Depending on the Z, the cap value can be as high as 1uF, so MKT caps will suffice with values like that.
Again, as a guide, to repeat in simplest term , add a filter, usually using a single cap and two resistors.
Aim to be -1.3dB to -1.5dB down at 20KHz.
See below. This is the "current" DAC output of a Sony CDP-XA5ES (I used to own one) and note the post DAC filter.
R535 10R - does not, maybe there for parasitic reasons
R537 10R Ditto
C531 2200pF (22n) filter cap
R540 10R - defines the R of the RC filter - because it works into a theoretical zero Z (Virtual Earth).
R539 - Same as R540
Note, if this was a "voltage" DAC, the roles of the resistors are reversed. The Low Z point of the circuit would now shift to the output of the DAC itself and R540 and R539 would no longer be needed.
You must define the Low Z point of the circuit.
All this was covered earlier. Only here we have an actual example from a commercial model - nobody is making any outrageous claims about something that has been done before. That misses the point.
The filter above is flat to 20KHz.
Increase the cap value to approx 0.33uF (we are talking about a value fifteen times larger. It will be significantly down at 20KHz.
The irony is that I did not think the type of filter was the issue (it never was in my mind) but the reticence of that degree of filtering.
Cheers, Joe
.
R535 10R - does not, maybe there for parasitic reasons
R537 10R Ditto
C531 2200pF (22n) filter cap
R540 10R - defines the R of the RC filter - because it works into a theoretical zero Z (Virtual Earth).
R539 - Same as R540
Note, if this was a "voltage" DAC, the roles of the resistors are reversed. The Low Z point of the circuit would now shift to the output of the DAC itself and R540 and R539 would no longer be needed.
You must define the Low Z point of the circuit.
All this was covered earlier. Only here we have an actual example from a commercial model - nobody is making any outrageous claims about something that has been done before. That misses the point.
The filter above is flat to 20KHz.
Increase the cap value to approx 0.33uF (we are talking about a value fifteen times larger. It will be significantly down at 20KHz.
The irony is that I did not think the type of filter was the issue (it never was in my mind) but the reticence of that degree of filtering.
Cheers, Joe
.
Attachments
It was actually Coris that claimed it was some special filter way back on the first page.
P.S. Thanks, Joe, for the rude PM. Too bad it was early and I didn't think to share it with the mods.
Last edited:
😀 😀
Sorry, but i have to leaf a little bit, thinking at what it may say now the bitter critics of this filtering technique. Specially they who meant such filtering it will never work, leads to ringing in audio signal, etc...
How we can easy see (thanks Joe for the picture in post 202), this filtering technique it were used before in a consumer device. It looks to me like is no any invention of today at all (but it were invented at its time of course...), and it was well known by few clever designers out there.
Joe still have his merit for find it out and making known to us about.
Now we have the prove that such filtering it works very well (as some of us have already experienced), many of us it were unknown with, and is to be used from now on of those who will want to implement it.
I think we may continue to discuss it further about, trying to find out more, improved it, and so on...🙂
Sorry, but i have to leaf a little bit, thinking at what it may say now the bitter critics of this filtering technique. Specially they who meant such filtering it will never work, leads to ringing in audio signal, etc...
How we can easy see (thanks Joe for the picture in post 202), this filtering technique it were used before in a consumer device. It looks to me like is no any invention of today at all (but it were invented at its time of course...), and it was well known by few clever designers out there.
Joe still have his merit for find it out and making known to us about.
Now we have the prove that such filtering it works very well (as some of us have already experienced), many of us it were unknown with, and is to be used from now on of those who will want to implement it.
I think we may continue to discuss it further about, trying to find out more, improved it, and so on...🙂
Last edited:
That's not a very accurate boast. In the example you reference, that Joe provided, there's a both isolation resistance and a much smaller capacitor value. There's no huge honkin' capacitor thrown directly across the output of a DAC or input of an op-amp. What's shown there will not ring. It probably won't ring even with the huge capacitor Joe suggests.
Again, you're comparing apples to oranges when you don't actually know enough to make such boasts. You're actually doing Joe a disservice every time you post. Regardless of what I (and others) may think of the validity of his "discovery," at least he knows how it works. You don't.
Last edited:
Hi Joe,
I guess this reply has been overtaken by changes to the topology, but nevertheless: the original topology shown was a cap across two wires that had zero voltage across them (the I/V inputs). I'm all open minded but hearing the effects of a cap with both wires at zero voltage that's stretching it to be mild.
NOW there's a change with two resistors in series to develop some voltage across the wires so that the cap can have an effect. NOW it may be possible to hear an effect.
Note the time-line: FIRST a cap across zero voltage wires that was hailed as an audible effect. NOW a change to the topology (series resistors) so that it MAY be audible. Still no idea whether it might be audible, because nobody has done (at least not shown) any serious controlled test.
Jan
Last edited:
Oops! Almost fooled me there! Correction: even with the two series resistors into the I/V converter, the cap still has no effect. Reason: both inputs of the I/V converter are at zero volts. So still no voltage across the cap, so still no difference with or without the cap.....
Jan
What has been done here does in my mind constitutes a controlled test. At the very least a blind test.
Still, the best methodology, with all the vagaries of human hearing (my late Father was an Acoustic Engineer and he left me with a book with lots of interesting examples), it still has to come down to what the human ear hears, but also stands the test of time where it has to satisfy a criteria that is often tougher than any controlled/blind tests.
Keep in mind too, the number of people involves here (you don't know how many, do you?), some as skeptical as you, that is a factor too.
Back in the 70's and 80's Martin Colloms conducted heaps of controlled listening tests on all sort of things. It is interesting how much his attitude has changed since then - he now feels they fall way short of being realistic tests, because they don't actually take in the state of mind of the listener and a lot of other things. A kind of Heisenberg moment I would think - you know, the observer influences the outcome just by being there
I have clients who make the living based on their good hearing. You think YOU are hard to convince. I don't get paid for nothing, trust me. When dollars are exchanged with professionals, then you better come up with the goods.
Back to the filter:
Of course the filter will be audible. That goes without saying, just make the filter too aggressive and all you will hear a rolled off HF.
But that is not what we are trying to achieve right? There has to be another payoff, or it is a waste of time.
Personally, as we have not heard it doing this to other than delta-sigma DACs, that it is related to noise shaping and dither. As somebody said earlier, the DAC is a bit like a Black Box because we are not told precisely what it is.
Maybe forcing that inbuilt output device into an unknown condition? Some say it may be parasitic? It might, except we have been able to exclude it with DACs that can be grounded directly. Hence they are immune, but it is a good start - at least we are thinking.
Whatever:
Maybe, somehow, the old adage still has to right, hearing is believing and a controlled test is also about hearing, is it not?
Somewhere along the line somebody's ears must make a decision.
It will not settle everything though - not with Black Boxes we have here.
Remember they call it "dark matter" - not because it is dark, but because we are in the dark as to what it really is. In fact we have no clue right now.
So we must deny it exists - back to the Dark Ages perhaps?
Cheers, Joe
PS: Isn't Customer Satisfaction a 'controlled test' or even worse, a 'test that is beyond your control' - which is a much more scary for guys like me.
.
Last edited:
That's why I asked Coris to post a scope trace of the DAC output, specifically the rise time of the output is interesting. I think the DAC outputs rather stiff pulses. If the rise time is known, along with the output impedance, then the output can be modelled rather accurately.
With a perfect opamp and zero inductance connections, the cap would have 0V across it as you say. Unfortunately, this is not the case : the impedance of the IV's input will rise with frequency, and the parasitic inductances have the same effect. Also opamps do not like RF at the inputs, at all. The situation is quite different between a voltage out and a current out.
Also I wonder if the effect could be even more pronounced by putting one cap to ground on each DAC output instead of across the outputs, which would also attenuate common mode noise.
If we want to understand what is going on rather than just talk, it's going to take some accurate measurements...
- DAC output voltage with and without cap
- Voltage at the IV input with and without cap (is the same as above if direct connection to DAC)
You may take very wrong that about doing services by my posts here and there... I`m not intend to do a service to anybody personally. I only think is all right enough to share here what, and if I found out on something, and eventual confirm the other findings, when it works as presented. That`s all.
Yes, I do not know very well how it works this filter. Actually, I did not special researches in this, but some raw measurements. My results was quite contradictory. I have only try it so far, experimented on it (with or without the resistors shows in the Sony schematic), and I fond it out that it works, it bring important improvements. And I`m not the only one who have experienced the improvements in using this filtering technique.
BTW, I do not use huge caps in my configuration. It does not work for me. The best way I found it so far for my circuit is 1nF film cap without any resistors.
Not again.
The cap has to see a voltage across it, indeed to do anything, has anybody suggested otherwise at any time? Did I not say to aim at -1.3dB at 20KHz relative to 1KHz - and guess where?
Across the cap of course. How else can it be down?
After all, is anybody doubting that we are talking about a filter? If it doesn't filter, it ain't no filter.
I could go back and give all the post numbers (there must be three or four at least) but I just don't want to waste the time... I clearly said that a filter cannot operate into a Zero Ohm or into a short, so can we please consider that case closed and I won't have to repeat it, maybe for the FIFTH time... please?
OK, that being establish (sigh) - can we actually deal with what is happening across that cap, as that was always supposed to be the discussion?
I agree that, in the post-DAC circuit, that it reduces slew rate induced distortion, a subject that I have repeatedly talked about and acutely aware off. But in this case we have already been able to dismiss it - because, with "voltage" DACS I use 1:1 transformers and getting that -1.3dB down at 20KHz shows the same result/effect.
The transformer then is the only post-DAC circuit and fed to a sero-feedback high-resolution tube amp which is slew rate distortion free.
So we can move on from there, OK. That also has been covered.
I think we DO need to look at the load and not just filtering of the DAC's output. What is heard is very similar to a 'damping' quality.
It also odd, this is limiting bandwidth, the circuit is indeed slower in that sense, yet is also sounder quite a bit faster with much cleaner edges of transients.
The other clue - we are definitely talking about delta-sigma DACs.
If you have a suggestion about modelling the filter effect on delta-sigma DACs, then please don't just talk, but do it. Make yourself useful and we will all be happy with a contribution.
The purpose of this thread was:
1. Come on board and help establish that the effect is real - in other words, just do it.
2. Once established that what we have is real and not imagined, then we can have a free and respectful discussion - and make suggestions as to how we can find answers (this will also have a non too hidden benefit - we can enjoy the sound we are getting while doing so - and ease tensions).
3. Discuss possible implementations, from "current" and "voltage" DACs as well as transformers. Again, they are not non-filters, but real filters.
We need to be either on board or not. If anybody feels this is a waste of time, then do the right thing and let others get on with it. Don't be a spoiler for the sake of being a spoiler. Don't take on a holier than thou attitude about which is the only way (there is never just one way) to sort this out - just because it happens to be your way? By all means suggest a way, but don't impose a singular view - especially if all you are doing is shwong off. You know, it is all that hard to tell the difference.
So ask yourselves: So are you in or out?
If you are, show consideration to others that are in. It is not much to ask for.
Cheers, Joe
It seems that your humility is not appreciated?
How is a single human thought generated? Where does that voice inside my head come from? Nothing that science has learnt in thousands of years has gotten us any closer to answering those questions.
Nobody knows everything, except maybe God?
.
I just wonder (quite a lot) why you (if you may own a DAC system), do not try by your self exactly what you propose here, or you are thinking about. Just solder a cap here and there, accordingly to your logic, hear the results and scope what you may think about in your configuration. It may/can not be simple than this...😕
Reading somewhere about something, sometimes quite contradictory things, my first reaction it may be to reproduce (if easier enough) the experiments, to find out my self what is actually going on. But I can clearly see that somebody else do not even think to try by themselves the simplest things to be done... I may say that it seems this to me very strange...😕
Is this the human being works... Some knows better, best in one area, somebody else it do in another area. Somebody can control very well the details of a small field, while somebody else can cover a larger detail area. Is like this it actually works in this human world. There was not "invented" yet that human being who can cover everything/everywhere in knowledge... But we can all together, or a large part of many entities interested in the same field or another, cover and do it/find out about exceptional things (together). About this it were proved many times in the human history, as everybody knows (I suppose) about...
But I can well enough understand that the human vanity it may be quite special in some circumstances... for some of us. Unfortunately.
Last edited:
Thanks Julf. I appreciate that - I don't carry grudges and it would be nice if we all just got our act together.
We will all feel a lot better.
.
- Status
- Not open for further replies.