Return-to-zero shift register FIRDAC

Jump to Latest
#2,084
There are valid reason why I don't normally publish gerbers. My layouts are not for normal diy as I use plenty of components that require a reflow oven and I don't have time to provide any assembly support. There are also too many copycats lurking here. But anybody interested and bravehearted can PM me for gerbers.
 
#2,085
bohrok2610 said:
I'm aware that measurements are now out of fashion in this thread but I've noticed that "stock" filter has somewhat elevated distortion at 10kHz.

This is stock filter (without 3rd stage).
View attachment 1278463

With a simple modification I could improve this by almost 20dB.

View attachment 1278464

Since I'm under NDA I may be able to share what I find with Marcel, but probably can't share details publicly. However anybody with a pair of ears should easily find out the solution just by listening.
Click to expand...
Definitely not out of fashion , just that some of us can't do them 🙁 Its a shame about all these NDA's , a 20db improvement would be nice to have
 
#2,089
Harmonic distortion at 10 kHz is just a measure of how well or how poorly a circuit can handle signals with a relatively large rate of change. The fact that the harmonics are above the hearing range of most humans (but well within the hearing range of cats and dogs) doesn't change that.
 
  • Like
Reactions: dkfan9
#2,091
Hmm.. in the profound past I had replaced the NJM2068D opamps (6V/usec) in my EMU1616m card's Dac ouput filter and ADC input stage sections - to OPA1612 (27V/usec) bipolar, low noise, moderate speed parts. The better slew rate cause less slewing distortion, which could become detrimental in an active filter/ IV position at the output of an S/D dac..
The first thing I noted was that while at 1kHz the results did not change too much, at 10kHz the distortion decreased an impressive way. Very similar to what Bohrok had shown here..🤔. The OPA2210 is having 6V/usec slew rate specification..
emu1616m_-18db_10KHz_victor'scr.PNG
 
Last edited:
#2,093
So, it would be my first stab at guessing what had been done under NDA here.. 🙂
I would not put up my life on it. Hope there are some even simpler solutions..🙂
Just a second note: in the Singxer SDA-6 output stage, after the AK4499EQ balanced current outputs first there are plain OPA1612 I/V converters then a Diff/Se converter stage and finally a Sallen Key active lowpass filter stage with AD8599.. maybe also a possible option?
 
#2,094
I will tell you this: I am probably not going to use the Marcel output stage board. Probably, but not fully decided yet. I will seek to find some other way. Andrea Mori has a nice discrete transistor output stage which I have heard with the Andrea DSD dac here now, and which might be worth trying for Marcel's dac. Also, I have some access to a small number of interesting prototype audio transformers which could be used for a sorta no-dac dac type of signal processing, etc. Otherwise, my concern would be that a string of multiple IC opamps eventually starts to sound like what it is. Which is to say, the sound can be colored by an excess of "opamp in RF environment" sound. Bill Whitlock used the terms "veiled" and or "grainy" in the context of ground noise. Those terms and some others might be applicable to what too many opamps in a dac can do.
 
Last edited:
#2,095
Markw4 said:
Which someone might presume you could hear, or was it only an issue for instrumentation use?
Click to expand...
Instrumentation use. I needed it to control my tweaking with composit amps..
And / but- could I hear, perceive a change in the sound? though it was not my principal dac already at that time but I was kind of fond of it's sound - I used to appreciate it before the mods.
Now that has changed after. I stopped listening to it afterwards.. 😂😜 Fortunately it was not the primary interest.
Edit: I place it here, because I can not edit my previous post anymore: The screenshot is the tweaked ADC input section with opa1612, and recording a Victor's 10kHz oscillator.
 
Last edited:
#2,099
The increased distortion at 10kHz seems to be caused by the common-mode noise suppressor circuit (around U8 & U15). My quick & dirty fix was to "eliminate" it by removing 3k9 resistors R46 & R89. This works as expected in differential mode but as @MarcelvdG promptly pointed out unfortunately not in single-ended mode as common-mode noise is increased and there will be DC offset unless 3rd filter stage is used. I mostly use differential mode so I failed to notice the DC offset.

I tried modifying the CM noise suppressor in other ways but with no tangible impact on 10kHz distortion. And it remains a mystery why the distortion is so much higher with higher frequencies. So unfortunately no solution exists yet.

MarcelvdG said:
Harmonic distortion at 10 kHz is just a measure of how well or how poorly a circuit can handle signals with a relatively large rate of change. The fact that the harmonics are above the hearing range of most humans (but well within the hearing range of cats and dogs) doesn't change that.
Click to expand...
I have to slightly disagree with this as distortion starts to rise at much lower frequencies than 10kHz. I just happened to notice it at that frequency. Also IMD will increase. Whether or not this is audible is another question but still it would be nice to have frequency independent distortion profile.
 
#2,100
gpapag said:
IMO, it takes great talent and skill to show face while claiming NDAs in an open forum, especially at an open design thread.
Click to expand...
That would be about half of the dac designers here. How many said they couldn't say certain things about ESS dacs due to NDAs? Everyone who had a legal datasheet and or other proprietary information had to say that, including me.
 
Last edited:
You must log in or register to reply here.
Top Bottom