I use the dam1021 in dual-mono configuration and listen to the buffered output on HD800 (with a global EQ gain of -4db to boost the bass). The volume control is typically set to -15db or higher, which means about 2.5 unused bits in the ladder. Still have more than 24 bits available (not that ANY earthly technology can use all 24 due to inevitable thermal noise). Only the top few bits (maybe 18?) are 0.01%, and the rest looks like 0.02/0.05, but it's not that much of a waste and the lower bits need exponentially less accuracy theoretically speaking. All of this problem could potentially go away COMPLETELY if we get a resistor calibration tool and software support.
Last edited:
That sounds fair. I guess the channels wouldn't sync perfectly but only to within one input clock cycle as the software is currently implemented, but maybe that's okay. Not a speaker person so idk
Yes... Sadly, I can corroborate that observation in this project and many others... The causes are very interesting. Maybe it's why most people need a hobby in the first place
I'm not bothered about clock power source at all any more. But it still makes me a little sad that the Salas shunt regulator in my build goes through the onboard LDO and another op-amp reg before it reaches the 4v rail. I guess that's why Soren uses SMPS - to avoid unnecessary heartbreak
Just a quick question, what rev are your dam1021, I have changed buffers a few times. Currently I'm using opa1678 on the dam1021, those should be good.
Indeed in my vision all the digital crap stay behind a fiber optic wall and yes, it will be a different thread for sure.
Silly mistake using the opa1602 with 600x input current noise on a high impedance input?
I am planning to use in my next project with the dam 1021 the digital volume control (HQ player) with the raw output and the 6386 LGP (triode with exponential transfer characteristics) for setting/adjusting the gain so that I won't lose bit depth.
I have also experimented with different psu solutions and ended up with this:
https://audioxpress.com/article/shun...tage-regulator
https://www.ebay.com/sch/i.html?_fro...latro&_sacat=0
IMO this is the easiest way to get exceptional SQ without modifying the DAC.
Silly mistake using the opa1602 with 600x input current noise on a high impedance input?
?? The input on the buffers on a dam1021 see 625 ohm, I wouldn't call that high impedance....
Actually 18 bits (28-10), but you're still 2 bits (or 12dB) better than the fabled TDA1541A.
And at -60dB volume most systems would be so low that it wouldn't matter anyway.
Hey don't dance around the issue. :|
The current noise on 1602 totally dominates voltage noise compared to 1678. TI says not to use bipolar input opamp in that case (because of noise).
I'm not an EE person or an op-amp expert so I can forgive someone for choosing the "SoundPlus High Performance Op-Amp" advertised to have "Superior sound quality" over something that's just "low-distortion"... Also if you count the 70% lower THD with 2V RMS, the numbers might actually be in favor of the 1602... It looks negligible to me either way. You probably picked 1678 for different reasons, CMOS, price, whatever...
Last edited:
I don't see any reason to lose even one bit with digital volume control while a good attenuator does the job without losing nothing, but you are free to swim in the digital crap while I prefer to stay away as possible from it.
Anyway I got this DAC just to compare it with the "fabled" Tda1541A and AD1862, my design is not yet ready, so we will see.
Although my main interest is to measure the timing quality of this device.
IMHO an analog attenuator that does lose nothing does not exist, but let's agree to disagree...
I would advise to AB the difference at least... fine if analog control is preferred then. Note that some digital attenuation is inevitable to avoid clipping in the digital filter for oversampling.
I would advise to AB the difference at least... fine if analog control is preferred then. Note that some digital attenuation is inevitable to avoid clipping in the digital filter for oversampling.
It's a question of points of view (or let say sonic result), maybe I'm a nostalgic since I use 1% THD (or more) feedback free tube amplifier while perhaps you run solid state 0.00000....% THD amp.
Although I usually go for NOS (as I said I would avoid any digital crap) I don't understand why the oversampling should cause clipping, you are adding samples not changing the amplitude.
Although I usually go for NOS (as I said I would avoid any digital crap) I don't understand why the oversampling should cause clipping, you are adding samples not changing the amplitude.
The newly added/interpolated fake samples may go past 0dB therefore clipping...
If it's so the interpolation does not work correctly, one more reason to avoid such artifacts.
Please, take a look at the PDM100 datasheet, there is no mention of clipping.
So the software does worse than an old integrated digital filter???
In our FIFO buffer we use dithering to reconstruct the LSBs in case of truncation, but we don't change the amplitude, there is no reason and it's a mistake.
Last edited:
Well, what you are "losing" are a lot of added zeros in most cases.
But you can compare for yourself once you've got the DAC set up.
This is an "alternative fact" as e.g. sampling of a sine wave with f=FS/4 and some phase offset shows something else.
@living sounds
I think you are wrong, you are losing data.
Below 20 dB of attenuation (4 bit) you lose 1 bit for each more 6dB attenuation.
Moreover I hope the LSBs were not truncated so you need dithering to avoid the truncation, that means add noise that is not present in the original signal.
This 1 Euro part performs better
I think you are wrong, you are losing data.
Below 20 dB of attenuation (4 bit) you lose 1 bit for each more 6dB attenuation.
Moreover I hope the LSBs were not truncated so you need dithering to avoid the truncation, that means add noise that is not present in the original signal.
This 1 Euro part performs better
Attachments
Only if there is actual information. With a CD rip there is nothing below 16 bit anyway. And a signal attenuated to -60 dB retains 18 bit accuracy while at the same time most of those bits are already below the hearing threshold or lost in ambient noise, if proper gain staging wrt to the amplifier is applied.
But try it.