Building the ultimate NOS DAC using TDA1541A

[Tazzz]

I did some playing around comparing some different level translator circuits for the 1541', assume the 10pf is the actuall input capacitance of the dac you can see how much is injected into it.

I think there might be room for some improvement with regard to the back to back diode solution.

Note tough the rise / fall times are intentionally set high (wich exacerbates things)

[marconi118]

Hi John,

Going back to the TDA1543 simple implementation, have you ever experimented with the TDA1387 or TDA1545(EIAJ)?

They are 8pin and simple as the TDA1543 but with the 0v output compliance of the TDA1541.
It could be better sounding than TDA1543 but less complicated than TDA1541 DAC.

Only use the input stage from TDA1543 schema and the output mosfet + 500r i/v stage of the TDA1541

TDA1387 is discontinued but can be found on all soundblaster awe64 and soundblaster 16 (CT2770)

[abraxalito]

I've been playing with all three varieties of 8-pin DACs you mentioned. Its quite a lot of hard work to get the TDA1545 and TDA1387 sound as good as the TDA1543, even though their measurements look better on paper. I take it because they're CMOS parts they play mean with their power supplies, whereas the 1543 being bipolar is fairly benign in this regard. Also the 1543's current source can be disabled whereas the others' cannot. TDA1387 is available on the second hand market here in China and extremely good bang for the buck at under 10cents a pop.

[roger57]

Originally Posted by -ecdesigns-

Hi roger57,

TDA1541A DEM oscillator is now soft-synced with 2.8224 MHz bit clock (attached picture). Sync signal is now taken from pin 2 (BCK) only.
The bit clock is derived from the masterclock using two UHS D flip-flops that form a synchronous divide-by-4 Johnson counter.
Both, new masterclock module and better performing DEM clock resulted in more transparency and more unmasked flaws.
....

The new voltage regulators and current buffers further increased transparency, now revealing resistor noise spectrum issues. Some critical resistors in volume control and Circlotron power amp had to be changed to hybrid resistors in order to fine-tune the system noise spectrum. The hybrid resistors consist of the right "mix" of bulk metal foil, carbon composition, metal film and wire-wound resistors.

EC Designs / John,

Thank you for the thorough response. I always wonder how you find the time to provide such answers to DIY members, and yet find the time to do all your design. Do you ever find time to sleep?
My present TDA1541A board uses an inverter from BCLK for the DEM clock input, which seems to be a good result. However, I was asking about the clocking circuit as I was actually going to build the Mk7 reclock circuit, but I guess that won't happen now, based on what you are saying. Also interested in your regulator specifics and what you have found there.
As with your previous designs, do you think you might post them here in the near future?

Regards,
Gary

[-ecdesigns-]

Hi roger57,

Also interested in your regulator specifics and what you have found there.

I attached schematics of common-mode shunt regulators I am currently experimenting with.

First diagram is the 3V6 shunt regulator for the sub miniature timing module. It is set @ 20mA. It consists of a common-mode ripple rejection circuit (T1, T2, R1, R2, C1...C3). This circuit provides a clean non-stabilized DC voltage.

Constant current sources are placed in feed and return lines, different types of CCS can be used here. The main function of these CCS is increasing impedance between load and transformer.

Regulator noise spectrum has big impact on sound quality, even if it's in the nanovolt range. I suspect this low level noise still manages to modulate masterclock and cause trigger uncertainty in logic circuits and D/A stage.

R3 and R5 (carbon composition resistors) are used as pink noise generators. Their signal is attenuated and summed to current sensing resistors R3 and R6. Regulator noise spectrum modulates the masterclock frequency. Since power supply noise cannot be reduced to zero, and the noise spectrum of power supply and masterclock combination may not be ideal, this noise spectrum tuning becomes necessary in order to approximate perfect tonal balance.

The shunt regulator is a straight-forward circuit consisting of a high gain Darlington (T6), filtered LED reference (L1, C4, and C5) and a temperature compensated transistor (T7).

The second shunt regulator is intended for TDA1541A +5V supply and low impedance reference voltage (DC-coupling), it is set @ 100mA. I used multiple JFET CCS in parallel here in order to achieve low voltage drop. It is also possible to use different CCS here, depending on input voltage / voltage drop.

Since these are "floating" supplies it is possible to connect them similar to batteries. This way the same circuit can be used for positive and negative power supplies.

The -15V shunt regulator has 2 LEDs in series and value for R24 is increased to approx. 3K.

The common mode ripple rejection circuit must be located as close to the transformer as possible in order to keep wiring carrying ripple / noise voltage as short as possible. This way ripple and interference is mainly concentrated on the mains power supply module.

The clean DC voltage is fed to the regulators using (shielded) twisted pairs. Shunt regulators must be placed as close to the load as possible, in order to achieve this they should be built as compact as possible, 10 x 20mm PCB for example. It is also desirable to fully screen the shunt regulator PCBs in order to maintain low noise levels (EMI).

[Audio curiousity]

Hi ecdesigns,

Interesting thread. Read till thread 70 and the last page 418. Hope to read the remaining pages soon. Might be my next building project. Keep up the good work.

A suggestion:
In post #4175 you mention a LED reference:
"The shunt regulator is a straight-forward circuit consisting of a high gain Darlington (T6), filtered LED reference (L1, C4, and C5) and a temperature compensated transistor (T7)." I suppose low frequency might give the most trouble.

An idea might be to replace the LED by a low noise NPN transistor used as a Zener with collecter lead clipped off, base connected to negative and emitter connected to positive voltage. This might result in better noise figures then LEDs with a main advantage in the lower frequency range. Example: BFP520. I haven't tried it myself yet, so no experience.

More interesting info about LED noise (maybe mentioned before?): Some noise measurements for LEDs and zener diodes

[tloftis]

Originally Posted by -ecdesigns-

Here is a photo of the sonic resonator, a 360 degree radiator with real-time correction system for sub-bass, bass and mid. This system needs no damping materials. Mid-high and High are placed in an open resonance chamber with conical deflectors and polarizer discs for incredible realistic sound, it actually reconstructs the acoustics of the place the recording was made, with pinpoint accuracy. Chesky audio recordings sound magnificent on these sonic resonators. The photo shows an early prototype of about 4 years ago. Components are handcrafted form MDF and sprayed with metallic paint. The new version has solid aluminum and plastic parts and is sonically optimized to the max. It has a highly optimized passive filter using some very exotic parts like NON-inductive stranded copper wire resistors (took a long time and many listening sessions to fine-tune). You cannot believe how much effect a simple component as a resistor, capacitor or coil can have on sound reproduction. As I already said, hearing is believing. The most common reaction I get from High-End enthousiasts is: this sound cannot possibly come from those speakers....but then, it does. Second is, where is the subwoofer? ,those speakers cannot produce that bass or can they... sure they can.

those speakers are too cool. would love to hear them

[Joshua_G]

Hi John,

Originally Posted by -ecdesigns-

Regulator noise spectrum has big impact on sound quality, even if it's in the nanovolt range. I suspect this low level noise still manages to modulate masterclock and cause trigger uncertainty in logic circuits and D/A stage.

R3 and R5 (carbon composition resistors) are used as pink noise generators. Their signal is attenuated and summed to current sensing resistors R3 and R6. Regulator noise spectrum modulates the masterclock frequency. Since power supply noise cannot be reduced to zero, and the noise spectrum of power supply and masterclock combination may not be ideal, this noise spectrum tuning becomes necessary in order to approximate perfect tonal balance.

So, are you tuning the amount of pink noise generated by the regulator for the sonic impact, or what?

Originally Posted by -ecdesigns-

The common mode ripple rejection circuit must be located as close to the transformer as possible in order to keep wiring carrying ripple / noise voltage as short as possible. This way ripple and interference is mainly concentrated on the mains power supply module.

The clean DC voltage is fed to the regulators using (shielded) twisted pairs. Shunt regulators must be placed as close to the load as possible, in order to achieve this they should be built as compact as possible, 10 x 20mm PCB for example. It is also desirable to fully screen the shunt regulator PCBs in order to maintain low noise levels (EMI).

RFI/EMI propagate also by radiation, therefore, it seems to me that a good approach may be to filter them as much as possible, right at the mains entrance. Also, power transformer(s) with a static shield between the primary and secondary may be of a benefit. Also, bypassing (Zobel) the rectifier diodes and RF LC or CLC filter right after the rectifying diodes. It seems that there is never too much caution, or over-doing, when it concerns filtering RFI/EMI.

[-ecdesigns-]

Hi Joshua_G,

So, are you tuning the amount of pink noise generated by the regulator for the sonic impact, or what?

The noise at the load cannot be reduced to zero at room temperature regardless of voltage regulator specs. This noise will have certain noise spectrum, and depending on this spectrum the noise can cause unacceptable degrading of perceived sound quality at these performance levels. Fine tuning this noise spectrum can improve perceived sound quality.

When using suitable ultra low noise pre-amp, noise spectrum tuning can be verified by a spectrum analyser.

RFI/EMI propagate also by radiation, therefore, it seems to me that a good approach may be to filter them as much as possible, right at the mains entrance. Also, power transformer(s) with a static shield between the primary and secondary may be of a benefit. Also, bypassing (Zobel) the rectifier diodes and RF LC or CLC filter right after the rectifying diodes. It seems that there is never too much caution, or over-doing, when it concerns filtering RFI/EMI.

Check schematics in post #4073, it shows the the filter circuits in the mains power supply.

I use multiple balanced LC filters to attenuate interference. The schematic also shows I am using Schottky diodes, these don't require Zobel networks.

Safety earth gets polluted because it runs along mains feed and return lines. Since the mains wires aren't twisted together but are usually simply placed in a plastic tube, interference on protective earth lead cannot be nulled. This means that when a static shield is connected to this polluted safety earth lead it can even inject more interference into the power supply this way. This of course depends on local conditions.

The reason a digital audio reproduction system is so extremely sensitive to noise is that system clock, bit clock and word clock act as carrier waves for lower frequency interference signals that are modulated on it by power supply, ground or digital audio source for example. We than have multiple FM / AM radio transmitters right on the DAC module that can spread interference over the entire DAC circuits where its locally demodulated. After demodulation we basically transported interference wirelessly from point A to point B.

In an analogue sound reproduction system we sometimes have problems with high frequency oscillation in the power amp. This usually results in audible hum and interference. This is a simple example how the interference (power supply ripple voltage) is modulated on a carrier wave (oscillating power amp) and then demodulated, feeding power supply ripple voltage to the speakers.

In analogue sound reproduction systems we can easily fix this problem by preventing oscillations that can then act as carrier wave. With digital audio reproduction systems we are stuck with the carrier waves and have to figure out different solutions.

[Joshua_G]

Thanks, John.
Do you connect somehow the mains ground to your Transporter-DAC?
If yes, where and how?