Buffalo II & transformers

[markusA]

Would it be ok to interpret it like this...
With a very low impedance load the dac operates as fairly close to a perfect curent source with very good numbers (THD and such).
As soon as the load impedance starts to rise the numbers start to drop but the drop is reversed exponential, i.e. the drop is most prenounced in the begining and when the load impedance starts to get into the high numbers the differences are very small.

The big difference is when the dac cannot operate as a ideal current source and the load impedance is lowish but not really low. As soon as the load impedance passes an arbitrary number it becomes not so critical.

The difference between 2ohm and 10ohm is big as well as 10 vs 100ohm but the difference between 1kohm and 10kohm is small while 10kohm vs 1Mohm doesn't really matter all that much?

[revintage]

About transformers you must remember RDC is added to the output impedance. So going 2:1 could be be a good thing. Must say I don´t think 600:600 is the best solution. Example using LL1690 that can be looked upon as a very good 10k:10k :

1:1, Zout from DAC=195ohm, RDC of transformer 150+150ohm each side, Zout will be 195+300+300=795ohm.

2:1, Zout from DAC=195ohm primary RDC=150+150ohm . This is overreduced to the secondary as 495/4=124ohm. The secondary windings are now parallelled giving a RDC of 150/2=75ohm. This gives a total of Zout 124+75=199ohm which is a useful figure.

Output level will be low though as DAC ouput is only 3,3Vpp. This means 0,6Vrms when going 2:1.

[Russ White]

Originally Posted by Marek

Now it we want to use BufII in voltage mode, imput impedance of transformer should be many times higher than 195ohm so maybe trafo 10k:10k will be better choice than 600:600?

The reason is that the big difference between using 600 and 10K is going to be voltage loss. You will have over 2VRMs either way. So nothing to worry about.

I am not here to try to persuade you that using transformers is a bad idea. Far from it, I am trying to explain what I might do if I were using one. But I would probably try lot of things. And so should you.

Cheers!
Russ

[revintage]

over 2VRMs either way

Hey Russ,
I used the Sabre datasheet to find my figures. Has 9018 higher output?

[Russ White]

No the output is the same. About 2.2VRMS(depending on AVCC) differential.

So you might come in just under 2VRMS depending on the transformer you use, but not enough to worry over.

[revintage]

Hey Russ,

Is the datasheet wrong?

This is from the 9008 datasheet running AVCC of 3,3V. How do you make 2Vrms out of that? The spec says 3,3Vpp that without a doubt corresponds to 1,17Vrms.

If the sheet is right, going for 5V should give us 1,77V.

[Russ White]

The datsheet is correct but you are summing (or rather differencing) the differential outputs.

[revintage]

This has gone OT and isn´t the most important thing in the world:

The datasheet says differential so the guys at ESS did a small spec mistake(but great DACs). They don´t single end. If we are kind we can call the specification diffuse.

I am still a little puzzled/impressed how they can make the single outputs 3,3Vpp swing fully up to the rails with 3,3V DACC.

Had to check a few others like Wolfson and CL and what they call differentiel output is between the phases(like it should) and in Wolfsons case is 2Vrms. Lets call that "Industry Standard".

Russ, I can admit one should be suspicious as the indicated Vout was very low. Great you could clarify. And also edited your first "not-so-very-selling" answer.

So back to basics, with a 2:1 we will get 1,2Vrms which is a perfectly acceptable.

I had a discussion with Kevin at K&K a while ago about what transformer to choose to my coming 9018 and he recommended LL1676 before LL1690 when going 2:1. But this one will give us a Zout of ca 600ohm due to higher RDC.

[Russ White]

Yes, I am thinking the ESS datasheet simply used the wrong term. Not your fault for believing it.

[io]

best sound, best DAC

http://img2.tar.hu/makki/img/74681784.jpg#3