I've seen the ecdesigns implementation
Building the ultimate NOS DAC using TDA1541A
post #2748
The reclock item is okay for me.
But why do the attenuation to I2S?
Building the ultimate NOS DAC using TDA1541A
post #2748
The reclock item is okay for me.
But why do the attenuation to I2S?
This is an old way of making the DEM. The latest is dem mod is shown in @6116. :
Building the ultimate NOS DAC using TDA1541A
The attenuator on the I2S is there in order to reduce the noise from the continous stream of the I2S. This is delt with if you use TDA1541a in stopped clock - simultanious mode. But I would recommend that you read at least the last 50-100 pages of this thread. That would answer a lot of your questions...
Hi Guys
I was wandering, if it was a good idea to replace the 2k2 that sinks 2 ma into the analog output of TDA1541a from the +5V supply, with a larger resistor and then make a dedicated low noise power-supply with a somewhat higher, say 25-30 V.
You could even make this power supply variable and use that to set the output from TDA1541a to 0 volt.
And a step further you could make a very slow DC servo thatmonitored the output and controlled the dedicated powersupply, to avoid long time drift of the DC out. The TDA1541a increases its distortion figures, if its output is not held to 0 V DC.
Any thoughts?
I was wandering, if it was a good idea to replace the 2k2 that sinks 2 ma into the analog output of TDA1541a from the +5V supply, with a larger resistor and then make a dedicated low noise power-supply with a somewhat higher, say 25-30 V.
You could even make this power supply variable and use that to set the output from TDA1541a to 0 volt.
And a step further you could make a very slow DC servo thatmonitored the output and controlled the dedicated powersupply, to avoid long time drift of the DC out. The TDA1541a increases its distortion figures, if its output is not held to 0 V DC.
Any thoughts?
If you want a truly low noise current source you can use my approach:
D12 is LM329.
ISS to Iout of TDA1541(A) which injects around 2 mA of current making it a bipolar swing (-2 mA to +2 mA).
I did use it within my DAC with TDA1541A:
You can regulate the injected current around 2 mA to get a perfect offset.
D12 is LM329.
ISS to Iout of TDA1541(A) which injects around 2 mA of current making it a bipolar swing (-2 mA to +2 mA).
I did use it within my DAC with TDA1541A:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
You can regulate the injected current around 2 mA to get a perfect offset.
It certainly does not make any sense, but all right. Also, high value resistor will not isolate anything better. That is just some ridiculous assumption based on audio vodoo which I'm not familiar with. Higher resistance gives you a higher noise and that is a physical law. The idea of such resistor is to inject a DC current which should be stable, so the power supply for that resistor should be as clean as possible.
Last edited:
3lite, thank you for the excellent CCS schematic and food for thought. I like how you referenced it to ground, and ground only. It seems to me that C71, as well as R101, should be connected to a very quiet / high quality ground for best performace.
Can you guys see and appreciate the difference between his CCS and a more common one that "hangs" from the positive rail through a small emitter resistor? This is a noise input from V+ into the CCS. It will need much cleaner rails to perform well.
In 3lite´s we have a sziklai pair providing a very clean ground referenced supply for the ccs.
Good choice of zener; voltage is in the "quiet region" for zeners. What about temperature dependence of the whole?
Thanks again,
Alex
Can you guys see and appreciate the difference between his CCS and a more common one that "hangs" from the positive rail through a small emitter resistor? This is a noise input from V+ into the CCS. It will need much cleaner rails to perform well.
In 3lite´s we have a sziklai pair providing a very clean ground referenced supply for the ccs.
Good choice of zener; voltage is in the "quiet region" for zeners. What about temperature dependence of the whole?
Thanks again,
Alex
My mistake, it´s not a 6.9V zener, it´s a precision voltage reference. LM329
http://www.analog.com/media/en/technical-documentation/data-sheets/129329fd.pdf
Thanks
Alex
No problem at all I thought it might be useful for someone and I do enjoy the fact that there are people who do do appreciate it. You are correct as well. Regarding the minor drift with temperature, I always had a DC servo on the output, so I didn't have any issues in general. In fact, I never tested it so I don't really know. However, I do not believe it will be an issue.
Anyway, a bit off topic, but I have a working prototype of my digital interpolation filter for TDA1540 / TDA1541A with 8x oversampling (8192 taps, two filters to choose from - minimum or linear):
The new PCB developed specifically for TDA1540 / TDA1541A is in production, but I did test it with my generic PCB designed for the 16x oversampling version and it is working quite nicely It accepts I2S stream from 44.1 kHz up to 384 kHz and interpolates them to 384 kHz for TDA1540 / TDA1541A which is the limit for the latter one. The filter outputs data in offset binary format for the simultaneous mode with a clock of 5.6448 MHz or 6.144 MHz (depending whether the stream is 44.1 kHz or 48 kHz based).
I will have it posted here when I assemble the new PCB
I thought it might be useful for someone and I do enjoy the fact that there are people who do do appreciate it. You are correct as well. Regarding the minor drift with temperature, I always had a DC servo on the output, so I didn't have any issues in general. In fact, I never tested it so I don't really know. However, I do not believe it will be an issue.Anyway, a bit off topic, but I have a working prototype of my digital interpolation filter for TDA1540 / TDA1541A with 8x oversampling (8192 taps, two filters to choose from - minimum or linear):
The new PCB developed specifically for TDA1540 / TDA1541A is in production, but I did test it with my generic PCB designed for the 16x oversampling version and it is working quite nicely
It accepts I2S stream from 44.1 kHz up to 384 kHz and interpolates them to 384 kHz for TDA1540 / TDA1541A which is the limit for the latter one. The filter outputs data in offset binary format for the simultaneous mode with a clock of 5.6448 MHz or 6.144 MHz (depending whether the stream is 44.1 kHz or 48 kHz based).I will have it posted here when I assemble the new PCB
That is just lovely!
Have a look yourself:
http://www.lampizator.eu/LAMPIZATOR/LINKS AND DOWNLOADS/DATAMINING/tda 1541A.pdf
Lets calculate an output voltage with an impedance of 150 Ohm:
150 = U / 0.0046 => U = 0.69~ V
You are heavily degrading the performance of TDA1541A.
If you truly want to have an I/V conversion using a resistor you might want to comply with datasheet to a certain degree:
First DIY DAC, Using TDA1541A
Another option is to lower the I/V resistor and use I/V transformer such as the 1465 from Sowter. I use its earlier sister, the 9545 and they worked very well for a pair of NOS TDA1541s.
Being bifilar wound, common mode noise is cancelled out and the inductance also filters out higher frequencies so I never had issues with aliasing despite of no additional filtering.
They recommend using a secondary load resistor which is reflected to the primary at the turns ratio squared, (ie 25 or 100, depending on how you connect the windings) thereby allowing for the use of a very low i/v resistor
without further amplification being needed.
DAC I/V CONVERSION OUTPUT TRANSFORMERS
Last edited:
This sounds a little like you have not been reading this thread very thoroughly.
Try, for a start to read these posts:
6069
6077
6098
6116
6351
I know about the compliance issue in TDA1541a , but if you solve one issue just to create worse issues (hearing wise not measure wise), you did not solve it properly. I have made TDA1541 with opamp I/V , CCS to inject current and plain passive solution.
To my ears the TDA1541a performs best with the passive soultion. I need to try the same circuit that I use for the TDA1543a , but it is quite another beast, so maybe that won´t work either. It is much more important, still in my ears, to avoid digital filters (resampling in a PC is another story) and reduce jitter and digital noise by using simultanious mode AND stopped clock operation.
Sorry for bringing such an old discussion up once more, but I got a little provoked by the "That is just lovely" and "it certainly makes no sense".
I am sure 3lite has done a good job on his DAC, but I feel he should hesitate with that kind of statements..