Which DACs bypass digital filtering?

On 10/21/2017 at 2:15 PM, George Hincapie said:

Add Metrum Acoustics to your list. Yes, I am a fan, but they are wonderful.

hot of the press, 6Moons first take on the new JADE http://www.6moons.com/audioreviews2/metrum6/1.html

Hi..i am a new user here. As per my knowledge most NOS DACs have components that serve as an analog low pass filter in order to do the reconstruction of the digital bitstream to analog music.  With Redbook input they thus reproduce the situation with the very earliest CD players, whose intermodulation distortion levels led to the use of the phrase "digital sound" in a negative way.

On 25/10/2017 at 6:03 PM, NoellEagan said:

Hi..i am a new user here. As per my knowledge most NOS DACs have components that serve as an analog low pass filter in order to do the reconstruction of the digital bitstream to analog music.  With Redbook input they thus reproduce the situation with the very earliest CD players, whose intermodulation distortion levels led to the use of the phrase "digital sound" in a negative way.

The idea is to have a D/A only DAC, leaving the upsampling and the filtering to the music file player/processor.

A quick update to the list:

- Phasure NOS1 DAC

- Teac UD-50x family

- Esoteric (some models?)

- Aqua DACs

- Lampizator DACs

- Audio-GD (some models)

- Spring AudioDACs (some models?)

- Denafrips DACs

- iFi micro (some models)

- Metrum Acoustics DACs

- RME ADI-2 Pro

And of course, if it wasn't mentioned before:

- T+A DAC8 DSD

6 hours ago, Miska said:

And of course, if it wasn't mentioned before:

 

- T+A DAC8 DSD

I wasn't sure about that one.

How about Audio Note DACs?  Although in the spirit of this thread, their limited inputs/rates  probably don't make for a practical proposition for these looking to do pre-DAC DSP.  I think the cheapest model 0.1x? has USB input and accepts  up to the giddy heights of 24/96.

7 hours ago, semente said:

- iFi micro (some models)

I believe all the iFi models bypass the DAC chip's internal filters at 352.8/384 kHz sample rate. The DSD1973 only supports those rates if the filters are bypassed.

3 hours ago, mansr said:

I believe all the iFi models bypass the DAC chip's internal filters at 352.8/384 kHz sample rate. The DSD1973 only supports those rates if the filters are bypassed.

The attached is from the tech notes of the iDAC2, which uses the DSD1793.

Does the mention of 705/768kHz mean that even 352/384kHz rates are internally up-sampled to 705/768kHz?

5 minutes ago, Em2016 said:

The attached is from the tech notes of the iDAC2, which uses the DSD1793.

Sorry, DSD1793 is what I meant above.

5 minutes ago, Em2016 said:

Does the mention of 705/768kHz mean that even 352/384kHz rates are internally up-sampled to 705/768kHz?

 

That figure isn't entirely accurate, but it's close enough. The iDAC2 doesn't support rates higher than 384 kHz (the limit of the DSD1793). There is no resampling other than that internal to the DAC chip.

6 minutes ago, mansr said:

That figure isn't entirely accurate, but it's close enough.

Hmm that’s from the iDAC2 tech notes.

So what do you think the mention of 705/768kHz is referring to?

A typo by iFi?

https://ifi-audio.com/wp-content/uploads/data/iDAC2 - Spilling the Sauce (2 of 6)_ 9 Nov 2015

3 minutes ago, Em2016 said:

Hmm that’s from the iDAC2 tech notes.

 

So what do you think the mention of 705/768kHz is referring to?

 

A typo by iFi?

Maybe. I trust the TI datasheet (http://www.ti.com/product/dsd1793) over iFi marketing material. The iDAC2 as a whole definitely tops out at 384 kHz. I've tried one myself.

6 minutes ago, mansr said:

Maybe. I trust the TI datasheet (http://www.ti.com/product/dsd1793) over iFi marketing material. The iDAC2 as a whole definitely tops out at 384 kHz. I've tried one myself.

Don't forget iFi have said the publicly available datasheet is not the whole story because they said they worked closely with BB Jap to customise that chip.

Hard to believe Thorsten would make a typo/oversight like that with their technical notes though with the mention of 705/768kHz, no?

9 minutes ago, mansr said:

Maybe. I trust the TI datasheet (http://www.ti.com/product/dsd1793) over iFi marketing material. The iDAC2 as a whole definitely tops out at 384 kHz. I've tried one myself.

Btw, yesterday, completely separate to this thread (huge coincidence) I asked iFi about this:

2 minutes ago, Em2016 said:

Don't forget iFi have said the publicly available datasheet is not the whole story because they said they worked closely with BB Jap to customise that chip.

Customised chip? I don't think so. Maybe if you ordered 100 million units.

2 minutes ago, Em2016 said:

Hard to believe Thorsten would make a typo/oversight like that with their technical notes though with the mention of 705/768kHz, no?

Hard to believe TI wouldn't market the chip as 768 kHz capable if it can actually handle that rate. The block diagram you posted also doesn't agree with the datasheet. My measurements say the datasheet is correct.

23 minutes ago, mansr said:

Hard to believe TI wouldn't market the chip as 768 kHz capable if it can actually handle that rate. The block diagram you posted also doesn't agree with the datasheet. My measurements say the datasheet is correct.

Actually TI chips can use 705.6/768 with external digital filter. But the built-in one is limited 352.8/384k. Ayre used the PCM1792A in 705.6/768 mode in the original QB-9. The datasheets are just not so clear about it.

23 minutes ago, mansr said:

Customised chip? I don't think so. Maybe if you ordered 100 million units.

I think they mean they use some undocumented features. I know ESS Sabre has such features too...

P.S. Yes, iDAC2 is limited to 352.8/384k and DSD256.

4 minutes ago, Miska said:

Actually TI chips can use 705.6/768 with external digital filter. But the built-in one is limited 352.8/384k.

The datasheet says the limit is half those rates. It's of course possible that many of the chips can be overclocked and TI simply doesn't guarantee it. If that's the case, you could test them yourself and discard any that didn't pass. With a high enough target price point, the added cost could be acceptable.

45 minutes ago, mansr said:

The block diagram you posted also doesn't agree with the datasheet.

The block diagram is from here:

https://ifi-audio.com/wp-content/uploads/data/iDAC2 - Spilling the Sauce (2 of 6)_ 9 Nov 2015

1 minute ago, Em2016 said:

The block diagram is from here:

 

https://ifi-audio.com/wp-content/uploads/data/iDAC2 - Spilling the Sauce (2 of 6)_ 9 Nov 2015

Yes, and the datasheet tells a different story. The basic idea is the same though.

3 minutes ago, mansr said:

Yes, and the datasheet tells a different story. The basic idea is the same though.

I know. Just posted a link in case anyone (not you) thought I was making this stuff up. Always nice to post links to the sources.

I'd love to know what THAT particular 705/768k (in the iDAC2 tech notes) is actually referring to though. Hopefully iFi can explain.

If IIRC PCM1704 is a NOS chip and all OS is made by the digital filter. In the BB PCM1704 datasheet they say: “8X OVERSAMPLING AT 96kHz” and ”Digital data words are read into the PCM1704 at eight times the standard DVD audio sampling frequency of 96kHz (e.g., 8 x 96kHz = 768kHz) to create a sinewave output of 1100Hz.”

On 03/12/2017 at 9:58 AM, Em2016 said:

 

The block diagram is from here:

 

https://ifi-audio.com/wp-content/uploads/data/iDAC2 - Spilling the Sauce (2 of 6)_ 9 Nov 2015

 

It looks like it was a fair assumption that the iDAC2 Technical Notes were correct and there is more happening than the DSD1793 datasheet shows, as iFi has said before.

From the horse’s mouth:

“Any sample rate from 88.2kHz to 192kHz is converted to 705.6/768kHz IF the minimum or standard filter is selected. If the bitperfect filter option is selected the DAC is fed directly to the actual DAC part of the chip and converted without any upsampling”

This explains their block diagram much more clearly.

1 hour ago, Em2016 said:

It looks like it was a fair assumption that the iDAC2 Technical Notes were correct and there is more happening than the DSD1793 datasheet shows, as iFi has said before.

 

From the horse’s mouth:

 

“Any sample rate from 88.2kHz to 192kHz is converted to 705.6/768kHz IF the minimum or standard filter is selected. If the bitperfect filter option is selected the DAC is fed directly to the actual DAC part of the chip and converted without any upsampling”

 

This explains their block diagram much more clearly.

I don't trust a word they say. Too many blatant lies. For example, the filter selection labelled "minimum phase" is anything but. It is the very much linear phase filter called "slow rolloff" in the datasheet.