• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 KHz

Jump to Latest
H

hop.sing

Member
Joined 2009
andrea_mori said:
The attached application note clearly explains the difference between resolution and accuracy (or precision).
Click to expand...
Thank you, but the practical consequences of missing accuracy are still not clear to me.
Am I right in assuming that less accuracy translates into more distortion? But that does not seem to be an issue with the dam, because as the signal goes down in level, the distortion goes down as well. Is that the advantage of the sign magnitude architecture of this dac?
So the full resolution seems to be available, as I can still measure tones right down to -110 dB full scale, but accuracy is only 14bit, but with low level signals, distortion disappears into the noise floor.
Is that right?
If so, what is the problem? I, at least, do not hear distortion on this DAC.
 
Z

zfe

Member
Joined 2014
hop.sing said:
Thank you, but the practical consequences of missing accuracy are still not clear to me.
Am I right in assuming that less accuracy translates into more distortion? But that does not seem to be an issue with the dam, because as the signal goes down in level, the distortion goes down as well. Is that the advantage of the sign magnitude architecture of this dac?
So the full resolution seems to be available, as I can still measure tones right down to -110 dB full scale, but accuracy is only 14bit, but with low level signals, distortion disappears into the noise floor.
Is that right?
If so, what is the problem? I, at least, do not hear distortion on this DAC.
Click to expand...

The low level behaviour is exactly the point with the sign-magnitude R2R architecture (in contrast to a single R2R ladder, where at zero transit the whole DAC bit-state is flipped). To quote from the PCM1704 DATA sheet:
The sign-magnitude architecture, which steps away from zero with small steps in both directions, avoids any glitching or large linearity errors, and provides an absolute current output. The low-level performance of the PCM1704 is such that true 24-bit resolution can be realized around the critical bipolar zero point.
Click to expand...
 
andrea_mori

andrea_mori

Banned
Joined 2005
hop.sing said:
So the full resolution seems to be available, as I can still measure tones right down to -110 dB full scale, but accuracy is only 14bit, but with low level signals, distortion disappears into the noise floor.
Is that right?
Click to expand...

Yes, you are right.
You have a perfect 14 bit accuracy dac, no doubt.

As I said previously, if the music you play has 14 bit of dynamic you got heaven.
But why 24 bit dac? Since you need only 14 bit dac?
In this case I recommend to use a TDA1541A or an AD1865. They implements better accuracy anyway, and both sound very fine.

Again, when the 15th bit switches the distortion will double, 16th bit double again and so on.

"The ladder operates as an array of voltage dividers
whose output accuracies are solely dependent
on how well each resistor is matched to the others.
Ideally, resistors within the ladder are
matched so that the voltage ratio for a given bit is
exactly half of that for the preceding bit."

Any error in resistors matching increases the distortion.
 
H

hop.sing

Member
Joined 2009
andrea_mori said:
Yes, you are right.
You have a perfect 14 bit accuracy dac, no doubt.

As I said previously, if the music you play has 14 bit of dynamic you got heaven.
But why 24 bit dac? Since you need only 14 bit dac?
In this case I recommend to use a TDA1541A or an AD1865. They implements better accuracy anyway, and both sound very fine.

Again, when the 15th bit switches the distortion will double, 16th bit double again and so on.

"The ladder operates as an array of voltage dividers
whose output accuracies are solely dependent
on how well each resistor is matched to the others.
Ideally, resistors within the ladder are
matched so that the voltage ratio for a given bit is
exactly half of that for the preceding bit."

Any error in resistors matching increases the distortion.
Click to expand...
But what happens to low level signals at the TDA1541A or the AD1865, does the distortion also disappear?
If not, the DAM still has an advantage....
 
S

soundcheck

Member
Joined 2005
soundcheck-audio.blogspot.com
oneoclock

oneoclock

Member
Joined 2010
1 Khz 0 dB tone shows DAM1021 -94 dB THD, similar TDA1541A

1 Khz 24 bit -60 dB, in DAM1021 harmonic distortion is -130 dB and noise -145 dB.
1 Khz 16 bit -60 dB TDA1541A harmonic distorsion -102 dB. noise 110 dB.

1 Khz 24 bit -90 dB in DAM1021 is a perfect sine wave and still 55 db dynamic range to -145 dB noise without harmonic distortion.
1 Khz 16 bit -90 dB TDA1541A signal is square waves and no have dynamic range.

Something is escaping to Andrea_mori and TT Electronics IRC reasoning about the R2R and the DAM1021 maximum performance and 0,01% R tolerance.
 
Last edited:
danny_66

danny_66

Member
Joined 2005
soundcheck said:
However. The sound character will very much depend on the sound footprint of the chosen resistors.
Click to expand...
Not so much as an I/V resistor.
The resistors in the R-2R network are used to reconstruct the analog signal, that's different than in an I/V conversion.
In the DAM1021 a single resistor in the network passes a simple single voltage if switched on.
In I\V you have the complete analog audio signal.
 
andrea_mori

andrea_mori

Banned
Joined 2005
glt said:
If we apply the "sliding window" theory, then in the TDA an AD, there is no space for the window to slide....
Click to expand...

Please, see also post #321
http://www.diyaudio.com/forums/vend...te-r-2r-sign-magnitude-24-bit-384-khz-33.html

I have no doubt that Sign magnitude is superior to two's complement notation, but there are no sliding doors when music dynamic exceed 13/14 bit. The lower the accuracy, the greater the distortion.
Ask yourself why PCM1704 ladder network used on chip laser trimmed resistors. It's a sign magnitude dac. And this process is very expensive.
And don't forget temp drift, aging, and switches resistance.
But ok, if you think I'm wrong no problem, i give up.
 
S

soundcheck

Member
Joined 2005
soundcheck-audio.blogspot.com
danny_66 said:
Not so much as an I/V resistor.
The resistors in the R-2R network are used to reconstruct the analog signal, that's different than in an I/V conversion.
In the DAM1021 a single resistor in the network passes a simple single voltage if switched on.
In I\V you have the complete analog audio signal.
Click to expand...

Don't you take the signal directly from the "ladder" at the raw-output? Basically a group of resistors/joints/traces/... ?
 
DEQ+TheEnd

DEQ+TheEnd

Member
Joined 2005
soundcheck said:
OK. That's a different story then. Then the accuracy is not the hottest topic/challenge anymore.

However. The sound character will very much depend on the sound footprint of the chosen resistors.
Click to expand...

Søren is not unfamiliar with sound quality of Z-fold resistors. He have already commended on it early in this thread. Guess he doesn’t foresee enough interest to justify building a single PBC DIY DAC at 5000$+ price point. And I don’t dare to think of necessary finances needed to bring it together. I've for sure would not pay that kind of money up front in a GP.
 
spzzzzkt

spzzzzkt

Disabled Account
Joined 2005
andrea_mori said:
But ok, if you think I'm wrong no problem, i give up.
Click to expand...

This is from the AES paper "20 Bit 'Collinear' DAC, a Solution to Low Level Problems" which discussed the PCM63's co-linear Sign Magnitude architecture:

After careful consideration of all the options, the preferred solution is one in which two separate DACs are combined in a complementary arrangement to provide
a current output which changes minimally around BPZ. The user is then free to follow this device with an amplifier of choice and not be concerned with switching or deglitching networks.

The only disadvantage of this approach, compared with a signed magnitude approach, is the added thermal noise of one DAC which is always on at BPZ. However, with careful design, this noise can be reduced to insignificance when compared to other noise sources. Figure 1 is a block diagram of this "Colinear" DAC and Figure 2 shows the code conversion between the input, binary twos complement (BTC) and that required for "Colinear" operation. This conversion is carried out on the chip and is transparent to the user.

From the coding it can be seen that only small bit changes occur either side of bipolar zero and since these bits can be matched to within one percent without any trimming, the linearity at BPZ will be accurate to 1/100 of one LSB or essentially perfect. At higher signal levels large bits are activated and the linearity degrades. But since the highest linearity is required for the lowest signal levels an increase in the resolution of the converter is appropriate. Noise, thermal and systematic will obviously place a limit on the achievable dynamic range and in the audio bandwidth, 20 bit resolution appears to be a practical limit.
Click to expand...

The DAM seems to be closer to the PCM63 architecture than the Sign Magnitude implementation the authors discuss, which relies on high speed amplifiers to switch between DAC halves.

BTW "linearity" is the correct term for what you are calling "precision".
I know the connotations attached to "precision" suits your agenda of discrediting the DAM better, but it is incorrect.

Figures 1 and 2 referred to in the quote are in the attached pdf.
 

Attachments

  • figures.pdf
    67.6 KB · Views: 176
Last edited:
andrea_mori

andrea_mori

Banned
Joined 2005
spzzzzkt said:
BTW "linearity" is the correct term for what you are calling "precision".
I know the connotations attached to "precision" suits your agenda of discrediting the DAM better, but it is incorrect.
Click to expand...

Call it "linearity", "precision" or better "monotonicity", always the same thing defined with different words. In any case there is a linear correlation between the ladder accuracy (resistors tolerance, thermal drift and so on) and distortion.
Burr Brown used a single ladder per channel in their Collinear dac, switching the voltage rails between plus and minus, according to the sign bit. There is a reason: using two ladder, the non-linearity will be doubled.

I believe you misunderstood my words, I have nothing to discredit, since I have no interest to discredit, I'm not a competitor. I said this previously, and I say this again, if it was not clear.
This is a commercial device, like Total Dac, MSB, AMR or my Naim CD3 TDA1541A based. There is nothing to do it yourself (box aside), no schematic, nothing shared.
Simply yet another commercial dac. Anyone is free to buy this dac, I have never suggested to not buy it. Maybe I will buy myself this dac, to compare it with other stuff, as soon as I'll find the time to do that. And Soeren's work is absolutely admirable.

But we are on a diy forum, so I believe we can debate around technical choice of any vendor and any device, without any preconcepts. That's the only thing I'm trying to do, either if we are talking about this dac, or if we are talking about the Naim CD3.
 
N

nightsky

Member
Joined 2013
some good news

I am extremely satisfied with this DAC as it outperforms everything I have had before (many high end & studio DACs,). I use it with a 6H6p tube output stage which really makes a huge difference compared to the raw output. The buffered output is not useable for music listening in my opinion.
As I use a mini SHARC module with remote volume control I have no problems with pops or other noises but am limited to 96/24 which is not a problem for me as I mostly listen anyway to CDs and 96/24 or 88/24 downloads.
I never had the on/off switch problem with the loud pop as I use a delay circuit that turns on all equipment in a certain order and off in exactly the opposite order with a different delay time. Even with a power outage there would be no problem and danger for the speakers.

Here is my system: JPLAY with mini Player, SOtmTX PCIe USB card with external psu, Wave I/O, miniSHARC, dam DAC, tube output stage. All connection I2S. Each stage has a separate psu.
 
N

new2hifi

Member
Joined 2010
Power outage?

nightsky said:
.......................
I never had the on/off switch problem with the loud pop as I use a delay circuit that turns on all equipment in a certain order and off in exactly the opposite order with a different delay time. Even with a power outage there would be no problem and danger for the speakers.................
Click to expand...


nightsky, it will be nice if you can posts from pictures.

Are you using a UPS on dam; how come you have protection from power outage pops?
 
You must log in or register to reply here.
Top Bottom