Sorry, I missed this.
You credit it me with too much knowledge; my grasp of electronics is quite elementary.
On one of my early no-DAC experiments, prompted my another member, I replaced the 47K resistor with a 50K Alps volume pot, which seemed to work well.
Ray
You credit it me with too much knowledge; my grasp of electronics is quite elementary.
On one of my early no-DAC experiments, prompted my another member, I replaced the 47K resistor with a 50K Alps volume pot, which seemed to work well.
Ray
I've been busy with moving coil pre-amps for the last few months and haven't done anything with my DAC except listen to it and enjoy it. But it was raining yetserday afternoon and I couldn't get out into the garden so I wired up a TVC using DIY inductors i got from eBay.
33 Step Transformers Volume Control Passive Preamp | eBay
I mounted the inductors into an old enclosure, and just for testing, mI wired up 9 volume steps (out of 33 steps total). As its just a test rig, the output of my pre-amp plugs into the input of the TVC box. This means, that for DAC play back, the the pre-amp is still taking its output from the Lundahl transformers, and passing this to the input of TVCs.
I wasn't sure what to expect - daisy-chaining transformers doesn't sound like a great idea to me. this was just to see if the relatively cheap TVCs would work. So I turned my passive attenuators up to full, and hit the play button.
I was not prepared for this - the sound has improved in every way!! The bass is strong and there is plenty of top end. The biggest improvement with the No-DAC was the sense of a sound stage that it produced, almost like analog, this expanded the sound stage even further. And pianos sound like there is a piano in the room. So right now I am really excited, and next steps are:
1. remove Lundahls transformer from the chain - take the DSD signal from flip flop directly to the TVC
2. remove passive attenuators from circuit.
So the final signal chain will look like:
digital file -> USB --> JLSounds USB2I2S --> flip flop --> TVC --> power amp.
I am expecting the TVC to perform LPF duty, exactly as Lundahls successfully did. And they do double duty as an attenuator, simplifying the signal chain even further. Will let you know how i go with the next steps. But I am expecting big things!!
33 Step Transformers Volume Control Passive Preamp | eBay
I mounted the inductors into an old enclosure, and just for testing, mI wired up 9 volume steps (out of 33 steps total). As its just a test rig, the output of my pre-amp plugs into the input of the TVC box. This means, that for DAC play back, the the pre-amp is still taking its output from the Lundahl transformers, and passing this to the input of TVCs.
I wasn't sure what to expect - daisy-chaining transformers doesn't sound like a great idea to me. this was just to see if the relatively cheap TVCs would work. So I turned my passive attenuators up to full, and hit the play button.
I was not prepared for this - the sound has improved in every way!! The bass is strong and there is plenty of top end. The biggest improvement with the No-DAC was the sense of a sound stage that it produced, almost like analog, this expanded the sound stage even further. And pianos sound like there is a piano in the room. So right now I am really excited, and next steps are:
1. remove Lundahls transformer from the chain - take the DSD signal from flip flop directly to the TVC
2. remove passive attenuators from circuit.
So the final signal chain will look like:
digital file -> USB --> JLSounds USB2I2S --> flip flop --> TVC --> power amp.
I am expecting the TVC to perform LPF duty, exactly as Lundahls successfully did. And they do double duty as an attenuator, simplifying the signal chain even further. Will let you know how i go with the next steps. But I am expecting big things!!
I've just noted 2 (at least) errors in my post above.
1. TVC does 3 tasks, not 2:
- low pass filter, to get rid of high frrquency DSD noise.
- attenuator
- DC blocker (to get rid of DC on flip flop output).
I missed one step in the signal path:
digital file --> HQPlayer upsampling to DSD256 --> USB output --> JLSounds USB2I2S --> flip flop --> TVC --> power amp.
1. TVC does 3 tasks, not 2:
- low pass filter, to get rid of high frrquency DSD noise.
- attenuator
- DC blocker (to get rid of DC on flip flop output).
I missed one step in the signal path:
digital file --> HQPlayer upsampling to DSD256 --> USB output --> JLSounds USB2I2S --> flip flop --> TVC --> power amp.
Hi hazard5000, was there attenuation circuit in your Dac ? i remember it as FF output to the lundahls.
I am in the mist of building my third dsd no dac, the previous 2 was using the DSD principle board which is acting as a FIR filter. Both sounded fantastic. I want to try another approached which is using the transformer as the LPF and see which is better....in any case i am using a TVC as my main preamp, so i might just implement the no dac into the preamp if it sounds better...
I am in the mist of building my third dsd no dac, the previous 2 was using the DSD principle board which is acting as a FIR filter. Both sounded fantastic. I want to try another approached which is using the transformer as the LPF and see which is better....in any case i am using a TVC as my main preamp, so i might just implement the no dac into the preamp if it sounds better...
Interesting developments Hazard, will look forward to seeing how it works out. I tried my Silk TVCs hooked up to a flipflop board and encountered some noise.
I'm currently working on a differential version of the Signalyst DSC1 which uses transformers so plan to try the Silks again in that context.
Ray
The Silk TVCs allow for differential or SE input and/or output. See the diagrams here;
Silk Audio 23 Step Supermalloy TVC (Pair)
Hi Flowerpot, my original DAC circuit didn't have any attenuation. I biamp my speakers, and have a stereo power amp sitting behind each speaker, and a passive attenuator (John Broskie mono balanced attenuator) sitting atop each "stereo block". So changing volume meant a trip across to each speaker. The beauty of the TVC, apart from the great sound, is that I can adjust volume for both speakers at once. And get rid of the passive attenuator out of circuit.
If you are using a TVC as your pre-amp, then you try this out yourself - but beware, if you don't have a flip flop output, there will be DC on the output of the USB Board (Vsupply/2, usually 1.65V), which could easily saturate the transformer. This will manifest as lack of bass, which is the problem I encountered (see Post #1) before someone suggested the flip flop to get differential output.
The Silks look like high quality units so I am surprised that they didn't work. I would love a pair but they are pricey, these eBay units are much cheaper and do a great job (so far). Let me report back when i direct connect the DSD into the TVC.
Like Naught Boy said. They are transformers, and they don't know the difference between single end and balanced. Both the input, and output, can be either configuration. Just be careful about DC offset if you are using them in a single end configuration.
DSC1 resistor values
I tried a 16-bit sine wave weighted moving average filter with my JLSounds board. I will build the 32-bit equal weight moving average DSD filter in the differential DSC1 kit I ordered from China using 0.1% or better 10K SMD resistors. The 15K resistor value was a concession to the AD844 IV op amp data sheet.
It is funny that someone is selling differential DSC1 kits without any review of how it sounds. For $45usd I can take a chance and at least get some ideas and my own test results.
Unless I hear the highest SQ music ever from this kit, I will also build a 32-bit sine wave weighted average filter. I will also test a JFET buffer between the 32-bit filters and the differential output transformer.
Because the DSD bit-streams were derived from music, which is primarily a complex blend of sine waves, it made sense to do sine wave averaging of the DSD bit stream.
I tried a 16-bit sine wave weighted moving average filter with my JLSounds board. I will build the 32-bit equal weight moving average DSD filter in the differential DSC1 kit I ordered from China using 0.1% or better 10K SMD resistors. The 15K resistor value was a concession to the AD844 IV op amp data sheet.
It is funny that someone is selling differential DSC1 kits without any review of how it sounds. For $45usd I can take a chance and at least get some ideas and my own test results.
Unless I hear the highest SQ music ever from this kit, I will also build a 32-bit sine wave weighted average filter. I will also test a JFET buffer between the 32-bit filters and the differential output transformer.
Because the DSD bit-streams were derived from music, which is primarily a complex blend of sine waves, it made sense to do sine wave averaging of the DSD bit stream.
The moving average filter / FIR was actually done a few years back... not sure if anyone ever remember or notice the electrart DSD principle board. Its an 8 tap or a 8 bit FIR filter... there is a 32 tap/bit version available from yanasoft...
i had both and funny thing is the 8tap version sounded better... till now i have not heard any dac better than the 8 tap filter, i am now building another 32 tap to go with the BBB botic... and also the transformer only version to see which is better.. transformer LPF or 32 tap filter...
i had both and funny thing is the 8tap version sounded better... till now i have not heard any dac better than the 8 tap filter, i am now building another 32 tap to go with the BBB botic... and also the transformer only version to see which is better.. transformer LPF or 32 tap filter...
Returning to the original very simple no-DAC for a moment...
I've made some new single-ended PCBs that are a bit different to the previous versions. The circuit remains as per the original;
plus the mute circuit that evolved on this thread, but I've now split it into two PCBs;
On the right is the 'no-DAC' board. This is no longer designed specifically to stack onto the JLSounds USB board so is a little more flexible. Features include;
I also still have a small number of the flipflop boards available.
Ray
I've made some new single-ended PCBs that are a bit different to the previous versions. The circuit remains as per the original;
An externally hosted image should be here but it was not working when we last tested it.
plus the mute circuit that evolved on this thread, but I've now split it into two PCBs;
An externally hosted image should be here but it was not working when we last tested it.
On the right is the 'no-DAC' board. This is no longer designed specifically to stack onto the JLSounds USB board so is a little more flexible. Features include;
- the option of ufl connections as well as conventional headers.
- smd components for LP filter section
- on-board pads for DC blocking caps
I also still have a small number of the flipflop boards available.
An externally hosted image should be here but it was not working when we last tested it.
Ray
Quick Update:
1. I removed my passive pre-amp from the signal path. Did it make an improvement? I'd be hard pressed to say that I heard a difference. I will settle with the observation that the TVC does a great job and sounds very transparent.
2. I bypassed the Lundahls and took the flip flop output direct to the TVC input. Outcome - I found lots of noise, just like Ray did, when he connected directly to his Silk TVC. What is going on here?
First hypothesis is that Lundahl has a lower f3 (more input capacitance) than the new TVC, and is acting as an (expensive) LPF. However, the noise is very wide band width. It is not hf noise, more like white noise and is quite obvious even in my midrange/woofer (which operates from 650 Hz down). And during quiet parts (eg fade in / fade out) there are whoosing type artefacts. Very strange.
I am upsampling to DSD256 so i would expect any DSD noise to be very hf so this is quite unexpected. I went back and re-added the Lundahl between the flip flop and TVC. Noise is gone again. So - the first hypothesis appears correct. But the Lundahl cannot be filtering noise that is within the audio range, and leaving the music behind. So second hypothesis - the upsampling process creates lots of HF noise which combines with audio signal to create intermodulation distortion (noise) in the audio range. So - I need to build a LPF and test this hypothesis.
The TVC has quite low DCR on the input (12 ohm) so I don't want to make an RC filter because it will attenuate the signal. Luckily i recently bought a few 10mH common mode chokes for use in power supply filters, these have DCR about 0.4 ohms, so I can make a balanced LPF with one of these between FF and TVC. Maybe next weekend.
On a final note - DIYINHK has just released a new USB to I2S board, isolated and capable of DSD512. The JLSounds only does DSD256, so i might get one of these try it out. Hopefully it will push the noise up so far that I can't hear the intermodulation artefacts!
Isolated XMOS 768kHz DXD DSD512(DSD1024) high-quality USB to I2S/DSD PCB - DIYINHK
What happened to this thread, did it just die or moved on to some new thread? I'm very interested in no-dac approach and real-time conversion to dsd, so could anyone point me in the right direction, what are the achievements so far in this field?
Also @nautibuoy - what USB board would you recommend to use with your boards, and do you have more to sell?
Also @nautibuoy - what USB board would you recommend to use with your boards, and do you have more to sell?
I only used the JLSounds I2SoverUSB board, which worked very well. I'm currently exploring the use of a Beaglebone Black/Botic combo do deliver DSD to my ongoing DSD experimentation (see the Signalyst DSC-1 thread).
I still have a small number of no-DAC PCBs available, see this post;
http://www.diyaudio.com/forums/digital-line-level/273474-best-dac-no-dac-160.html#post4834249
Note that the flipflop board is designed to mount onto the headers of the JLSounds board.
PM me if you're interested.
Ray
Hi blackhole.
I've been looking at your audio-Linux site this morning and see you have added your updated no-dac design to the front page;
http://www.diyaudio.com/forums/digital-line-level/273474-best-dac-no-dac-156.html#post4774920
Does that imply that you sorted out the noise problems you were experiencing?
Ciao
Ray
I've been looking at your audio-Linux site this morning and see you have added your updated no-dac design to the front page;
http://www.diyaudio.com/forums/digital-line-level/273474-best-dac-no-dac-156.html#post4774920
Does that imply that you sorted out the noise problems you were experiencing?
Ciao
Ray
Not yet. I don't know what could be the problem. I was thinking to various possibilities:
1) Use the new card with isolation
Isolated XMOS 768kHz DXD DSD512(DSD1024) high-quality USB to I2S/DSD PCB - DIYINHK
2) Disconnect the ground on the output (the 47k resitor)
3) There is sometime a costant low volume hiss and some noise. Maybe the Russian capacitor with metal housing?
Maybe I will post an image of the assembled dac.
1) Use the new card with isolation
Isolated XMOS 768kHz DXD DSD512(DSD1024) high-quality USB to I2S/DSD PCB - DIYINHK
2) Disconnect the ground on the output (the 47k resitor)
3) There is sometime a costant low volume hiss and some noise. Maybe the Russian capacitor with metal housing?
Maybe I will post an image of the assembled dac.
This morning I assembled one of my new single-ended no-dac PCBs. The circuit and values are exactly the same as those you posted. I plan to hook it up to my BeagleboneBlack/Botic HQP NAA a little later so will be interesting to see how it works out.
An externally hosted image should be here but it was not working when we last tested it.
