Looking at the filter table - I like the "theorem" and thus, a sinc filter would be my choice, with warts (ringing) and all... 🙂
My wish for Linear Phase filter would read:
Pass band: -1 dB at 0,45Fs (20k@44,1)
Stop band: -140 at 0,4999Fs
GD: dont care
Flatness: 0,01 dB OK
It's there a possibility for user defined filters at all in the chip?
IVX - can you provide such a configuration?
Or, could you provide means for us to load our own filters?
//
My wish for Linear Phase filter would read:
Pass band: -1 dB at 0,45Fs (20k@44,1)
Stop band: -140 at 0,4999Fs
GD: dont care
Flatness: 0,01 dB OK
It's there a possibility for user defined filters at all in the chip?
IVX - can you provide such a configuration?
Or, could you provide means for us to load our own filters?
//
ES9822 is actually ES9842 chip internally tied to pairs 1-3 and 2-4. In the current FW I've not implemented a custom FIR.
Thanks! Is 1-3 = Left or Right as per the front text?
"Current FW"... I read that as not entirely impossible in the future ;-D
After all, it's Christmas time and one may wish for whatever one want 🙂
I did some scrutiny of the filter plots... here its looks like that the Lin Phase FRO breaks the fs/2 theorem rule already at say -5 dB but the APO stays below down to say - 60 dB... but not so hirez pdf so...
I will use the APO for the time being.... It was set as default if I remember correctly.
//
"Current FW"... I read that as not entirely impossible in the future ;-D
After all, it's Christmas time and one may wish for whatever one want 🙂
I did some scrutiny of the filter plots... here its looks like that the Lin Phase FRO breaks the fs/2 theorem rule already at say -5 dB but the APO stays below down to say - 60 dB... but not so hirez pdf so...
I will use the APO for the time being.... It was set as default if I remember correctly.
//
Why is this important... check:
Nyquist, aliasing and folding in Labview.mp4YouTube · Angstrom Designs6 minuter, 33 sekunder4 feb. 2013
What this video misses out is the fact that the steepness of the filter is very important - he doesn't say what filter is used but I guess its not very steep letting > fs/2 into the sampling process. Beacause if it was swept enough, fs/2 would be enough - now he needs fs*10 to get the distortion out. And, is his problem in the AD or in the DA or in both? But the concept is clear - folding is evil and we get back frequencies from 'fs/2 -> fs' as transposed into the base band - ugly!!!
And once it is in the data from the AD process, it can never be washed out as it is now in-printed in the base band (0-fs/2, i.e. the music).
This is why we shall not try to get every single Hz nearby fs/2 but see to that noting above fs/2 slips through.
//
Nyquist, aliasing and folding in Labview.mp4YouTube · Angstrom Designs6 minuter, 33 sekunder4 feb. 2013
What this video misses out is the fact that the steepness of the filter is very important - he doesn't say what filter is used but I guess its not very steep letting > fs/2 into the sampling process. Beacause if it was swept enough, fs/2 would be enough - now he needs fs*10 to get the distortion out. And, is his problem in the AD or in the DA or in both? But the concept is clear - folding is evil and we get back frequencies from 'fs/2 -> fs' as transposed into the base band - ugly!!!
And once it is in the data from the AD process, it can never be washed out as it is now in-printed in the base band (0-fs/2, i.e. the music).
This is why we shall not try to get every single Hz nearby fs/2 but see to that noting above fs/2 slips through.
//
Last edited:
Right on target:
At 11:00 is the reason why I asked if the filters could be run on only a 1/4 of the Fs (Fs = 384 but filters run at 96 etc)
//
At 11:00 is the reason why I asked if the filters could be run on only a 1/4 of the Fs (Fs = 384 but filters run at 96 etc)
//
the problem is CT7601 source code seems to mixed up these two sampling frequencies 352 and 705.
Hi IVX,
I had a look at your circuit. It looks like a modified Sallen Key notch, or 2T structure. What is the advantage of your circuit against the normal one?
Thanks
Alfred
I had a look at your circuit. It looks like a modified Sallen Key notch, or 2T structure. What is the advantage of your circuit against the normal one?
Thanks
Alfred
Guys, I have one question about Win11+Cosmos ADC issue. The user just reported that his new HP laptop with Win11 records ONLY the left channel from Cosmos ADC, and sends that to the right channel as well, like a mono ADC interface. Again, it is not about my MONO mode(i.e. not about internal ADC regime), the user can switch that by Windows volume slider and see the LEDs become green, so the problem isn't about that, but seems about UAC2 implementation in Win11, at least in that particular build. Another W11 PC of the same user works ok with the same ADC. Please leave a short text about your experience with W11 if any. Thank you.
the out-of-band noise hump is reduced, just by the way.
I think you guys remember that hump if used 384k of sampling rate, it starts from 100k and higher so that hump is reduced. Furthermore, 10-20k H3 is also dropped to 3db or so. A-weighted noise may be .1db improved, so the new FW is about out-of-band noise mostly.
I think you guys remember that hump if used 384k of sampling rate, it starts from 100k and higher so that hump is reduced. Furthermore, 10-20k H3 is also dropped to 3db or so. A-weighted noise may be .1db improved, so the new FW is about out-of-band noise mostly.
Some users report 1db(A) less noise with the new FW, I tested two samples one had .1db(A) and another one .5db(A) of the improvement. In other words, the new FW may lift the grade of your ADC to the next level(if the improvement is .5-1db that's for sure) 😱
Nice. If you could fix the FW so that it would be compatible with the majority of the USB isolators, it would be good.
//
//