As you know I’ve been using the 1632 filter stage , I recently did try the SE mod and preferred it , it sounded a little better focused to me . The musics a little easier to follow , I wanted to spend more time listening before posting anything.
Tonight I decided to dig out the original active filter board . First I tried it as standard , I still prefer this filter board without the third stage so just bypassed it .I then tried the SE input which made more of a difference with this stage than it did with the 1632 for some reason . I’m quite surprised Marcel didn’t hear any difference . Again it’s better focused but now widens the soundstage and creates more of an enveloping sound .
I’ll spend more time comparing to see which I prefer between this or the 1632 .
Obviously you get a dc offset now so I added 100uf nichicon KZ coupling caps after the output resistors . I also used a modified pin socket for the SE input between the boards to allow quick and easy comparisons .
Regarding the 1632 filter stage (post #2471) grounding the -In input just makes the filter stage a single-ended to balanced converter. Other than that the filter stage works more or less the same as with differential input. Only major difference comes from DAC output common noise which is not cancelled with single-ended input. So if this single-ended input mode with OPA1632 filter stage is preferred there are 2 likely explanations: 1) common noise is preferred or 2) confirmation bias. I'll leave it to Marcel to explain why 1) might be the case with this DAC.
With the OPA1632 filter connected to only one DAC output, 390 ohm from the unconnected filter input to ground, unused DAC output open, there will be some extra second harmonic distortion due to signal on the reference. It's about 0.13 % D2 at 0 dB DSD at the worst frequency (240 Hz...300 Hz), if I did the calculation properly - not a lot compared to what the old HiFi standards allow, quite a lot compared to the distortion levels bohrok2610 and others have measured.
The experiment I tried was Mark's earlier slow common-mode loop experiment, not a real single-ended connection.
Last edited:
That is quite a lot of added 2nd HD. It is a plausible explanation of audible difference and preference as well.
The strange thing is it doesn’t sound noisier or brighter . Trying both stages again I still feel the SE input makes more difference with marcels filter stage than it did with Bohrok2610 OPA1632 stage . Both sound good with balanced input it’s just that some prefer SE input . I normally find with higher noise things tend to sound more distorted , not the case here .
Is there anything else worth trying ? Great thing with this dac is it’s easy to try things and revert back if needed .
Is there anything else worth trying ? Great thing with this dac is it’s easy to try things and revert back if needed .
I dunno about people sometimes. I keep trying to explain I know the theory. I have also said many times that I hate distortion, including 2nd harmonic. What you guys don't seem to get is there is something else going on whereby low level musical details are being canceled out as common mode noise. I also know the theory doesn't mention that. However, all one has to do is listen and its quite clear what is being preferred is not distortion at all. What is being preferred is less loss of low level musical details. The loss happens because the + and - dac outputs are not perfectly matched at low levels.
Anyway, this type of misunderstanding is what happens when all people know how to do is measure distortion with FFTs and they don't know how to listen. So they form erroneous theories based on what happens to be easiest to measure. Other stuff which is harder to measure is ignored and even worse, its the possibility of its existence rejected without much careful thought.
Anyway, this type of misunderstanding is what happens when all people know how to do is measure distortion with FFTs and they don't know how to listen. So they form erroneous theories based on what happens to be easiest to measure. Other stuff which is harder to measure is ignored and even worse, its the possibility of its existence rejected without much careful thought.
Last edited:
Instead of speculating with alternative theories why don't you just measure your SE output DAC to show that it does not have elevated 2nd HD.
You still don't understand. I will tolerate a little distortion if I have to. I don't like it, but I would rather hear all of the music.
Its like choosing whether to throw out the baby with the bathwater or keep a little bathwater if that's the only way to save the baby. But people keep saying, "But you can still measure some remaining bathwater!!!" So you must like bathwater!
Its like choosing whether to throw out the baby with the bathwater or keep a little bathwater if that's the only way to save the baby. But people keep saying, "But you can still measure some remaining bathwater!!!" So you must like bathwater!
No, I do understand your speculation. The problem with it is that your "all of the music" contains common mode noise and 2nd HD which are actually not part of real "all of the music". But you have not shown that any part of real "all of the music" is missing with differential output.
All one has to do is learn how to listen in terms of discrimination of sound stage. Preference can't and won't help with that.
Jumping to conclusions that distortion preference accounts for everything just because distortion is so easy to measure doesn't make sense.
Jumping to conclusions that distortion preference accounts for everything just because distortion is so easy to measure doesn't make sense.
Last edited:
This is not about jumping to conclusions. Just show that your SE output does not have added common mode noise and 2nd HD and that some detail is missing with differential output. Otherwise this is just speculation.
I'll never understand why some people doubt everything except the validity of uncontrolled listening tests, but anyway, it's too unimportant to get worked up about.
Let me just add that if added common mode noise and 2nd HD make this DAC sound better I would be very hesitant to call it SOTA. No offense to Marcel.
Okay, Marcel. However, I am getting PMs expressing disbelief that some people don't hear sound stage so they can't discriminate the important difference with SE. HD is just a red herring.
Either way the dac still sounds great and provided us with fun trying this stuff out 😁
I wish I could do all the needed measurements, if somethings simple and reversible to try I’ll try it . If I don’t like the results I’ll revert it back .
Still no denying this dac sounds the best I’ve tried so far
I have no issues with hearing soundstage so I don't see the relevance of this comment. Single-ended does not mean better soundstage.
I will attempt to do a measurement later comparing the normal and SE modes of the 1632 filter. Although my measurement setup is not optimal it should be hopefully acceptable for a comparative function. I am a bit pushed for time though so might not be able to get to it.
I can't go into details, but surprisingly, an audio DAC for cheap car radios that I designed on my previous job has essentially all the features that Mark prefers. If anyone should want to try it, one of its versions (TEF6686) is stocked by Mouser and its full datasheet can be found when you look around on the internet, even though it is marked confidential (TEF6687.pdf, the family is TEF668X).
Last edited:
I have attempted a few measurements of one of my RTZ DACs to compare balanced and "SE" modes quickly. Disclaimer: I am new to the software and the equipment to do the tests so please don't shoot the messenger! 🤓
The only ADC I have is a Behringer UMC202HD so that had to do. I'm not using any fancy setup to try to isolate the PC from the measurements so as far as absolute readings go I have no idea if we're seeing the test device or the ADC/PC or external influences..
However, since this is a comparative test then I hope the figures will mean something to someone more than they do to me.
First up is the opa1632 filter configured as balanced but with the 100uF output coupling capacitors still fitted since it would've taken me longer to remove and re-apply them.
Next up is the "SE" mode where the -ve input to the filter op-amps are disconnected from the DAC and connected to ground through a resistor.
And finally, I added 2 x 200R resistors on the un-used DAC outputs to ground to approximately match the loading on the used ones.
Interestingly 3rd harmonic seems to increased while 2nd harmonic reduces. I am listening now with the loading resistors but I haven't had enough time to really hear whether it has affected the sound in any way.
These are not really the results I was expecting. I can't help think my testing process is not capturing the reality.
OK, I'll return to my bunker and put my tin hat on. 😏
The only ADC I have is a Behringer UMC202HD so that had to do. I'm not using any fancy setup to try to isolate the PC from the measurements so as far as absolute readings go I have no idea if we're seeing the test device or the ADC/PC or external influences..
However, since this is a comparative test then I hope the figures will mean something to someone more than they do to me.
First up is the opa1632 filter configured as balanced but with the 100uF output coupling capacitors still fitted since it would've taken me longer to remove and re-apply them.
Next up is the "SE" mode where the -ve input to the filter op-amps are disconnected from the DAC and connected to ground through a resistor.
And finally, I added 2 x 200R resistors on the un-used DAC outputs to ground to approximately match the loading on the used ones.
Interestingly 3rd harmonic seems to increased while 2nd harmonic reduces. I am listening now with the loading resistors but I haven't had enough time to really hear whether it has affected the sound in any way.
These are not really the results I was expecting. I can't help think my testing process is not capturing the reality.
OK, I'll return to my bunker and put my tin hat on. 😏