Non Oversampling DAC-complementing CD-PRO

[ThorstenL]

Koinichiwa,

Originally posted by Peter Daniel
They should be fine. Now I understand the window, but how I/V resisors and ref. resistor fits there?

Okay, let's do this one more time. This is actually quite clear from the datasheet if you ask me.

I have already run through this on the gainclone Board, but the search engine there sucks, so I'll run this again. Someone check if I get the same numbers twice... ;-)

Anyway, our absolute, maximum output swing is 1.8V to +V -1.2V. So for 5V this is 5V -1.2V -1.8V = 2V.

The DAC's output full scale current is rated as 2.3mA typhcial, +/-0.35mA. If we assume for a moment the typhical TDA1543 we have 2V peak-peak permissable Voltage swing and 2.3mA peak-peak current swing.

The I/V resistor hence becomes 2V/2.3mA = 869 Ohm. NPV is 910 Ohm, Kusonoki San obviously got away using 1k.

Now the next step. Our "midpoint" voltage must be 1.8V+swing/2 and thus 1.8V+2V/2V =2.8V.

This means we need to have the output produce a current giving 2.8V across a 910 Ohm resistor for digital silence or 1.15mA output current. So the total current must be 3.07mA of which 1.15mA are supplied by the DAC.

Looking at page 5 of the TDA1543 Datasheet we see the DAC output composed out of a variable current sink (the actual DAC) and a fixed current source (controlled by Rref).

Now Vref is fixed at nominal 2.2V and the "gain" for the reference current drawn out of the Vref pin is nominally 2.

As our DAC sinks 1.15mA and 3.07mA are delivered to our external I/V resistor to set the midpoint we need to make 1.15mA+3.07 = 4.22mA flow through our current source and thus 4.22mA/2 = 2.11mA through the resistor from Vref to ground.

With 2.2V reference Voltage and 2.11mA current the Rref resistor should be 1.04k, 1k NPV.

If we want to maximise the DAC's dynamic range, we can ramp up the Supply Voltage to 8V. According to the Datasheet this fully permissable as typhical operation. The ABSOLUTE MAX supply Voltage rating is 9V.

So lets assume we are like Austin Powers and "like to live dangerous". This means we set +V at 8.6V by using a 7808 and a diode.

Our output Voltage swing can now be 1.8V to 8.6V - 1.2V = 7.4V or 5.6V peak-peak (2V RMS). Our full scale current is still only 2.3mA, so our I/V resistors become 5.6V/2.3mA=2.43k NPV = 2k4.

To get the output to the new midpoint of 1.8V+(5.6V/2) = 4.6V. We will have to have 4.6V/2k4 = 1.92mA flowing in our 2k4 resistor, plus our 1.15mA DAC digital silence current, so 3.07mA. In our reference resistor 3.07mA/2 = 1.535mA need to flow, so 2.2V/1.535mA=1.43k, NPV 1k5.

This will of course totally max out the output voltage range and I would suspect that given the permissable tolerance in series production of the chip you would have to select the chips, maybe 1 out of 10 will make the grade.

Does that clear up any remaning certainties?

Sayonara

PS, Kusonokis first DAC used 4 Chips in Parallel, 9V Supply and 620 Ohm IV, plus 390 Ohm Rref. Scaled to signle chip this would be using NPV E12 Resistors the same values I calculated above and is for the I/V resistor consistent with Stereophiles measurements on the Progression DAC....

[ThorstenL]

Koinichiwa,

Originally posted by RichardJones


Use 3 seperate 6V SLA batteries and those TL431 shunt regulators.

Marginal. You may better off in this case with really low dropout, low power, low noise 3-Pin regs. Can I suggest simply switching to 12V SLA's? This will give plenty of voltage headroom to play with and then use LM334 or J-Fet CCS's.

Originally posted by RichardJones

Use the triplet of 100uf BG FK series, 270n, and 10nF caps at the receiver and DAC supply pins.

If you absolutely MUST have BG's, use NX-HiQ. The FK are not as good IME as the SH Os-Cons.

Originally posted by RichardJones

What type of 10nF cap should I use at the input to the receiver on RXP and RXN???

I'd say Silver Mica with short leads or COG ceramic SMD. Even better is a nice pulse transformer instead of the capacitors. Make that a 1:2 one so you get more trigger signal and less jitter (the "eye in the "eyepattern" is opened "wider").

Originally posted by RichardJones

The negative of all 3 batteries will be tied together, and all grounds will be routed back to this point.

BAD IDEA. Return each batteries negative pole to the chip it supplies and bus-ground from there. The analogue Receiver ground links to the digital receiver ground and that links to DAC ground.

Originally posted by RichardJones

> Anything special for the filter on the FILT pin?

I like doing things in SMD, using a cap of a few nf directly across the pins and the R&C in SMD solered to the Pin's forming a V.

No PCB traces on the Filt Pin please and cut of excessive lead-length too, this node is high impedance and a good aerial for any stray crap floating around. Don't give this a chance to be picked up by the PLL and to modulate the recovered clock.

Sayonara

[RichardJones]

Do you have a schematic for the JFET CCS?

12V SLA batteries would be more user friendly I guess.

NX-HiQ BG's? I will look into them.

Do you have a recommended place for those 1:2 pulse transformers?

Thankyou!

[ThorstenL]

Koinichiwa,

Originally posted by RichardJones
Do you have a schematic for the JFET CCS?

Shortcircuit gate and source and select from a large batch of suitable devices for desired current.

Originally posted by RichardJones
12V SLA batteries would be more user friendly I guess.

A nearly discharged 6V battery will leave you with under 0.5V for the CCS or resistor in series with the PSU line. At the same time the 7.2V maximum voltage for the batteries fully charged is more than you can apply to the CS receiver directly, without risking damage, so you must regulate somewhere.

Shuntregulators are not very efficient, they just sound very good.

Originally posted by RichardJones
Do you have a recommended place for those 1:2 pulse transformers?

Try:

http://www.scientificonversion.com/digital_audio.html

Sayonara

[fedde]

KYW: It is very cool you can spend so much time again here on Diyaudio and the Amp discussion board. Way to go !

BTW: I just tried a BG-NX 6.3 V 22 uF in a 5V TDA1543 DAC. IMVHO it rocks!!! This weekend I'll do some further experiments, but I am ***VERY*** satisfied with the DAC I've been listening to for the last two hours (and partly also the last two days).

Fedde

[Peter Daniel]

I also listened to mine today and it improved a lot and it sounds great. I don't think I'll be trying TDA1541

[jean-paul]

5V TDA1543 DAC

[jean-paul]

Originally posted by Peter Daniel
I also listened to mine today and it improved a lot and it sounds great. I don't think I'll be trying TDA1541

It is better though.

[Peter Daniel]

I was joking about not trying it. I know it has quite a number of caps aroung the chip. I was thinking about using BG N, what is a recommended value here?

[fedde]

Don't rub it in Jean-Paul

I still can't understand why I like my old rubbed up DAC more than my new one

After this weekend I'll know more. What I know now is that I can play full volume with this DAC without getting headaches or any other diseases
With 5V it still have some slam left, some decent controlled slam. The sound I have here reminds me a little of the sound of the original 47 labs gear I heard at Wilco's place in Utrecht (audio-cube, dutch distributor). Very crispy&resonant guitars and slapping basses. Good stereo placing. Addictive and seductive...

And now I will postpone the rest of the message till a point of time that I am more objective and with my feet on the ground. (they are now tapping you know )

Fedde