I don't think so, and with the variations it's certainly more. John (and his brother) are giving a lot : 16 TDA variations + 2 furthests + tweaks. I readed twice the whole, often came back again and again on some posts. And if you add TDA 1543 as well at the beginning, well the links list would be very long. It shakes a lot and links will direct you towards many things around a dac... from Shakespeare to Shakira. A good book
that needs google to find your kitty. or your triode tube tiger version (sorry that's Max's fault if I try poetry here but without his soul )Defintly one of the coolest thread whatever you discover it lately and need time to read.
Edit : it has evolved towards both the discrete R2R newer ECDESIGNS products and the delicate problem of the way to feed the dac circuitry.
Last edited:
Hi,
Where do you situe the load in that context please and which pins exactly do you talk about ?
IIRC, so take it carefuly and forgett if you already know, I don't know what you readed already but felows will correct me here if not enough acurate if you need more:
ECdesign gave a shematic and values of the I2S network changed, there were diodes in reverse position as well for the pin 2. Also a 100% resistors, also partially atenuated...
The goal was to give around 0.3V, at least less than 0.5 V. So all is about the level of your source. Most of them has already resistors at the outputs.
The idea was given up, despite it gave good results, John wanted more improvments. To go beyond , the TDA 1541A itself was given up to go towards real R2R with discrete construction due to the aviability of smd precision resistors at democratic price since few years. And the way to feed the dac stage was also improved with a better galvanic isolation and the power reduction of the digital signal coming from the hardwares in order to reduce the noise at its best.
With these "fragile" signals, it was shown the layout is important : impedance matching between the boards in a diy multiboards mecano or best one board is a good idea. Screening the I2S signals after few inches is mandatory as if possible individual return paths that can be made for instance with uf-l parts and cables -seen also more often in clock areas-. More than that is also the voltage level of the source you have. You may beginn with zero ohms then reduce voltage inputt and listen to with a good recording as a reference, then increase. Also all that things play as a filter... a simple solder change the result... (I vote for straight copper, a single board, 2 layers can be enough with the speeds involved with the tda 1541A if not more secure if care is taken in the layout pcb drawing around the dac chip and the ground(s) than multilayers made not very carrefully.) Some says those extra care in the I2S input is too much when measuring with a scope... all about the level of the step and th etrade offs I surmise.
John gave some inputs not so long ago (last years iirc about a short road map of the tda1541A signal feeding) so maybe read the last year or the last two years, it should suffice if you need more than the resistors value in front of the I2S . (And I readed the whole thread last year, shame on me)
I'm not a tech so my understanding is low... ! I would like to know if people tried 1:1 RF traffos, butI'm not aware of it, as it may cost a lot. And isolation can be made before with some modern chips that are certainly more linear but it's already above my simple head .
So beginn with no resistors if you can then try good resistors (Sussumu low noise smd, wirewound ayrton perry type if classic size...
just my two cents, your needs level may be higher than mine. Also remember with I2S the shorter the better 4" to 6" distance at max should be your targett... not always feasible due to the complexity (simultaneous mode and all the refinment around the Ws and Bck refinments seen here and there)
ECdesign gave a shematic and values of the I2S network changed, there were diodes in reverse position as well for the pin 2. Also a 100% resistors, also partially atenuated...
The goal was to give around 0.3V, at least less than 0.5 V. So all is about the level of your source. Most of them has already resistors at the outputs.
The idea was given up, despite it gave good results, John wanted more improvments. To go beyond , the TDA 1541A itself was given up to go towards real R2R with discrete construction due to the aviability of smd precision resistors at democratic price since few years. And the way to feed the dac stage was also improved with a better galvanic isolation and the power reduction of the digital signal coming from the hardwares in order to reduce the noise at its best.
With these "fragile" signals, it was shown the layout is important : impedance matching between the boards in a diy multiboards mecano or best one board is a good idea. Screening the I2S signals after few inches is mandatory as if possible individual return paths that can be made for instance with uf-l parts and cables -seen also more often in clock areas-. More than that is also the voltage level of the source you have. You may beginn with zero ohms then reduce voltage inputt and listen to with a good recording as a reference, then increase. Also all that things play as a filter... a simple solder change the result... (I vote for straight copper, a single board, 2 layers can be enough with the speeds involved with the tda 1541A if not more secure if care is taken in the layout pcb drawing around the dac chip and the ground(s) than multilayers made not very carrefully.) Some says those extra care in the I2S input is too much when measuring with a scope... all about the level of the step and th etrade offs I surmise.
John gave some inputs not so long ago (last years iirc about a short road map of the tda1541A signal feeding) so maybe read the last year or the last two years, it should suffice if you need more than the resistors value in front of the I2S . (And I readed the whole thread last year, shame on me)
I'm not a tech so my understanding is low...
! I would like to know if people tried 1:1 RF traffos, butI'm not aware of it, as it may cost a lot. And isolation can be made before with some modern chips that are certainly more linear but it's already above my simple head .So beginn with no resistors if you can then try good resistors (Sussumu low noise smd, wirewound ayrton perry type if classic size...
just my two cents, your needs level may be higher than mine. Also remember with I2S the shorter the better 4" to 6" distance at max should be your targett... not always feasible due to the complexity (simultaneous mode and all the refinment around the Ws and Bck refinments seen here and there)
Last edited:
Thanks diyiggy! I want to go with simple resistors (Dale RN55D). I have already pcb boards of my own design. Each TAD1541 + voltage regulators on its own board, CS8414 receiver on another board. All boards 2 layer, upper copper for a ground plane. Plan to connect the I2S by screened 3-wire cables, ground planes by Cu foil.
Please avoid this Dale, go for RN60 or better Yageo precision resistor (cheaper, better - less noisy). Lighty soldering quantity but brillant and clean . (Joke, it made me years to acheive that, lol ), leads must be in contact with the traces in order to minimize the soldering quantity between the leads and the pcb copper traces.
Edit: make it short and take 50 ohms for the coaxs wires if your pcbs are normalized... and keep it short beteen the USB receiver an the TDA. You can play also with the input of the galvanic isolation before the CS8414 : if your source before is clean enough and has already a not direct output, try with or without the 4 leads isolation traffo. Sometimes with some pcbs and systems the direct inputt of the cs8414 gives noticeable better sounding but the chip is rare and it can fried if a short !
Sorry guys for disgress.
), leads must be in contact with the traces in order to minimize the soldering quantity between the leads and the pcb copper traces.Edit: make it short and take 50 ohms for the coaxs wires if your pcbs are normalized... and keep it short beteen the USB receiver an the TDA. You can play also with the input of the galvanic isolation before the CS8414 : if your source before is clean enough and has already a not direct output, try with or without the 4 leads isolation traffo. Sometimes with some pcbs and systems the direct inputt of the cs8414 gives noticeable better sounding but the chip is rare and it can fried if a short !
Sorry guys for disgress.
Last edited:
Why ? My own experience, subjective ! the best will be wirewound in serie, Vishay has a lot of choice. Do you care about such inductance difference with so short distance ? It's a true question as I don't know but I assume you may not hear the difference in that context... but Dale rn55... I can always hear somthing wrong with it despite I always bought mines from official resselers and it was advised in many forums... maybe an allergy
. try both, take the rn55 if no difference is heard, or better go directly with the Yageos.You don't need autobahn for the digital traces
, tiny & short is ok...
Last edited:
Yes.
But if we consider the info passed on from others, is that still true?.
I dont know, just asking ..
Something to do with -5V as analogue ground?, or a combined ground depending on frequency (and guided by impedance of traces and the traces themselves)..
Someone will know, perhaps they are all gone - as we shall be too.
Tick Tock.. and not that stupid thing.
Ref: Black Mirror (yeh Netflix, but is that not where it is?).
HK
PS: I love your poetry 'iggy, you say 'you think not', I see it differently, take care brother.
But if we consider the info passed on from others, is that still true?.
I dont know, just asking ..
Something to do with -5V as analogue ground?, or a combined ground depending on frequency (and guided by impedance of traces and the traces themselves)..
Someone will know, perhaps they are all gone - as we shall be too.
Tick Tock.. and not that stupid thing.
Ref: Black Mirror (yeh Netflix, but is that not where it is?).
HK
PS: I love your poetry 'iggy, you say 'you think not', I see it differently, take care brother.
Last edited:
Yes I think so, as long as we are talking about the need for a very low impedance to ground should be the only way to treat a TDA1541's analog output. And understandable, as the data sheet from Phillips says so, but listening tests and measurements has shown , that the TDA1541 performs very well, even with more than 25 mV AC on the output. So the cure (high voltage amplification _ high transconductance) turned out to be worse than the disease.
But you are right, there has been some thoughts about using - 5V as ground reference for the analog output and this maybe had some benefits
Thank you so much Hanze, same for you. Purchase the little boat, tda 1541 inside, fishing with wifey, life is short.
At Koldby , There is sure something special about the ground and the tda 1541, i can see from experience what is wrong just looking at a pcb with it, and played a lot where few went about the three voltages and decoupling of it with caps .... to death... but on the digital side of the pcb...that of course has to see inside the chip and the way the ground is shared and the analog side inside and outside the chip. But not too much experience but passive parts on the IV side halas but some circuitry (I stick with opa 861 for IV by choice of no hassle and its low cost)
At Koldby , There is sure something special about the ground and the tda 1541, i can see from experience what is wrong just looking at a pcb with it, and played a lot where few went about the three voltages and decoupling of it with caps .... to death... but on the digital side of the pcb...that of course has to see inside the chip and the way the ground is shared and the analog side inside and outside the chip. But not too much experience but passive parts on the IV side halas but some circuitry (I stick with opa 861 for IV by choice of no hassle and its low cost)
Last edited:
I have two practical questions, sorry if that was discussed to death:
- Supply pins decoupling. Which voltage to which ground? I plan to solder decoupling caps directly to the pins of DIP28 socket. Shall I decouple all 3 voltages do AGND (pin 5)?
- 2mA sinking: I plan to solder two 2.5k resistors directly from pin 28 (+5V) to outputs (pins 6 and 25). Is that ok or shall I take +5V before the last RC filtering cell?
Thanks!
- Supply pins decoupling. Which voltage to which ground? I plan to solder decoupling caps directly to the pins of DIP28 socket. Shall I decouple all 3 voltages do AGND (pin 5)?
- 2mA sinking: I plan to solder two 2.5k resistors directly from pin 28 (+5V) to outputs (pins 6 and 25). Is that ok or shall I take +5V before the last RC filtering cell?
Thanks!
Last edited:
You'd have to make a solid ground that connects analog and digital ground right at the DAC. That is the point all the returns from the local decoupling caps have to go.
Weissi, I have a whole upper PCB layer as my gnd plane. But I want to place the last filtering caps right at the pins. From what I understand, all 3 supplies should be decoupled to AGND, is that correct?
