Hi JOSI1,
SD8-transport, 240mm x 90mm x 9mm.
TDA1541A-MK10, 120mm x 50mm x 9mm.
TDA1543-MK10, 120mm x 50mm x 9mm.
Stepped shunt volume control, 120mm x 50mm x 9mm.
SD8-transport, DAC and volume control are housed in a tilted vertical panel measuring 160mm x 260mm x 15mm. It's mounted to a base measuring 180mm x 180mm x 40mm containing power supplies and connectors.
I attached some sketches.
Housing consists of sandwich construction of aluminum sheets and wood.
The shunt regulators have different schematic and they are no longer powered through CCS.
I now use passive balanced 10th order RC filters between rectifier bridges and shunts. The resistors in the filters also limit the current for the connected shunt so no additional current limiting resistors are required.
I use RF JFETs (BF862). The gate connects to the wiper of a 1K trimmer, one side of the trimmer connects to GND, the other to -15V through 15K series resistor. Trimmer is adjusted for approx. 0 ... -20mV DC on the TDA1541A output. Each JFET has a separate trimmer plus 15K series resistor.
SD8-transport, 240mm x 90mm x 9mm.
TDA1541A-MK10, 120mm x 50mm x 9mm.
TDA1543-MK10, 120mm x 50mm x 9mm.
Stepped shunt volume control, 120mm x 50mm x 9mm.
SD8-transport, DAC and volume control are housed in a tilted vertical panel measuring 160mm x 260mm x 15mm. It's mounted to a base measuring 180mm x 180mm x 40mm containing power supplies and connectors.
I attached some sketches.
Housing consists of sandwich construction of aluminum sheets and wood.
The shunt regulators have different schematic and they are no longer powered through CCS.
I now use passive balanced 10th order RC filters between rectifier bridges and shunts. The resistors in the filters also limit the current for the connected shunt so no additional current limiting resistors are required.
I use RF JFETs (BF862). The gate connects to the wiper of a 1K trimmer, one side of the trimmer connects to GND, the other to -15V through 15K series resistor. Trimmer is adjusted for approx. 0 ... -20mV DC on the TDA1541A output. Each JFET has a separate trimmer plus 15K series resistor.
Attachments
Hello John,
thank you for your informations. Concerning the current buffer
I'm not sure if the following components from former designs
are still necessary:
Resistor for bias current: 2M2 remains
C=1uF from wiper to GND ??
resistor between wiper and Gate of FET ??
C=150pF from Source to GND ??
thank you for your informations. Concerning the current buffer
I'm not sure if the following components from former designs
are still necessary:
Resistor for bias current: 2M2 remains
C=1uF from wiper to GND ??
resistor between wiper and Gate of FET ??
C=150pF from Source to GND ??
Hello John,
is it possible to post a schematic of the new shunt regulator and the balanced RC filter?
John, I have built the regulators of post 4008. The -15V rail seems ok, but the + and - 5V rails are + and - 6.5V. If I'm not mistaken, this is outside the tolerance for the tda1541a, isn't it? Is the schematic of post 4008 correct?
Dear ec
Excuse me, but
I don't see anybody talking about 24 bit systems.
The 1541A still got 16 bit theoretical.
To the facts:
Your argumentation is again totally wrong.
Edl of typical 0.5 LSB sounds very nice but it isn't.
It is true that 0.5 LSB can be ignored in a 16 bit system at full scale.
Unfortunately the Edl has not much to do with full scale.
It is specially related to low level signals for a good reason.
At -60 dB the effective DAC resolution is reduced to 9-10 bits and things look very different.
0.5 LSB is blown up and becomes an unacceptable value if your goal is a high end system.
Calculation for 1 LSB change in a 9 bit system:
1,000,000 / 512 = 1953 cm = 0,02 km on a distance of 10 kilometer.
Remember that audio uses a logaritmic scale, because of the extremely high dynamic range of the ear.
Philips quotes -42 dB distortion for the 1541A at -60 dB which is not very impressive.
I got clearly better results with a few lucky 1541A chips, but still far from the optimum.
Edl must be at least reduced to < 0.02 LSB.
The "right" DAC chips allow to adjust Edl to zero.
From a Burr Brown DAC datasheet:
"DLE is important in audio applications because excessive DLE at Bipolar Zero can result in audible crossover distortion for low level output signals."
For a critical listener, DLE does not need to be excessive to be audible.
If you playback a full scale signal, perhaps.
You will understand that I feel myself obligated to make corrections when I see that dangerous halfknowledge is spread in the forum.
Excuse me, but
I don't see anybody talking about 24 bit systems.
The 1541A still got 16 bit theoretical.
To the facts:
Your argumentation is again totally wrong.
Edl of typical 0.5 LSB sounds very nice but it isn't.
It is true that 0.5 LSB can be ignored in a 16 bit system at full scale.
Unfortunately the Edl has not much to do with full scale.
It is specially related to low level signals for a good reason.
At -60 dB the effective DAC resolution is reduced to 9-10 bits and things look very different.
0.5 LSB is blown up and becomes an unacceptable value if your goal is a high end system.
Calculation for 1 LSB change in a 9 bit system:
1,000,000 / 512 = 1953 cm = 0,02 km on a distance of 10 kilometer.
Remember that audio uses a logaritmic scale, because of the extremely high dynamic range of the ear.
Philips quotes -42 dB distortion for the 1541A at -60 dB which is not very impressive.
I got clearly better results with a few lucky 1541A chips, but still far from the optimum.
Edl must be at least reduced to < 0.02 LSB.
The "right" DAC chips allow to adjust Edl to zero.
From a Burr Brown DAC datasheet:
"DLE is important in audio applications because excessive DLE at Bipolar Zero can result in audible crossover distortion for low level output signals."
For a critical listener, DLE does not need to be excessive to be audible.
If you playback a full scale signal, perhaps.
You will understand that I feel myself obligated to make corrections when I see that dangerous halfknowledge is spread in the forum.
Last edited:
My posting was mostly about DA conversion theory in general.
Pardon me for mentioning a "right" chip.
If you not wish to read my postings, you can add me to your ignore list.
User CP --> Settings and options --> Edit ignore list.
Completely agree, and i like to read these statements.
But in my view, these (yes, i mean several) posts which are telling us and especially to the thread opener, that his main and central design decision (i.e. using TDA1541A) is wrong do not belong here.
In my eyes it's simply impolite and does not help the discussion about his development, which is about implementing this chip and not another one. It belongs elsewhere.
As no one would burst into the Pass Forum, eg. the BA-2 thread, and tell everyone and Nelson Pass that MOS-FET's in the output stage are a bad thing.
But in my view, these (yes, i mean several) posts which are telling us and especially to the thread opener, that his main and central design decision (i.e. using TDA1541A) is wrong do not belong here.
In my eyes it's simply impolite and does not help the discussion about his development, which is about implementing this chip and not another one. It belongs elsewhere.
As no one would burst into the Pass Forum, eg. the BA-2 thread, and tell everyone and Nelson Pass that MOS-FET's in the output stage are a bad thing.
Because of the thread title.
Building the ultimate NOS DAC using TDA1541A
At least three essential limitations of the TDA1541A raise serious concerns and the question if the goal could be achieved.
More appropriate would have been:
Building the ultimate TDA1541A DAC using NOS
You say it yourself: The goal and core theme is clearly "Building the ultimate NOS DAC"
The main and central design decision is the choice of the DAC chip TDA1541A.
"ultimate" is a very big claim. As this is a discussion forum, not a sweet talk-blindly believing everything-place, criticism about the primary design choice must be expected. If the criticism is ignored or responded with false/wrong/weak arguments, the criticism will continue.
I'm not aware of a Nelson Pass thread titled "Building the ultimate classA amplifier using MOSFET output stage."
Last edited:
Dear Bernhard,
I think the problem might be related with reading comprehension...
"Building the ultimate NOS DAC using TDA1541A"
Means to me: "creating the best DAC that can be made with the TDA1541 chip".
(which might also be the best DAC ever)
There is no reason to intrude endlesly on the thread to critisize the DAC chip that the Author has chosen.
Unless you love to see your name written...
Yours,
M.
I think the problem might be related with reading comprehension...
"Building the ultimate NOS DAC using TDA1541A"
Means to me: "creating the best DAC that can be made with the TDA1541 chip".
(which might also be the best DAC ever)
There is no reason to intrude endlesly on the thread to critisize the DAC chip that the Author has chosen.
Unless you love to see your name written...
Yours,
M.
For those who lost the overview, the latest discussion is about this statement:
If some of the LSBs ( 2 or 3 ? ) are fully swamped by audio set distortion / noise anyway, how can I see the LSB toggling very clearly ?
Further, how can the signal be so extremely clean that it is possible to adjust the DLE, watching this signal on a scope ?
Remember, the signal runs through two or more opamps inside the CD player ( I/V & LPF ), then passes a 40x = 32 dB preamp ( RF capable preamp, noise & thd performance unknown ), and then a scope input amplifier.
How can it be
If some of the LSBs ( 2 or 3 ? ) are fully swamped by audio set distortion / noise anyway, how can I see the LSB toggling very clearly ?
Further, how can the signal be so extremely clean that it is possible to adjust the DLE, watching this signal on a scope ?
Remember, the signal runs through two or more opamps inside the CD player ( I/V & LPF ), then passes a 40x = 32 dB preamp ( RF capable preamp, noise & thd performance unknown ), and then a scope input amplifier.
How can it be
Analyzer with 500mV full scale and 40x preamp:
PCM56 8x os Yamaha CDX-5000 good adjusted MSB track 15
PCM56 8x os very bad adjusted MSB track 15
Dear maxlorenz
most of us are not native english speakers, but I think that
"creating the best DAC that can be made with the TDA1541 chip"
would be "Building the ultimate TDA1541A DAC"
most of us are not native english speakers, but I think that
"creating the best DAC that can be made with the TDA1541 chip"
would be "Building the ultimate TDA1541A DAC"
Remarks like this are totally expendable.