...
So, forgive me for not reading everything, but 546 pages is a little too much for meIs there a complete, working DAC project with good description, which I could build at home (etch my own PCB) ? I will appreciate any help
Could be NOS or not.
I hear ya!
http://www.diyaudio.com/forums/digi...e-nos-dac-using-tda1541a-499.html#post3758887
Whew. Back on Track!
and by that, I mean this!!!
http://www.diyaudio.com/forums/digi...any-good-tda1541a-dac-kit-55.html#post3505444
Welcome Holgin friend!
(may the 1541 bite yo ***!)
Last edited:
caps around TDA 1541
Hello,
I did get interested in buiding Doede Douma DDDAC 1794 but there are some changes iun the pipeline that take longer than expected so i did decide to modify my Curcio dac that use the tda1541.
This one uses a tube output stage. The high voltage supply was pretty straight forward 220volts ac secundairy transformer, 4 standard diodes, 100mF cap, 470 ohm and another 100 mF before entering the regulation part which is on the mainboard.
Did put a new bigger transformer with a center tap, a 6x4 tube rectifier, choke input with a common mode connected lundahl LL2743, 22mF Black Gate, a ll1668 and then another 22mF BG BIG IMPROVEMENT.
The other voltages needed for the circuit will also get a new power supply with choke input.
Some resistors on the mainboard ( 3 watt metal oxide) will be replaced by Mills wirewound. That did turn out very good in my Daniel preamp ( also Curcio design using similar topology)
Wanna change the 3 electrolytic caps close to the dac chip into Oscon.
The14 100nF film caps are wima mkp10. Some people say they should be higher value but space is limited. Can get extra '' space '' by stuffing them on both side of the circuit board.
In the dddac 1794 some people are raving about Kemet PPS caps. Because of newer? technology and lower voltage ratings you can have more nF than the wima for the same dimensions. You can 680nF 63 volts dc or 270nF 100 volts dc with can be soldered without any problem.
I did read some companies did put bigger values like 1mF but only for some of the caps. Do i need good sounding caps or do they have to be technically advanced to give proper results? Maybe some paper in oil caps might not be good because of leakage current which tyhis chip might not tolerate at all.
Any thoughts are welcome.
Sincere greetings, Eduard
Hello,
I did get interested in buiding Doede Douma DDDAC 1794 but there are some changes iun the pipeline that take longer than expected so i did decide to modify my Curcio dac that use the tda1541.
This one uses a tube output stage. The high voltage supply was pretty straight forward 220volts ac secundairy transformer, 4 standard diodes, 100mF cap, 470 ohm and another 100 mF before entering the regulation part which is on the mainboard.
Did put a new bigger transformer with a center tap, a 6x4 tube rectifier, choke input with a common mode connected lundahl LL2743, 22mF Black Gate, a ll1668 and then another 22mF BG BIG IMPROVEMENT.
The other voltages needed for the circuit will also get a new power supply with choke input.
Some resistors on the mainboard ( 3 watt metal oxide) will be replaced by Mills wirewound. That did turn out very good in my Daniel preamp ( also Curcio design using similar topology)
Wanna change the 3 electrolytic caps close to the dac chip into Oscon.
The14 100nF film caps are wima mkp10. Some people say they should be higher value but space is limited. Can get extra '' space '' by stuffing them on both side of the circuit board.
In the dddac 1794 some people are raving about Kemet PPS caps. Because of newer? technology and lower voltage ratings you can have more nF than the wima for the same dimensions. You can 680nF 63 volts dc or 270nF 100 volts dc with can be soldered without any problem.
I did read some companies did put bigger values like 1mF but only for some of the caps. Do i need good sounding caps or do they have to be technically advanced to give proper results? Maybe some paper in oil caps might not be good because of leakage current which tyhis chip might not tolerate at all.
Any thoughts are welcome.
Sincere greetings, Eduard
The 1541A prefers the polyester filter caps - MKS in Wima, not MKP - current values are 1uF - SMD ones can easily be fitted - 16 volts rating okay - they take some hours to settle in.
Those Kemet PPS ones would probably be Rifa - the 1794 has different filtering requirements to the 1541A - different mechanism
All the best.
Those Kemet PPS ones would probably be Rifa - the 1794 has different filtering requirements to the 1541A - different mechanism
All the best.
Hello James,
The original caps by Curcio were mkp10 100nf. I know the 1794 is another chip but maybe because it is all about digital information '' handled '' inside the chip the caps working well with one chip might be ok in the other chip as well.
I think in general smd parts are not that well regarded in audiophile circles.
I did check for the kemet pps caps but there are a lot of them that require minimum number of 4000 to order them at mouser so i will have to search for stock items.
Greetings, Eduard
The original caps by Curcio were mkp10 100nf. I know the 1794 is another chip but maybe because it is all about digital information '' handled '' inside the chip the caps working well with one chip might be ok in the other chip as well.
I think in general smd parts are not that well regarded in audiophile circles.
I did check for the kemet pps caps but there are a lot of them that require minimum number of 4000 to order them at mouser so i will have to search for stock items.
Greetings, Eduard
Actually, many of the new SMD caps are just the same old caps but with the SMD ends, similar to resistors - most of the talk in audiophile circles is misinformation due to preconceived ideas, unfortunately, particularly about the Nickel used in the terminal metal, etc - many of those same caps actually 'work better' because they don't have the flying leads - open minds, own homework, and so forth - caps and/in dacs is nearly always a personal preference and the 1541A seems to be no different ...
Try Elfa in Sweden, but need to check if the Kemet listing is the old Rifa ones via the spec sheets - as usual in Europe, there's generally an equivalent.
Try Elfa in Sweden, but need to check if the Kemet listing is the old Rifa ones via the spec sheets - as usual in Europe, there's generally an equivalent.
Dual Mono DAC update:
Now all shunt-regs and also TDA1543 DAC chips have Wima polyester caps.
The shunt-reg for the SD receiver chip is in series with the TeddyReg.
I am waiting for a better clock.
I soldered a small Bybee on the VCC of the DAC chips to see how it sounds. I have to wait for proper burn-in of the caps first.
Cheers,
M.
Now all shunt-regs and also TDA1543 DAC chips have Wima polyester caps.
The shunt-reg for the SD receiver chip is in series with the TeddyReg.
I am waiting for a better clock.
I soldered a small Bybee on the VCC of the DAC chips to see how it sounds. I have to wait for proper burn-in of the caps first.
Cheers,
M.
Attachments
It's me.
It is just that I'm feeling generous and want to retribute some of the kindness and friendship received all over the years in this marvelous forum.
BTW, my other DAC uses as output coupling cap a combination of big and small Russian PETP military caps and a smaller polystyrene, in parallel, and it works great!
Cheers,
M.
It is just that I'm feeling generous and want to retribute some of the kindness and friendship received all over the years in this marvelous forum.
BTW, my other DAC uses as output coupling cap a combination of big and small Russian PETP military caps and a smaller polystyrene, in parallel, and it works great!
Cheers,
M.
Yeah, this thread has been a real gem, and still doing okay.
I've shifted to an Ayre Acoustic QB-9 dac but still intend to use the 'ryanj' new boards to complete the project - there's just something about the 1541a that's attractive sound, and also the 'little ones'
I must say those Russian K73s are a big surprise - even recently a few guys were very complimentary about their benefits - have some old Rifa pfe216 styrenes and must try them in //.
I've shifted to an Ayre Acoustic QB-9 dac but still intend to use the 'ryanj' new boards to complete the project - there's just something about the 1541a that's attractive sound, and also the 'little ones'
I must say those Russian K73s are a big surprise - even recently a few guys were very complimentary about their benefits - have some old Rifa pfe216 styrenes and must try them in //.
experiment with a ripple eater
I have experimented with a ripple eater based on the BUZ11 with the gate on a 120K/100K divider; and a source resistance of say 200 ohms decoupled; a small 4.7 uF to the gate from the input. Drop resistor varies from voltage (5V is lower than 30V power line).
Now the interesting part:
I had a small 33 ohms trimpot, and turned it to null the input noise; and I tested it on a LM317. [Because that is my "lab-supply". ]
The nulling nulled out the LM317 supply noise!
The ripple-eater does not take out much DC level (but destroys the impedance of course).
It looks like it can work at very high frequencies.
I tested it to be in the feed line of a TDA1541 shunt stack before the CCS.
generally a normal capacity multiplier is more effective (20 dB) so I made this for now; but still I wonder how John would like this, as the ripple eater can go quite high in frequencies with careful trimming and that is where such a circuit could excel.
2¢.
I have experimented with a ripple eater based on the BUZ11 with the gate on a 120K/100K divider; and a source resistance of say 200 ohms decoupled; a small 4.7 uF to the gate from the input. Drop resistor varies from voltage (5V is lower than 30V power line).
Now the interesting part:
I had a small 33 ohms trimpot, and turned it to null the input noise; and I tested it on a LM317. [Because that is my "lab-supply". ]
The nulling nulled out the LM317 supply noise!
The ripple-eater does not take out much DC level (but destroys the impedance of course).
It looks like it can work at very high frequencies.
I tested it to be in the feed line of a TDA1541 shunt stack before the CCS.
generally a normal capacity multiplier is more effective (20 dB) so I made this for now; but still I wonder how John would like this, as the ripple eater can go quite high in frequencies with careful trimming and that is where such a circuit could excel.
2¢.
Hi triode_al,
Single capacitance multiplier or ripple filter still passes some ripple. Even with the base tied to a battery reference voltage, there will still be ripple voltage on the emitter of the transistor.
You have to cascade at least two ripple filters in order to obtain low ripple. For sensitive circuits you need to use at least 3 cascaded ripple filters.
Also use a resistive divider (add resistor across the electrolytic capacitor) for sufficient headroom. Required headroom depends on input ripple voltage amplitude
Problem with these ripple filters is that the output impedance is not low enough so the output voltage will vary with the load current.
So you need a circuit with considerable gain and use a global feedback system in order to achieve low output impedance.
I still use quatrode series regulators in my projects.
Shunt regulators need to be over dimensioned (bias current) in order to handle peak currents. This leads to higher ripple voltage and switching noise levels in the rectifier and requires higher capacity transformers and related higher stray capacitance. Then there is also the problem with cooling.
It is also important to keep transformers away from the electronics, do not put mains transformers inside the housing that also contains sensitive circuits. Transformers produce magnetic fields that can still couple ripple and noise into nearby circuits. Mains interference can also enter nearby circuits when routed into the housing.
I use following power supply for the Mosaic DAC:
12V external linear power supply with quatrode regulator -> 6V5 quatrode pre regulator -> 4V5 quatrode post regulator -> Mosaic DAC matrix.
I use heavily filtered low noise voltage reference (LM329) for quatrode pre and post regulators. I tried many other voltage references but this is my favourite.
Low noise integrated voltage regulators are rare. However I do use a TPS7A4700 programmed at 3V3 for the SD-transport.
Voltage regulators can pick up EMI, this will degrade the output voltage (noise, interference). It is therefore important to use good screening and EMI absorbers. This is why I use solid GND planes on all PCBs and very compact, solid aluminum housings that absorb EMI and act as Faraday’s cage.
Wiring between voltage regulator and circuit increases effective voltage regulator output impedance, so keep wiring as short as possible. In order to reduce power supply wiring inductance use twisted pairs (cancelling of magnetic fields). In some cases it is even required to use screened, twisted pairs power supply wiring in order to minimise EMI pick-up.
Single capacitance multiplier or ripple filter still passes some ripple. Even with the base tied to a battery reference voltage, there will still be ripple voltage on the emitter of the transistor.
You have to cascade at least two ripple filters in order to obtain low ripple. For sensitive circuits you need to use at least 3 cascaded ripple filters.
Also use a resistive divider (add resistor across the electrolytic capacitor) for sufficient headroom. Required headroom depends on input ripple voltage amplitude
Problem with these ripple filters is that the output impedance is not low enough so the output voltage will vary with the load current.
So you need a circuit with considerable gain and use a global feedback system in order to achieve low output impedance.
I still use quatrode series regulators in my projects.
Shunt regulators need to be over dimensioned (bias current) in order to handle peak currents. This leads to higher ripple voltage and switching noise levels in the rectifier and requires higher capacity transformers and related higher stray capacitance. Then there is also the problem with cooling.
It is also important to keep transformers away from the electronics, do not put mains transformers inside the housing that also contains sensitive circuits. Transformers produce magnetic fields that can still couple ripple and noise into nearby circuits. Mains interference can also enter nearby circuits when routed into the housing.
I use following power supply for the Mosaic DAC:
12V external linear power supply with quatrode regulator -> 6V5 quatrode pre regulator -> 4V5 quatrode post regulator -> Mosaic DAC matrix.
I use heavily filtered low noise voltage reference (LM329) for quatrode pre and post regulators. I tried many other voltage references but this is my favourite.
Low noise integrated voltage regulators are rare. However I do use a TPS7A4700 programmed at 3V3 for the SD-transport.
Voltage regulators can pick up EMI, this will degrade the output voltage (noise, interference). It is therefore important to use good screening and EMI absorbers. This is why I use solid GND planes on all PCBs and very compact, solid aluminum housings that absorb EMI and act as Faraday’s cage.
Wiring between voltage regulator and circuit increases effective voltage regulator output impedance, so keep wiring as short as possible. In order to reduce power supply wiring inductance use twisted pairs (cancelling of magnetic fields). In some cases it is even required to use screened, twisted pairs power supply wiring in order to minimise EMI pick-up.
Thank you, dear John, for the excellent sumary.
I'll certainly do that when the project circuits will be in a "final" state.
I want to make an update from my last tweak but I must recognize that I made a mistake: at the base of the Darlington from your LED-shunt regulator I wrongly read 1uF so I put big red Wimas and not the 1nF that you recommend. Probably some of the following is due to that mistake.
I made two simultaneous tweaks last time (against my custom), swapping the Epcos that was in place for a big 1uF/250V Wima capacitor on the LED shunt reg for the DAC chips, plus soldering two red 1uF/65V Wima's at the TDA1543 power pins.
The change in sound was similar to what I experienced when playing with "Black Gates" on power supplies: two goods, instant blacker back ground and powerfull deep bass, and one bad, HF very laid back, which made it sound like "plastic". It sounded like junk during 5-6 days, then slowly, the HF returned. Now it is sounding pretty decently, and with more impact than previously.
Yesterday I listened Britten's "War Requiem" (Chandos) and it was scary. For example Tuba mirum had live-like dynamics. Very recommended music, BTW.
Now I want to mod all my other sources.
I understand that you are accepting orders for the Mosaic on your site...
Now that I am on the mood for building amps I probably will attempt to build your floating-bridge or ciclotron amp, but I must study first.
Very grateful as always,
M.
I'll certainly do that when the project circuits will be in a "final" state.
I want to make an update from my last tweak but I must recognize that I made a mistake: at the base of the Darlington from your LED-shunt regulator I wrongly read 1uF so I put big red Wimas and not the 1nF that you recommend. Probably some of the following is due to that mistake.
I made two simultaneous tweaks last time (against my custom), swapping the Epcos that was in place for a big 1uF/250V Wima capacitor on the LED shunt reg for the DAC chips, plus soldering two red 1uF/65V Wima's at the TDA1543 power pins.
The change in sound was similar to what I experienced when playing with "Black Gates" on power supplies: two goods, instant blacker back ground and powerfull deep bass, and one bad, HF very laid back, which made it sound like "plastic". It sounded like junk during 5-6 days, then slowly, the HF returned. Now it is sounding pretty decently, and with more impact than previously.
Yesterday I listened Britten's "War Requiem" (Chandos) and it was scary. For example Tuba mirum had live-like dynamics. Very recommended music, BTW.
Now I want to mod all my other sources.
I understand that you are accepting orders for the Mosaic on your site...
Now that I am on the mood for building amps I probably will attempt to build your floating-bridge or ciclotron amp, but I must study first.
Very grateful as always,
M.
Yes, thank you. I already saw it...faulty keyboard here...
It seems you no longer offer SD card player integrated to DAC but Transport separated from DAC. In that case, if you feel confident your solution to deal with multi-clocks does not degrade sound, I'll go for the 24bit/192Khz DAC, provided we are still on the introductory price time line
Could you show picture or link to the USB interconnect needed?
I'll send email.
M.
Having had the previous sd player sd 1 for 2 or 3 years, I can really say that from my point of view, sd cards are a pain in the ***
I'm very happy to be able to use Jriver, vst EQ for late night music, and all the advantage of a usb dac.
but I understand your opinion!
Hi maxlorenz,
The Mosaic 16 SD-player replaces the SD2 player, both have built-in SD-transport.
The “Transport” is a BeagleBone Black computer that connects the Mosaic 16 or 24 USB DAC to a computer network so one can use a music server.
There is only one fuzzy logic masterclock in the Mosaic DAC or Mosaic SD-player that provides all critical audio timing signals.
We are now testing some minor modifications based on customer feedback. One of these is lower output impedance.
The Mosaic DAC can now directly drive a beyerdynamic DT-990 premium studio Headphone with 250 Ohm impedance.
Now we basically listen to the music produced by electronic switches and a passive resistor matrix. Then one hears how much distortion and degrading is introduced by active circuits like (pre) amplifiers.
The Mosaic 16 SD-player replaces the SD2 player, both have built-in SD-transport.
The “Transport” is a BeagleBone Black computer that connects the Mosaic 16 or 24 USB DAC to a computer network so one can use a music server.
There is only one fuzzy logic masterclock in the Mosaic DAC or Mosaic SD-player that provides all critical audio timing signals.
We are now testing some minor modifications based on customer feedback. One of these is lower output impedance.
The Mosaic DAC can now directly drive a beyerdynamic DT-990 premium studio Headphone with 250 Ohm impedance.
Now we basically listen to the music produced by electronic switches and a passive resistor matrix. Then one hears how much distortion and degrading is introduced by active circuits like (pre) amplifiers.
Hi -EC-
Thank you for the clarification...somehow the music input for the player is not evident. Maybe some good connection diagrams would be revealing on your site. For example, "transport" means autonome functioning. Does the BeagleBone Transport provides music data without being connected to a server?
Is the SD card meant only for booting, or has it mixed function?
Excellent.
I can wait a little, to have the perfect and ultimate (?) solution.
I am very entertained with my dual-mono DAC, which taught my a lot...
Most amps have 10K input impedance anyway...
Best wishes,
M.
Thank you for the clarification...somehow the music input for the player is not evident. Maybe some good connection diagrams would be revealing on your site. For example, "transport" means autonome functioning. Does the BeagleBone Transport provides music data without being connected to a server?
Is the SD card meant only for booting, or has it mixed function?
Excellent.
I can wait a little, to have the perfect and ultimate (?) solution.
I am very entertained with my dual-mono DAC, which taught my a lot...
Most amps have 10K input impedance anyway...
Best wishes,
M.