I'm not sure if following issue was discussed here, but I have following observation:
The distortion components of the 1kHz -45dB signal sampled at 48kHz played from SPDIF depends on previous operation of SW volume control.
I'm using Toslink -> Buffalo3 -> IVY I/V unbalanced -> +40dB amplifier with OPA627 -> ESI Juli@ soundcard.
The distortion components (under -100dBV) changes when volume is decreased and then returned back to the maximum. They does not changes when output of DAC is disconnected and connected back to the signal chain.
Can someone confirm this?
Here are two measurements with difference that only digital volume was decreased and then returned back. The y axis is calibrated.
The distortion components of the 1kHz -45dB signal sampled at 48kHz played from SPDIF depends on previous operation of SW volume control.
I'm using Toslink -> Buffalo3 -> IVY I/V unbalanced -> +40dB amplifier with OPA627 -> ESI Juli@ soundcard.
The distortion components (under -100dBV) changes when volume is decreased and then returned back to the maximum. They does not changes when output of DAC is disconnected and connected back to the signal chain.
Can someone confirm this?
Here are two measurements with difference that only digital volume was decreased and then returned back. The y axis is calibrated.
Attachments
More info: mine ES9018 changes harmonics distortion pattern on any change of any volume register (0..7) to different value via I2C.
These harmonics varies in the range -105dBV to -130dBV.
The pattern does not change with DAC reset and neither on short-term power down.
Harmonics pattern does not change on stop/start playback and also does not change on volume change in the signal.
It changes only when value of volume register is changed to different value.
Following command loop produces dynamic harmonic spectrum on sin 1kHz -45dB:
- V0=1; 100ms delay; V0=0; 100ms delay
- V1=1; 100ms delay; V1=0; 100ms delay
- V2=1; 100ms delay; V2=0; 100ms delay
- V3=1; 100ms delay; V3=0; 100ms delay
- V4=1; 100ms delay; V4=0; 100ms delay
- V5=1; 100ms delay; V5=0; 100ms delay
- V6=1; 100ms delay; V6=0; 100ms delay
- V7=1; 100ms delay; V7=0; 100ms delay
- repeat
These harmonics varies in the range -105dBV to -130dBV.
The pattern does not change with DAC reset and neither on short-term power down.
Harmonics pattern does not change on stop/start playback and also does not change on volume change in the signal.
It changes only when value of volume register is changed to different value.
Following command loop produces dynamic harmonic spectrum on sin 1kHz -45dB:
- V0=1; 100ms delay; V0=0; 100ms delay
- V1=1; 100ms delay; V1=0; 100ms delay
- V2=1; 100ms delay; V2=0; 100ms delay
- V3=1; 100ms delay; V3=0; 100ms delay
- V4=1; 100ms delay; V4=0; 100ms delay
- V5=1; 100ms delay; V5=0; 100ms delay
- V6=1; 100ms delay; V6=0; 100ms delay
- V7=1; 100ms delay; V7=0; 100ms delay
- repeat
Random harmonics pattern clarification and visualization:
My setup: Buffalo 3 with ES9018S C500, 100MHz clock, 8ch I2S input, chip default register settings, sinus 1kHz -45dB.
The harmonics are different on each start: http://i.imgur.com/CAzdfTl.gif
The harmonics change after alteration of volume registers: http://i.imgur.com/X8QercG.gif
My setup: Buffalo 3 with ES9018S C500, 100MHz clock, 8ch I2S input, chip default register settings, sinus 1kHz -45dB.
The harmonics are different on each start: http://i.imgur.com/CAzdfTl.gif
The harmonics change after alteration of volume registers: http://i.imgur.com/X8QercG.gif
Seems like this issue of varying distortion patterns of harmonics is not related to ES9018 but to the OPA1632, because:
- it happens also on balanced outputs of IVY
- it happens also with different (non-shunt) power supply for IVY
- it does not happen with different I/V stage (discrete Legato from Twisted Pear Audio)
With the Legato there are no harmonics at all with -45dB signal level and -150dBV depth of FFT!
- it happens also on balanced outputs of IVY
- it happens also with different (non-shunt) power supply for IVY
- it does not happen with different I/V stage (discrete Legato from Twisted Pear Audio)
With the Legato there are no harmonics at all with -45dB signal level and -150dBV depth of FFT!
The FFT from the IVY seem to vary a lot even when you don't change the volume...
Has it got the capacitors in the OPA1632 feedback ? Can you post your schematics ? Bonus for scope shot of output waveform of first opamp before passive filter, at 20mV/div, 100ns..1µs/div timebase, playing low amplitude sinewave. Look for staircases and slew rate.
Has it got the capacitors in the OPA1632 feedback ? Can you post your schematics ? Bonus for scope shot of output waveform of first opamp before passive filter, at 20mV/div, 100ns..1µs/div timebase, playing low amplitude sinewave. Look for staircases and slew rate.
There are no capacitors in the feedback, only 750 Ohm resistor in each polarity - http://www.twistedpearaudio.com/docs/linestages/ivy3_schematic_1_0.pdf
I can try to create the "bonus" later.
I can try to create the "bonus" later.
Once you change the volume, you should change input signal at the same time.
For example, input -45dBV, volume 0dB. You get result 1.
Input -44dBV, volume -1dB, you get result 2.
Then, compare the result 1 and 2. To see whether the row result is same or not.
Otherwise, you can not get the same test condition when you are comparing.
BR
Paul
The purpose of +40dB amplifier is to amplify the noise bottom of the DAC which is normally immersed under and masked by the sound card noise. It is possible to design a decent amplifier for these purposes, OPA627 is not a bad candidate, AD797 would be even better. I use the same method to measure S/N of the link level preamplifiers.
Yikes... that's a lot of gain!
You are testing your 40db amplifier as much as anything else in the rig.
Also you should read back all the chip registers and make sure you have the DAC truly configured as you expect.
1) "The effect" is there with +40dB amplifier, but it is also present without it (with less details).
2) Ouput with other I/V is OK.
3) DAC registers has chip default values ... only volume registers (0-7) are changed from value 0 to 1 and back (i.e. 0.5dB)
Because it has been found not directly related to ES9018, I'm going to continue with my investigation in the TPA support forum and I'll post short summary here on DIYA after the issue will be resolved.
Varying distortion patterns of IVY III - IVY I/V Stage - Twisted Pear Audio Support
Miero, it seems like the noise you see doesn't depend on changing the volume control value or not, as the initial measurements all seem to have quite a bit of noise/distortion already... can you confirm this ?
Can you post scope traces ?
Bonus : Here is FFT of ES9018 output, from Rigol DS2000 300MHz scope sampling at 2Gsps. As you can see, MCLK is feeding through (as expected), nice HF spikes up to the scope bandwidth limit... some passive filtering might not be such a bad idea after all.
Can you post scope traces ?
Bonus : Here is FFT of ES9018 output, from Rigol DS2000 300MHz scope sampling at 2Gsps. As you can see, MCLK is feeding through (as expected), nice HF spikes up to the scope bandwidth limit... some passive filtering might not be such a bad idea after all.
Attachments
Last edited:
That's the "magic" of the digital sources. Most people concentrate their efforts on audio band distortion, but in fact it is the HF interference that makes a difference in sound. Takes a long time to get enough experience to evaluate what is important and what is unimportant
.What's worse, the design of instrument boxes regarding shielding and wiring that guys use for their DACs is usually horrible.