DAC project

[dirkwright]

How does this look for a USB DAC? I'm using the Twisted Pear USB receiver and COD DAC, a pair of Jensen step down transformers, and a line stage I got off of eBay. This DAC is going to be only connected to a computer via USB so I don't need other digital inputs. The output of the line stage is powerful enough to drive some headphones, which is a plus for me. The transformers provide I/V conversion and also provide a low source impedance for the line stage. This is the first time I've considered putting together my own DAC.

 

[beau2317]

You could always use a tube stage instead of a transformer..

 

[dirkwright]

You could always use a tube stage instead of a transformer..

Hahaha!!!

 

[qusp]

transformers are usually a pretty bad choice for directly coupling to a current out dac imo, would you not be better off using a discrete or chip SS IV stage and simply using the transformer to buffer the output of it? usually transformers are put to use on voltage out dacs, or current out dacs after a passive resistive or SS IV stage

I wont mention tubes =) i also dont really approve, so i get it =P

 

[dirkwright]

I thought transformers were the ideal I/V converter? I think Audio Note has them in their DAC. The output of the Twisted Pear COD is a balanced current source, which could be converted to a voltage with a resistor and a capacitor on each side and channel (4 sets total), but the advantages of the transformer are much more than that in my opinion. They provide RFI filtering (so that the line stage doesn't need it), digital noise filtering, ground isolation, common mode noise rejection and a perfect balanced to unbalanced conversion, all in one unit. The transformer I selected has outstanding response as well as providing a really low source impedance for the line stage, which means it's going to be very quiet. The only down side that I can see is that I should keep the power transformers out of the same box as the audio transformers to prevent any possible hum pickup.

 

[qusp]

i'm not saying you shouldnt use the TX, although ive been there done that with sabre and while it was kinda nice and effortless, i just found it made everything sound like that, if you know what i mean, kinda lacking zing when its called for. but regardless yes they do all you mention and in a kinda elegant way (except the IV conversion), but replacing the actual IV resistor is not one of those things imo. it will work, but you are replacing a pretty predictable and linear component with something that is anything but that.

as part of the IV sure, to provide isolation, impedance matching, BAL-SE conversion, rolling off the junk in the HF etc that other than the added distortion and non-linearity it does all that quite well, especially with a decent component like the jensens; but as a resistor they make a pretty good transformer. all imo of course

 

[dirkwright]

It's a dang good transformer:
http://www.jensen-transformers.com/datashts/10kbd.pdf
THD@ 20Hz = 0.015%
THD@ 1kHz = <0.001%
-3dB @ 0.5Hz & 180kHz

 

[qusp]

but its still a transformer and thus not linear in the slightest, theres no way around that due to all the interwinding capacitance and parasitics.

i'm intrigued to hear a friends urushi painted Feastrex finemet line trannies. a friend of mine bought some for his sabre for a rather insane price, but i still want to hear them. he seems to love painting everything in urushi ha

 

[dirkwright]

Well, I'm new to designing DAC's so who knows, maybe a transformer isn't a good idea. I'm going to give it a try anyway though. I don't feel they are as nonlinear as you suggest though. The data Jensen provides is extremely good in my opinion, for this purpose.

 

[dirkwright]

Here's the schematic from Audio Note showing their transformer I/V conversion circuit. They state:
"I/V Transformers are used to maximize the energy transfer during the Current-to-Voltage phase of the conversion; resulting in increased dynamics."
and:
"The output of most resistor ladder DAC chips is in the form of a current rather than a voltage. There are many ways to convert the current into a voltage but the most commonly used system is that of an op-amp connected as an I to V converter. This system requires the use of a high degree of feedback, and as a result there are problems associated with it. One of those is internal slew rate limiting of the op-amp itself. The rate of change of current at the output of even an audio DAC is very fast indeed. Even modern fast op-amps will slew limit internally and that affects sound quality. Some engineers have found that using extremely fast op-amps improves the sound quality, but we have completely sidestepped the issue by using a transformer I/V system. The transformer not only provides an I to V function but the way it is used in our latest DACs transfers maximum energy from the DAC chip itself. This in itself reduces overshoot and ringing and because the system is slightly overdamped the rise time is reduced to an acceptable rate as well."

 

[qusp]

yes and as you can see, the IV resistor is BEFORE the TX. even so audionote has been rather specifically vague on the performance of their own solution and that of the generalised unnamed competitors product, not really very useful. i'm afraid i dont take on advertising spiels as very meaningful

 

[dirkwright]

yes and as you can see, the IV resistor is BEFORE the TX. even so audionote has been rather specifically vague on the performance of their own solution and that of the generalised unnamed competitors product, not really very useful. i'm afraid i dont take on advertising spiels as very meaningful

The resistor and capacitor on the primary and secondary of the transformer are for damping, not I/V conversion, as I understand it. Also, I think you misunderstand my intention here. I'm not trying to convince you or anyone else that this is better. I just think it's interesting.

 

[BrianDonegan]

The outpit of a Sabre DAC is very different feom the output of PCM1794. As such, the results of using a transformer will be very different. I think this approach will probably make you very happy.

 

[dirkwright]

The outpit of a Sabre DAC is very different from the output of PCM1794. As such, the results of using a transformer will be very different. I think this approach will probably make you very happy.

Thanks. That's very encouraging. I did build the USB receiver board last night, except that I made a mistake and ordered 4 wrong resistors. So, I have to get the right ones to finish it. This is the first time I've soldered SMD's and it's a challenge, for sure. I hope it works when it's done!

 

[qusp]

ah ok, i just wondered why the back and forth. as i mentioned very early in the thread, i did this myself but with a common gate Fet input and no tubes quite some time ago, it works reasonably well, just not to my taste and not really inline with my end goal IE providing a transparent output with which to experiment with amps, i do not want colour from the dac.

a cap for damping? .... thats a LP filter and it will traditionally be part of IV conversion as well as setting the load line characteristic. the sabre will see this as an impedance across its output and will be putting out primarily voltage. still sounds quite well, just not for my taste, nothing says you have to have my taste though of course

 

[Ken Newton]

The resistor and capacitor on the primary and secondary of the transformer are for damping, not I/V conversion, as I understand it. Also, I think you misunderstand my intention here. I'm not trying to convince you or anyone else that this is better. I just think it's interesting.

Hi, dirkwright,

I feel that the sense of adventure and discovery which you are showing is one of the greatest things about this hobby. Might I might make a few comments regarding your DAC box plans?

I suppose the first thing is, that the resistor connected from the DAC chip to ground in that Audio Note schematic does, indeed, perform i/v conversion. I've never understood why AN says that it is the transformer is performing the i/v function. I suppose, one could argue that because the transformer primary requires current it is performing i/v conversion, but that would seem a bit too cute for me.

The capacitor in parallel with the i/v resistor forms a low-pass filter in conjunction with the DAC chip's output impedance, which is 1.7k ohms for the AD1865 DAC utilized by Audio Note. The i/v resistor, in parallel with the DAC output impedance, forms the source impedance seen by the transformer's primary. The lower the driving source impedance the lower will be the transformers distortion.

The second thing is, the transformer you've chosen is a 4:1 step down. Given the small signal which will be generated at the primary, you don't want to have the circuit working against itself by having to amplify the signal more than is necessary. So, if you already have possession of this transformer, I suggest that you try reversing the primary and secondary. That is, wire it as a 1:4 step up. You then could reduce the active amplification by a factor of 16! Which doesn't necessarily mean it will produce better sound, but you won't know untill you try .

Most of all, continue to have fun.

 

[dirkwright]

I'm not using their DAC chip. The one I'm using has differential outputs. I won't be using that exact circuit anyway. I was merely illustrating one example of using a transformer in I/V conversion. The "gnd" point on their DAC chip does not appear to be connected to ground, or at least not the ground for the part of the circuit on the secondary side of the transformer. They appear to be floating the "gnd" connection on the DAC.

 

[dirkwright]

If I reverse the transformer, then bad things start to happen, like loss of bandwidth. What Jensen shows is a unity gain circuit with an op amp. The gain of the op amp is about 4 times, so over all it's a wash. It will have nothing, zero, nadda, between the output of the DAC I'm using and the transformer. So, what I'm using isn't exactly like what Audio Note shows.

thanks for your interest and kind remarks.

 

[qusp]

it really doesnt matter, the resistor still forms part of an RC filter across the primary and performs the IV conversion, or rather in the case of sabre the impedance causes the sabre to perform its own. not the transformer in either case 1865 or 9018

 

[Ken Newton]

I'm not using their DAC chip. The one I'm using has differential outputs. I won't be using that exact circuit anyway. I was merely illustrating one example of using a transformer in I/V conversion. The "gnd" point on their DAC chip does not appear to be connected to ground, or at least not the ground for the part of the circuit on the secondary side of the transformer.

Yes, I realized that. The Sabre DAC chips, and nearly all others today, utilize differential outputs which are referenced not to ground but rather, to a D.C. voltage midway between ground and the chip's positive supply rail. The DAC chip industry made this move mostly as a cost saving measure. Only a unipolar power supply is required, rather than the bipolar supply which used to be required.