• 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.

DAC11001: New 20-bit, R-2R Precision DAC

Anyway, the distortion plots that show rising distortion with frequency look suspiciously similar to distortion plots of standalone Sallen Key filters using opamps. This is due to the non-linear input capacitance of FET input opamps. Yes, I know that TI has gone to great lengths to minimize this effect, but it doesn't take much when using an opamp in unity gain non-inverting applications for this to take place.

Walt Jung has discussed this before. Much of the information is in various ADI publications online.

Scott Wurcer also discussed this in the AD743 data sheet, with an example shown there. He may have preceded Walt, in fact.

I doubt it. Paul already told you what its origin is. OPA828 or OPA1611 or whatever they are using will not perform this poorly, even in a SK filter. Not exactly top secret information that would be out of the knowledge base of TI apps engineers.
 
Last edited:
Really the only reason I did it in hardware is that I did not want to get into the XMOS firmware. We have developed an XMOS image that made it a basic USB-I2S bridge. It might have made the solution more elegant, but I am a lousy EP programmer :)

I also thought having the discrete digital logic would make the design more reproducible, because you can use any 32-bit RJ I2S input. For example, you can use a DIR and skip the MCU all together.
 
I am new to the diy audio hobby (but have experience in related fields of electronics). I'd like to try my luck on designing an analog circuit around this chip (or any similar chip). Any general design guides, spice models available, and options for ics for buffering etc? The only audio based analog devices whose design I have somewhat understood (by looking at the schematic) are some first watt amplifiers and Amb beta 22.
 
Last edited:
DAC11001A Stereo DAC DIY?

Anyone want to DIY a stereo DAC based on DAC11001A?
As a new 20bit-R2R DAC, I think it would be interesting and worth of trying.

If we can build a project based on USB-I2S-SPI solution that Paul proposed, it would be compatible with many I2S product like Ian's FIFO or some reclocking.

I think I can try to make a simple PCB that have some 74 logics with I2S interface and runs at NOS mode, is there anyone interested and/or willing to help, support this?

It would be as simple as it could, and it reminds me that there's another project that aims to use AD768, DAC729 instrumetation DAC as audio DAC.

Regards,

Jen
 

Attachments

  • AD768.png
    AD768.png
    383.7 KB · Views: 428
Anyone want to DIY a stereo DAC based on DAC11001A?
As a new 20bit-R2R DAC, I think it would be interesting and worth of trying.

If we can build a project based on USB-I2S-SPI solution that Paul proposed, it would be compatible with many I2S product like Ian's FIFO or some reclocking.

I think I can try to make a simple PCB that have some 74 logics with I2S interface and runs at NOS mode, is there anyone interested and/or willing to help, support this?

It would be as simple as it could, and it reminds me that there's another project that aims to use AD768, DAC729 instrumetation DAC as audio DAC.

Regards,

Jen

Please use an FPGA/CPLD with clock synchronized outputs. It's not 1990 anymore. A Lattice MachXO2-256 (TinyFPGA AX1) can probably do this in a single device, for example.
 
Last edited:
I just watched a video featuring Jason at Schiit saying they're close to releasing a new Yggy DAC using a TI 20bit R2R chip. Which must be the DAC11001, right? He also mentioned one using a TI 16bit chip - which could that be?

DAC81001 is likely if the statement is correct. There's a significant enough price difference between the two ($66 vs $33) that they might create a different SKU for the 81001. Both of them are 16ish bit SINAD/ENOB parts due to THD (5th harmonic? check the datasheet) being at -105dB, but the 11001 has better FSE/INL specs, likely process binned chips. Total speculation - VCR is impractical to trim, and they're likely binning the lowest VCR chips as 11xxxx and the higher VCR silicon as 9/8xxxx devices.
 
Last edited:
All,

Yay the logic required to do I2S->SPI is pretty basic and a simple CPLD would probably do the trick. You could probably do that in schematic capture using the Altera/Intel stuff for EPLD 600 or something like that and then convert the file to jedec and use an Atmel/Microchip 7500 part (24pin I think) and have reclocking and synchronous output.

So basically something like XMOS XU208->CPLD->DAC chip->Analog output.

Thanks,
Gordon
 
So they're quartering the BOM cost for increased glitch energy. I'm assuming the 8812 is used for both channels. $20 in quantity vs $110 for the DAC11001. Both have the same THD/ENOB specs, but the 8812 glitch energy is much larger by ~25-100x. Is that what they're selling as a perceivable difference?
 
Member
Joined 2007
Paid Member
@paul Frost if you are still around, and are ok with a question ;) ...

If so, I have noticed that TI has updated the DAC11001 to a "B" version which has improved specifications, however, I am puzzled by the THD specs shown in figure 6-47. If one extrapolates the graph towards DC it looks as if it might be close to - 118 dB (and a similar spec at 20 kHz) but then it drops quite a bit to around - 103.5 dB at 12 kHz. And it is an unusual bell shaped THD curve ... Any chance you can say why the THD spec looks as it does - and if anything can be done to amend this bell shape?

I also notice that the VREFPF & VREFNF voltages are +/- 4.5 volts for these measurements. I might have missed it but I did not find any information about how the THD specs change when e.g. increasing the VREF** voltages towards the maximum voltages (in practice it looks as if they are +/- 20 VDC). Any chance you have such information available?

Also, I notice that you use a sixth-order filter in relation to e.g. the FFT specs in fig. 6-45. Typically a quite steep filter in an audio context - does this imply that the DAC itself is quite noisy upwards in frequency?

And then, here hoping that dreams may come true :angel:, I personally (and maybe other audio related diy'ers / professionals?) would be very interested in something like a DAC11001"C" capable of 1.536 MHz sampling at similar specs and with 22 - 24 bit resolution. Might this appear some day - soon?

Cheers & thanks,

Jesper
 
Hi Jesper,
1. THD is impacted by the ratio of generated tone and sample rate, so we would expect the best THD at a low frequency tone and high frequency sample rate. So you would see a positively increasing THD due to the harmonics getting larger, which accounts for the positive slope side of the bell curve. The negative slope side is due to the filter removing the harmonics from the measurement. For example, at 12kHz, the 3rd harmonic is now beyond the 30kHz brick wall filter.
1643037353472.png

2. I am confirming with our design team what impact we see due to reference voltage. I know at some larger references, are distortion will increase. On the DAC11001A, a +3V reference was the best case performance (THD wise). We improved it on the rev B design. I will update when I have a number to share.

3. Yes the sixth order is a bit steep :). This device is not really design for audio, so the standard AES17 filters would not make sense. That being said, the brick wall filters used in the Audio Precision measurement systems are also pretty high order, and the 20kHz BW and 40kHz BW are standard in audio DAC specification, so it not totally out of place to use a high order filter after it. Ultimately, we use the filter to simplify our specifications, because every customer is a bit different.

4. The real issue with creating higher resolution DACs is not just the resolution (which is a very hard problem, obviously) but also how you measure it! Our industry standard, highest precision meters top out at 8.5 digits (HP3458), so we have to get very creative on finding ways to test our devices. Higher resolution really means we also have to fight the noise floor in applications as well!
 
Member
Joined 2007
Paid Member
Hi Paul,

& many thanks for getting back on this with a helpful and clarifying feedback ;)

Considering what you have written, however, I would say that it would not fit into my requirements - saying this promptly so unless you might yourself wish to have the information about the reference vs. distortion specs you do not need to look into this on my behalf .. But thanks again for the reply & should you & TI some day find a "path" through to an almost equal-frequency-distortion DAC11001"C" - measurement equipment challenges notwithstanding - I for one will be very interested in hearing about it!

Cheers,

Jesper