Blumlein 88
Grand Contributor
- Joined
- Feb 23, 2016
- Messages
- 22,972
- Likes
- 42,653
Had some time to look at the linearity of a couple of my devices. Was surprised at how well the devices measured out. No reason to post full results. Both were within +/- .02 db FS down to -120 db. I made some signals that go the full distance beyond just the top 20 bits. One DAC was a bit more up and down with at least 1 db swings for the bits below -120 db. The other was even better.
Here is a screen shot of the LSB. Theoretically it would be -138.47 db. Here there is a .07 db offset due to gain settings on the ADC. So it shows -138.74 which is actually -138. 67. Of course at those levels the noise of the gear cause some fluctuations of a .5 db or sometimes more. This is more or less the average over 10 seconds.
I used a 12 khz sine wave for the test signal. So the black cursor at 12 khz is what the readings on the side reference. 128 K FFT. The DAC is my Focusrite Forte being measured by an Antelope Audio Zen Tour. The Zen Tour DAC was also very good though having larger fluctuations below the 20th bit.
Here is the same Forte showing the 20th bit.
So here are the correct levels and the average levels for my Forte.
.......................Forte
120.41......... 120.48
126.43..........126.57
132.45..........132.67
138.47..........138.67
Now these lower levels fluctuate around and this is my eyeball averaging. I should probably run them thru REW and let it do a real longer term average. But I was surprised these are so close you basically couldn't tell it from perfect for the noise. It appears to indicate excellent linearity all the way to the Least Bit though not low enough noise one should call it of 24 bit resolution. The gear might qualify for 19 bit resolution looking at noise levels. Of course this is what sigma delta DACs are able to provide. Good luck getting multi-bit R2R ladder DACs to approach something like this.
I'm not suggesting any need to regularly measure below -120db. For one thing the noise floors corrupt exact values below that point. The measurements of both DACs were very steady at -120db and above. Below that noise increasingly had some influence. This was just a lark to satisfy my own curiosity.
Here is a screen shot of the LSB. Theoretically it would be -138.47 db. Here there is a .07 db offset due to gain settings on the ADC. So it shows -138.74 which is actually -138. 67. Of course at those levels the noise of the gear cause some fluctuations of a .5 db or sometimes more. This is more or less the average over 10 seconds.
I used a 12 khz sine wave for the test signal. So the black cursor at 12 khz is what the readings on the side reference. 128 K FFT. The DAC is my Focusrite Forte being measured by an Antelope Audio Zen Tour. The Zen Tour DAC was also very good though having larger fluctuations below the 20th bit.
Here is the same Forte showing the 20th bit.
So here are the correct levels and the average levels for my Forte.
.......................Forte
120.41......... 120.48
126.43..........126.57
132.45..........132.67
138.47..........138.67
Now these lower levels fluctuate around and this is my eyeball averaging. I should probably run them thru REW and let it do a real longer term average. But I was surprised these are so close you basically couldn't tell it from perfect for the noise. It appears to indicate excellent linearity all the way to the Least Bit though not low enough noise one should call it of 24 bit resolution. The gear might qualify for 19 bit resolution looking at noise levels. Of course this is what sigma delta DACs are able to provide. Good luck getting multi-bit R2R ladder DACs to approach something like this.
I'm not suggesting any need to regularly measure below -120db. For one thing the noise floors corrupt exact values below that point. The measurements of both DACs were very steady at -120db and above. Below that noise increasingly had some influence. This was just a lark to satisfy my own curiosity.
Last edited:
It is not hard or impossible to do very well here.
