AD1896 In slavemode how??

[pred]

Yes something like that. The thing abot current sources, although it makes sence at first glance, I do not know if it is necessary. But if you like it I am willing to go for it.

Pred

PS
Have you decided to go for dual mono?

 

[kimschips]

Hey

Thanks for your mail.

Yes let´s go dual mono

All the best

Kim

 

[agent.5]

I think you can pre-regulate with a Jung superreg, and then all individual chips, and different voltage supplies, will have their own mini shunt regs. An ultra low noise voltage reference is used (e.g. MAX 6126), and this voltage reference is filtered by an op amp low pass filter (See Edn article @

http://www.diyaudio.com/forums/showthread.php?postid=712325#post712325.
)

The reference voltage is shared by all the mini shunt regs.

Each mini shunt reg will likely need only 4 additional parts (assuming that there are already some capacitors near the power pins)

2 resistors
1 op amp
1 pass element

 

[kimschips]

Hello agent.5

Thanks for your input.

Could you draw a fast schematic for the principle you are talking about?

What is the purpose with the opamp together with the shuntreg?

Best regards
Kim

 

[agent.5]

Hello agent.5

Thanks for your input.

Could you draw a fast schematic for the principle you are talking about?

What is the purpose with the opamp together with the shuntreg?

Best regards
Kim

a TL431 is nothing more than a voltage reference, and a comparator.

http://www.tubecad.com/2005/January/blog0031.htm

This is for tube, but the principle is the same. The very last schematic is a TL431-less shunt. The op amp gets the voltage from the zener diode (30V) connected to the cathode of the tube. The op amp is used to control the pass element. In this case, a tube. But it can be a transistor.

There is a voltage divider (1M resistors) to create a 15V voltage reference. But I think we are better off just use a ultra low noise precision voltage reference.

Also look here (botton of page). The noise floor of this regulator is due pretty much due to the op amp. You can cut it way down by using an even lower noise op amp.

http://www.edn.com/contents/images/62305di.pdf

 

[kimschips]

Hello

Thanks agent.5
You could also look here Lcaudio XO3 clock psu with adr423 as a voltage reference

http://www.lcaudio.dk/xo3dia.gif

But I still think all three suggestions are too complicated for what pred and me have planned.

I would go with with a sulzer reg by each chip if we are in that league

Kim

 

[agent.5]

the xo3 schematic is pretty much what I have in mind. Granted I have never seen this schematic until today.

 

[pred]

There is a voltage divider (1M resistors) to create a 15V voltage reference. But I think we are better off just use a ultra low noise precision voltage reference.

Also look here (botton of page). The noise floor of this regulator is due pretty much due to the op amp. You can cut it way down by using an even lower noise op amp.

http://www.edn.com/contents/images/62305di.pdf

I know and agree with you on the noise. But to be frank I think that ideas of using "jung like" regulators for each digital and analogue stage is overkill, which can at the end give more problems then benefit. I worked several times with low noise electronics that was supplied from batteries or regulated supplies. It was not audio so we can not talk about the quallity, but the differences in performance were so small that batteries only advantage was no 50Hz. In those cases, like in this project PSSR is big so fight for 10-20db of rail noise is questionable. Working with tubes, specialy in se configurations is different since they have small PSSR so anything happening on rails is directly transformed to the output. Just to calm things down, I adore tubes and just finished my tube preamp and I love it.

Due to the price of components we have to be more or less sure that we get DAC working from the first trial. At the end we are trying here to pull the project on the "buget", not to sell it in huge quantities so several versions is out of option. Size of the pcb is important and anything more than euro size (100x160mm) is overkill. So if you ask me I'll stik to tl's, simple and effective.

Pred

 

[kimschips]

Hey

About ultra low noise race. It can be done in the pre regs board and external I-V so yes there are free hands

Pred I agree with you about prices on components.
But the CS8416 and AD1896 can you get as free samples so it helps a little.

We could also replace TL431 with a 2V7 zener diode and use a BCP51 transistor as shunt reg?

The zener diode have 100ohm dynamic impedance, Tl431 has 0.2ohm

What do you think?

Kim

 

[pred]

I would keep tl431 and if you want to use a shunt transistor let us do it. I always thought that the project should be a compormise to achieve something.

I am just worried about the stability of such regulator. But, since these components will not be SMD (or you plan this SMD as well) small changes are possible, as well as rewiring.

Take a look for your self.

TNT audio regulators

The achieved performance with the shunt and zener is -110db of noise and around 1Ohm of otput impedance. It is worse 10 times for the output resistance and better 20db for noise than simple tl431. Now if we look at the spikes coming from digital circuits, maybe the output resistance of the regulator is more important than noise, since the spikes will be translated to noise. I have never done this, so I am just wild guessing and maybe someone is more expert on this to help us (guido?). What I wanted to say all along is that in my opinion the PCB layout will be a bigger contributor to noise IMHO. My goal is to arrive to -110db of performance if it is possible not 130db wich is the chip specs. Adding a lot more components complicates the design and I am affraid will maybe lead to the worse performance.

Now I am really going on vacation so I leave you to continue without me till october.

Pred

 

[pred]

And I forgot...

For me money is not an object or priority. I work for my livinig so I can afford my hobby. Since I found out long time ago that when I include my familly, work and other obligations that I have a chance for one project per year (if I am lucky).

As I told you just the cost of aluminium for the case and transformers will be in the order of magnitude of 2 DAC (1794). So small components that you put inside like tl's are really cheap. I found CS for around 7EUR a chip (Cirrus wan't give me the samples ) so that is ok. Analog devices are much better they send me the samples, and btw in switzerland I did not find AD1896 to buy in small quantities.

So complicate the design as you wish, and finally as I plan to do the PCB along with you I can always change my mind and do the PSU differently. Just always bear in mind that you know what are you donig and that the chance of first time success must be kept pritty high.

Pred

 

[kimschips]

Yes cirrus demands a little work for samples I had to write to cirrus and explain what to do with em
So 7Euro is cheap-> try with 25Euro in DK


I know I´am pushing the thing with PSU a bit. I just know how big difference decoupling caps and psu make for an audio circuit.

But let´s keep to a simpel TL431 for the digital circuits

Lcaudio uses a shuntreg without decoupling caps for the zapfilter 2 I-V stage (no capacitor sound)

I think I will push the psu a bit for the two Dac ships-> a very critical stage in the dac construction

I will think about it an will now begin to draw a schematic
All the best and have a very nice holiday
Best regards
Kim

 

[kimschips]

Hello

Thanks for all your inputs

The schematic for the project can be seen here:

http://www.diyaudio.com/forums/show...1045#post731045

Please make your inputs
We can then start making the pcb!!

Thanks for all

Kim