Buffalo II & transformers

markusA · 2010-04-01 4:41 pm

Let's talk transformers and the B-II.

Running the B-II in current output mode and using a trafo and resistor for I/V.
The Lundahl LL1674 seams like a popular choice.

How do you hook it up and what are your results?

What impedances are you seeing, gains, voltages, ...?
Pros and cons for using a trafo?

Other transformers are welcome as well

markusA · 2010-04-01 5:28 pm

The Rakk Dac do it this way... (LL1674)
I guess that'll work if you want SE output?

alboran · 2010-04-01 6:02 pm

Hello
I can tell my personal experience with buffalo dac (es9018) + transformer.
I have tried many Transformer, Current and Voltage ..
Stevens & Billington TX-103 - Audio Tekne LT-8310 - UTC HA-108 (UTC A20) - UTC HA-101 - Jensen JT11-P-1 and others ...
the best sound I got with the two Jensen, I recommend trying these very Jensen.
wishes for the tests, good plays

fierce_freak · 2010-04-01 6:03 pm

I've read in a few places that some feel the sabre chip sounds better through transformers when it's forced into voltage-out mode. I think something like a 10K:10K transformer with 2-5k resistance across the primaries would work well, but I'm sure there are others way more knowledgeable about this that can correct me.

ianp · 2010-04-01 7:29 pm

Just reported this in the main thread

Just like to report that my Buffalo II DAC using Sowter 9454E transformers is now alive and well.

I have the DAC output going directly into the Sowters. All seems well.

The 9545 transformers were previously used in a modified SRC2496. Actually, I have another SRC2496 with upgraded 4396 DAC chip and (more) Sowter 9545 transformers. I will be doing a direct comparison between this and the Buffalo II.

ianp

Marek · 2010-04-01 8:08 pm

fierce_freak said:
I've read in a few places that some feel the sabre chip sounds better through transformers when it's forced into voltage-out mode. (...)

Thats interesting. the intuition tell me that it shouldn't be true although I do not have Sabre DAC...yet. Transformer is a current element because magnetic flux inducted in primary winding is proportional to current flowing through it.
But as I said this is theory...

I recomend to read this lecture http://www.analog.com/static/imported-files/application_notes/AN_912.pdf

markusA · 2010-04-01 9:16 pm

Isn't the wole idea of using a trafo to have the dac in current mode?
If used in voltage mode you don't need it?

And the reason for that is improved THD which need a really low impedance load, as close to zero ohms as possible?

markusA · 2010-04-02 9:26 am

Some comments made by Russ White about current output and load.
http://www.diyaudio.com/forums/twisted-pear/160782-buffalo-ii-10.html#post2107223 (About loading)
http://www.diyaudio.com/forums/twisted-pear/160782-buffalo-ii-12.html#post2111740 (Output impedance)
http://www.diyaudio.com/forums/twisted-pear/160782-buffalo-ii-12.html#post2111970 (Current vs V mode and THD)

And please read the link Marek supplied.
http://www.analog.com/static/imported-files/application_notes/AN_912.pdf (A paper on transformers)

After reading all this I'm confused.
Acording to Russ any loadimpedance will be bad for the distortion figures, even a 10ohm load.
Somewhere in the 700ohms region the dac shifts from current out to voltage out.
Using a transformer the load impedance will be reflected through the trafo onto the dac. Reaching zero ohms or even fairly close doesn't really seam like an option with a trafo?
So if a load quickly pushes the THD up in current mode, why even bother?
It sounds a little like "go op-amp or use voltage out"?
And using voltage out you don't really need the trafo if you have a buffer.

Also, reading the paper they talk about R0, would that be the dac output impedance (195ohms)?

Marek · 2010-04-02 10:35 am

markusG,

R0 is termination resistors of DAC. It seems that those resistors change impedance seen from secondary (output impedance) because Zs=2*N^2*R0.
For now I'm not using those resistors but I'm thinking about that - maybe it will decrease THD, but for sure will decrease output level

. For me output level is a problem because I have a AD1853 based DAC which has only 3mA output current and I use LL1517 (1:1) transformer. Unfortunetally I haven't LL1674 or LL7903/5.....

markusA · 2010-04-04 8:21 pm

Yeah, ok? Termination resistors... I'm a little electronically challenged, trying to learn though.
In the model, output impedance of the dac isn't mentioned.
Couldn't one say the output impedance of the dac is referenced to ground?
Ergo it becomes R0 in the model?

I'm a little worried using a LL1674 will give me to high output level? 1:4 would make the 1.65VAC into +6VAC and you only need 2V to drive most amps to clipping. The buffer will take care of the current.
So, I was thinking a 1:1 or 1:2 ratio would be better? However the lower the ratio the higher the load impedance to dac will see.
It feels like going in a circle and no one's the winner?

How high should the target output voltage be? (Passive vs into a buffer?)
Is it even possible to use the dac sucessfully in current mode with a trafo?
Since I'm adding a buffer, do I even need the trafo or is it redundant? Is the trafo only needed for truly passive output?
Is voltage output really any worse than curent out? -120dB to -108dB THD is really only a factor 4. The distortion is 4 times larger with voltage out, still in the 0.000x% THD range.
My tube buffer will surly have more than that?

Marek · 2010-04-05 12:06 am

hi,

Connecting LL1674 trafo directly to output of DAC puts it automatically to curent mode because od low DCR of primary windings.
In the model DAC is treated as an ideal current source which has infinite output impedance. Real DAC isn't ideal current source and has finite impedance and this value should be mentioned in datasheet. For example PCM1704 has 1kOhm.
LL1674 should be fine with 1:4 because the higher turns ratio, the lower impedance is seen by DAC - lower THD. If you got too high output level - great! - decrease load resistance on secondary - you let the DAC to see even lower impedance.
If you use trafo with 1:4 ratio you do not have to use any active buffers.

BTW What resistor you use on secondary?

markusA · 2010-04-05 9:51 am

AN_912.pdf, right hand paragraph, top of the page: "Equation 4 through Equation 7 can be used to predict the impedance seen by each DAC output (ZNORM and ZCOMP)".
Znorm=Zcomp=R0*RL/(2RL+4R0*N^2).

If no R0 is used it should be comparable to a resistor with infinite resistance?
RL<<R0 -> Znorm = RL/4N^2 (Very approximate)
Given the LL1674, N=4 -> Znorm = RL/64

Vs= (sqrt(2)*Imax/2)*(N*R0*RL/(RL+2R0*N^2))
Approximate R0 with infinity and we get
Vs=(sqrt(2)*Imax/2)*(RL/2N)
We want Vs=2 and Imax is 4mA I think?
RL=5657ohm

Plug this in and Znorm=88ohm seen by the dac. Do we just add the DC resistance to this, making Znorm=121ohm?

I'm waiting for the guys over at TPA to start taking orders for the B-II.

Now, either I've done the equations wrong or else it doesn't seam like such a perfect match any more?
In the thread discussing the B-II it's said the load seen by the dac is very important. It needs to be very low if the advantage of current out mode is to be gained. 121ohm isn't very low in this regard.

markusA · 2010-04-05 10:23 am

I made a ****, Imax should be 16mA.
I was just thinking of my last post.
If you don't have R0 the equations behave somewhat odd.
The LL1674 has a 33ohm dc resistance on the primaries.
33//R0 = 33 if the R0 appoaching infinity approximation is used.
This way we would get very different figures.
For the models purposes we would have R0=33 in this case.
Vs approaches 1.6V when RL becomes very large.
Not good but manageble I guess? However, Vs drops when RL decreases...

So if these equations is to be of any use I need to know what to use as R0.
And what to to with the real life values? R dacoutput=195ohm & R dcll1674=33ohm.

Marek · 2010-04-05 12:36 pm

markusG,

I think you should do it this way:
take your ll1674, put on secondary some fixed resistor (eg. 2,2k) then calculate R0 with equation (9) for Rl=2,2k it will be 68ohm and use this value as termination resistor. In this case impedance seen by DAC will be 55ohm (including DCRs of pri and sec ofcourse) and output level about 2V. It is quite managable.
You must strike a happy medium between acceptable output level end lowest possible impedance seen by DAC.

I don't know how we should treat real output impedance od DAC and situation when R0 is infinite..

markusA · 2010-04-05 1:02 pm

I don't know if I'll even be needing a trafo?
I would like to understand the math though so I can make an educated decision.

Although I don't quite understand how you got 55ohm? I'll accept it at face value and make the following comment.
Russ claims (see link in one of myprevious posts) that THD will be affected by such a large value.
The 12dB THD difference between current vs voltage output will be much less using the calculated values.
The question becomes whether one should really care if the dac is working in current mode or voltage mode? (Since we cant get the Znorm below 10ohm)
If the dfferences are small enough it becomes more important get a good impedance match to the amp and the right gain/voltage level.

Russ White · 2010-04-05 1:15 pm

If I were going to use a transformer with the BUF II it would be a 1:1 600:600 type. Because your *always* using the DAC as a voltage source when you use a transformer. This is because the DAC is truly a voltage source with an output impedance of 195R. 195R is not a lot of impedance. So it is nothing close to an ideal current source. To compensate you need a load with *extremely* low impedance to make the "non-idealness" have as little impact as possible

. Much lower than you could achieve with any transformer I know of. I have said before even a 10 ohm load modulates the outputs enough to cause 10-12db THD gain.

You also have to look at the transformer as a reactive component and your not likely going to get the impedance down low enough where THD will be stellar across the audio band. Good, yes. But, not as good as possible. I don't mean to make that sound too negative. It is not meant so. It just to set expectations. Also the distortion (not just harmonic) of the transformer itself will be much worse than -108db. So I would not even let that number come into play.

So if I were using a transformer I would not think of the DAC as a current source. As for it to be an effective current source requires an input impedance at AC and DC which is simply too low for any transformer (any many active stages) to achieve.

Now it is perfectly acceptable to add resistors to GND at the outputs of the DAC which then really just become voltage dividers. This is a very effective way to manage the output swing.

Cheers!
Russ

markusA · 2010-04-05 1:26 pm

Thanks for chiming in Russ, you kind of make my point.
Current vs Voltage mode becomes a moot point and when you don't have to focus on superlow load impedance it's possible to focus on other aspects of the build.

Marek · 2010-04-05 2:01 pm

Russ White said:
If I were going to use a transformer with the BUF II it would be a 1:1 600:600 type. Because your *always* using the DAC as a voltage source when you use a transformer. This is because the DAC is truly a voltage source with an output impedance of 195R. (...)
Russ

I don't understand why BUFII will work as a voltage source..it was mentioned many times that load less than some value (700ohm?) will switch it to the current mode so load about 50ohm will not?

marcusG,

about 55ohm. to the Rl value you should add DCR of secondary winding (605ohm) so with Rl=2,2k you will get 2,8k (-> R0=87ohm), then it is reflected to primary and with equation (10) you get Znorm=22ohm, but there is DCR of primary winding 33ohm so 33+22=55ohm

Russ White · 2010-04-05 5:17 pm

Marek said:
I don't understand why BUFII will work as a voltage source..it was mentioned many times that load less than some value (700ohm?) will switch it to the current mode so load about 50ohm will not?

marcusG,

about 55ohm. to the Rl value you should add DCR of secondary winding (605ohm) so with Rl=2,2k you will get 2,8k (-> R0=87ohm), then it is reflected to primary and with equation (10) you get Znorm=22ohm, but there is DCR of primary winding 33ohm so 33+22=55ohm

Your missing the whole point. It is both a voltage and a current source. it does not switch to either it is always both.

Its just that for it to appear to be an ideal current source (infinitely high relative output impedance) the load it is driving needs to be close to zero ohms.

For it to appear to be any ideal voltage source(infinitely low relatively output output impedance.) the load need to be close to infinite impedance.

So you see using it as a voltage or a current source if purely dependent on the relative impedance ratio between the source and the load.

At 195R 55R is not a very low impedance at all.

The ratio is what is important and it needs to be orders of magnitude higher.

I think I have explained this as well as I can.

Cheers!
Russ

Marek · 2010-04-06 8:30 am

It makes more sense to me now. I thought that there are 2 separate modes switched automagically - two diffrent modes like in other DACs like AD1865, but it usually it has separated pins.

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?

One thing I am thinking about is the lowest THD do not guarantee best sound. I have DAC based on AD1853 (current balanced output) orginally designed to operate with opamp output stage (one of the best OPA627BP and AD727) so DAC sees near zero impedance but sound was worse then this when using transformers. So what is the point? NFB -> IMD distorsion?

Buffalo II & transformers

Member

Member

Attachments

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member