I recently purchased a non-commerical version of WinAudioMLS. It is a very powerful tool. In order to support this powerful tool, a decent audio interface is needed. We can buy off-the-shelf USB audio interface supporting 24-bit, 192 kHz. Not all of them give you best analog performance.
Here is my plan: USBStreamer from miniDSP + home made high-performance analog front. USBStreamer is based on XMOS USB audio solution. It supports 24-bit, 192 kHz streaming. Driver of USBStreamer supports ASIO. High-performance 24-bit, 192 kHz ADCs are available from Digikey, Mouser, etc. We can add opamp signal conditioning circuit and attenuator to have a complete analog front end. Currently, I using WM8788 for evaluation. Here is WinAudioMLS and USBStreamer running at 96 kHz sample frequency. Input of WM8788 is connected to a 1 kHz low-distortion signal source.
FFT shows harmonics of 60 Hz mains. 8 kHz and 16 kHz are external interference.
Here is my plan: USBStreamer from miniDSP + home made high-performance analog front. USBStreamer is based on XMOS USB audio solution. It supports 24-bit, 192 kHz streaming. Driver of USBStreamer supports ASIO. High-performance 24-bit, 192 kHz ADCs are available from Digikey, Mouser, etc. We can add opamp signal conditioning circuit and attenuator to have a complete analog front end. Currently, I using WM8788 for evaluation. Here is WinAudioMLS and USBStreamer running at 96 kHz sample frequency. Input of WM8788 is connected to a 1 kHz low-distortion signal source.
FFT shows harmonics of 60 Hz mains. 8 kHz and 16 kHz are external interference.
Attachments
Here is THD of loopback.
Right figure: default setting, THD+N = 0.0077%, THD = 0.0027%
Left figure: 64-bit FFT, THD+N = 0.0008 %, THD = 0.0000 %
Attachments
Last edited:
The free one seems to have poor performance as shown in the figure. It has much higher noise floor and frequency offset.
Frequency was set to 777 Hz.
Attachments
Last edited:
Windows function affects THD result
The choice of FFT window function affects the THD result.
Left: THD+N is -105 dB for Blackman window.
Right: THD+N is -144 dB for Rife-Vincent window.
Both are with 64-bit FFT enabled and looback.
The choice of FFT window function affects the THD result.
Left: THD+N is -105 dB for Blackman window.
Right: THD+N is -144 dB for Rife-Vincent window.
Both are with 64-bit FFT enabled and looback.
Attachments
This is Extended 'Non-Commercial'. The results shown are digital I2S loopback on the miniDSP USBStreamer (USB to I2S, SPDIF board).
tanks for posting this I have TrueRTA and found it wanting
I am using Audiophile 192 sound card (internal) and loop back suffers from peack at 50 Hz Floor level is about 70dB
Most probably my mistake I shuld have used an externall one.
So I will have to think 2 times before trowing more money on what then wuld not be more useful than toy.
Can you post more details for the interface please?
I am using Audiophile 192 sound card (internal) and loop back suffers from peack at 50 Hz Floor level is about 70dB
Most probably my mistake I shuld have used an externall one.
So I will have to think 2 times before trowing more money on what then wuld not be more useful than toy.
Can you post more details for the interface please?
The loopback test simply had the USBStreamer (miniDSP) connected to a USB port of my notebook. The I2S data 1&2 OUT were connected to I2S data 1&2 IN. ASIO was used instead of Windows driver. There was no analog-digital conversion in the path.
I have a Roland US-25 which does not good at all for measurement purposes. I wish to use WinAudioMLS, USBStreamer and a custom analog front-end board building a complete high-performance audio analyzer.
Last edited:
I made a test to evaluate loopback (USBStreamer data out pin connected to data in pin) FFT for using two different connecting modes: directsound (DS) or ASIO. For DS, WinAudioMLS signal generator, Audacity and foobar were used. Foobar output the 1 kHz sine generated by Audacity. System is Win 7, 64-bit Pro. FFT plot was obtained by WinAudioMLS with 64-bit FFT enabled.
Did I make a proper test for intended purpose? What is the root cause of sidebands associated with DS connection?
Did I make a proper test for intended purpose? What is the root cause of sidebands associated with DS connection?
Attachments
WM8740 DAC board measured data
Here are measured data for WM8740 DAC board. The differential output of the DAC chip was directly connected to AP S1 input since the output filter/buffer has not been populated. AP S1 analog generator and digital generator (D/A to output amplifier) were also measured to set reference levels.
I2S input of the DAC board was from USBStreamer. WinAudioMLS (with ASIO connection selected) was used to provide 1 kHz tone.
Analog generator provides the best output. It can be clear seen from the bandreject FFT plots. WM8740+USBStreamer+WinAudioMLS seems to be as good as S1 digital generator+D/A+analog output. In fact, WM8740 combination has fewer high order harmonics.
Here are measured data for WM8740 DAC board. The differential output of the DAC chip was directly connected to AP S1 input since the output filter/buffer has not been populated. AP S1 analog generator and digital generator (D/A to output amplifier) were also measured to set reference levels.
I2S input of the DAC board was from USBStreamer. WinAudioMLS (with ASIO connection selected) was used to provide 1 kHz tone.
Analog generator provides the best output. It can be clear seen from the bandreject FFT plots. WM8740+USBStreamer+WinAudioMLS seems to be as good as S1 digital generator+D/A+analog output. In fact, WM8740 combination has fewer high order harmonics.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.