No.
I'd like to, but the transformers I have roll off far above the audio band. Still seems to work well for not generating spurious high harmonics inside the audio band. 20-20K FR isn't really affected.
I'd like to try a transformer that gives me a LP of about 50KHz, but so far it have not.
I'd like to, but the transformers I have roll off far above the audio band. Still seems to work well for not generating spurious high harmonics inside the audio band. 20-20K FR isn't really affected.
I'd like to try a transformer that gives me a LP of about 50KHz, but so far it have not.
I've heard a number of reasons why digital doesn't sound as good as it should. One software engineer thought the problem was that many CDP and Dacs are primarily designed by people whose background is in computers and as such, concentrate on Data Retrieval, while such mundane matters as power supplies, proper grounding, etc., are given little (or no) thought, and any old PS will do.
Many of those who do aftermarket mods to commercial gear, actually address these overlooked areas with upgraded and refined power supplies, grounding and coupling, which are the areas that an electrical engineer, unlike his software counterpart, knows to be very critical.
Best Regards,
Terry
Last edited:
Hey, it's digital! No need for a good PSU, right? 😉
What convinced me was a lowly Goldstar* CD player that Jean Hiraga had modified back in the mid-80s. He gave it a tube output stage. Drilled holes in the top for venting, you could see the tubes. When I asked him why the Goldstar, he laughed and said "The digital section was good, just the analog section was bad."
That was a great sounding little player. Not a Wadia killer, but it sure held its own. Made a lot of folks mad, too because it was so cheap. It was a game changer for me.
*remember Goldstar? Famous for cheap microwave ovens and such in the 80s. They are now LG.
Interesting, Pano, because this what I almost posted (it was still on my clipboard):
<That's my biggest complaint with digital - the sometimes edgy harshness mostly in the upper-mid/lower-high region. It's a big reason why I sometimes like to use my tube amp for listening. Maybe it's the transformer as much as the tubes that I'm hearing? I actually never considered it much.>
Anyway, I was gonna go on about the possibility that this issue was being addressed by designers or DSP folks, etc. In other words, getting that transformer benefit without the transformer.
<That's my biggest complaint with digital - the sometimes edgy harshness mostly in the upper-mid/lower-high region. It's a big reason why I sometimes like to use my tube amp for listening. Maybe it's the transformer as much as the tubes that I'm hearing? I actually never considered it much.>
Anyway, I was gonna go on about the possibility that this issue was being addressed by designers or DSP folks, etc. In other words, getting that transformer benefit without the transformer.
At the last VSAC, a couple of years back, I was cooped up with Dan Wright (ModWright) in the Demo room we shared with him and the Cardas girls.
He mentioned almost the same thing (word for word!) that your pal, Jean Hiraga stated about the Goldstar.
It seems that the fundamentals are still...fundamental.😱
Best Regards,
Terry
Sure, running a digital audio device from the USB Vbus is good enough - otherwise it's not a "properly implemented" device 😀 Don't the bits get delivered correctly & that's all that matters - everything else is just placebo effects 😉
Getting to good enough in digital is easy - getting beyond this is a bit more difficult.
What is the transformer for? Is it just 1:1 turns ratio for converting a differential output to single ended?
Does it guarantee the correct load for the DAC (as specified in the data sheet) regardless of the following cable and amplifier input loading, as an op amp would?
Does it guarantee the correct load for the DAC (as specified in the data sheet) regardless of the following cable and amplifier input loading, as an op amp would?
Correct. I just use a 1:1, 1:1.4 or 1:2 ratio. Usually 1:1.
It acts as a DC blocker to keep the DAC's common DC out of whatever follows and it offerers good common mode rejection. What's common? A lot of digital hash, it seems.
You can use it balanced or unbalanced on the secondary.
By placing a 3.3K across the output the DAC sees that as its load. Most are rated into a 1K load. With whatever stage follows also loading the output the 3.3K is a nice compromise point. Seems to work well.
It acts as a DC blocker to keep the DAC's common DC out of whatever follows and it offerers good common mode rejection. What's common? A lot of digital hash, it seems.
You can use it balanced or unbalanced on the secondary.
By placing a 3.3K across the output the DAC sees that as its load. Most are rated into a 1K load. With whatever stage follows also loading the output the 3.3K is a nice compromise point. Seems to work well.
Not to drag this too far off topic, but how does an output transformer offer common mode rejection? It offers a balanced output (so does the RC mod that I used), but CMR will be determined by the input of the next device downstream. To me, it makes more sense to put the transformer there, and use a passively implemented balanced output.
Because it can be used on the voltage style DAC chips that have differential outputs. They are the most common type these days.
The DAC +/- outputs have DC in common, as well as switching and other noises, best as I can determine. Those don't get thru the transfo. Audio is the differential signal.
You're on your own at the other end. With short runs of a meter or 2, I haven't found this to be a problem.
The DAC +/- outputs have DC in common, as well as switching and other noises, best as I can determine. Those don't get thru the transfo. Audio is the differential signal.
You're on your own at the other end. With short runs of a meter or 2, I haven't found this to be a problem.
But since that's balanced already, all you need is simple RC filters. I can see the transformers helping if the DAC output is single ended, but that's not the case for the DAC that you and I use. The CMR is then determined by the CMR of the input transformer at the far end, which can be very high. That also takes care of galvanic isolation and DC offset wrt ground.
Surprisingly, no. I think Jensen has a white paper about this- I'll take a look and post a link if I can find it. Basically, an input transformer can be made with a much better primary balance and CMR than the balance in an output transformer secondary. And, of course, if your preamp is like 95% of consumer preamps, it will have an unbalanced input (i.e., no CMR).
I begin to understand, I think. The idea is that the audio signal can be decontaminated of digital nastiness, only if it has been passed through some steampunk-like arrangement of coils, vacuum tubes and vibrating stylus. Intuitively I can see how that 'belief system' works: the digital sharpness is absorbed by magnetic cores, dissipated as heat in the vibrating stylus, and doesn't make it across the gap in the vacuum. And I can also imagine a notion that says mechanical and magnetic devices can't move in 'unnatural' ways (like a pendulum can't be forced) so that unnatural, digital errors are filtered out, leaving only the pure, honest-to-goodness analogue.
The audio game. For those in the business, it's the gift that keeps on giving!!!
The audio game. For those in the business, it's the gift that keeps on giving!!!
Last edited:
A friend of mine (Lee) used an output TX on the dac with good results, but I believe these depend highly on the quality of the TX... and these are expensive.
IMO a bessel filter followed by a jfet buffer is very good for the output.
As for the rest we only need to address local decoupling, earthing, clocking and isolating psu for each chip to have world class effects.
In the analog sections of the CDP (output stage and dac analog) we need to be very carefull with the psu.
An amplifier is mainly used to modulate the psu output so the faster and cleaner the better.
In the digital sections we need steady power but most importantly we need to avoid psu contaminations. (Clocks are terrible poluters)
Good results can be had by carefully decoupling and if possible, independent psu for each +v 🙂
IMO a bessel filter followed by a jfet buffer is very good for the output.
As for the rest we only need to address local decoupling, earthing, clocking and isolating psu for each chip to have world class effects.
In the analog sections of the CDP (output stage and dac analog) we need to be very carefull with the psu.
An amplifier is mainly used to modulate the psu output so the faster and cleaner the better.
In the digital sections we need steady power but most importantly we need to avoid psu contaminations. (Clocks are terrible poluters)
Good results can be had by carefully decoupling and if possible, independent psu for each +v 🙂
SY, you're missing the point, I think.
Let me start at the beginning:
I don't feel that I need an active analog section after the DAC chip. The modern chips supply the all voltage and current I need.
However -
The DC on the outputs of the chip needs to be blocked. That can be done with caps, a transformer or maybe a servo. I choose the transformer because it's an "all in one" for me. It blocks the DC, it massively attenuates the common mode noise from the chip and it provides galvanometric isolation. The output can be used either balanced or unbalanced very easily. All provided right in the same component. It's really as simple as that.
You can use just a cap on one leg of the output, like the positive. I've tried it but don't like it, some others love it. You could also use a cap on each leg and a resistive balanced to unbalanced network. That works well, but not as well to my ear as the transformer.
The disadvantage of the transformer? Size, cost, weight. And if they are too small, distortion in the bass. That's why you don't see them often in modern circuits. They don't fit well with our modern faster, lighter, cheaper way of life.
Let me start at the beginning:
I don't feel that I need an active analog section after the DAC chip. The modern chips supply the all voltage and current I need.
However -
The DC on the outputs of the chip needs to be blocked. That can be done with caps, a transformer or maybe a servo. I choose the transformer because it's an "all in one" for me. It blocks the DC, it massively attenuates the common mode noise from the chip and it provides galvanometric isolation. The output can be used either balanced or unbalanced very easily. All provided right in the same component. It's really as simple as that.
You can use just a cap on one leg of the output, like the positive. I've tried it but don't like it, some others love it. You could also use a cap on each leg and a resistive balanced to unbalanced network. That works well, but not as well to my ear as the transformer.
The disadvantage of the transformer? Size, cost, weight. And if they are too small, distortion in the bass. That's why you don't see them often in modern circuits. They don't fit well with our modern faster, lighter, cheaper way of life.
- Status
- Not open for further replies.