DAC design: the I/V stage

[opus112]

OK. So what type of filtering is needed then, with those DACs? Is it as Ken Newton said earlier?

The recommended filtering is done with 2nd or 3rd order active stages around opamps. However those filters (shown in most of the DSs) are designed to give ultra-low noise for the spec sheets (lower noise is one of the marketing highlighted features) and hence use lowish value resistors to minimize noise. In my experience loading the outputs of DACs in this way squashes dynamics - I prefer to buffer the DAC prior to filtering its output.

 

[Ric Schultz]

I use a 5K resistor in series and a 1meg to ground (along with very small cap for the filter)......therefore giving more than one meg load.....really easy on the DAC. After that the buffer has multi megaohm input impedance but its output impedance is low in order drive anything easily. The recommended filters are made up by numbers guys. They don't listen. My aural experience tells me that more filtering sounds more filtered. Me no likey filtered sound. Me want direct goosebumping transparency.

 

[opus112]

If your time-constant is determined by the 5k and the very small cap then that'll be a 7k (roughly) load at the -3dB point. For a 30k roll-off the 'very small cap' will be about 1nF? Or have I misunderstood?

 

[DonH50]

What about the images?

 

[Ric Schultz]

If your time-constant is determined by the 5k and the very small cap then that'll be a 7k (roughly) load at the -3dB point. For a 30k roll-off the 'very small cap' will be about 1nF? Or have I misunderstood?

correctamundo

 

[ack]

Care to enlighten us on these calculations? What's a "time-constant"

 

[opus112]

A 'time-constant' is engineer-speak for the product of (result of multiplying) two component values together. In this case its the value of a resistor multiplied by the value of a capacitor. The units of the product are seconds, hence the 'time' part of the phrase.

For a simple RC low pass filter, the R is the series component and the C is the shunt component (from the output to gnd). To calculate the 'corner frequency' (that's the frequency where by convention we say the filter begins to take effect, its output is 3dB lower than the input at this frequency) multiply the R and C together and then multiply that result by 2*PI. Take the reciprocal and this gives you the corner frequency.

To pick up on Don's hinted point - the single RC low pass filter is a very gentle filter, meaning it doesn't hit high frequencies very hard (attenuate them strongly). For every doubling of frequency it merely doubles the attenuation, hence its called a 'first order' filter. To attenuate DAC images we normally need a steeper (higher order) filter than this. Ric's view seems to be that the penalties of steeper filters in terms of phase aren't worth paying in this instance.

Images may or may not be a problem depending on a person's system. For example this week I've been playing with a chip which sounded awesome when driving headphones directly but markedly less glorious when its output was taken to my integrated amp and fed through speakers. It turned out that images were the problem - the images are ultrasonic and the headphones just ignored them but the amp was somewhat sensitive to them.

 

[ack]

Got it, thanks

 

[Folsom]

To calculate the 'corner frequency' (that's the frequency where by convention we say the filter begins to take effect, its output is 3dB lower than the input at this frequency) multiply the R and C together and then multiply that result by 2*PI. Take the reciprocal and this gives you the corner frequency.

I use this I love online calculators.

 

[Ken Newton]

Additional interesting questions:

1. What's the effect on SQ on chaining Opamps in the output section (same slew rate or different slew rates)

The basic effects are increased noise and the cumulative multiplication of distortion product order by each succeeding stage's non-linearity. The subjective effect on SQ will be, well, subjective.

2. What's the 'slew rate' of a tube output section?

Given any single gain stage, tube or transistor, that is not enclosed within a feedback loop, slew rate is primarily a function of that stage's net output resistance and the amount of shunt capacitance seen by that output. Decrease the output resistance and the slew rate increases. Decrease the shunt capacitance and the slew rate increases. Of course, the inverse effect is true for both of those factors.

 

[PeterJohn]

Hi...i am a new user here. I used a simple resistor and got great results. But then in my experiments I noticed that making the power supply to the DAC chip better made the sound even better. I scratched my head for a very long time as to why this might be and eventually came to the realization that the ability of the particular DAC chip I was using to reject power supply noise was dependent on the impedance presented by the following I/V stage to its output.

 

[Empirical Audio]

Hi...i am a new user here. I used a simple resistor and got great results. But then in my experiments I noticed that making the power supply to the DAC chip better made the sound even better. I scratched my head for a very long time as to why this might be and eventually came to the realization that the ability of the particular DAC chip I was using to reject power supply noise was dependent on the impedance presented by the following I/V stage to its output.

Actually, the sound quality of any D/A chip is highly reliant on both the power supply and POWER DELIVERY. By that I mean cabling, voltage regulation and decoupling caps.

If you want to hear a good I/V, try this:

https://www.ebay.com/itm/Teramoto-FINEMET-Premium-Transformers-for-AK4399-DAC/200641346566?hash=item2eb727fc06:m:m12r51rpqCwtCltu_c694qg

https://www.ebay.com/itm/Teramoto-FINEMET-Multi-tap-Premium-Transformer-Buffalo-ES9018-DAC-1-1-1-1-1/200641347079?hash=item2eb727fe07:g:RycAAOSw2s1U0EH~

These are without a doubt, the best audio transformers on the planet.

Steve N.
Empirical Audio

 

[microstrip]

Actually, the sound quality of any D/A chip is highly reliant on both the power supply and POWER DELIVERY. By that I mean cabling, voltage regulation and decoupling caps.

If you want to hear a good I/V, try this:

https://www.ebay.com/itm/Teramoto-FINEMET-Premium-Transformers-for-AK4399-DAC/200641346566?hash=item2eb727fc06:m:m12r51rpqCwtCltu_c694qg

https://www.ebay.com/itm/Teramoto-FINEMET-Multi-tap-Premium-Transformer-Buffalo-ES9018-DAC-1-1-1-1-1/200641347079?hash=item2eb727fe07:g:RycAAOSw2s1U0EH~

These are without a doubt, the best audio transformers on the planet.

Steve N.
Empirical Audio

Do you have experience with their multi-tap transformer volume control? https://www.ebay.com/itm/Teramoto-FINEMET-Premium-Transformer-Volume-Control-TVC-Custom-Seiden-Switch/200870484580?hash=item2ec4d05a64:g:qsAAAMXQMsRQ29p8

 

[Empirical Audio]

Do you have experience with their multi-tap transformer volume control? https://www.ebay.com/itm/Teramoto-FINEMET-Premium-Transformer-Volume-Control-TVC-Custom-Seiden-Switch/200870484580?hash=item2ec4d05a64:g:qsAAAMXQMsRQ29p8

I don't, but I have ordered custom transformers from them for years. Simply the best. No-brainer.

Steve N.

 

[analogsa]

Interesting. I have tried a few finemet parts and while they generally have abundant detail, dynamics and frequency extension there is also a certain relentless quality they seem to impart on music. Especially if there is more than one occurrence of finemet in the system.

Perhaps the Teramoto parts have no such shortcomings.

 

[Sablon Audio]

I’ve recently been playing around with finemet in some usb cables and had a wide range of views back from testers, ranging from slightly soporific to amazing. It could be one of those YMMV things rather than an unequivocal recommendation.