DIYHiFi.org

Ok, thank you! I don't want a new hobby, this is my hobby. I think I'm a guy who is always looking to reach the end of my projects, but sometimes I need help because I do not understand how it works or why it not work. I'm confused because Glen tell me the he used this DAC a few years ago and he had 2.5V at output. Who is right? I don't want burn my DAC. Maxpou

Hi rbroer,
I have no samples, Jocko have many samples. This is a Jocko's post. I will buy a PCM1794 for 22$ it's why I don't want burn it. Maxpou

Look at the schematic for the ezDAC. Runs off a 25ohm R. I read somewhere that this schematic comes from an manufacturer app-note, but can't find the post anymore.

I used different values to run off +-2.5V rails. 22.1ohm R. It is sensitive to op-amp choice, but works nicely.

And no, I haven't heard a R-2R to compare it to. That will be a project in the near future.

Found the reference:

http://www.diyaudio.com/forums/digital-source/79671-suggestions-balanced-dac-output.html#post918060

Sure, why not? We like to build DAC's with super-high distortion and boatloads of noise. Why not?

http://stereophile.com/digitalprocessor ... ndex6.html

The distortion was below 0.01% at full scale. That's with zero feedback and no shitty op-amps. Look at Figure 14 to understand how low the noise is. Just a little bit of 60 Hz and 120 Hz hum -- at -115 dBFS. But you are going to do better when you can't figure out how to read a data sheet? Good luck!



When you get your circuit made with the zero ohms input impedance, let me know. I'd like to see the schematic for that. Then send me the measurements showing how much better it is than the QB-9. Then borrow a QB-9 from a friend and let me know which one sounds better.

Hi Charles,
sorry but it's not my words but it's a jocko's words I read here, last post of this page. viewtopic.php?f=15&t=262&p=27598&hilit=DACs+work+best+into+zero+ohms#p27598

I think I can learn from you and no you of me. I hope I did not offend you? Maxpou

Charlie knows that I know how to do that. He also knows how low of a load impedance I can get.

But for you guys...........................different story.

Go back and read the data sheet, some more.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

Hi,

@maxpou:
Your circuits "IV converter-1" to "IV converter-3" won´t work.
First: the current sources Q2 and Q5, resp. Q6 and Q1 need to run on different currents. If they run on the same current this current runs from Q5 via Q3 through Q2. No idle current would be left for Q9. So either increse Q5´s current by reducing R2 or decrease Q2´s current by increasing R9. The difference of currents needs to be greater than the DAC´s signal current.
Second: You need to understand that this circuit doesn´t work differentially but as a parallel connection of complementary legs and as such it sums the currents in Riv.
This means that +Idac and -Idac need to be equal (in amplitude and phase!) and free from centre currents. Then the signal currents through Q9 and Q10 are the same but with different polarity of the idle currents. The idle currents cancel out over Riv just leaving the sum of the signal currents. If the DAC signal currents were out of phase the net current through Riv would canvel to zero.

Regarding the output offset of currentsteering DACs like PCM1794A and PCM1796. They want to see a low ohmic termination or virtual gnd centered around 0V. The key word beeing the ´voltage compliance´ which depends on the kind of current sources/sinks within the DAC, but roughly is +-1V for current steering DACs (compare SLAA399of BB/TI "passive termination for current output DACs Nov.2008, page4). If You hold the output at 2.5V the internal protection diodes (capable of 10mA) would conduct and eventually and probabely fail. These protection diodes start conduction just above 100mV which shows in increasing distortion values. If the impedance the DAC sees is below ~22Ohms the distortion values are as in the PCM1794As Datasheet or lower. taking the outputs single ended leaves of course the uneven harmonics left in the outut signal, but this means a more ´natural´ spectrum and who should care about anyway when the values of those harmonics remain below -100dB??

jauu
Calvin

Hi Clavin,
thank you for your explanations
I understand the first point and easy to resolve, but the second point is more hard. The only solution that I see is replace the folded cascode by a differential amp if I understand correctly or maybe you have another solution for me. Maxpou

Hi,

regarding the second point: it could work only if the DAC´s centre currents were of equal amplitude but different polarity.
This is not the case in any of the DACs I know.
As a recommendation for a nice IV for DACs with centre currents ... just stay with the Jocko and add a buffer stage.

jauu
Calvin

If you want to have a real discrete circuit, why would not mind using the op-amps in the servo?

Hi guys,
maybe this circuit works? maxpou

Hi,

no, it won´t.
You could throw another thousand circuit schematics at us, but please learn some basics before.
As a hint. LED9, LED12, Q14/Vout, supply-voltages

jauu
Calvin

Hi,
I don't understand! what's the problem? Maxpou

I spoke to TI engineers years ago WRT DAC degradation versus non ideal virtual gnd.
From memory it was about 10 ohms before DAC's specs were compromised at all.

This leaves it pretty open for any type of grounded base stage I-V. The compromises will
definitely be lesewhere in the circuit.

WRT zero FB I_V's and linearity - the latest one I am working on for Sabre DAC simulates close to zero distortion <-120dB at 20kHz 0dBFS.
No local FB such as CFP's either (as used in twisted pears discrete I-V). Allowing for simulator to reality
conversion factor of 10, it should do 0.001 or better.

cheers

Terry