Hi, It can be a good Idea to have a differential R2R analog output before OP to cancel DC offset, but the number of IC will be doubled...
The low pass output filter has a 12 db/octave. It is a 2th order filter.
Regards
Stefano
The low pass output filter has a 12 db/octave. It is a 2th order filter.
Regards
Stefano
Saluto Stefano. Thanks for joining in. Nice DAC 🙂
Please feel welcome to let us know about your general design aims and choices for your elegant discrete R2R.
Please can you kindly consider sharing any measurements? If you can upload any FFT/IHD plots it will help @batteryman by supplying a reference to help with his own measurements. This may save us confounding unknowable measurement uncertainties with the actual DAC performance.
Are you working on any new designs or different resistors for the ladder?
Grazie amico mio.
I use a Sowter bifilar wound I/V transformer (balanced in, balanced or SE out) this means the magnetising effects from non zero dac outputs cancels out.
It works well on my balanced TDA1541 with no need to inject the 2ma offset.
Presumably the output at zero signal is 2.5v?
What is the best load impedance for the direct ladder output?
IT WORKS.
Finally reached the listening and testing phase. (left the supplied TL082 opamps fitted but have some LME49720s to try)
It worked perfectly first time (so the DIR9001 was set correctly) and with both 44 and 48k coax inputs.
You can see my Lipo psu in the photo, which supplies + - 8 and 16v to the dac. Not yet fitted the sockets for charging the eight 3.7v 8ah cells.
I have to say I am impressed with how clean the audio is considering there is no oversampling and only the dac's 12db 24Khz filter. (Solo male classical vocals are great for showing up artefacts.)
(CD source is a Cambridge Azur 651c and 48k was from a Nvidia Shield Pro and SML USB-spdif adaptor streaming my favourite professionally recorded Youtube videos (Voces 8, Voices of Music, Apollo's Fire etc)
The Dir9001 is run from the dac's 5v supply but I will give it a separate supply
It does have a hard act to follow - the Dacmagic 1 (two TDA1305 CC bitstream dacs in balanced arrangement) is a very sweet sounding dac
Finally reached the listening and testing phase. (left the supplied TL082 opamps fitted but have some LME49720s to try)
It worked perfectly first time (so the DIR9001 was set correctly) and with both 44 and 48k coax inputs.
You can see my Lipo psu in the photo, which supplies + - 8 and 16v to the dac. Not yet fitted the sockets for charging the eight 3.7v 8ah cells.
I have to say I am impressed with how clean the audio is considering there is no oversampling and only the dac's 12db 24Khz filter. (Solo male classical vocals are great for showing up artefacts.)
(CD source is a Cambridge Azur 651c and 48k was from a Nvidia Shield Pro and SML USB-spdif adaptor streaming my favourite professionally recorded Youtube videos (Voces 8, Voices of Music, Apollo's Fire etc)
The Dir9001 is run from the dac's 5v supply but I will give it a separate supply
It does have a hard act to follow - the Dacmagic 1 (two TDA1305 CC bitstream dacs in balanced arrangement) is a very sweet sounding dac
Attachments
Last edited:
A pity that Stefano says there is some restrictions shipping electronics out of EU at this time. I see that he has nice products at nice price.
There was no problem shipping to the UK and we are not in the EU now.
If his items are on ebay, the ebay Global Shipping Program should make it possible to ship to many countries.
Yup i contacted him through ebay last week and he informed me about the restriction.
Dear Batteryman, thanks for you words.
You can run DAC also at 24 bit 96 Khz via USB to spdif converter or a USB to I2S.
With this solution you can play Hires files in NOS pure 24bit.
Thanks again.
Stefano
Dear All,
because the URSS<->Ucraine war we have a lot of restriction/verification at the border for electronic parts.
Many time they send it back.
Often I had to ship twice times with more time costs at my charge.
Because I do this job not for to gain, but for passion, we have to check how we can organize shipping outside EUR zone.
DHL appear better than UPS but it cost a lot.
Another solution is I provide you the project then you will build directly on your site.
In the next time I will answer to your answer on Ebay.
Regards
Stefano
Thank you. You have made a brilliant dac - it sounds better than I ever expected.
I cannot get over just how clean and pure the audio is and I now realise that over sampling and steep LP filters are nothing but tricks to overcome poor dac chip sound quality.
The sound is more transparent and detailed wihtout adding anything: it gave me goose bumps today listening to Saint Saens Dance Macabre by the Polish 'Orchestra Kamerton', which I use as one of my reference recordings. (optical spdif from my LG Oled tv)
I will post some Picoscope screenshots when the ADC arrives in a few days.
Are you using an output filter at all? If not, the Picoscope should show finely stairstepped waveforms coming out of the dac. Of course stairsteps in the time domain appear as harmonic distortion when viewed in the frequency domain. A steep LP filter may be needed to correct the distortion. The good news is that a much less steep filter would probably be fine if you oversample the dac a little. Say, maybe upsample 44kHz to 88kHz...
Of course I might sort of contradicting your above hypothesis, depending. If so, sorry. Not meaning to be offensive. Just trying to explain some technical considerations.
Of course I might sort of contradicting your above hypothesis, depending. If so, sorry. Not meaning to be offensive. Just trying to explain some technical considerations.
There is an onboard 24kHz LP 12db filter, that's all, apart from passive filtering in the Arcam Alpha 2 and modified ML Aeons, + ancient ears. (12khz cut off).
Upsampling would defeat the whole point of the dac - minimum digital processing, hence no CPLD, and no digital filters.
I am only interested in the sound and so far, very impressed. I plan to replace the 5v regulators (LM317 based) with AMS1117-5 low noise, low dropout type at the weekend. (one for the dac and one for the Dir90001.)
Might turn out the LM317 regulators sound better than AMS1117. If so, then NJM7805 might sound a bit better yet. Is this for Vref in particular?
Regarding upsampling, have one of Andrea's DAC_Lite dacs here. Have played CD rips and hi-res on it.
High quality upsampling of CD rips is similar to playing hi-res. Not exactly so, mainly because of bit-depth. But the dac sounds good playing at higher sample rates too. Nothing wrong with that. In truth, modern ADCs used to create digital audio sample analog audio with only a few bits of resolution, maybe 4 or 5 bits, but at a very high sample rate. That's called 'raw' mode. From there its converted to hi-res, CD, DSD, etc. So there is always some DSP involved in almost any digital music. Given that understanding, as long as its done very well there shouldn't be anything wrong with upsampling. Seems to me the things you don't like about the sound of upampling and digitally filtered dacs aren't necessarily in entirely a function of those two factors. Ever hear of 'substrate coupled noise?' How do you know that isn't responsible for most of what you don't like about upsampled dacs? Or maybe its something else, or a combination of things you mostly don't know about?
Why would I suggest that? Because I also have an AK4499 dac here with DSP that upsamples and does hardware conversion of CD to DSD. Andrea and his friends came to visit. We listened to both dacs, his and mine, on large panel electrostatic speakers (Sound Lab) which are very revealing. When we played back some hi-res material Andrea remarked that both dacs sounded almost exactly the same. Of course, he also said he never heard an oversampling dac like mine before. Perhaps he heard some bad ones and assumed that oversampling was the problem. Maybe other people have had similar experiences?
Regarding upsampling, have one of Andrea's DAC_Lite dacs here. Have played CD rips and hi-res on it.
High quality upsampling of CD rips is similar to playing hi-res. Not exactly so, mainly because of bit-depth. But the dac sounds good playing at higher sample rates too. Nothing wrong with that. In truth, modern ADCs used to create digital audio sample analog audio with only a few bits of resolution, maybe 4 or 5 bits, but at a very high sample rate. That's called 'raw' mode. From there its converted to hi-res, CD, DSD, etc. So there is always some DSP involved in almost any digital music. Given that understanding, as long as its done very well there shouldn't be anything wrong with upsampling. Seems to me the things you don't like about the sound of upampling and digitally filtered dacs aren't necessarily in entirely a function of those two factors. Ever hear of 'substrate coupled noise?' How do you know that isn't responsible for most of what you don't like about upsampled dacs? Or maybe its something else, or a combination of things you mostly don't know about?
Why would I suggest that? Because I also have an AK4499 dac here with DSP that upsamples and does hardware conversion of CD to DSD. Andrea and his friends came to visit. We listened to both dacs, his and mine, on large panel electrostatic speakers (Sound Lab) which are very revealing. When we played back some hi-res material Andrea remarked that both dacs sounded almost exactly the same. Of course, he also said he never heard an oversampling dac like mine before. Perhaps he heard some bad ones and assumed that oversampling was the problem. Maybe other people have had similar experiences?
Last edited:
Dear, attached you will see the R2R Dac running @176Khz 24Bit No Oversampling 25Khz Full Scale Sine Wave generated with Audacity -> USB -> Spdif -> I2S -> R2R -> Analog. Basically the concept of this DAC is: No Oversampling, No Noise Shape Feedback, No Algorithm or CPU. Pure Bit as the TDA154x family but featuring up to 24 bits. As out italian Grandmaster of Sound Bartolomeo Aloia he thought (and I believe): Feedback is good for instrument but bad for ears. 🙂
Attachments
Why all the format conversions? They add nothing to the quality of the sound. Why not just use the 24 or 32 bit parallel output from Audacity or other digital audio source and dispense with all the conversions and shift register nonsense?
Maybe i’m missing something but accurate clocking would require something between to gate the delay in serialising the parallel pin programming. Then you want something to isolate noise and counter any isolation delay.
So how do you explain the clean scope screenshots? (The dac requires Philips I2S which has one bck clock delay before data stream)