I think there are some original Philips engineers that hang out here at DIYA, so perhaps they should comment first 
Anyway, back in Oct. 2015, on head-fi.org, Jason Stoddard (of Schiit Audio), made the following comments about one of the Schiit D/A products:
So, how much of the above is hype and how much is legit? (Again, our Dutch friends may respond first!)
Note that Schiit uses $$$ monolithic Analog Devices multi-bit DACs (not specifically designed for audio).
The AD5547 is $13 - $20 per chip!
The AD5791 (used in the Schiit top-of-the-line DAC) is $38 - $166 per chip!
Anyway, back in Oct. 2015, on head-fi.org, Jason Stoddard (of Schiit Audio), made the following comments about one of the Schiit D/A products:
So, how much of the above is hype and how much is legit? (Again, our Dutch friends may respond first!)
Note that Schiit uses $$$ monolithic Analog Devices multi-bit DACs (not specifically designed for audio).
The AD5547 is $13 - $20 per chip!
The AD5791 (used in the Schiit top-of-the-line DAC) is $38 - $166 per chip!
About those modern AD multi-bit DACs (AD5547, AD5791) ...
It's a bit surprising that here on DIYA of all place (where multi-bits have survived and thrived while most of the high-end audio industry went sigma-delta), these DACs have been totally ignored.
In fact, I think only Braxy and some a-hole called "hollowman" has posted on these chips:
Schitt's flagship DAC uses a not-for-audio DAC chip. - diyAudio
Schiit Audio using AD5791 (modern R2R/multibit) - diyAudio
Come on guys .... Schiit's made a killing off their multi-bit dacs .. and no one in the DIY circle has experimented with the AD chip?
Don't tell me it's the up-front cost? Especially when compared to eBay listing of TDA1541a (double crowns), PCM63's, etc.
It's a bit surprising that here on DIYA of all place (where multi-bits have survived and thrived while most of the high-end audio industry went sigma-delta), these DACs have been totally ignored.
In fact, I think only Braxy and some a-hole called "hollowman" has posted on these chips:
Schitt's flagship DAC uses a not-for-audio DAC chip. - diyAudio
Schiit Audio using AD5791 (modern R2R/multibit) - diyAudio
Come on guys .... Schiit's made a killing off their multi-bit dacs .. and no one in the DIY circle has experimented with the AD chip?
Don't tell me it's the up-front cost? Especially when compared to eBay listing of TDA1541a (double crowns), PCM63's, etc.
Simply put, proof is in the pudding, and there's a bunch of modern dac chips that supersede these from Analog Devices, especially for audio applications.
Given the multibit obsession on diyaudio has not been one pushed by the folks with the technical wherewithal, its no surprise that no one has worked with these chips.
Given the multibit obsession on diyaudio has not been one pushed by the folks with the technical wherewithal, its no surprise that no one has worked with these chips.
They require a lot of extra engineering like de-glitching to use them for audio. This was tried way back in 1986 or so and failed. The TDA/BurrBrown/AD1862/5 style multi-bit converters won out until S-D appeared.
abraxalito's post is a good read, I personally don't like the tone of some of the Schiit posts and as he points out they do contain some errors.
abraxalito's post is a good read, I personally don't like the tone of some of the Schiit posts and as he points out they do contain some errors.
I forgot to mention I have some data on those DAC's using the dreaded sample and hold de-glitcher, but I can't share the results. Quite impressive though there is extra circuitry needing careful layout, nothing's for free.
@hollowman - I did better than being banned, I got a personal dressing down that my daughter found and asked "Why do these people hate you"?
@hollowman - I did better than being banned, I got a personal dressing down that my daughter found and asked "Why do these people hate you"?
Last edited:
Amir made some measurements (on ASR : Measurement and Review of Schiit BiFrost Multibit DAC | Audio Science Review (ASR) Forum) of one of Schitt's DACs that uses the AD5547. The result of interest to me was the linearity, which looks quite appalling. Mike Moffat said that the result shows a broken DAC but another guy on a different forum found pretty much the same issue with his measurements.
Pcm1704 is already obsolete and now almost gone from the market. I was trying to make generic pcm1704 (not original but same function) for several years. My first choice was LTC2642, which has better glitch performance(0.5nVs) than AD5547(3.5nVs). This is tolerable in an audio application without using deglicher by S/H. My current audio system has both pcm1704 and LTC2642. I compared them carefully and concluded my preference is pcm1704 because it has soft sound like a film camera. The difference exists between them. LTC2642 has clearness like a video camera. Some people may prefer the clarity. I don't know glitch can make the difference. It was a little bit surprising to me they had the difference. Anyway, my first attempt to make generic pcm1704 was not successful.
The second approach was from last year. I needed to find glitch free DAC except for pcm1704. High-speed DAC for RF application is basically glitch-free in the audio band because their clock is very high. DAC2904(14bit 125MSPS) has 0.002nVs glitch. The best solution for audio application is AD9117(14bit 125MSPS) as long as I have tested. AD9117 has at least 15bit equivalent accuracy after internal calibration. Parallelling by two channels of AD9117 has better THD+N performance than pcm1704 by slight DSM to extend bit resolution up to analog noise floor. Up to 15-bit resolution, it operates in multi-bit. From 16bit to 20bit(analog noise floor) is done by DSM.
The second attempt has almost finished. I'm now trying to design PCB. High-speed DAC has another advantage of coexistence with DSM. AD9117 has two possibilities, multi-bit and DSM because of its high sampling rate. DSM(5-bit quantizer) has better THD than multi-bit because it can have digital calibration which multi-bit can't do. My prototype board has large-scale FPGA(xc7a100). It takes some time to shape up the size to small-size one(probably xc6slx25). Unfortunately, I was forced to give up QFP package which is DIY friendly because I wanted to implement both multi-bit and DSM. But now I'm sure the goal is around the corner.
The second approach was from last year. I needed to find glitch free DAC except for pcm1704. High-speed DAC for RF application is basically glitch-free in the audio band because their clock is very high. DAC2904(14bit 125MSPS) has 0.002nVs glitch. The best solution for audio application is AD9117(14bit 125MSPS) as long as I have tested. AD9117 has at least 15bit equivalent accuracy after internal calibration. Parallelling by two channels of AD9117 has better THD+N performance than pcm1704 by slight DSM to extend bit resolution up to analog noise floor. Up to 15-bit resolution, it operates in multi-bit. From 16bit to 20bit(analog noise floor) is done by DSM.
The second attempt has almost finished. I'm now trying to design PCB. High-speed DAC has another advantage of coexistence with DSM. AD9117 has two possibilities, multi-bit and DSM because of its high sampling rate. DSM(5-bit quantizer) has better THD than multi-bit because it can have digital calibration which multi-bit can't do. My prototype board has large-scale FPGA(xc7a100). It takes some time to shape up the size to small-size one(probably xc6slx25). Unfortunately, I was forced to give up QFP package which is DIY friendly because I wanted to implement both multi-bit and DSM. But now I'm sure the goal is around the corner.
Looking forward to learn from you guys and the development of this discussion.
I wish there was a measurement that correlates with listening satisfaction. I for one am still listening to the TDA1543 and it is an absolute winner for me. John (ECDesigns) says the jitter signature of this particular chip makes it a winner. He also said it resembles an analog tape machine (I agree with his impressions). It´s somewhere on that giant thread of his.
BTW, I and Richard (Abraxalito) had been discussing small differences between the TDA1543A and the TDA1543. I just wanted to say for the record that the non-A chip is the magic one for me. I haven´t been as happy with the TDA1543A, and believe me when I say I wanted to be happier. Because it fixes some issues with the internal logic of the chip. But it just didn´t give me the emotional response I was used to.
I wish there was a measurement that correlates with listening satisfaction. I for one am still listening to the TDA1543 and it is an absolute winner for me. John (ECDesigns) says the jitter signature of this particular chip makes it a winner. He also said it resembles an analog tape machine (I agree with his impressions). It´s somewhere on that giant thread of his.
BTW, I and Richard (Abraxalito) had been discussing small differences between the TDA1543A and the TDA1543. I just wanted to say for the record that the non-A chip is the magic one for me. I haven´t been as happy with the TDA1543A, and believe me when I say I wanted to be happier. Because it fixes some issues with the internal logic of the chip. But it just didn´t give me the emotional response I was used to.
I have a hunch... The missing measurement is the "time aberration" (my idea of a name ). The Dutch know all about it
See for example:
5534 audio amp
And:
https://www.by-rutgers.nl/ME6211-PRO37R.html
). The Dutch know all about it See for example:
5534 audio amp
And:
https://www.by-rutgers.nl/ME6211-PRO37R.html
To finish my "case"
Example of a successful / well-regarded product using both TDA1543 and OPA134:
The Altmann Attraction DAC
Example of a successful / well-regarded product using both TDA1543 and OPA134:
The Altmann Attraction DAC
Last edited:
It's easier to design I/V stage of high-speed DACs rather than audio DACs like pcm1792 because bandwidth is beyond the capability of an OP amp. Analog Device recommends transformer conversion since they are RF and easy to have high-performance SMD transformer. But my target is the audio band and not allowed to use a transformer. Audio band transformer is like OPT for a tube amp. The only solution is a passive one. One register is the best selection which has excellent performance up to 100MHz irrelevant to a current value. Fortunately, DACs for RF have relatively wide compliance voltage. From +0.5V to -0.5V has no performance degradation. You can achieve -120dB THD by DSM. But you need five-minute warming up after power on. Even a monolithic ICs, -120dB requires such stabilizing time. I think a passive I/V is superior to active one If you want to use 11.2MHz DSD in pcm1792.
As to current value, I have tested eight or nine RF DACs(14bit and 16bit with 2mA to 30mA). AD9747 is 16bit with 30mA and has the best SNR in RF as long as I have tested. But it has almost the worst SNR in the audio band. I'm sure it has high performance in RF but not in the audio band. AD9117 is low power version of AD9767 which is a junior version of AD9747. So, AD9117 is far junior to AD9747 but is the best in the audio band. RF DACs have strange inherent performance in the audio band. The more current doesn't mean good performance, usually have an inverse effect. Unexpectedly, low current AD9117(I use it with 6mA) has the best noise power in the audio band. Paralleling two channels achieves 108dB(without A filter). Pcm1792, which is the champion for me, is 112dB in the same circumstance. If you parallel four channels(two AD9117), you can achieve 111dB.
Yggy metrics
Stereophile's measurements of Schiits Yggy - -which uses the high $$ AD DAC -- are summarized here :
Dunno ... maybe their DACs were designed by ear?? Still, with Mike Moffat (co-founder of Theta), and with old Theta's measuring much better, the poor lab performance is confusing. That said, non-oversampling DACs (Zanden, etc.) also have tend to produce poor lab metrics.
Stereophile's measurements of Schiits Yggy - -which uses the high $$ AD DAC -- are summarized here :
Dunno ... maybe their DACs were designed by ear?? Still, with Mike Moffat (co-founder of Theta), and with old Theta's measuring much better, the poor lab performance is confusing. That said, non-oversampling DACs (Zanden, etc.) also have tend to produce poor lab metrics.
AD1862 is superior DAC (still be in fair price on ebay, DIP packages are dismounted from older electronic, this can be guarantee of genuine) ...
PCM1704 (also superior DAC but overpriced) ... from ebay si problem because they are dismounted using hot-air and this process often degrade parameters or destroy smd chip ...
PCM1704 (also superior DAC but overpriced) ... from ebay si problem because they are dismounted using hot-air and this process often degrade parameters or destroy smd chip ...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.