So it looks at least in my case tha aliasing was ot from the DAC at all, as all DACs I have tested show the same aliased signal.
I confirmed this be chnging the sample rate and I notice that the alias changes, sometimes producing two peaks.
So this seems to be something to do with the sound card or the software at least in my case.
Here is an example:
Green is buf/ivy Red is Apogee.
Sample rate is 81920khz
I confirmed this be chnging the sample rate and I notice that the alias changes, sometimes producing two peaks.
So this seems to be something to do with the sound card or the software at least in my case.
Here is an example:
Green is buf/ivy Red is Apogee.
Sample rate is 81920khz
Attachments
Aliasing
The sampling rate of the sound card does have an effect, to some degree. If the input spectrum contains components above half the sampling frequency, then these components are folded down into baseband. If the sampling frequency changes, then the position of the folded residuals changes too. When sampling at 96 kHz, this affects only components located above 48 kHz, where you would normally only expect negligible content due to the filtering. See also the attached FFT of the HD 760 output, measured with the same sound card and same sampling rate. No aliasing here.
In your FFTs, it seems to me that the sum of the fundamental and the aliasing component adds up to 96 kHz, which could have something to do with upsampling. In my case (sampling the sound card at the same 96 kHz as you), it adds up to 88.2 (=2x44.1), which would indicate no upsampling. I don't know if your SPDIF source is already upsampled or if this takes place in the sabre chip.
The sampling rate of the sound card does have an effect, to some degree. If the input spectrum contains components above half the sampling frequency, then these components are folded down into baseband. If the sampling frequency changes, then the position of the folded residuals changes too. When sampling at 96 kHz, this affects only components located above 48 kHz, where you would normally only expect negligible content due to the filtering. See also the attached FFT of the HD 760 output, measured with the same sound card and same sampling rate. No aliasing here.
In your FFTs, it seems to me that the sum of the fundamental and the aliasing component adds up to 96 kHz, which could have something to do with upsampling. In my case (sampling the sound card at the same 96 kHz as you), it adds up to 88.2 (=2x44.1), which would indicate no upsampling. I don't know if your SPDIF source is already upsampled or if this takes place in the sabre chip.
Attachments
Re: Aliasing
No, there was no upsampling involved.
I think my ADC is the source of the aliasing. As it shows up on anything I measure.
I did not user SPDIF into my Buffalo, I used I2S. Specially SPDIF into MUX then I2S to buffalo.
I will try later with SPDIF directly into buffalo..
Cheers!
Russ
Javin5 said:
No, there was no upsampling involved.
I think my ADC is the source of the aliasing. As it shows up on anything I measure.
I did not user SPDIF into my Buffalo, I used I2S. Specially SPDIF into MUX then I2S to buffalo.
I will try later with SPDIF directly into buffalo..
Cheers!
Russ
Aliasing
Hi Russ
Thanks for your replies. I really start to feel bad about consuming so much of your time with this problem.
Obviously the amount of alising supression depends on the filtering. I tried to change rolloff with the dip-switch, but nothing changed. I then tried to also recycle power; still no change. On my Buffalo board version 1.1, positions 2 and 3 of the dip switch are missing on one side. Should I still be able to change filter rolloffs?
Hi Russ
Thanks for your replies. I really start to feel bad about consuming so much of your time with this problem.
Obviously the amount of alising supression depends on the filtering. I tried to change rolloff with the dip-switch, but nothing changed. I then tried to also recycle power; still no change. On my Buffalo board version 1.1, positions 2 and 3 of the dip switch are missing on one side. Should I still be able to change filter rolloffs?
Re: Aliasing
If you pull the pads which are adjacent to the switches (on the microcontroller side) up or down it will have the same effect as setting the switch. Since you have a 1.1 board, it is possible you need to pull position 3 low.
There were a small set of boards that had firmware to use 9bit quantization. For some reason it did not work correctly. If you have such a module then setting switch 3 low will solve that issue.
Cheers!
Russ
Javin5 said:
If you pull the pads which are adjacent to the switches (on the microcontroller side) up or down it will have the same effect as setting the switch. Since you have a 1.1 board, it is possible you need to pull position 3 low.
There were a small set of boards that had firmware to use 9bit quantization. For some reason it did not work correctly. If you have such a module then setting switch 3 low will solve that issue.
Cheers!
Russ
I now have MUX into Buffalo (via I2S and using Volumite) into IVY all working well. Thanks Russ and Brian.
The dip switches on the MUX are all set in mid position. Is this right? I am only inputting spdif from a cd player and from Squeezebox at present.
There is one little glitch, however, - every now and again one channel goes very loud for a fraction of a second. Ideas very welcome.
I have tried various configurations with different power amps. Here's my findings, for what they're worth, which only go to show that I'm a firm believer in preamps.
1) DAC (balanced out) straight into PA03 balanced power amp.
2) DAC (SE out) straight into Symasym 5.3 SE monoblocks.
3) DAC into DNM 3C preamp into Symasym 5.3 SE monoblocks - all SE.
1 probably has the most detail but 3 is by far the best. The Symasyms are very quick power amps. The DNM is also a quality piece of kit. This combination really kicks *** - very powerful and dynamic - so much more enjoyable to listen to than 1 and 2 that there's no contest. Unfortunately the PA03 is the only balanced power amp I have to try.
The dip switches on the MUX are all set in mid position. Is this right? I am only inputting spdif from a cd player and from Squeezebox at present.
There is one little glitch, however, - every now and again one channel goes very loud for a fraction of a second. Ideas very welcome.
I have tried various configurations with different power amps. Here's my findings, for what they're worth, which only go to show that I'm a firm believer in preamps.
1) DAC (balanced out) straight into PA03 balanced power amp.
2) DAC (SE out) straight into Symasym 5.3 SE monoblocks.
3) DAC into DNM 3C preamp into Symasym 5.3 SE monoblocks - all SE.
1 probably has the most detail but 3 is by far the best. The Symasyms are very quick power amps. The DNM is also a quality piece of kit. This combination really kicks *** - very powerful and dynamic - so much more enjoyable to listen to than 1 and 2 that there's no contest. Unfortunately the PA03 is the only balanced power amp I have to try.
FFT with 19.6khz input -8db
FWIW I got pretty good results out of the buffalo/IVY at 19600hz.
I did have some aliasing from CODEC, but I am quite sure that is the sound cards fault as every source I tested showed the same alias.
What is nice is I don't see any hint of IMD.
Here is the FFT:
FWIW I got pretty good results out of the buffalo/IVY at 19600hz.
I did have some aliasing from CODEC, but I am quite sure that is the sound cards fault as every source I tested showed the same alias.
What is nice is I don't see any hint of IMD.
Here is the FFT:
Attachments
Aliasing
This is really simple to verify. Play your 20 kHz tone and check the output of the Buffalo/IVY with a scope. If you see a clean sine wave, you have very little or no aliasing.
Since no ADC or sampling is involved this way, you can be sure that what you see is what you get. Attached is the output I got this way on an old Tektronix scope. The upper trace is the output of the HK player and the lower trace is the output of the Buffalo/IVY. This is in agreement with my FFT's.
This is really simple to verify. Play your 20 kHz tone and check the output of the Buffalo/IVY with a scope. If you see a clean sine wave, you have very little or no aliasing.
Since no ADC or sampling is involved this way, you can be sure that what you see is what you get. Attached is the output I got this way on an old Tektronix scope. The upper trace is the output of the HK player and the lower trace is the output of the Buffalo/IVY. This is in agreement with my FFT's.
Attachments
Re: Aliasing
I believe that you are getting it, I just can't explain exactly why.
I am not seeing the same thing.
What I do know is that with the firmware that had the 9bit quantizer setting we saw the same sort of distortion. Switching it to 6-bit cleared it up.
Cheers!
Russ
Javin5 said:
I believe that you are getting it, I just can't explain exactly why.
I am not seeing the same thing.
What I do know is that with the firmware that had the 9bit quantizer setting we saw the same sort of distortion. Switching it to 6-bit cleared it up.
Cheers!
Russ
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.