ARTA?
but you are telling something that is absolutely wrong.
If you do not match ARTA to the sound card, you can get distortion of any magnitude, 100% are no problem. You can only know it if you test the sound card and adjust levels.
Just installing and measuring will never give a good result.
Did you ever use ARTA?
I use Arta quite often (but REW more). Calibration is for levels shown (and frequency response etc., but that is not discussed here). Distortion has nothing to do with calibration, it is created by the soundcard hardware, not by the software analysing the output samples.
Measuring in dBFS scale requires no calibration, it is just ratio of the max. possible sample level. Other voltage-related scales would be incorrect without calibration, but still the THD (i.e. distortions ratio to the fundamental) would be the same, calibration is just a scaling factor in the software.
Measuring in dBFS scale requires no calibration, it is just ratio of the max. possible sample level. Other voltage-related scales would be incorrect without calibration, but still the THD (i.e. distortions ratio to the fundamental) would be the same, calibration is just a scaling factor in the software.
It's all packed away now, so maybe I will get to measure it later. I think I will skip the diodes, since this will be permanently installed with whatever speakers I use.
When it comes to the calibration, I'm agree with your reasoning. I see the same tendencies on the scope FFT (but resolution is not as good) as in Arta, and the scope is as calibrated as it gets.
Maybe there is a point to do some calibration for the % number reading in the bottom of the Arta window, and one could also do a loopback on the soundcard and subtract the sound card distortion from the measurement, but the loopback is relatively clean on the sound card, compared to the amps I usually measure.
Last edited:
If you do not match the levels of your sound card you will not get the best signal noise ratio or produce distortion. If you do not know that, you have never used ARTA. Any sound card has a sweet spot, where distortion and signal noise level have an optimum. If you do not adjust this for your sound card, the performance will be worse than the object you measure, which means useless.
Have a look in the ARTA tutorial, where you find out about the quality of your sound card.
REW has just the same, you have to match levels. This is not done automatically, you have to adjust in the audio mixer. Any time you change levels, you have to repeat this.
Have a look in the ARTA tutorial, where you find out about the quality of your sound card.
REW has just the same, you have to match levels. This is not done automatically, you have to adjust in the audio mixer. Any time you change levels, you have to repeat this.
I am sorry but I have been working on soundcard distortion compensation for over a year and I do know how this stuff works. Please go through that thread I linked. You will see lots of arta measurements (before I moved to RWE). Much better figures than posted here of that TPA. Of course never calibrated, no reason for relative measurements.
It is true that every soundcard has a "sweet spot" but the improvement is only minor. You can measure properly up to slightly below the limitation level of the input (you will see the limitation clearly on the spectrum). Distortion of that TPA amp is way above internal distortion of any decent soundcard at any input level up to limitation. I have measured my TPA3116 a few times before.
Calibration just establishes relation between sample values and real output/input voltage shown in the chart axes/some absolute-level calculations. THD is relative, just like the chart itself.
Of course if the input signal exceeds the input range, the limitation distorts the spectrum heavily. But to avoid that you do not need any calibration, limitation is distinctly visible and easy to avoid by lowering the level until the "harmonics skirt" drops.
No need to discus the calibration anymore in this thread, it is really off topic here.
It is true that every soundcard has a "sweet spot" but the improvement is only minor. You can measure properly up to slightly below the limitation level of the input (you will see the limitation clearly on the spectrum). Distortion of that TPA amp is way above internal distortion of any decent soundcard at any input level up to limitation. I have measured my TPA3116 a few times before.
Calibration just establishes relation between sample values and real output/input voltage shown in the chart axes/some absolute-level calculations. THD is relative, just like the chart itself.
Of course if the input signal exceeds the input range, the limitation distorts the spectrum heavily. But to avoid that you do not need any calibration, limitation is distinctly visible and easy to avoid by lowering the level until the "harmonics skirt" drops.
No need to discus the calibration anymore in this thread, it is really off topic here.
Ok, let's move on. When I connected it to have a listen, I had hum.
Never had a hum problem with my home setup before, and never ran balanced. Now I connected it with the balanced inputs to the unbalanced sound card, and there was a loud hum. If I connected the input - side (shield on the rca from sound card) to the supply ground, the hum went away. I seem to remember reading something like this before, so I guess that's what I need to do to get rid of the hum.
Never had a hum problem with my home setup before, and never ran balanced. Now I connected it with the balanced inputs to the unbalanced sound card, and there was a loud hum. If I connected the input - side (shield on the rca from sound card) to the supply ground, the hum went away. I seem to remember reading something like this before, so I guess that's what I need to do to get rid of the hum.