Unfortunately the PM bounces back.
It appears that account is inactivated.
Do we have any members here from Taiwan who can try to find his contact info??
It appears that account is inactivated.
Do we have any members here from Taiwan who can try to find his contact info??
Last edited:
Entirely up to you.
Not here, plus buying it is much more complicated.
Thor
easy to solve that with mulltiple entry on the board : header for your board anda one more for the JLsounds, and few SMA/uf-L, it is a trade off view on the EMC/EMI side, although one can layout smd jumpers to enable the front-end entry he wishs. So male header on the third party USB board and female header on the core TDA1541A pcb for minimal one story stack (should be 5 mm total heigth with the flatest headers.
That means also one has to know the spacing between the chinesse boards. I bet it is 2.54 mm 🙂 . As we have a picture of it it is easy to layout the I2S and I2StoSim lines in the rigth order vis à vis of the third party pcb.
That means also one has to know the spacing between the chinesse boards. I bet it is 2.54 mm 🙂 . As we have a picture of it it is easy to layout the I2S and I2StoSim lines in the rigth order vis à vis of the third party pcb.
Last edited:
Is he chineese speaker or englih one's, both ? I know few members living in south Asia area whom can try !
???? You either have isolation, or you do not.
Sure that's how I do it.
CMOS is 74XX74. There are many logic families and devices to pick from.
If using ECL reclocking anyway, then a 74LVC161 would probably be the best choice for a divider.
I don't have the logic symbols for ECL in TINA-TI, so someone else would have to draw out this logic.
PSU needs two diodes thermally coupled to the TDA1541 and the ECL Latches doing reclocking and driving TDA1541 need RCR on each output (as show before from the AMR CD-77).
With MC100EP52 as re-clocker we need 5 pcs (but that is enough for a balanced DAC with dual TDA1541) and 3 pcs MC100EPT22 or 6pcs MC100EPT20.
If used they should be on the core module with TDA1541 and can share the TDA1541 logic supply, at the cost of an extra ~300mA+ current draw (ECL is hungry) at 3.3V or 5V.
I sourced multiple R Core transformers from Ali, 115V/230V with screen. Inexpensive. Appear decently made too.
No SPDIF, so not useful for me.
I looked at all these and only bought chinese AFTER determining that cost/hassle here in South east asia were just not sensible.
Thor
I follow every development and "non-textbook" ideas with great interest. Thanks for their contribution to all.
My medium-term goal is to build a dual balanced DAC on @batteryman PCB (done), build a low noise PSU, finish the half-ready I/U permalloy transformers, put everything in a neat box, and enjoy. I have also a USB to dual I2S to SIM board. I will add also the I2S TTL to CML attenuator as advised by @ThorstenL . Perhaps I will add the DNL compensation, too, just for fun. At the moment I don't aim for more. Perhaps I will be interested in an "open source" PCB in the future, but for now I want to finish what I have started.
My medium-term goal is to build a dual balanced DAC on @batteryman PCB (done), build a low noise PSU, finish the half-ready I/U permalloy transformers, put everything in a neat box, and enjoy. I have also a USB to dual I2S to SIM board. I will add also the I2S TTL to CML attenuator as advised by @ThorstenL . Perhaps I will add the DNL compensation, too, just for fun. At the moment I don't aim for more. Perhaps I will be interested in an "open source" PCB in the future, but for now I want to finish what I have started.
I would just do SMA. They are small enough to be easy to fit and chunky enough to stand some use.
"Core Board" concept:
So SMA for:
BCK
LE
DL
DR
CLK (to be divided down for DEM, select MCK or BCK to taste)
MCK (for reclocking)
All CMOS/TTL level
On board:
74LVC/LCX161 Divider for CLK (programmable 2...16, can handle 50MHz clocks)
3pcs MC100EPT22
or
6pcs MC100EPT20 as level translators
5pcs MC100EP52 Flip-Flop's (preferred for differential inputs)
or
5pcs MC100EP31 Flip-Flop's
Signal conditioning RC after the ECL Flip Flops
TDA1541 with all decoupling capacitors etc., up to 1,000uF(is) Os-Con's
Power supplies either fully external, or a 20V TL431 & BD140 et al chain for Audio (+5V....-15V) and separate 7805 etc. for the digital side including the ECL section connected between DGND and +5V - DGND & AGND linked with choke and diodes and optional footprints for resistors and capacitors.
Plug in option for analogue section if it shares the +5...-15V supply.
This makes a stand alone DAC section that can be paired with any frontend using BIY modules on Veroboard.
The DAC section is basically maximally tricked out, clocks, input multiplexing and input devices are up the implementer.
Maybe add a "veroboard" like section with breakoff and SMA footprints (and perhaps some drivers) to limit what can be bolloxed up.
Such a PCB would have to manufactured with all SMD on board and tested, so it would need to cost a substantial amount. It's a bit much for DIY assembly.
I'd call it likely 100 Bux each more as someone has to put in a lot of legwork.
I just don't see much demand for that sort of thing...
Thor
Last edited:
Many thinks it as only gnd isolation, then because it is async so reclocked say it is isolated because of a language trick as it is only the time domain that is really isolated!
So there are differences... should we have a chip that isolate the I2S lines ? I readed the full speed USB isolators indiustrial ICs from Tithat isolate also the two signals lines (the differential) can be percieved not for the best.
No, not if there is any way to avoid this. I worked with them and prefer to keep them far away (breaks out the garlic necklace, holy water, hammer and wooden stake, silver bullets for his gun and makes the sign of the cross).
That a load of Tosh.
USB has 3 speeds. Low speed (1.5MbpS), Full speed (12MbpS) and High Speed (480MbpS).
USB has two types of "Audio Class", the original Audio Class (limited to 96/32) and USB Audio Class 2 (not as such related to USB 2.0) which needs AT LEAST USB 2.0 High Speed operation to work (single endpoint currently limited to 768kHz/32Bit) but can scale to faster USB systems (currently no ready implementation OTS available).
All USB2.0 Isolator IC's isolate the data lines only. NON isolate grounds or Vusb. As the latter is a totally trivial job IC manufacturers leave that part to those who implement the IC.
Until very recently however most USB isolation IC's were limited to USB 2.0 Full Speed meaning they do not work for USB Audio Class 2.0 devices.
I have also seen Schenese "USB Isolators" that linked input and output ground with a ground plane and thusly not isolated at all, clearly these were designed by illiterate peasants who had no clue what they are doing (actually that lot designs most PCBs in Scheena).
As a rule, any USB isolator IC is a mix of USB receiver, Isolation barrier and re-drivers (bidirectional due to the nature of USB).
So electrically they appear as USB Hub/Re-driver (aka USB Regen). If you hear a difference from inserting them, then either:
1) The re-driver improved signal integrity leading to fewer corrupted packets at a level where the difference between original signal and interpolation becomes material
2) The isolator broke the only link to earth in the USB system and now the USB side is floating and subject to severe interference beyond design limits causing enough packet corruption and interpolation of corrupt data that difference becomes audible.
3) The isolator contains isolated power which is used by the USB device for clocking etc. and the isolated power is badly designed (e.g. using a OTS 5V ISO module with several 100mV PP Output noise and not applying extra filtering)
That's about it. Isolators are not magic. Used where non is needed they can make the situation worse.
As a results iFi's USB Isolator had a switch that allowed the selection of "Full Iso", "Soft ground" (linked via RC) and "Hard Ground" (e.g. no isolation at all). It was also USB 3.0...
Thor
Last edited:
not cheap are the SMA wires if you can not source them in Asia (at Mouser the least honerous is 15 euros and if you needs 3 or 4 of them... it beecomes enormous ! But easy to mak the foot prints on the boards : everyone chooses, free world = smiles !
@thorten, I of course tallked about the 480 Mbps.
@thorten, I of course tallked about the 480 Mbps.
Back when I designed a clock PCB for diyhifisupply I made a footprint for a combo that could take 2-pin XH or SMA.
Thor
When I asked him (many years ago) he said he didn't have it anymore.
Given the level of interest, this document HAS to surface, somebody has to find it, come on!
Sure , normally a serious job is done on serious bases , how someone would change the rules if he don't know ( perfectly ) them in the first place ??
.
.
Sorry I mismatched the values on the schematic. 🙁
Dem pin Cmsb is 1uF (all others are correspond) to the 176.4KHz Fo value...
.
That is what i can recommend to, But that is another topic conected more to the input interface, not to the dac board?
I tried many of the USB isolator so called hi speed and full speed.
JUST one worked. This one from Silanna semiconductors.
This was many years before new USB isolators from TI and i think that we have one new too from LT or AD, not sure...
I can report that the sound was improved with USB digital islation. Everything was more "analog", As a place different Phono head into the headshell...
I had 2 versions of Silanna isolators, one origimnal PCB with switching DC-DC convertor on the board. And other with custom separate classic power sypply.
BUT the silanna digital isolators are obsolete now.
.
I think that the same sonic different, better results could be obtained with the recent non-obsolete parts?
But the tis the topic for say Amanero Nest Board? 🙂
.
I am talking about digital isolation just before DAC chip inputs (with also isolated all incoming sygnals that DAC system using.
I tried it also with ADUM1100 isolator for each line BEFORE recklocking prior to DAC and the sound became also more analog.
This circuits are isolating grounds and signal. And of course that can be done with little HF transformers too. Actually ADUM1100 have one "transformer inside.
I isolate the MCK too besides SCK, DATA, and LE.
.
In this way the isolation of computer noises and "pollution" from the usb ground is addressed to USB-I2S cpld or FPGA noisy interfaces, and futhure in the paths I2S to TS transcriber module. That can be power suplied by the same transfprmer and PS unit.
On that part is MCK oscilators too...
.
Next isolation is for DAC only. To use own supply as it should be. And the recklocking is welcome because too...
.
I will scethch the concept schematic of that...
The signals are just cleaner when no optical disc is used because there is no need of errors corrections. When using a sd card or a HDD, you're already not Red Book anymore. And if one is just talking about the 16/44.1, there were also 16/98, more often than one thinks, they are often in all digital libraries.
I2S is just a transport protocol (local and short distance), whatever the source, what you want is no clandestine passengers . Anyway it is async nowadays so it is "reclocked" at the streamer place close to the DAC pcb. No doubt it is cleaner than optical CDs. The SD Trans transport you use is so not Red Book. If the question is why it is assumed to sound better than an USB streamer or not is another question. And yes the SD Trans is an async "local" streamer, just the source (sdcard) is closer than with a streamer (HDD NAS for instance)
I2S is just a transport protocol (local and short distance), whatever the source, what you want is no clandestine passengers . Anyway it is async nowadays so it is "reclocked" at the streamer place close to the DAC pcb. No doubt it is cleaner than optical CDs. The SD Trans transport you use is so not Red Book. If the question is why it is assumed to sound better than an USB streamer or not is another question. And yes the SD Trans is an async "local" streamer, just the source (sdcard) is closer than with a streamer (HDD NAS for instance)