DC coupling for DACs

[jan.didden]

If you tie the center tap directly to ground (as opposed to through a capacitor) you will indeed have DC current flowing through the primary.

If each half of the primary has a DCR of 20 ohms, and the output offset on AOUTP and AOUTN is 2.5 volts each, then there will be (assuming the DAC chip can deliver it) 2.5/20 or 125mA flowing through each half of the primary to ground through the center tap.

Under AC conditions, current will flow through the center tap to ground depending on the asymmetry between AOUTP and AOUTN.

se

Sure, but you would NEVER connect the center to ground, wouldn't you? In fact the only reason to havea center tap on the primary is because it just happens to be on the xformer. Don't use it, it only brings problems and it doesn't do anything good!

Same with a cap from center to ground. Don't do it, you only upset the balance and don't gain anything.

Jan Didden

 

[Steve Eddy]

Sure, but you would NEVER connect the center to ground, wouldn't you? In fact the only reason to havea center tap on the primary is because it just happens to be on the xformer. Don't use it, it only brings problems and it doesn't do anything good!

Same with a cap from center to ground. Don't do it, you only upset the balance and don't gain anything.

Er, exactly what problems does a center tap bring and how exactly does it "upset the balance"?

se

 

[Bernhard]

I just would not use the center tap, so you get rid of those 2,5V and convert balanced to single ended.

It is something different if the DAC has current output and a single resistor is used for I/V.

The DACs might want to output slightly different currents, where does the differential current of a few LSB go

 

[Bernhard]

I want to built a completely passive DAC:

n x chip >> R_I/V >> step up transformer >> n x parallel (to keep output impedance down) passive brickwall filters >> output.

 

[_atari_]

I want to built a completely passive DAC:

n x chip >> R_I/V >> step up transformer >> n x parallel (to keep output impedance down) passive brickwall filters >> output.

Is not that a complete different approach?
I wanted to keep my DAC as passive as possible, using the hopefully easier to use voltage output of the WM8740.

You approach sounds a much more like a non oversampling dac.

Did I get you right?

 

[Bernhard]

Is not that a complete different approach?
I wanted to keep my DAC as passive as possible, using the hopefully easier to use voltage output of the WM8740.

You approach sounds a much more like a non oversampling dac.

Did I get you right?

My DAC is non os but could be os too.

The DAC chips have built-in-chip-opamp for I/V conversion that could be used optional but I use current output.

Difference is that it has passive I/V conversion which is the most important stage to be passive in a DAC because an active I/V will have to deal with the unfiltered high slew rated staircase signal

Your DAC will already be mostly active because it has the built-in-chip-opamp for I/V conversion that can not be bypassed and will have to deal with the... see above

If you wanted to keep your DAC as passive as possible the WM8740 is maybe the wrong DAC.

 

[Cauhtemoc]

Your DAC will already be mostly active because it has the built-in-chip-opamp for I/V conversion that can not be bypassed and will have to deal with the... see above

If you wanted to keep your DAC as passive as possible the WM8740 is maybe the wrong DAC.

Modern DACs with voltage out are not just DACs with current out and a built in I/V conversion stage, they are generic voltage out designs that generate the output as a voltage straight from the sigma delta modulator. It's a completely different architecture from the current output DACs.

 

[Bernhard]

generate the output as a voltage straight from the sigma delta modulator.

Sounds good.

 

[Bernhard]

I tried 1:5 and 1:10 transformers with the passive Filter Zout 5k and subjective sound was very good, a little dark maybe but without any grain, it added another -3dB @ 20kHz rolloff to the DAC which already got -3,5dB @ 20kHz with opamp output, did not correct the filter values yet...

 

[_atari_]

I tried 1:5 and 1:10 transformers with the passive Filter Zout 5k and subjective sound was very good, a little dark maybe but without any grain, it added another -3dB @ 20kHz rolloff to the DAC which already got -3,5dB @ 20kHz with opamp output, did not correct the filter values yet...

But to get you right: You are doing it on a current output dac? Because as the datasheet suggests the WM8740 likes to see a high output impedance. No big difference but sill big enough

 

[Bernhard]

Yes current output with 22 ohm on primary.
I have to try parallel passive filters to get a lower Zout.

 

[_atari_]

Yes current output with 22 ohm on primary.
I have to try parallel passive filters to get a lower Zout.

Did you simulate your construction with several filters in parallel, since I can assume that it will have side effects (read something like that in a paper about grounding (not exactly the same but also in digital circuits a net with a lot of resistors and capacitors in parallel ).
I would be really curious.

 

[jan.didden]

Does your DAC have unipolar output current (say, from zero to 2mA) or bipolar (say, from -1 to +1 mA)? If unipolar, putting a cap in parallel with that 22 ohms will charge the cap to some DC voltage and will offset the DAC from ground, which generally makes it less linear.

Jan Didden

 

[_atari_]

Wrapup

Hi,

this thread got a little bit fuzzy the last days. So I try to wrap up all findings I personally got from this thread:

1. The idea using transformers is doable
2. The sound quality depends directly from the quality and suitability of the transformer.
3. There is a little dispute about using a capacitor between the middle pin of the primary and ground, alternatively it can/should be left unconnected.
4. There are currently no "hot" tranformers which perfectly fit the task - so a lot of calculation, datasheet reading and believing other reviews has to be performed.
5. transformers will be expensive !
6. The Wolfson voltage output DACs can be good. It can even be assumed. But no final settlement on this dispute too.

I think the next logical step will be:
1) perhaps using AD1955 as DAC instead of WM8740 - bit more complicated but easier to source and at least as good as the WM8740.
2) breadborading my circuit.

I will keep you informed (breadboarding will take some time)

Best regards & thanks for you support.

_atari_

 

[Bernhard]

Did you simulate your construction with several filters in parallel, since I can assume that it will have side effects (read something like that in a paper about grounding (not exactly the same but also in digital circuits a net with a lot of resistors and capacitors in parallel ).
I would be really curious.

Yes, simulation with four filters looks exactly like with one ideally but 125 ohm source impedance adds some 0,8 dB ripple which could be cured fiddling with values.

The filters are R-C-L-C-L-C-L-C.

IMHO the input series resistor of 4,8kohm decouples the rest of the network from each other network when the source impedance is very low in comparison.

 

[Bernhard]

Does your DAC have unipolar output current (say, from zero to 2mA) or bipolar (say, from -1 to +1 mA)? If unipolar, putting a cap in parallel with that 22 ohms will charge the cap to some DC voltage and will offset the DAC from ground, which generally makes it less linear.

Jan Didden

My DAC is bipolar, no caps in parallel.

No idea if it is worth trying without opamp.
Other alternative is opamp + buffer like BUF 634 or LH0033CG.