How Long you sleep everyday John 
Hiii Max
It was superb to see the all of the answer
John has do it first before i answer to him
Oh yes ecsdesign if you maybe take the
dac near final maybe you can give us partslist
for DI4T because some exotic parts near the
end to us like BGN.
And like we did for other project, if you part available
in the store we want to order for our project maybe
we can order you parts together
Looks like your dacs near or more than 500 parts
OMIGOD
Regards, Jeffry
Hiii Max
It was superb to see the all of the answer
John has do it first before i answer to him
Oh yes ecsdesign if you maybe take the
dac near final maybe you can give us partslist
for DI4T because some exotic parts near the
end to us like BGN.
And like we did for other project, if you part available
in the store we want to order for our project maybe
we can order you parts together
Looks like your dacs near or more than 500 parts
OMIGOD
Regards, Jeffry
balanced output
Hmm...converting to balanced output seems more complicated than I thought. Wouldn't a transformer coupled output solve the problem more simply? I know our bias against transformers but some really feel this is a better way to go than coupling caps. I guess it depends on the quality of the transformer.
Hmm...converting to balanced output seems more complicated than I thought. Wouldn't a transformer coupled output solve the problem more simply? I know our bias against transformers but some really feel this is a better way to go than coupling caps. I guess it depends on the quality of the transformer.
Hi pocoyo,
Yes that figure comes pretty close. The DI8M even had much more parts.
Ok, I thought about this, and DI4T cost, and come up with an idea.
I figured-out some tricks to get pretty good performance from TDA1543 chips (almost as good as TDA1541A). Think of performance very close to DI4T.
The idea is as follows:
Design a replacement module for the existing DA1541A modules. This module holds 2 x TDA1543, I2S attenuators, and 4.6V stabilizers. One of the 4 modules generates a 2.5V reference voltage for all other modules. Cooling wouldn't be required either, so the heatsink construction could be left-out as well.
Now a cheap, single 10V / 15VA unregulated power supply can be used. The existing DI16PS module would be fine too.
All electrolytic capacitors for the -15V and -5V can be left-out.
The DEM clock distribution circuit can also be left-out.
The DA1543 modules would provide approx. 2 x 9.2mA (differential output) full-scale current / channel.
Amplification could be done using the existing TUBEDIF modules, or discrete all-JFET amplifiers. People could also experiment with other amplifier circuits, or perhaps use suitable step-up transformers.
This could significantly reduce DI4 project cost and complexity. The TDA1543 chips are quite cheap and still available.
Let me know what you think of this idea.
Yes that figure comes pretty close. The DI8M even had much more parts.
Ok, I thought about this, and DI4T cost, and come up with an idea.
I figured-out some tricks to get pretty good performance from TDA1543 chips (almost as good as TDA1541A). Think of performance very close to DI4T.
The idea is as follows:
Design a replacement module for the existing DA1541A modules. This module holds 2 x TDA1543, I2S attenuators, and 4.6V stabilizers. One of the 4 modules generates a 2.5V reference voltage for all other modules. Cooling wouldn't be required either, so the heatsink construction could be left-out as well.
Now a cheap, single 10V / 15VA unregulated power supply can be used. The existing DI16PS module would be fine too.
All electrolytic capacitors for the -15V and -5V can be left-out.
The DEM clock distribution circuit can also be left-out.
The DA1543 modules would provide approx. 2 x 9.2mA (differential output) full-scale current / channel.
Amplification could be done using the existing TUBEDIF modules, or discrete all-JFET amplifiers. People could also experiment with other amplifier circuits, or perhaps use suitable step-up transformers.
This could significantly reduce DI4 project cost and complexity. The TDA1543 chips are quite cheap and still available.
Let me know what you think of this idea.
Hiii
For TDA 1543 maybe can use for your new DAC
For trial version before he buy the " ultimate "
Congratulation !
For me TDA 1541 A still bigger size & in my hearts
for my ultimate dacs in my stereo system someday
DIY who build the ultimate are really brave to spend
their money for one reason ... satisfaction that money
couldnt buy
For TDA 1543 maybe can use for your new DAC
For trial version before he buy the " ultimate "
Congratulation !
For me TDA 1541 A still bigger size & in my hearts
for my ultimate dacs in my stereo system someday
DIY who build the ultimate are really brave to spend
their money for one reason ... satisfaction that money
couldnt buy
advanced I/V conversion
Hi,
I've readed your article. It is very interesting. I only have a thing to add, and this is the best performance for the I/V part and it is:
Please, do not use OPA627, it is slow, and it does not sound good. Try THS4031/4031 mono/dual. It is superfast, super high quality, ultra low noise and the best op amp for the I/V part that I'v tried in my life.
Regards, galeb
Sarte audio elite's technical department
Hi,
I've readed your article. It is very interesting. I only have a thing to add, and this is the best performance for the I/V part and it is:
Please, do not use OPA627, it is slow, and it does not sound good. Try THS4031/4031 mono/dual. It is superfast, super high quality, ultra low noise and the best op amp for the I/V part that I'v tried in my life.
Regards, galeb
Sarte audio elite's technical department
Hi galeb,
You are correct about OPA627, that's why I started experimenting with the LM4562.
Some of the last I/V Op-amps I tested were THS4031 and THS4032, results weren't as impressive as I hoped for.
The DI4T now has differential passive I/V conversion (non-inductive copper wire I/V resistors), and tube-only amplification.
The DI4MJ has twin-differential passive I/V conversion (4 I/V resistors / channel), and discrete DC-coupled all-JFET summing amplifiers.
You are correct about OPA627, that's why I started experimenting with the LM4562.
Some of the last I/V Op-amps I tested were THS4031 and THS4032, results weren't as impressive as I hoped for.
The DI4T now has differential passive I/V conversion (non-inductive copper wire I/V resistors), and tube-only amplification.
The DI4MJ has twin-differential passive I/V conversion (4 I/V resistors / channel), and discrete DC-coupled all-JFET summing amplifiers.
Hi maxlorenz,
Discrete (JFET) amplifiers perform excellent.
I am quite happy with the new DI4MJ that runs on 4 x TDA1543 @ 4.6V power supply. It performs much better than my DI16 (8 * 4) DAC that had 32 x TDA1543 installed.
Yes it's still IN, but it's different from the previous versions. It's greatly simplified, and it provides delay for all DAC chips, therefore allowing fully synchronous DAC chip clocking (previous DI DAC versions used BCK for the first DAC chip and NBCK for the others).
Discrete (JFET) amplifiers perform excellent.
I am quite happy with the new DI4MJ that runs on 4 x TDA1543 @ 4.6V power supply. It performs much better than my DI16 (8 * 4) DAC that had 32 x TDA1543 installed.
Yes it's still IN, but it's different from the previous versions. It's greatly simplified, and it provides delay for all DAC chips, therefore allowing fully synchronous DAC chip clocking (previous DI DAC versions used BCK for the first DAC chip and NBCK for the others).
Hi MGH,
The TUBEDIF module has a differential input stage (ECC83S), and two cathode followers in parallel (ECC82) to lower output impedance.
The two ECC82 triodes could be used for balanced output. This will increase output impedance to approx. 450 Ohm for each output. It requires two cathode load resistors of approx. 27 K Ohm, and one extra coupling cap, that's all.
The only configuration I could think of, is passive I/V conversion, high-performance step-up transformers and DC-coupled JFET buffers.
The TUBEDIF module has a differential input stage (ECC83S), and two cathode followers in parallel (ECC82) to lower output impedance.
The two ECC82 triodes could be used for balanced output. This will increase output impedance to approx. 450 Ohm for each output. It requires two cathode load resistors of approx. 27 K Ohm, and one extra coupling cap, that's all.
The only configuration I could think of, is passive I/V conversion, high-performance step-up transformers and DC-coupled JFET buffers.
Hi pocoyo,
I bought the computer-controlled milling machines from EMC Elektronik:
http://www.emc-webline.de/produkte.htm
These are reasonably priced "desktop" milling machines.
Length: 420mm (16.53")
Width: 325mm (12.79")
Height: 54mm (1.12")
I bought the computer-controlled milling machines from EMC Elektronik:
http://www.emc-webline.de/produkte.htm
These are reasonably priced "desktop" milling machines.
Length: 420mm (16.53")
Width: 325mm (12.79")
Height: 54mm (1.12")
Hi -EC-
What value of DIY I/V R do you use?
Or is it active I/V?
I have eight "towers" already...I could try it in both my DACs...lowering to 4.6V should be no problem (there is cold down here; winter is getting long).
I hope you could post a diagram of the timing chain someday
You are using BCK attenuators for TDA1543 also, right?
Regards,
M
What value of DIY I/V R do you use?
Or is it active I/V?
I have eight "towers" already...I could try it in both my DACs...lowering to 4.6V should be no problem (there is cold down here; winter is getting long).
I hope you could post a diagram of the timing chain someday
You are using BCK attenuators for TDA1543 also, right?
Regards,
M
Hi folks
i was lately went out from this amaizing thread
I have a Q
about
Improved I2S attenuators with bias voltage?
*
maybe it is better solution
for USB input
PCM2... and after that transmiter Wolfson WM8..4 ?
booth 12MHz osc.
mayby not crystal only, but
some beter clock source?
of course better PS...
*
I have the chips and will try theese days,
sending report back...
*
i was lately went out from this amaizing thread
I have a Q
about
Improved I2S attenuators with bias voltage?
*
maybe it is better solution
for USB input
PCM2... and after that transmiter Wolfson WM8..4 ?
booth 12MHz osc.
mayby not crystal only, but
some beter clock source?
of course better PS...
*
I have the chips and will try theese days,
sending report back...
*
Hi Zoran,
It's about attenuators / band-limiters that are placed close to the TDA1543 or TDA1541A I2S DATA and WS input pins.
These Philips DAC chips have fast current steering logic, and use transistors (no CMOS logic). This means that the inputs can accept lower signal levels. This in turn would limit the amount of HF interference (from the I2S input signals) being dumped on the chip substrate. The bias voltage is required because of digital input circuit design (TDA1543 data sheet page 5).
The practical effect of these attenuators is more refined sound quality due to reduced on-chip interference levels. I use these I2S attenuators in all DI-DAC designs.
The BCK (timing) input only requires a damping resistor (series resistor) of approx. 22 Ohm (depends on PCB lay-out).
First of all, USB audio requires bit-perfect playback, even the best digital audio receiver won't help if playback isn't bit-perfect. Failing to achieve bit-perfect playback usually results in loss of resolution, and a "flatter" sound stage.
Bit-perfect playback is usually achieved by tweaking and hacking the computer OS, drivers and applications (bit-perfect playback is usually not achieved with standard installation / applications). In order to achieve bit-perfect playback, the software DSP that's used for volume control, mixing and effects has to be completely bypassed.
When you succeed in achieving bit-perfect playback, there is no guarantee that it stays that way (software updates, patches and service packs). So this has to be re-checked after every significant software change.
Even if you managed to correctly apply all tweaks, you will never be 100% sure if playback is actually bit-perfect, unless you perform a file compare.
The other issues like jitter, crosstalk, ground loops and (wireless) interference can be tackled next.
I have been working on these "issues" for quite some time now, and tested most circuits (sample rate conversion, reclocking, slaving the digital audio source, PLL, and so on).
I arrived at the following configuration that seems to provide best performance of all tested configurations:
- Toslink optical interface.
- CS8416 SPDIF receiver in slave-clock mode.
- VCXO/PLL with LC / Zobel loop filter.
- LC filtered VCXO & clock buffer power supply.
- Ultra-high speed Clock buffers.
It's about attenuators / band-limiters that are placed close to the TDA1543 or TDA1541A I2S DATA and WS input pins.
These Philips DAC chips have fast current steering logic, and use transistors (no CMOS logic). This means that the inputs can accept lower signal levels. This in turn would limit the amount of HF interference (from the I2S input signals) being dumped on the chip substrate. The bias voltage is required because of digital input circuit design (TDA1543 data sheet page 5).
The practical effect of these attenuators is more refined sound quality due to reduced on-chip interference levels. I use these I2S attenuators in all DI-DAC designs.
The BCK (timing) input only requires a damping resistor (series resistor) of approx. 22 Ohm (depends on PCB lay-out).
First of all, USB audio requires bit-perfect playback, even the best digital audio receiver won't help if playback isn't bit-perfect. Failing to achieve bit-perfect playback usually results in loss of resolution, and a "flatter" sound stage.
Bit-perfect playback is usually achieved by tweaking and hacking the computer OS, drivers and applications (bit-perfect playback is usually not achieved with standard installation / applications). In order to achieve bit-perfect playback, the software DSP that's used for volume control, mixing and effects has to be completely bypassed.
When you succeed in achieving bit-perfect playback, there is no guarantee that it stays that way (software updates, patches and service packs). So this has to be re-checked after every significant software change.
Even if you managed to correctly apply all tweaks, you will never be 100% sure if playback is actually bit-perfect, unless you perform a file compare.
The other issues like jitter, crosstalk, ground loops and (wireless) interference can be tackled next.
I have been working on these "issues" for quite some time now, and tested most circuits (sample rate conversion, reclocking, slaving the digital audio source, PLL, and so on).
I arrived at the following configuration that seems to provide best performance of all tested configurations:
- Toslink optical interface.
- CS8416 SPDIF receiver in slave-clock mode.
- VCXO/PLL with LC / Zobel loop filter.
- LC filtered VCXO & clock buffer power supply.
- Ultra-high speed Clock buffers.