Transformer Output Dac's - Yesterdays news?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
With the help of the great minds on this forum I managed to implement Transformer output (UTC A-20's) to a CS4398 DAC a few years back.

I now have an AK4399 chinese (Weiliang) DAC and I like the sound but we always want to tinker and try to improve don't we...

So i started to search the web for transformer output on the AK4399 but there is much less of a fan base now on this topology... Very little info posted in the last 2 years!

Am I missing something or are transformer outputs no longer the way to go?

Stu
 
I'm saying that with opamps built in it would (as far as my experience goes) be desirable to either buffer the opamps first (to get the best PSRR from the opamp) or at a pinch, bias them into classA with an external CCS to avoid them creating more noise on the supplies from their classAB output stage. But on reflection perhaps you're using a trafo with a very high shunt inductance so the load current is insignificant, whereas I use el-cheapo ferrite trafos with lower inductance.

Just as a general perspective - with the attention to SQ that's implied by using trafos (meter readers in general don't have any kind of soft spot for trafos here on DIYA) why start out with a DAC that contains opamps (and is designed by the numbers rather than for SQ)? I'm curious to know the rationale.
 
Does this mean that the AK4399 has opamps built in? I have UTC A-20 transformers and would like to use them to avoid the NE5344 opamp output stage...

Abraxalito, do you think i would not gain anything by doing this regarding SQ?

With the CS4398 chinese DAC board i Just found a place on the board to tap directly from the CS4398 before it used the opamp output stage, I was hoping to do the same thing with the AK4399 board from Weiliang.

Stuart
 
Simply the isolation (hence common-mode noise reduction) that comes from using trafos makes them worthwhile in my experience. As far as I know, yes all voltage out DAC chips contain internal opamps.

I just took a look at the AK's DS - yes, see the block marked 'SCF' it means 'switched capacitor filter' - as far as I'm aware it has opamps, CMOS ones at that. Ditto the CS4398.
 
Thanks for the encouragement, I checked the data sheet and see the switched capacitor filter. I presume that this then has pins

AOUTLP and AOUTLN for left channel
AOUTRP and AOUTRN for right channel

and I would simply connect the transformers to these pins using the diagram below?
 

Attachments

  • tamura.jpg
    tamura.jpg
    41.6 KB · Views: 557
I would pull out all the resistors. You don't need or want series resistance between the DAC and transformer. Nor do you want the secondary loading resistor.

DAC data sheet specs regarding minimum resistance apply to DC conditions. With the transformer, you have no DC conditions, and the AC impedance is solely determined by the load. Ideally this load is very large, certainly not the 680 ohm resistors selected.

Been done many times before - take out all resistors.
 
Definitely agree there - you don't want any series resistance. Perhaps they're included to ensure stability for the driving circuit - I'd substitute ferrite beads for the 1ks which add inductance but almost no series resistance. Also you don't want pure capacitance (the 4.7nF) but some resistance in series with this (say 10k) might do good as a primary snubber.
 
You are in the realm of personal opinion and subjectivity, just understand that. Polarizing comments to follow.
I for one love the transformer output, and when design allows I will always pursue it. Elimination of capacitors, either on the DAC side or the receiving end, can only be a good thing IMO. Big fan of the Jensen JT-11-DMPC and similar with 80% nickel core. You could also try Cinemag or Lundahl.

On the non-subjective side, you gain galvanic isolation with xfmr output, which can be very effective in eliminating ground loops. You also obtain a balanced or unbalanced output, if you need either one. Finally, your xfmr provides a DC path for bias currents to flow, again if needed in your system.

You definitely gain something with xfmrs, the question is what if anything do you lose, and does the scale tip either way in your personal preferences to dictate whether xfmrs are right for you or not.
 
you will saturate Ni core xfmr with DC current

and the historical "audiophile" use of xfmr with DAC has mostly been for current out DAC - stepping up the tiny V from the few Ohm shunt R loading the I_out of the DAC

so I'd suggest people may want to discount your "advice" as being profoundly ignorant of the engineering principles that are involved
 
Start with DACs (two or more, to run balanced, digitally invert the signal to one set) which don't contain internal opamps, then build a filter and output stage for this with discrete transistors running classA. Then attach the trafo (via caps) to do bal-SE conversion. That's what I currently have and it sounds, naturally enough, superb.
 
Well I'm learning lots thanks to you guys, am beginning to think that the ak4399 is not a suitable chip for the utc a-20,s

Stu

You won't need a DC blocking cap as there will not be a voltage offset at the transformer.

Transformers are a perfectly viable choice for a DAC such as the AK4399. They have advantages such as very high CMRR (much higher than the average op-amp implementation) and, in a sense, infinite PSRR. They have disadvantages too, but a quality part, well implemented, can sound very good indeed.
 
OK, well lots of mixed views, I guess I well let me ears be the judge but wanted to run it by people on this forum to make sure there are no obvious signs that I could damage the Dac.

So I will lose all the resistors?

What is the purpose of the 0.0047uf cap across the primaries and is it necessary?

cheers

Stu
 

Attachments

  • tamura.jpg
    tamura.jpg
    41.6 KB · Views: 302
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.