Sallen & Key 2-Order unity gain filter distortion and noise
- Thread starter Tompson17
- Start date
Help Support GroupDIY Audio Forum:
Active circuitry WILL introduce noise. Comparing the spectra between the "nothing connected" and the "bad channel", i see a rise of about 20dB or so (from under -120dBFS to a bit under -100dBFS).
The dynamic range of CD's, with their 16 bits, is 96dB; 24bit audio will net you, in practice, about 120dB of dynamic range.
Have you tried switching your interface to 96kHz, so you can fit in more harmonics?
PS: If you're worrying about noise in line-level signals under -100dBFS, you probably won't even want to try measuring the noise on the outputs of the power amps ;D
And then there's the perception side of things - https://en.wikipedia.org/wiki/Auditory_masking
The dynamic range of CD's, with their 16 bits, is 96dB; 24bit audio will net you, in practice, about 120dB of dynamic range.
Have you tried switching your interface to 96kHz, so you can fit in more harmonics?
PS: If you're worrying about noise in line-level signals under -100dBFS, you probably won't even want to try measuring the noise on the outputs of the power amps ;D
And then there's the perception side of things - https://en.wikipedia.org/wiki/Auditory_masking
The measuring system is 196kHz, but I have "full range" channels and high pass channels from 2,5kHz cutted both on 15 kHz, so the next harmonics are also attenuated.
btw, the measuring system gents self noise on -130dBu and dynamic range is up to 40 dBu, so I cannot say I am measuring with dBFS. And, for all, previous stages (like the input + attenuator) produces very low noise and little distortion. The two signal followers in attenuator produces wery low noise, but the followers used in Sallen Key filters "do", so the final noise is filtered, but why the noise is produced, dont know... one signal follower can rise the noise about 1 dB or what I know, with noise level at -110 dBu, but this time, it is more than 20 dB worse...
btw, the measuring system gents self noise on -130dBu and dynamic range is up to 40 dBu, so I cannot say I am measuring with dBFS. And, for all, previous stages (like the input + attenuator) produces very low noise and little distortion. The two signal followers in attenuator produces wery low noise, but the followers used in Sallen Key filters "do", so the final noise is filtered, but why the noise is produced, dont know... one signal follower can rise the noise about 1 dB or what I know, with noise level at -110 dBu, but this time, it is more than 20 dB worse...
Like i said - the power amps will be a LOT noisier than that, anyway (and let's not even get into the over-1% distortion of speaker drivers themselves, at any meaningful SPL)... 
I mean, i assume that IS the end-goal for this circuit, right?
I mean, i assume that IS the end-goal for this circuit, right?
Yes, power amplifiers are many times worse, I know, but there is some dynamic range and I know most of gear powered with +/-15V can work well with +26 dBu with inner level +22dBu, or +20, doesn't matter, but I have THD + N ratio about 70 dB in worst case with input signal +20dBu and inner signal +16 dBu...
I know it cannot make sense, but I need full dynamic range for full power amplifier excitation
I know it cannot make sense, but I need full dynamic range for full power amplifier excitation
I have simulated the currents from all op amp outputs and in every stage from every filter I get about 1,5mA max. Douglas self use in his crossovers resistors like 1kOhm and smaller (with 5532's) and get nice results.
What I tried now is to put 100R resistor between IC11D and the following cap and the distortion is still the same.
What I tried now is to put 100R resistor between IC11D and the following cap and the distortion is still the same.
Interesting... if I measure the distortion on op amp output, I see the distortion, but when I measure after 100R resistor, the distortion is much lower... ??? don't understand it...
Could be the direct connection between followers the problem source?
I have some experience with one design from somebody, and he put 100nF caps between op amp outs and ground. It was about 5 caps somewhere in the design (realisation, the mistake was 100nF instead of 100pF). Interesting was the white noise which I saw on the screen of my analyzer. Another interesting thing is that the noise was measurable right after input amplifier, but the first cap was after 4 or 5 op amps... I remember it looked like white noise, but the noise was of course permanent regardless of the presence of the signal.
This case seems to be quite similar in the principle of risen noise (but the significantly noise is atfer 2,5k - 15kHz filter, distortion is "anywhere"). On the other side, the noise is rising olny when the signal is just say "high", so it can be like op amp current limiting. But where?... I dont see any trouble on the PCB, it is fabricated with solder mask, no resistance is too low to produce current limiting of an op amp. And the smallest value of resistance is 820R which is on the input block and here is the many called distortion not measurable...
Could be the direct connection between followers the problem source?
I have some experience with one design from somebody, and he put 100nF caps between op amp outs and ground. It was about 5 caps somewhere in the design (realisation, the mistake was 100nF instead of 100pF). Interesting was the white noise which I saw on the screen of my analyzer. Another interesting thing is that the noise was measurable right after input amplifier, but the first cap was after 4 or 5 op amps... I remember it looked like white noise, but the noise was of course permanent regardless of the presence of the signal.
This case seems to be quite similar in the principle of risen noise (but the significantly noise is atfer 2,5k - 15kHz filter, distortion is "anywhere"). On the other side, the noise is rising olny when the signal is just say "high", so it can be like op amp current limiting. But where?... I dont see any trouble on the PCB, it is fabricated with solder mask, no resistance is too low to produce current limiting of an op amp. And the smallest value of resistance is 820R which is on the input block and here is the many called distortion not measurable...
Similar threads
