Twas just a pin, sorry.
Yes, this is possible - moreover, if you use a relatively slow triode with common grid as the load on the secondary, the result may be pretty impressive. I've been doing this for prehistoric chips like tda1541.
Now available on ebay;
Differential Version DSC1 DAC DSD decoding hardware kit Italy Amanero XMOS | eBay
I just ordered one.
Twas just a pin, sorry.
Hmmm...
It's difficult to be sure from the pictures on the ebay listing but it appears that the differential board only has transformers on the output, I cant see anything that looks like an I/V resistor?
How are your listening impressions developing with your build?
Ray
First of all, there are quite a few approaches to the task. One must clearly imagine the whole track, so I'm taking a pause to think the things out.
First, I've got some advises from professionals about reworking the window filter (shift regs and resistors). There is a way to create a full-fashioned FIR filter with non-square window function, that can reduce the noise by some orders. This is the very first topic of research.
Next, I'll obviously replace the CFA to an integrating stage. Too much HF passing through is no good.
And afterwards there's some space for engineering the postfilter.
Alternatively, a common grid tube could be used as the I/V. Loaded by a neat transformer it can behave (in addition) as a first-order lowpass filter.
Also a transformer can be used as in the new Chinese design, followed by I/V resistor, or common grid/base/gate stage, or traditional conversion stage. But don't try those pea-sized modem transformers, it's waste of time.
First, I've got some advises from professionals about reworking the window filter (shift regs and resistors). There is a way to create a full-fashioned FIR filter with non-square window function, that can reduce the noise by some orders. This is the very first topic of research.
Next, I'll obviously replace the CFA to an integrating stage. Too much HF passing through is no good.
And afterwards there's some space for engineering the postfilter.
Alternatively, a common grid tube could be used as the I/V. Loaded by a neat transformer it can behave (in addition) as a first-order lowpass filter.
Also a transformer can be used as in the new Chinese design, followed by I/V resistor, or common grid/base/gate stage, or traditional conversion stage. But don't try those pea-sized modem transformers, it's waste of time.
I'm thinking of using my Silk TVCs on the output, configured for Balanced to SE operation and of course they will act as a LP filter too.
Can I ask if you're HQPlayer? Are you using DSF files or upsampling. What DSD rate are you feeding the DAC?
Also, with your Chinese kit, did you use any of the parts provided or did you just assume they're bad and replace them parts from a known source? I would like to give the supplier the benefit of the doubt but it feels safer to replace the parts.
Cheers
Ray
Ray,
I'm using Audirvana+ on a hackintosh and JLsound's usb board as the source, then chinese AK4137 board as the DSD converter. Tried converting with A+ directly, same noise figure.
As my resistor values differ from the authors, I bought my own bag of Susumu. Caps are NP0 ceramic, Wima polyprop, Nichicon KG electrolytic. 74AHCT chips are from China, look pretty real. Do you suspect them? Opamps bought separately.
I'm using Audirvana+ on a hackintosh and JLsound's usb board as the source, then chinese AK4137 board as the DSD converter. Tried converting with A+ directly, same noise figure.
As my resistor values differ from the authors, I bought my own bag of Susumu. Caps are NP0 ceramic, Wima polyprop, Nichicon KG electrolytic. 74AHCT chips are from China, look pretty real. Do you suspect them? Opamps bought separately.
What DSD rate is the AK4137 delivering?
I've seen enough bad tales to have an inherent mistrust of Chinese parts so for a few extra pennies I replaced the chips.
Ray
Interesting, at DSD256 the noise should be further from the audio band.
I need to try and get my build working to see if I have a similar noise issue. I plan to use HQPlayer at DSD512 streaming to a Beaglebone Black NAA running miero's botic distro and an isolator/reclocker between the Beaglebone and the DAC.
Ray
Without I-V there will be much more distortion.
Whether it sounds subjectively better or worse is another matter.
Has anyone actually got one of these running, listened to it, measured it.
Terry
I have one on order. The first thing I will do is try to derive a schematic from the PCB, particularly the output stage around the transformers.
Holla!
I was in trouble when i see this DSC1 version in comparaison to the first DSC1 board.
On the picture there is twice more 74AHCT595 than the first one.
8 parallel output bits are available for each shift register.
8 shift register totalize 64 output bits.
Maybe i wrong because it's not in my skills but Signalyst told this dac was especially make for running with his HQ player.
Soft features permit "64-bit floating point processing".
I don't think adding more shift register is a good choice in my opinion.
Last edited:
There are no more shift registers, this new board is configured for differential operation.
FWIW I wouldn't mind knowing how Jussi feels about these Chinese copies. If
they actually contacted him. I guess ultimately the hardware will only help sell
the software - so maybe it's a good thing... still it's amazing how many great
ideas from Diya or elsewhere end up as pcb's / kits on ebay in pretty short
order.
Maybe this place is just an R + D lab for Chinese cottage industry.
T
they actually contacted him. I guess ultimately the hardware will only help sell
the software - so maybe it's a good thing... still it's amazing how many great
ideas from Diya or elsewhere end up as pcb's / kits on ebay in pretty short
order.
Maybe this place is just an R + D lab for Chinese cottage industry.
T
I obviously can't speak for Jussi but he is on record as having no interest in producing hardware and he placed his original DSC1 under the CERN Open Hardware Agreement
Signalyst
which says "anyone should be able to see the source (the design documentation in case of hardware), study it, modify it and share it."
CERN Open Hardware Licence - Cernohl - Open Hardware Repository
The Chinese kit producer seems to credit Jussi through reference to his original schematics and recommendation to use HQPlayer. Where they're not playing ball is by failing to respect the CERN licence terms with respect to sharing they're modifications; they should be distributing their updated schematics and, I believe, board layouts.
Ray