True. It would be the "Final TDA1541".
Project Working Title: "Rememberance of sounds past".
To get the most out of WM8805 needs software. I think I have something workable somewhere but I prefer to avoid anything not "pure hardware" hardware.
Mind you, I bought three working sony CD-Players for under 50 Bux each in less than a week, one TDA1541 in each. They are 750 and 950. There is still some mining for TDA1541 to be done.
Minimal. Before our output Latch (74AUC16374) we add a 74LVC86 etc. to invert data in the digital domain.
As I suggest a 16-Bit latch to reclock the outputs (before signal conditioning) we have many available pin's.
As we have two banks, we make one bank for Stereo Single chip with the second bank disabled.
For balanced, we set a few jumpers etc. and now each chip is mono balanced, OR to avoid channel differences, we use one chip as stereo + and the second as stereo - .
I think crosstalk is overrated. Try LP. Best soundstage ever, crosstalk at best -26dB mid band.
BTW, my aim for this would to be ending up with something anyone can order from JLCPCB or similar Chinese small QTY PCBA houses and add his/her/it/whatever TDA1541, decoupling capacitors etc, et al.
So a semi-kit in the public domain under GPL or similar. We design it, make some and test, because we want to, then we let the child fight for itself...
Thor
Seeing that -5/6V is always the least cared for supply, doubtful.
I really want to know what this MF -5V does in the chip. It bugs me totally to not know.
Thor
If i find dead tda chip, i may be able to decap it, and under microscope we can actually see whats going on in that domain. Should bring one step closer to bettering the project.
Yeah, there are devices for them to be mined from, but i'm a firm beliver desoldering these buggers degrades/takes a toll on the chip. For example pcm63p-y sounds wastly different to NOS pcm63p-k, on the same board, and it shouldn't as much. And it's only found desoldered, hence the theory. Perhaps careful snipping/dremel of the pcb around the pins, and fine file the pins.
Yeah, there are devices for them to be mined from, but i'm a firm beliver desoldering these buggers degrades/takes a toll on the chip. For example pcm63p-y sounds wastly different to NOS pcm63p-k, on the same board, and it shouldn't as much. And it's only found desoldered, hence the theory. Perhaps careful snipping/dremel of the pcb around the pins, and fine file the pins.
I just had a look-see for a simple current conveyor + resistor to 5V I/U conversion.
We need around 8mA bias current and this allows 250 Ohm for I/U conversion. This then has very low HD.
So we get around -2.5V bias below +5V and around 350mV RMS signal.
That SHOULD be enough to add simple gain circuits according to taste to get any specific sound you like.
Thor
We need around 8mA bias current and this allows 250 Ohm for I/U conversion. This then has very low HD.
So we get around -2.5V bias below +5V and around 350mV RMS signal.
That SHOULD be enough to add simple gain circuits according to taste to get any specific sound you like.
Thor
Yes, I'd love to have a good, high resolution colour look at the die and bondwires.
Use the right desoldering gear.
Burn in?
Seems excessive. Old PCB's also used low temperature Tin/Lead solder.
Thor
Those kinda device cost millions , more than many will ever earn in their life , and you should have the skills to get it right , if I were Rudy I will laugh as much as I do , sorry folks 😎
reinventing the wheel is a human classic behavior , but the wheel is still a wheel 🙂
.
Last edited:
By NOS i mean never soldered. Shouldn't be affected by burning in thousands of hours played.
One i'm talking about was desoldered properly by a well known respected member here. So shouldn't be an issue of bad desoldering. I think it thermally stresses it once soldered.
@fabrice63 not really, for these purposes acid should be fine, without damaging the die and bond wires. Doesn't hurt to try on dead chips in any case.
No. A bunch of typewriter pages with hand drawn (professionally mind you) graphics covering the TDA1541 internal design including a full schematic and explanations of detailed function of block.
Thor
I used many tears ago previous version of the same manufactured many times before. Without saa chip. It was with another cpld. And without option of -data. For tda1540and for tda1541a. also I tried for balanxed mode eith inverter on data line since it was not an option. And it was always working good. I think that for tda1540 limit was 384KHz without any issue. And limit for tda1541A was 192KHz.
I can say from listener point, fpga and cplds, introducing the specific sound signatures. I think that is because they are complex, and too densed in a very small space. The consumption of energy is high. Internal clocks "sharing" is present, probably ground bounces. In these audio applications these devices are employed very very small percent of their real potential. My guess... Not an expert...
On the other hand, they are small, space savers, handy and available.
I’ve built several tda1541a dacs about a decade (or more) ago, and left this game due to multiple intercontinental moves. My personal reference was a ‘borrowing’ greatly from ECdesigns, plus 6072a in Gomes stage configuration (thorsten may recognize). It worked well and sounded good but I packed too much into a single enclosure so had to deal with hum and RF. Source was a squeezebox connected via i2s.
Given European weather I’m getting back to it and following this thread with interest. The dialog of the past couple of days is quite exciting as I was wondering what the state of the art is since I left in 2013 or so. Seems to me that not that much has changed, other than consensus on i2s conditioning and superiority of DEM sync to WS (grundig).
New surrounding hardware has arrived, like fifo buffers. Biggest change in the deployment of the tda1541a seems to be the move to simultaneous mode. Unlanded seems to be decoupling caps sizing, form factor and type (still under debate). The recent suggestions on current paths into -15v seem novel and interesting.
I was wondering what the latest thinking is around using multiple (2, 4 of the same batch) tda1541a in various setups, assuming SE output. Parallel, balanced or the Sony ‘interpolated’ setup. Or do none of these bring significant added value?
Given European weather I’m getting back to it and following this thread with interest. The dialog of the past couple of days is quite exciting as I was wondering what the state of the art is since I left in 2013 or so. Seems to me that not that much has changed, other than consensus on i2s conditioning and superiority of DEM sync to WS (grundig).
New surrounding hardware has arrived, like fifo buffers. Biggest change in the deployment of the tda1541a seems to be the move to simultaneous mode. Unlanded seems to be decoupling caps sizing, form factor and type (still under debate). The recent suggestions on current paths into -15v seem novel and interesting.
I was wondering what the latest thinking is around using multiple (2, 4 of the same batch) tda1541a in various setups, assuming SE output. Parallel, balanced or the Sony ‘interpolated’ setup. Or do none of these bring significant added value?
"superiority of DEM sync to WS (grundig)"
There is a common misbelief that DEM frequency should be >= 4x WS. I think the DEM decoupling capacitors are there to filter whatever frequency the DEM oscillator runs on. If you prefer low DEM oscillator frequency (50 Hz) simply use higher value capacitors. In NOS mode where the DEM is synced to the 44.1 kHz WS frquency as per the "Grundig" solution, just use a suitable film capacitor. Note the impedance of the DEM decoupling pins are in the order of 10 to 100 Mohms. It is just a simple filter function with the internal R and external C. I would like to see measurements or experiences of others about this sync/no sync/DEM frequency question. For now I use the so-called "Grundig" style DEM oscillator in NOS mode (44.1 kHz frequency synced to WS) and the capacitors are 220 nF film type. It works well.
Whoever stated first that DEM oscillator frequency should be >= WS frquency is due to give a technical explanation.
There is a common misbelief that DEM frequency should be >= 4x WS. I think the DEM decoupling capacitors are there to filter whatever frequency the DEM oscillator runs on. If you prefer low DEM oscillator frequency (50 Hz) simply use higher value capacitors. In NOS mode where the DEM is synced to the 44.1 kHz WS frquency as per the "Grundig" solution, just use a suitable film capacitor. Note the impedance of the DEM decoupling pins are in the order of 10 to 100 Mohms. It is just a simple filter function with the internal R and external C. I would like to see measurements or experiences of others about this sync/no sync/DEM frequency question. For now I use the so-called "Grundig" style DEM oscillator in NOS mode (44.1 kHz frequency synced to WS) and the capacitors are 220 nF film type. It works well.
Whoever stated first that DEM oscillator frequency should be >= WS frquency is due to give a technical explanation.
If you are starting again, I recommend a pair in balanced operation with simultaneous mode and bit clock stopped after the data is latched, plus the other developments like dem clocking and digital attenuators.
I/V for me is Sowter transformers (model 1465) but too expensive for most so gets slagged off for other reasons to justify not trying one from those who have a closed mind.
https://www.sowter.co.uk/specs/1465.php
Given the current street value of the tda1541a’s I have in the drawer, I may have to accept the costs of getting the most out of them (4*tda1541a 1998 Taiwan, 2*s1 1989 Holland, 2*s2 1998 taiwan)….
But indeed clever implementation comes first…